RTI Connext DDS Core Libraries User's Manual Users

User Manual: Pdf

Open the PDF directly: View PDF PDF.
Page Count: 1093

DownloadRTI Connext DDS Core Libraries User's Manual Users
Open PDF In BrowserView PDF
RTI Connext DDS
Core Libraries
User’s Manual
Version 5.2.3

© 2016 Real-Time Innovations, Inc.
All rights reserved.
Printed in U.S.A. First printing.
April 2016.
Trademarks
Real-Time Innovations, RTI, NDDS, RTI Data Distribution Service, DataBus, Connext, Micro DDS, the RTI logo,
1RTI and the phrase, “Your Systems. Working as one,” are registered trademarks, trademarks or service marks of
Real-Time Innovations, Inc. All other trademarks belong to their respective owners.

Copy and Use Restrictions
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form (including
electronic, mechanical, photocopy, and facsimile) without the prior written permission of Real-Time Innovations,
Inc. The software described in this document is furnished under and subject to the RTI software license agreement.
The software may be used or copied only under the terms of the license agreement.

Third-Party Copyright Notices
Note: In this section, "the Software" refers to third-party software, portions of which are used in Connext
DDS; "the Software" does not refer to Connext DDS.
This product implements the DCPS layer of the Data Distribution Service (DDS) specification version 1.2
and the DDS Interoperability Wire Protocol specification version 2.1, both of which are owned by the
Object Management, Inc. Copyright 1997-2007 Object Management Group, Inc. The publication of these
specifications can be found at the Catalog of OMG Data Distribution Service (DDS) Specifications. This
documentation uses material from the OMG specification for the Data Distribution Service, section 7.
Reprinted with permission. Object Management, Inc. © OMG. 2005.
Portions of this product were developed using ANTLR (www.ANTLR.org). This product includes software developed by the University of California, Berkeley and its contributors.
Portions of this product were developed using AspectJ, which is distributed per the CPL license. AspectJ
source code may be obtained from Eclipse. This product includes software developed by the University of
California, Berkeley and its contributors.
Portions of this product were developed using MD5 from Aladdin Enterprises.
Portions of this product include software derived from Fnmatch, (c) 1989, 1993, 1994 The Regents of the
University of California. All rights reserved. The Regents and contributors provide this software "as is"
without warranty.
Portions of this product were developed using EXPAT from Thai Open Source Software Center Ltd and
Clark Cooper Copyright (c) 1998, 1999, 2000 Thai Open Source Software Center Ltd and Clark Cooper
Copyright (c) 2001, 2002 Expat maintainers. Permission is hereby granted, free of charge, to any person
obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions: The above copyright notice and this permission
notice shall be included in all copies or substantial portions of the Software.
Copyright © 1994–2013 Lua.org, PUC-Rio.

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without
limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the
Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions
of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Technical Support
Real-Time Innovations, Inc.
232 E. Java Drive
Sunnyvale, CA 94089
Phone: (408) 990-7444
Email: support@rti.com
Website: https://support.rti.com/

Available Documentation
To get you up and running as quickly as possible, the RTI® Connext™ DDS documentation is
divided into several parts.
l

RTI Connext DDS Core Libraries Getting Started Guide — This document describes
how to install Connext DDS. It also lays out the core value and concepts behind the product
and takes you step-by-step through the creation of a simple example application. Developers
should read this document first. Addendums cover:
l RTI Connext DDS Core Libraries Getting Started Guide Addendum for Android Systems
l

l

l

l

l

l

l

RTI Connext DDS Core Libraries Getting Started Guide Addendum for Database
Setup
RTI Connext DDS Core Libraries Getting Started Guide Addendum for Embedded
Systems
RTI Connext DDS Core Libraries Getting Started Guide Addendum for Extensible
Types
RTI Connext DDS Core Libraries Getting Started Guide Addendum for iOS Systems

RTI Connext DDS Core Libraries Whats New in 5.2.0 — This document describes
changes and enhancements in the most recent major release of Connext DDS. Those upgrading from a previous version should read this document first. (Note: For what's new in maintenance and patch releases, see the RTI Connext DDS Core Libraries Release Notes.)
RTI Connext DDS Core Libraries Release Notes —This document describes system
requirements, compatibility, what's fixed, and known issues.
RTI Connext DDS Core Libraries Platform Notes — This document provides platformspecific information, including specific information required to build your applications using
Connext DDS, such as compiler flags and libraries.

iv

l

l

l

RTI Connext DDS Core Libraries User's Manual — This document describes the features of the
product and how to use them. It is organized around the structure of the Connext DDS APIs and certain common high-level tasks.
API Reference HTML Documentation (README.html) — This extensively cross-referenced
documentation, available for all supported programming languages, is your in-depth reference to
every operation and configuration parameter in the middleware. Even experienced Connext DDS
developers will often consult this information.
The Programming How To's provide a good place to begin learning the APIs. These are hyperlinked code snippets to the full API documentation. From the README.html file, select one of the
supported programming languages, then scroll down to the Programming How To’s. Start by
reviewing the Publication Example and Subscription Example, which provide step-by step examples
of how to send and receive data with Connext DDS.

Many readers will also want to look at additional documentation available online. In particular, RTI recommends the following:
l

l

l

Use the RTI Customer Portal (http://support.rti.com) to download RTI software, access documentation and contact RTI Support. The RTI Customer Portal requires a username and password.
You will receive this in the email confirming your purchase. If you do not have this email, please
contact license@rti.com. Resetting your login password can be done directly at the RTI Customer
Portal.
The RTI Community portal (http://community.rti.com) provides a wealth of knowledge to help you
use Connext DDS, including:
l Best Practices
l

Example code for specific features, as well as more complete use-case examples,

l

Solutions to common questions,

l

A glossary,

l

Downloads of experimental software,

l

And more.

Whitepapers and other articles are available from http://www.rti.com/resources.

v

vi

About this Document
Paths Mentioned in Documentation

xxxviii

Programming Language Conventions

xxxix

Traditional vs. Modern C++

xxxix

Extensions to the DDS Standard

xl

Environment Variables

xl

Additional Resources
Part 1: Welcome to RTI Connext DDS

xli
1

Chapter 1 Overview
1.1 What is Connext DDS?

2

1.2 Network Communications Models

3

1.3 What is Middleware?

6

1.4 Features of Connext DDS

7

Chapter 2 Data-Centric Publish-Subscribe Communications
2.1 What is DCPS?
2.1.1 DCPS for Real-Time Requirements

10
11

2.2 DDS Data Types, Topics, Keys, Instances, and Samples

12

2.3 Data Topics — What is the Data Called?

13

2.3.1 DDS Samples, Instances, and Keys

14

2.4 DataWriters/Publishers and DataReaders/Subscribers

15

2.5 DDS Domains and DomainParticipants

18

2.6 Quality of Service (QoS)

19

2.6.1 Controlling Behavior with Quality of Service (QoS) Policies
2.7 Application Discovery
Part 2: Core Concepts

19
20
22

Chapter 3 Data Types and DDS Data Samples
3.1 Introduction to the Type System

25

3.1.1 Sequences

26

3.1.2 Strings and Wide Strings

28

3.1.3 Introduction to TypeCode

29

3.1.3.1 Sending TypeCodes on the Network
3.2 Built-in Data Types

30
30

3.2.1 Registering Built-in Types

30

3.2.2 Creating Topics forBuilt-in Types

31

3.2.2.1 Topic Creation Examples
3.2.3 String Built-in Type

31
33

vii

3.2.3.1 Creating and Deleting Strings

33

3.2.3.2 String DataWriter

33

3.2.3.3 String DataReader

35

3.2.4 KeyedString Built-in Type

38

3.2.4.1 Creating and Deleting Keyed Strings

39

3.2.4.2 Keyed String DataWriter

40

3.2.4.3 Keyed String DataReader

43

3.2.5 Octets Built-in Type

46

3.2.5.1 Creating and Deleting Octets

47

3.2.5.2 Octets DataWriter

48

3.2.5.3 Octets DataReader

50

3.2.6 KeyedOctets Built-in Type

53

3.2.6.1 Creating and Deleting KeyedOctets

55

3.2.6.2 Keyed Octets DataWriter

55

3.2.6.3 Keyed Octets DataReader

59

3.2.7 Managing Memory for Built-in Types

62

3.2.7.1 Examples—Setting the Maximum Size for a String Programmatically

64

3.2.7.2 Unbounded Built-in Types

67

3.2.8 Type Codes for Built-in Types
3.3 Creating User Data Types with IDL

68
69

3.3.1 Variable-Length Types

70

3.3.1.1 Sequences

71

3.3.1.2 Strings and Wide Strings

71

3.3.2 Value Types

72

3.3.3 Type Codes

73

3.3.4 Translations for IDL Types

73

3.3.5 Escaped Identifiers

111

3.3.6 Namespaces In IDL Files

111

3.3.7 Referring to Other IDL Files

114

3.3.8 Preprocessor Directives

115

3.3.9 Using Custom Directives

115

3.3.9.1 The @key Directive

116

3.3.9.2 The @copy and Related Directives

117

3.3.9.3 The @resolve-name Directive

119

3.3.9.4 The @top-level Directive

120

3.4 Creating User Data Types with Extensible Markup Language (XML)

121

viii

3.4.1 Primitive Types

128

3.5 Creating User Data Types with XML Schemas (XSD)

128

3.6 Using RTI Code Generator (rtiddsgen)

138

3.7 Using Generated Types without Connext DDS (Standalone)

139

3.7.1 Using Standalone Types in C

139

3.7.2 Using Standalone Types in C++

140

3.7.3 Standalone Types in Java

140

3.8 Interacting Dynamically with User Data Types

141

3.8.1 Type Schemas and TypeCode Objects

141

3.8.2 Defining New Types

141

3.8.3 Sending Only a Few Fields

143

3.8.4 Sending Type Codes on the Network

143

3.8.4.1 Type Codes for Built-in Types

143

3.9 Working with DDS Data Samples

145

3.9.1 Objects of Concrete Types

145

3.9.2 Objects of Dynamically Defined Types

147

3.9.3 Serializing and Deserializing Data Samples

149

3.9.4 Accessing the Discriminator Value in a Union

150

Chapter 4 DDS Entities
4.1 Common Operations for All DDS Entities

152

4.1.1 Creating and Deleting DDS Entities

153

4.1.2 Enabling DDS Entities

154

4.1.2.1 Rules for Calling enable()

155

4.1.3 Getting an Entity’s Instance Handle

157

4.1.4 Getting Status and Status Changes

157

4.1.5 Getting and Setting Listeners

158

4.1.6 Getting the StatusCondition

158

4.1.7 Getting, Setting, and Comparing QosPolicies

158

4.1.7.1 Changing the QoS Defaults Used to Create DDS Entities: set_default_*_qos()

160

4.1.7.2 Setting QoS During Entity Creation

160

4.1.7.3 Changing the QoS for an Existing Entity

161

4.1.7.4 Default QoS Values

162

4.2 QosPolicies

162

4.2.1 QoS Requested vs. Offered Compatibility—the RxO Property

167

4.2.2 Special QosPolicy Handling Considerations for C

168

4.3 Statuses

169

ix

4.3.1 Types of Communication Status

170

4.3.1.1 Changes in Plain Communication Status

173

4.3.1.2 Changes in Read Communication Status

174

4.3.2 Special Status-Handling Considerations for C
4.4 Listeners

176
177

4.4.1 Types of Listeners

177

4.4.2 Creating and Deleting Listeners

179

4.4.3 Special Considerations for Listeners in C

180

4.4.4 Hierarchical Processing of Listeners

180

4.4.4.1 Processing Read Communication Statuses
4.4.5 Operations Allowed within Listener Callbacks
4.5 Exclusive Areas (EAs)
4.5.1 Restricted Operations in Listener Callbacks
4.6 Conditions and WaitSets

181
182
182
185
187

4.6.1 Creating and Deleting WaitSets

188

4.6.2 WaitSet Operations

189

4.6.3 Waiting for Conditions

190

4.6.3.1 How WaitSets Block

191

4.6.4 Processing Triggered Conditions—What to do when Wait() Returns

192

4.6.5 Conditions and WaitSet Example

193

4.6.6 GuardConditions

194

4.6.7 ReadConditions and QueryConditions

195

4.6.7.1 How ReadConditions are Triggered

196

4.6.7.2 QueryConditions

197

4.6.8 StatusConditions
4.6.8.1 How StatusConditions are Triggered
4.6.9 Using Both Listeners and WaitSets

197
199
199

Chapter 5 Topics
5.1 Topics

200

5.1.1 Creating Topics

202

5.1.2 Deleting Topics

204

5.1.3 Setting Topic QosPolicies

204

5.1.3.1 Configuring QoS Settings when the Topic is Created

206

5.1.3.2 Comparing QoS Values

207

5.1.3.3 Changing QoS Settings After the Topic Has Been Created

207

5.1.4 Copying QoS From a Topic to a DataWriter or DataReader

208

x

5.1.5 Setting Up TopicListeners

208

5.1.6 Navigating Relationships Among Entities

209

5.1.6.1 Finding a Topic’s DomainParticipant

209

5.1.6.2 Retrieving a Topic’s Name or DDS Type Name

209

5.2 Topic QosPolicies
5.2.1 TOPIC_DATA QosPolicy

209
209

5.2.1.1 Example

210

5.2.1.2 Properties

210

5.2.1.3 Related QosPolicies

211

5.2.1.4 Applicable DDS Entities

211

5.2.1.5 System Resource Considerations

211

5.3 Status Indicator for Topics
5.3.1 INCONSISTENT_TOPIC Status
5.4 ContentFilteredTopics

211
211
212

5.4.1 Overview

212

5.4.2 Where Filtering is Applied—Publishing vs. Subscribing Side

213

5.4.3 Creating ContentFilteredTopics

214

5.4.3.1 Creating ContentFilteredTopics for Built-in DDS Types

217

5.4.4 Deleting ContentFilteredTopics

218

5.4.5 Using a ContentFilteredTopic

219

5.4.5.1 Getting the Current Expression Parameters

219

5.4.5.2 Setting an Expression’s Filter and Parameters

220

5.4.5.3 Appending a String to an Expression Parameter

220

5.4.5.4 Removing a String from an Expression Parameter

221

5.4.5.5 Getting the Filter Expression

221

5.4.5.6 Getting the Related Topic

221

5.4.5.7 ‘Narrowing’ a ContentFilteredTopic to a TopicDescription

222

5.4.6 SQL Filter Expression Notation

222

5.4.6.1 Example SQL Filter Expressions

222

5.4.6.2 SQL Grammar

224

5.4.6.3 Token Expressions

225

5.4.6.4 Type Compatibility in the Predicate

227

5.4.6.5 SQL Extension: Regular Expression Matching

228

5.4.6.6 Composite Members

229

5.4.6.7 Strings

229

5.4.6.8 Enumerations

230

xi

5.4.6.9 Pointers

230

5.4.6.10 Arrays

230

5.4.6.11 Sequences

231

5.4.7 STRINGMATCH Filter Expression Notation

231

5.4.7.1 Example STRINGMATCH Filter Expressions

232

5.4.7.2 STRINGMATCH Filter Expression Parameters

232

5.4.8 Custom Content Filters

233

5.4.8.1 Filtering on the Writer Side with Custom Filters

233

5.4.8.2 Registering a Custom Filter

234

5.4.8.3 Unregistering a Custom Filter

236

5.4.8.4 Retrieving a ContentFilter

237

5.4.8.5 Compile Function

237

5.4.8.6 Evaluate Function

238

5.4.8.7 Finalize Function

239

5.4.8.8 Writer Attach Function

239

5.4.8.9 Writer Detach Function

239

5.4.8.10 Writer Compile Function

239

5.4.8.11 Writer Evaluate Function

240

5.4.8.12 Writer Return Loan Function

241

5.4.8.13 Writer Finalize Function

241

Chapter 6 Sending Data
6.1 Preview: Steps to Sending Data

242

6.2 Publishers

243

6.2.1 Creating Publishers Explicitly vs. Implicitly

248

6.2.2 Creating Publishers

249

6.2.3 Deleting Publishers

250

6.2.3.1 Deleting Contained DataWriters
6.2.4 Setting Publisher QosPolicies

251
251

6.2.4.1 Configuring QoS Settings when the Publisher is Created

252

6.2.4.2 Comparing QoS Values

254

6.2.4.3 Changing QoS Settings After the Publisher Has Been Created

254

6.2.4.4 Getting and Setting the Publisher’s Default QoS Profile and Library

255

6.2.4.5 Getting and Setting Default QoS for DataWriters

256

6.2.4.6 Other Publisher QoS-Related Operations

257

6.2.5 Setting Up PublisherListeners

257

6.2.6 Finding a Publisher’s Related DDS Entities

259

xii

6.2.7 Waiting for Acknowledgments in a Publisher

260

6.2.8 Statuses for Publishers

260

6.2.9 Suspending and Resuming Publications

261

6.3 DataWriters

261

6.3.1 Creating DataWriters

266

6.3.2 Getting All DataWriters

268

6.3.3 Deleting DataWriters

268

6.3.3.1 Special Instructions for deleting DataWriters if you are using the ‘Timestamp’ APIs and BY_
SOURCE_TIMESTAMP Destination Order:
268
6.3.4 Setting Up DataWriterListeners

269

6.3.5 Checking DataWriter Status

270

6.3.6 Statuses for DataWriters

271

6.3.6.1 APPLICATION_ACKNOWLEDGMENT_STATUS

272

6.3.6.2 DATA_WRITER_CACHE_STATUS

272

6.3.6.3 DATA_WRITER_PROTOCOL_STATUS

273

6.3.6.4 LIVELINESS_LOST Status

276

6.3.6.5 OFFERED_DEADLINE_MISSED Status

277

6.3.6.6 OFFERED_INCOMPATIBLE_QOS Status

277

6.3.6.7 PUBLICATION_MATCHED Status

278

6.3.6.8 RELIABLE_WRITER_CACHE_CHANGED Status (DDS Extension)

279

6.3.6.9 RELIABLE_READER_ACTIVITY_CHANGED Status (DDS Extension)

281

6.3.7 Using a Type-Specific DataWriter (FooDataWriter)

281

6.3.8 Writing Data

283

6.3.8.1 Blocking During a write()

286

6.3.9 Flushing Batches of DDS Data Samples

287

6.3.10 Writing Coherent Sets of DDS Data Samples

287

6.3.11 Waiting for Acknowledgments in a DataWriter

288

6.3.12 Application Acknowledgment

288

6.3.12.1 Application Acknowledgment Kinds

289

6.3.12.2 Explicitly Acknowledging a Single DDS Sample (C++)

290

6.3.12.3 Explicitly Acknowledging All DDS samples (C++)

290

6.3.12.4 Notification of Delivery with Application Acknowledgment

290

6.3.12.5 Application-Level Acknowledgment Protocol

291

6.3.12.6 Periodic and Non-Periodic AppAck Messages

293

6.3.12.7 Application Acknowledgment and Persistence Service

293

6.3.12.8 Application Acknowledgment and Routing Service

294

xiii

6.3.13 Required Subscriptions

294

6.3.13.1 Named, Required and Durable Subscriptions

295

6.3.13.2 Durability QoS and Required Subscriptions

295

6.3.13.3 Required Subscriptions Configuration

296

6.3.14 Managing Data Instances (Working with Keyed Data Types)

296

6.3.14.1 Registering and Unregistering Instances

297

6.3.14.2 Disposing of Data

299

6.3.14.3 Looking Up an Instance Handle

299

6.3.14.4 Getting the Key Value for an Instance

299

6.3.15 Setting DataWriter QosPolicies

300

6.3.15.1 Configuring QoS Settings when the DataWriter is Created

303

6.3.15.2 Comparing QoS Values

305

6.3.15.3 Changing QoS Settings After the DataWriter Has Been Created

305

6.3.15.4 Using a Topic’s QoS to Initialize a DataWriter’s QoS

306

6.3.16 Navigating Relationships Among DDS Entities

309

6.3.16.1 Finding Matching Subscriptions

309

6.3.16.2 Finding the Matching Subscription’s ParticipantBuiltinTopicData

311

6.3.16.3 Finding Related DDS Entities

311

6.3.17 Asserting Liveliness

311

6.3.18 Turbo Mode and Automatic Throttling for DataWriter Performance—Experimental Features

312

6.4 Publisher/Subscriber QosPolicies
6.4.1 ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension)

312
313

6.4.1.1 Properties

314

6.4.1.2 Related QosPolicies

314

6.4.1.3 Applicable DDS Entities

314

6.4.1.4 System Resource Considerations

315

6.4.2 ENTITYFACTORY QosPolicy

315

6.4.2.1 Example

316

6.4.2.2 Properties

317

6.4.2.3 Related QosPolicies

317

6.4.2.4 Applicable DDS Entities

317

6.4.2.5 System Resource Considerations

317

6.4.3 EXCLUSIVE_AREA QosPolicy (DDS Extension)

318

6.4.3.1 Example

319

6.4.3.2 Properties

320

6.4.3.3 Related QosPolicies

320

xiv

6.4.3.4 Applicable DDS Entities

320

6.4.3.5 System Resource Considerations

320

6.4.4 GROUP_DATA QosPolicy

320

6.4.4.1 Example

321

6.4.4.2 Properties

322

6.4.4.3 Related QosPolicies

322

6.4.4.4 Applicable DDS Entities

323

6.4.4.5 System Resource Considerations

323

6.4.5 PARTITION QosPolicy

323

6.4.5.1 Rules for PARTITION Matching

325

6.4.5.2 Pattern Matching for PARTITION Names

325

6.4.5.3 Example

326

6.4.5.4 Properties

329

6.4.5.5 Related QosPolicies

329

6.4.5.6 Applicable DDS Entities

329

6.4.5.7 System Resource Considerations

329

6.4.6 PRESENTATION QosPolicy

330

6.4.6.1 Coherent Access

331

6.4.6.2 Ordered Access

332

6.4.6.3 Example

333

6.4.6.4 Properties

334

6.4.6.5 Related QosPolicies

335

6.4.6.6 Applicable DDS Entities

336

6.4.6.7 System Resource Considerations

336

6.5 DataWriter QosPolicies
6.5.1 AVAILABILITY QosPolicy (DDS Extension)

336
337

6.5.1.1 Availability QoS Policy and Collaborative DataWriters

338

6.5.1.2 Availability QoS Policy and Required Subscriptions

339

6.5.1.3 Properties

340

6.5.1.4 Related QosPolicies

340

6.5.1.5 Applicable DDS Entities

341

6.5.1.6 System Resource Considerations

341

6.5.2 BATCH QosPolicy (DDS Extension)

341

6.5.2.1 Synchronous and Asynchronous Flushing

343

6.5.2.2 Batching vs. Coalescing

344

6.5.2.3 Batching and ContentFilteredTopics

344

xv

6.5.2.4 Turbo Mode: Automatically Adjusting the Number of Bytes in a Batch—Experimental
Feature

344

6.5.2.5 Performance Considerations

345

6.5.2.6 Maximum Transport Datagram Size

345

6.5.2.7 Properties

345

6.5.2.8 Related QosPolicies

346

6.5.2.9 Applicable DDS Entities

346

6.5.2.10 System Resource Considerations

346

6.5.3 DATA_WRITER_PROTOCOL QosPolicy (DDS Extension)

347

6.5.3.1 High and Low Watermarks

352

6.5.3.2 Normal, Fast, and Late-Joiner Heartbeat Periods

353

6.5.3.3 Disabling Positive Acknowledgements

354

6.5.3.4 Configuring the Send Window Size

355

6.5.3.5 Propagating Serialized Keys with Disposed-Instance Notifications

356

6.5.3.6 Virtual Heartbeats

357

6.5.3.7 Resending Over Multicast

357

6.5.3.8 Example

358

6.5.3.9 Properties

358

6.5.3.10 Related QosPolicies

359

6.5.3.11 Applicable DDS Entities

359

6.5.3.12 System Resource Considerations

359

6.5.4 DATA_WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension)

359

6.5.4.1 Example

362

6.5.4.2 Properties

362

6.5.4.3 Related QosPolicies

363

6.5.4.4 Applicable DDS Entities

363

6.5.4.5 System Resource Considerations

363

6.5.5 DEADLINE QosPolicy

363

6.5.5.1 Example

364

6.5.5.2 Properties

365

6.5.5.3 Related QosPolicies

365

6.5.5.4 Applicable DDS Entities

365

6.5.5.5 System Resource Considerations

365

6.5.6 DESTINATION_ORDER QosPolicy

365

6.5.6.1 Properties

367

6.5.6.2 Related QosPolicies

368

xvi

6.5.6.3 Applicable DDS Entities

368

6.5.6.4 System Resource Considerations

368

6.5.7 DURABILITY QosPolicy

368

6.5.7.1 Example

370

6.5.7.2 Properties

371

6.5.7.3 Related QosPolicies

371

6.5.7.4 Applicable Entities

371

6.5.7.5 System Resource Considerations

372

6.5.8 DURABILITY SERVICE QosPolicy

372

6.5.8.1 Properties

374

6.5.8.2 Related QosPolicies

374

6.5.8.3 Applicable Entities

374

6.5.8.4 System Resource Considerations

374

6.5.9 ENTITY_NAME QosPolicy (DDS Extension)

374

6.5.9.1 Properties

375

6.5.9.2 Related QosPolicies

375

6.5.9.3 Applicable Entities

375

6.5.9.4 System Resource Considerations

376

6.5.10 HISTORY QosPolicy

376

6.5.10.1 Example

379

6.5.10.2 Properties

379

6.5.10.3 Related QosPolicies

380

6.5.10.4 Applicable Entities

380

6.5.10.5 System Resource Considerations

380

6.5.11 LATENCYBUDGET QoS Policy

380

6.5.11.1 Applicable Entities

381

6.5.12 LIFESPAN QoS Policy

381

6.5.12.1 Properties

382

6.5.12.2 Related QoS Policies

382

6.5.12.3 Applicable Entities

382

6.5.12.4 System Resource Considerations

382

6.5.13 LIVELINESS QosPolicy

382

6.5.13.1 Example

385

6.5.13.2 Properties

385

6.5.13.3 Related QosPolicies

386

6.5.13.4 Applicable Entities

386

xvii

6.5.13.5 System Resource Considerations
6.5.14 MULTI_CHANNEL QosPolicy (DDS Extension)

386
386

6.5.14.1 Example

389

6.5.14.2 Properties

389

6.5.14.3 Related Qos Policies

389

6.5.14.4 Applicable Entities

389

6.5.14.5 System Resource Considerations

389

6.5.15 OWNERSHIP QosPolicy

389

6.5.15.1 How Connext DDS Selects which DataWriter is the Exclusive Owner

391

6.5.15.2 Example

391

6.5.15.3 Properties

392

6.5.15.4 Related QosPolicies

392

6.5.15.5 Applicable Entities

393

6.5.15.6 System Resource Considerations

393

6.5.16 OWNERSHIP_STRENGTH QosPolicy

393

6.5.16.1 Example

393

6.5.16.2 Properties

393

6.5.16.3 Related QosPolicies

394

6.5.16.4 Applicable Entities

394

6.5.16.5 System Resource Considerations

394

6.5.17 PROPERTY QosPolicy (DDS Extension)

394

6.5.17.1 Properties

397

6.5.17.2 Related QosPolicies

397

6.5.17.3 Applicable Entities

397

6.5.17.4 System Resource Considerations

397

6.5.18 PUBLISH_MODE QosPolicy (DDS Extension)

397

6.5.18.1 Properties

399

6.5.18.2 Related QosPolicies

399

6.5.18.3 Applicable Entities

400

6.5.18.4 System Resource Considerations

400

6.5.19 RELIABILITY QosPolicy

400

6.5.19.1 Example

403

6.5.19.2 Properties

403

6.5.19.3 Related QosPolicies

404

6.5.19.4 Applicable Entities

404

6.5.19.5 System Resource Considerations

404

xviii

6.5.20 RESOURCE_LIMITS QosPolicy

405

6.5.20.1 Configuring Resource Limits for Asynchronous DataWriters

406

6.5.20.2 Configuring DataWriter Instance Replacement

407

6.5.20.3 Example

407

6.5.20.4 Properties

408

6.5.20.5 Related QosPolicies

408

6.5.20.6 Applicable Entities

408

6.5.20.7 System Resource Considerations

408

6.5.21 SERVICE QosPolicy (DDS Extension)

408

6.5.21.1 Properties

409

6.5.21.2 Related QosPolicies

409

6.5.21.3 Applicable Entities

409

6.5.21.4 System Resource Considerations

409

6.5.22 TRANSPORT_PRIORITY QosPolicy

409

6.5.22.1 Example

410

6.5.22.2 Properties

410

6.5.22.3 Related QosPolicies

411

6.5.22.4 Applicable Entities

411

6.5.22.5 System Resource Considerations

411

6.5.23 TRANSPORT_SELECTION QosPolicy (DDS Extension)

411

6.5.23.1 Example

412

6.5.23.2 Properties

412

6.5.23.3 Related QosPolicies

412

6.5.23.4 Applicable Entities

412

6.5.23.5 System Resource Considerations

412

6.5.24 TRANSPORT_UNICAST QosPolicy (DDS Extension)

412

6.5.24.1 Example

415

6.5.24.2 Properties

415

6.5.24.3 Related QosPolicies

415

6.5.24.4 Applicable Entities

415

6.5.24.5 System Resource Considerations

415

6.5.25 TYPESUPPORT QosPolicy (DDS Extension)

416

6.5.25.1 Properties

416

6.5.25.2 Related QoS Policies

417

6.5.25.3 Applicable Entities

417

6.5.25.4 System Resource Considerations

417

xix

6.5.26 USER_DATA QosPolicy

417

6.5.26.1 Example

418

6.5.26.2 Properties

418

6.5.26.3 Related QosPolicies

418

6.5.26.4 Applicable Entities

419

6.5.26.5 System Resource Considerations

419

6.5.27 WRITER_DATA_LIFECYCLE QoS Policy

419

6.5.27.1 Properties

421

6.5.27.2 Related QoS Policies

422

6.5.27.3 Applicable Entities

422

6.5.27.4 System Resource Considerations

422

6.6 FlowControllers (DDS Extension)

422

6.6.1 Flow Controller Scheduling Policies

424

6.6.2 Managing Fast DataWriters When Using a FlowController

426

6.6.3 Token Bucket Properties

426

6.6.3.1 max_tokens

427

6.6.3.2 tokens_added_per_period

427

6.6.3.3 tokens_leaked_per_period

427

6.6.3.4 period

427

6.6.3.5 bytes_per_token

428

6.6.4 Prioritized DDS Samples

428

6.6.4.1 Designating Priorities

429

6.6.4.2 Priority-Based Filtering

430

6.6.5 Creating and Configuring Custom FlowControllers with Property QoS
6.6.5.1 Example

431
432

6.6.6 Creating and Deleting FlowControllers

433

6.6.7 Getting/Setting Default FlowController Properties

434

6.6.8 Getting/Setting Properties for a Specific FlowController

435

6.6.9 Adding an External Trigger

435

6.6.10 Other FlowController Operations

435

Chapter 7 Receiving Data
7.1 Preview: Steps to Receiving Data

437

7.2 Subscribers

440

7.2.1 Creating Subscribers Explicitly vs. Implicitly

444

7.2.2 Creating Subscribers

445

7.2.3 Deleting Subscribers

446

xx

7.2.3.1 Deleting Contained DataReaders
7.2.4 Setting Subscriber QosPolicies

447
447

7.2.4.1 Configuring QoS Settings when the Subscriber is Created

448

7.2.4.2 Comparing QoS Values

450

7.2.4.3 Changing QoS Settings After Subscriber Has Been Created

450

7.2.4.4 Getting and Settings Subscriber’s Default QoS Profile and Library

451

7.2.4.5 Getting and Setting Default QoS for DataReaders

452

7.2.4.6 Subscriber QoS-Related Operations

453

7.2.5 Beginning and Ending Group-Ordered Access

453

7.2.6 Setting Up SubscriberListeners

454

7.2.7 Getting DataReaders with Specific DDS Samples

456

7.2.8 Finding a Subscriber’s Related Entities

457

7.2.9 Statuses for Subscribers

458

7.2.9.1 DATA_ON_READERS Status
7.3 DataReaders

458
459

7.3.1 Creating DataReaders

463

7.3.2 Getting All DataReaders

465

7.3.3 Deleting DataReaders

466

7.3.3.1 Deleting Contained ReadConditions

466

7.3.4 Setting Up DataReaderListeners

466

7.3.5 Checking DataReader Status and StatusConditions

468

7.3.6 Waiting for Historical Data

469

7.3.7 Statuses for DataReaders

470

7.3.7.1 DATA_AVAILABLE Status

471

7.3.7.2 DATA_READER_CACHE_STATUS

471

7.3.7.3 DATA_READER_PROTOCOL_STATUS

472

7.3.7.4 LIVELINESS_CHANGED Status

475

7.3.7.5 REQUESTED_DEADLINE_MISSED Status

476

7.3.7.6 REQUESTED_INCOMPATIBLE_QOS Status

477

7.3.7.7 SAMPLE_LOST Status

478

7.3.7.8 SAMPLE_REJECTED Status

479

7.3.7.9 SUBSCRIPTION_MATCHED Status

482

7.3.8 Setting DataReader QosPolicies

482

7.3.8.1 Configuring QoS Settings when the DataReader is Created

485

7.3.8.2 Comparing QoS Values

487

7.3.8.3 Changing QoS Settings After DataReader Has Been Created

487

xxi

7.3.8.4 Using a Topic’s QoS to Initialize a DataWriter’s QoS
7.3.9 Navigating Relationships Among Entities

488
489

7.3.9.1 Finding Matching Publications

489

7.3.9.2 Finding the Matching Publication’s ParticipantBuiltinTopicData

490

7.3.9.3 Finding a DataReader’s Related Entities

490

7.3.9.4 Looking Up an Instance Handle

490

7.3.9.5 Getting the Key Value for an Instance

491

7.4 Using DataReaders to Access Data (Read & Take)

491

7.4.1 Using a Type-Specific DataReader (FooDataReader)

491

7.4.2 Loaning and Returning Data and SampleInfo Sequences

492

7.4.2.1 C, Traditional C++, Java and .NET

492

7.4.2.2 Modern C++

493

7.4.3 Accessing DDS Data Samples with Read or Take

493

7.4.3.1 Read vs. Take

494

7.4.3.2 General Patterns for Accessing Data

496

7.4.3.3 read_next_sample and take_next_sample

497

7.4.3.4 read_instance and take_instance

497

7.4.3.5 read_next_instance and take_next_instance

498

7.4.3.6 read_w_condition and take_w_condition

500

7.4.3.7 read_instance_w_condition and take_instance_w_condition

500

7.4.3.8 read_next_instance_w_condition and take_next_instance_w_condition

501

7.4.3.9 The select() API (Modern C++)

501

7.4.4 Acknowledging DDS Samples

502

7.4.5 The Sequence Data Structure

502

7.4.6 The SampleInfo Structure

504

7.4.6.1 Reception Timestamp

506

7.4.6.2 Sample States

506

7.4.6.3 View States

506

7.4.6.4 Instance States

507

7.4.6.5 Generation Counts and Ranks

508

7.4.6.6 Valid Data Flag

510

7.5 Subscriber QosPolicies

510

7.6 DataReader QosPolicies

510

7.6.1 DATA_READER_PROTOCOL QosPolicy (DDS Extension)

511

7.6.1.1 Receive Window Size

515

7.6.1.2 Round-Trip Time For Filtering Redundant NACKs

516

xxii

7.6.1.3 Example

516

7.6.1.4 Properties

517

7.6.1.5 Related QosPolicies

517

7.6.1.6 Applicable Dds Entities

517

7.6.1.7 System Resource Considerations

517

7.6.2 DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension)

517

7.6.2.1 max_total_instances and max_instances

522

7.6.2.2 Example

522

7.6.2.3 Properties

523

7.6.2.4 Related QosPolicies

523

7.6.2.5 Applicable Dds Entities

523

7.6.2.6 System Resource Considerations

523

7.6.3 READER_DATA_LIFECYCLE QoS Policy

523

7.6.3.1 Properties

525

7.6.3.2 Related QoS Policies

525

7.6.3.3 Applicable Dds Entities

525

7.6.3.4 System Resource Considerations

526

7.6.4 TIME_BASED_FILTER QosPolicy

526

7.6.4.1 Example

528

7.6.4.2 Properties

528

7.6.4.3 Related QosPolicies

528

7.6.4.4 Applicable Dds Entities

528

7.6.4.5 System Resource Considerations

528

7.6.5 TRANSPORT_MULTICAST QosPolicy (DDS Extension)

529

7.6.5.1 Example

531

7.6.5.2 Properties

531

7.6.5.3 Related QosPolicies

531

7.6.5.4 Applicable DDS Entities

532

7.6.5.5 System Resource Considerations

532

7.6.6 TYPE_CONSISTENCY_ENFORCEMENT QosPolicy

532

7.6.6.1 Properties

534

7.6.6.2 Related QoS Policies

534

7.6.6.3 Applicable Entities

534

7.6.6.4 System Resource Considerations

535

Chapter 8 Working with DDS Domains
8.1 Fundamentals of DDS Domains and DomainParticipants

536

xxiii

8.2 DomainParticipantFactory
8.2.1 Setting DomainParticipantFactory QosPolicies
8.2.1.1 Getting and Setting the DomainParticipantFactory’s Default QoS Profile and Library

539
543
544

8.2.2 Getting and Setting Default QoS for DomainParticipants

545

8.2.3 Freeing Resources Used by the DomainParticipantFactory

546

8.2.4 Looking Up DomainParticipants

546

8.2.5 Getting QoS Values from a QoS Profile

547

8.3 DomainParticipants

547

8.3.1 Creating a DomainParticipant

556

8.3.2 Deleting DomainParticipants

558

8.3.3 Deleting Contained Entities

559

8.3.4 Choosing a Domain ID and Creating Multiple DDS Domains

559

8.3.5 Setting Up DomainParticipantListeners

560

8.3.6 Setting DomainParticipant QosPolicies

562

8.3.6.1 Configuring QoS Settings when DomainParticipant is Created

564

8.3.6.2 Comparing QoS Values

565

8.3.6.3 Changing QoS Settings After DomainParticipant Has Been Created

566

8.3.6.4 Getting and Setting DomainParticipant’s Default QoS Profile and Library

567

8.3.6.5 Getting and Setting Default QoS for Child Entities

568

8.3.7 Looking up Topic Descriptions

568

8.3.8 Finding a Topic

569

8.3.9 Getting the Implicit Publisher or Subscriber

569

8.3.10 Asserting Liveliness

570

8.3.11 Learning about Discovered DomainParticipants

571

8.3.12 Learning about Discovered Topics

571

8.3.13 Other DomainParticipant Operations

571

8.3.13.1 Verifying Entity Containment

571

8.3.13.2 Getting the Current Time

571

8.3.13.3 Getting All Publishers and Subscribers

572

8.4 DomainParticipantFactory QosPolicies
8.4.1 LOGGING QosPolicy (DDS Extension)

572
572

8.4.1.1 Example

572

8.4.1.2 Properties

573

8.4.1.3 Related QosPolicies

573

8.4.1.4 Applicable DDS Entities

573

8.4.1.5 System Resource Considerations

573

xxiv

8.4.2 PROFILE QosPolicy (DDS Extension)

573

8.4.2.1 Example

574

8.4.2.2 Properties

575

8.4.2.3 Related QosPolicies

575

8.4.2.4 Applicable Entities

575

8.4.2.5 System Resource Considerations

575

8.4.3 SYSTEM_RESOURCE_LIMITS QoS Policy (DDS Extension)

575

8.4.3.1 Example

576

8.4.3.2 Properties

576

8.4.3.3 Related QoS Policies

577

8.4.3.4 Applicable Dds Entities

577

8.4.3.5 System Resource Considerations

577

8.5 DomainParticipant QosPolicies
8.5.1 DATABASE QosPolicy (DDS Extension)

577
577

8.5.1.1 Example

579

8.5.1.2 Properties

579

8.5.1.3 Related QosPolicies

579

8.5.1.4 Applicable Dds Entities

580

8.5.1.5 System Resource Considerations

580

8.5.2 DISCOVERY QosPolicy (DDS Extension)

580

8.5.2.1 Transports Used for Discovery

581

8.5.2.2 Setting the ‘Initial Peers’ List

581

8.5.2.3 Adding and Removing Peers List Entries

581

8.5.2.4 Configuring Multicast Receive Addresses

582

8.5.2.5 Meta-Traffic Transport Priority

583

8.5.2.6 Controlling Acceptance of Unknown Peers

583

8.5.2.7 Example

583

8.5.2.8 Properties

584

8.5.2.9 Related QosPolicies

584

8.5.2.10 Applicable Entities

584

8.5.2.11 System Resource Considerations

584

8.5.3 DISCOVERY_CONFIG QosPolicy (DDS Extension)

585

8.5.3.1 Resource Limits for Builtin-Topic DataReaders

589

8.5.3.2 Controlling Purging of Remote Participants

591

8.5.3.3 Controlling the Reliable Protocol Used by Builtin-Topic DataWriters/DataReaders

592

8.5.3.4 Example

592

xxv

8.5.3.5 Properties

593

8.5.3.6 Related QosPolicies

593

8.5.3.7 Applicable Dds Entities

593

8.5.3.8 System Resource Considerations

593

8.5.4 DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension)

593

8.5.4.1 Configuring Resource Limits for Asynchronous DataWriters

600

8.5.4.2 Configuring Memory Allocation

600

8.5.4.3 Example

601

8.5.4.4 Properties

602

8.5.4.5 Related QosPolicies

602

8.5.4.6 Applicable DDS Entities

602

8.5.4.7 System Resource Considerations

602

8.5.5 EVENT QosPolicy (DDS Extension)

602

8.5.5.1 Example

603

8.5.5.2 Properties

604

8.5.5.3 Related QosPolicies

604

8.5.5.4 Applicable DDS Entities

604

8.5.5.5 System Resource Considerations

604

8.5.6 RECEIVER_POOL QosPolicy (DDS Extension)

604

8.5.6.1 Example

606

8.5.6.2 Properties

606

8.5.6.3 Related QosPolicies

606

8.5.6.4 Applicable Dds Entities

606

8.5.6.5 System Resource Considerations

606

8.5.7 TRANSPORT_BUILTIN QosPolicy (DDS Extension)

606

8.5.7.1 Example

607

8.5.7.2 Properties

607

8.5.7.3 Related QosPolicies

607

8.5.7.4 Applicable DDS Entities

607

8.5.7.5 System Resource Considerations

608

8.5.8 TRANSPORT_MULTICAST_MAPPING QosPolicy (DDS Extension)

608

8.5.8.1 Formatting Rules for Addresses

609

8.5.8.2 Example

610

8.5.8.3 Properties

610

8.5.8.4 Related QosPolicies

610

8.5.8.5 Applicable DDS Entities

610

xxvi

8.5.8.6 System Resource Considerations
8.5.9 WIRE_PROTOCOL QosPolicy (DDS Extension)

610
610

8.5.9.1 Choosing Participant IDs

611

8.5.9.2 Host, App, and Instance IDs

613

8.5.9.3 Ports Used for Discovery

613

8.5.9.4 Controlling How the GUID is Set (rtps_auto_id_kind)

614

8.5.9.5 Example

618

8.5.9.6 Properties

618

8.5.9.7 Related QosPolicies

619

8.5.9.8 Applicable DDS Entities

619

8.5.9.9 System Resource Considerations

619

8.6 Clock Selection

619

8.6.1 Available Clocks

619

8.6.2 Clock Selection Strategy

619

8.7 System Properties

620

Chapter 9 Building Applications
9.1 Running on a Computer Not Connected to a Network

623

9.2 Connext DDS Header Files — All Architectures

623

9.3 UNIX-Based Platforms

624

9.3.1 Required Libraries

625

9.3.2 Compiler Flags

625

9.4 Windows Platforms
9.4.1 Using Visual Studio
9.5 Java Platforms

625
626
627

9.5.1 Java Libraries

627

9.5.2 Native Libraries

627

Part 3: Advanced Concepts

628

Chapter 10 Reliable Communications
10.1 Sending Data Reliably

629

10.1.1 Best-effort Delivery Model

629

10.1.2 Reliable Delivery Model

630

10.2 Overview of the Reliable Protocol

631

10.3 Using QosPolicies to Tune the Reliable Protocol

635

10.3.1 Enabling Reliability
10.3.1.1 Blocking until the Send Queue Has Space Available
10.3.2 Tuning Queue Sizes and Other Resource Limits

637
637
638

xxvii

10.3.2.1 Understanding the Send Queue and Setting its Size

639

10.3.2.2 Understanding the Receive Queue and Setting Its Size

642

10.3.3 Controlling Queue Depth with the History QosPolicy

644

10.3.4 Controlling Heartbeats and Retries with DataWriterProtocol QosPolicy

645

10.3.4.1 How Often Heartbeats are Resent (heartbeat_period)

645

10.3.4.2 How Often Piggyback Heartbeats are Sent (heartbeats_per_max_samples)

647

10.3.4.3 Controlling Packet Size for Resent DDS Samples (max_bytes_per_nack_response)

649

10.3.4.4 Controlling How Many Times Heartbeats are Resent (max_heartbeat_retries)

650

10.3.4.5 Treating Non-Progressing Readers as Inactive Readers (inactivate_nonprogressing_readers) 650
10.3.4.6 Coping with Redundant Requests for Missing DDS Samples (max_nack_response_delay)

651

10.3.4.7 Disabling Positive Acknowledgements (disable_positive_acks_min_sample_keep_duration) 652
10.3.5 Avoiding Message Storms with DataReaderProtocol QosPolicy

653

10.3.6 Resending DDS Samples to Late-Joiners with the Durability QosPolicy

653

10.3.7 Use Cases

654

10.3.7.1 Importance of Relative Thread Priorities

654

10.3.7.2 Aperiodic Use Case: One-at-a-Time

655

10.3.7.3 Aperiodic, Bursty

659

10.3.7.4 Periodic

664

10.4 Auto Throttling for DataWriter Performance—Experimental Feature

668

Chapter 11 Collaborative DataWriters
11.1 Collaborative DataWriters Use Cases

671

11.2 DDS Sample Combination (Synchronization) Process in a DataReader

672

11.3 Configuring Collaborative DataWriters

673

11.3.1 Assocating Virtual GUIDs with DDS Data Samples

673

11.3.2 Assocating Virtual Sequence Numbers with DDS Data Samples

673

11.3.3 Specifying which DataWriters will Deliver DDS Samples to the DataReader from a Logical Data
Source

673

11.3.4 Specifying How Long to Wait for a Missing DDS Sample

673

11.4 Collaborative DataWriters and Persistence Service

674

Chapter 12 Mechanisms for Achieving Information Durability and Persistence
12.1 Introduction

675

12.1.1 Scenario 1. DataReader Joins after DataWriter Restarts (Durable Writer History)

676

12.1.2 Scenario 2: DataReader Restarts While DataWriter Stays Up (Durable Reader State)

677

12.1.3 Scenario 3. DataReader Joins after DataWriter Leaves Domain (Durable Data)

679

12.2 Durability and Persistence Based on Virtual GUIDs

680

12.3 Durable Writer History

681

12.3.1 Durable Writer History Use Case

682

xxviii

12.3.2 How To Configure Durable Writer History
12.4 Durable Reader State
12.4.1 Durable Reader State With Protocol Acknowledgment
12.4.1.1 Bandwidth Utilization
12.4.2 Durable Reader State with Application Acknowledgment
12.4.2.1 Bandwidth Utilization

683
686
687
688
688
689

12.4.3 Durable Reader State Use Case

689

12.4.4 How To Configure a DataReader for Durable Reader State

690

12.5 Data Durability
12.5.1 RTI Persistence Service

692
692

Chapter 13 Guaranteed Delivery of Data
13.1 Introduction

695

13.1.1 Identifying the Required Consumers of Information

697

13.1.2 Ensuring Consumer Applications Process the Data Successfully

698

13.1.3 Ensuring Information is Available to Late-Joining Applications

699

13.2 Scenarios

700

13.2.1 Scenario 1: Guaranteed Delivery to a-priori Known Subscribers

701

13.2.2 Scenario 2: Surviving a Writer Restart when Delivering DDS Samples to a priori Known Subscribers

703

13.2.3 Scenario 3: Delivery Guaranteed by Persistence Service (Store and Forward) to a priori Known Subscribers
704
13.2.3.1 Variation: Using Redundant Persistence Services

706

13.2.3.2 Variation: Using Load-Balanced Persistent Services

707

Chapter 14 Discovery
14.1 What is Discovery?

710

14.1.1 Simple Participant Discovery

710

14.1.2 Simple Endpoint Discovery

711

14.2 Configuring the Peers List Used in Discovery

711

14.2.1 Peer Descriptor Format

713

14.2.1.1 Locator Format

714

14.2.1.2 Address Format

715

14.2.2 NDDS_DISCOVERY_PEERS Environment Variable Format

716

14.2.3 NDDS_DISCOVERY_PEERS File Format

717

14.3 Discovery Implementation
14.3.1 Participant Discovery
14.3.1.1 Refresh Mechanism

717
718
722

xxix

14.3.1.2 Maintaining DataWriter Liveliness for kinds AUTOMATIC and MANUAL_BY_
PARTICIPANT

724

14.3.2 Endpoint Discovery

728

14.3.3 Discovery Traffic Summary

733

14.3.4 Discovery-Related QoS

734

14.4 Debugging Discovery

735

14.5 Ports Used for Discovery

738

14.5.1 Inbound Ports for Meta-Traffic

739

14.5.2 Inbound Ports for User Traffic

740

14.5.3 Automatic Selection of participant_id and Port Reservation

740

14.5.4 Tuning domain_id_gain and participant_id_gain

740

Chapter 15 Transport Plugins
15.1 Builtin Transport Plugins

743

15.2 Extension Transport Plugins

744

15.3 The NDDSTransportSupport Class

745

15.4 Explicitly Creating Builtin Transport Plugin Instances

746

15.5 Setting Builtin Transport Properties of Default Transport Instance—get/set_builtin_transport_properties() 746
15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

748

15.6.1 Setting the Maximum Gather-Send Buffer Count for UDPv4 and UDPv6

763

15.6.2 Formatting Rules for IPv6 ‘Allow’ and ‘Deny’ Address Lists

765

15.7 Installing Additional Builtin Transport Plugins with register_transport()

765

15.7.1 Transport Lifecycles

766

15.7.2 Transport Aliases

767

15.7.3 Transport Network Addresses

768

15.8 Installing Additional Builtin Transport Plugins with PropertyQosPolicy

768

15.9 Other Transport Support Operations

769

15.9.1 Adding a Send Route

769

15.9.2 Adding a Receive Route

770

15.9.3 Looking Up a Transport Plugin

771

Chapter 16 Built-In Topics
16.1 Listeners for Built-in Entities

772

16.2 Built-in DataReaders

773

16.2.1 LOCATOR_FILTER QoS Policy (DDS Extension)

782

16.3 Accessing the Built-in Subscriber

783

16.4 Restricting Communication—Ignoring Entities

784

16.4.1 Ignoring Specific Remote DomainParticipants

785

16.4.2 Ignoring Publications and Subscriptions

786

xxx

16.4.3 Ignoring Topics

788

16.4.4 Resource Limits Considerations for Ignored Entities

788

16.4.5 Supervising Endpoint Discovery

788

Chapter 17 Configuring QoS with XML
17.1 Example XML File

791

17.2 QoS Libraries

792

17.3 QoS Profiles

793

17.3.1 Built-in QoS Profiles

794

17.3.2 Overwriting Default QoS Values

796

17.3.3 QoS Profile Inheritance

797

17.3.4 Topic Filters

799

17.3.5 QoS Profiles with a Single QoS

802

17.4 Configuring QoS with XML

803

17.4.1 QosPolicies

803

17.4.2 Sequences

804

17.4.3 Arrays

807

17.4.4 Enumeration Values

808

17.4.5 Time Values (Durations)

808

17.4.6 Transport Properties

808

17.4.7 Thread Settings

809

17.4.8 Entity Names

809

17.5 How to Load XML-Specified QoS Settings
17.5.1 Loading, Reloading and Unloading Profiles
17.6 XML File Syntax
17.6.1 Using Environment Variables in XML

810
811
812
813

17.7 XML String Syntax

814

17.8 URL Groups

814

17.9 How the XML is Validated

815

17.9.1 Validation at Run-Time

815

17.9.2 XML File Validation During Editing

816

17.10 Using QoS Profiles in Your Connext DDS Application

817

17.10.1 Retrieving a List of Available Libraries

823

17.10.2 Retrieving a List of Available QoS Profiles

823

17.11 Configuring Logging Via XML

823

Chapter 18 Multi-channel DataWriters
18.1 What is a Multi-channel DataWriter?

825

xxxi

18.2 How to Configure a Multi-channel DataWriter
18.2.1 Limitations

828
829

18.3 Multi-Channel Configuration on the Reader Side

830

18.4 Where Does the Filtering Occur?

832

18.4.1 Filtering at the DataWriter

832

18.4.2 Filtering at the DataReader

832

18.4.3 Filtering on the Network Hardware

833

18.5 Fault Tolerance and Redundancy

833

18.6 Reliability with Multi-Channel DataWriters

834

18.6.1 Reliable Delivery

834

18.6.2 Reliable Protocol Considerations

834

18.7 Performance Considerations

835

18.7.1 Network-Switch Filtering

835

18.7.2 DataWriter and DataReader Filtering

835

Chapter 19 Connext DDS Threading Model
19.1 Database Thread

837

19.2 Event Thread

838

19.3 Receive Threads

839

19.4 Exclusive Areas, Connext DDS Threads and User Listeners

841

19.5 Controlling CPU Core Affinity for RTI Threads

842

19.6 Configuring Thread Settings with XML

842

19.7 User-Managed Threads

844

Chapter 20 DDS Sample-Data and Instance-Data Memory Management
20.1 DDS Sample-Data Memory Management for DataWriters

846

20.1.1 Memory Management without Batching

847

20.1.2 Memory Management with Batching

849

20.1.3 Writer-Side Memory Management when Using Java

851

20.1.4 Writer-Side Memory Management when Working with Large Data

851

20.2 DDS Sample-Data Memory Management for DataReaders

853

20.2.1 Memory Management for DataReaders Using Generated Type-Plugins

854

20.2.2 Reader-Side Memory Management when Using Java

856

20.2.3 Memory Management for DynamicData DataReaders

857

20.2.4 Memory Management for Fragmented DDS Samples

859

20.2.5 Reader-Side Memory Management when Working with Large Data

859

20.3 Instance-Data Memory Management for DataWriters

861

20.4 Instance-Data Memory Management for DataReaders

861

xxxii

Chapter 21 Troubleshooting
21.1 What Version am I Running?

863

21.1.1 Finding Version Information in Revision Files

863

21.1.2 Finding Version Information Programmatically

864

21.2 Controlling Messages from Connext DDS
21.2.1 Format of Logged Messages

865
868

21.2.1.1 Timestamps

868

21.2.1.2 Thread identification

869

21.2.1.3 Hierarchical Context

869

21.2.1.4 Explanation of Context Strings

869

21.2.2 Configuring Logging via XML

871

21.2.3 Customizing the Handling of Generated Log Messages

872

Part 4: Request-Reply Communication Pattern

873

Chapter 22 Introduction to the Request-Reply Communication Pattern
22.1 The Request-Reply Pattern

875

22.1.1 Request-Reply Correlation

877

22.2 Single-Request, Multiple-Replies

877

22.3 Multiple Repliers

878

22.4 Combining Request-Reply and Publish-Subscribe

879

Chapter 23 Using the Request-Reply Communication Pattern
23.1 Requesters

881

23.1.1 Creating a Requester

882

23.1.2 Destroying a Requester

883

23.1.3 Setting Requester Parameters

883

23.1.4 Summary of Requester Operations

884

23.1.5 Sending Requests

885

23.1.6 Processing Incoming Replies with a Requester

886

23.1.6.1 Waiting for Replies

886

23.1.6.2 Getting Replies

887

23.1.6.3 Receiving Replies

889

23.2 Repliers

890

23.2.1 Creating a Replier

890

23.2.2 Destroying a Replier

891

23.2.3 Setting Replier Parameters

891

23.2.4 Summary of Replier Operations

892

23.2.5 Processing Incoming Requests with a Replier

893

xxxiii

23.2.5.1 Waiting for Requests

894

23.2.5.2 Reading and Taking Requests

894

23.2.5.3 Receiving Requests

895

23.2.6 Sending Replies
23.3 SimpleRepliers

896
896

23.3.1 Creating a SimpleReplier

897

23.3.2 Destroying a SimpleReplier

897

23.3.3 Setting SimpleReplier Parameters

897

23.3.4 Getting Requests and Sending Replies with a SimpleReplierListener

898

23.4 Accessing Underlying DataWriters and DataReaders
Part 5: RTI Secure WAN Transport

898
900

Chapter 24 Introduction to Secure WAN Transport
24.1 WAN Traversal via UDP Hole-Punching
24.1.1 Protocol Details

902
903

24.2 WAN Locators

907

24.3 Datagram Transport-Layer Security (DTLS)

908

24.3.1 Security Model

909

24.3.2 Liveliness Mechanism

909

24.4 Certificate Support

909

24.5 License Issues

911

Chapter 25 Configuring RTI Secure WAN Transport
25.1 Example Applications

914

25.2 Setting Up a Transport with the Property QoS

915

25.3 WAN Transport Properties

917

25.4 Secure Transport Properties

925

25.5 Explicitly Instantiating a WAN or Secure Transport Plugin

930

25.5.1 Additional Header Files and Include Directories

931

25.5.2 Additional Libraries

931

25.5.3 Compiler Flags

931

Part 6: RTI Persistence Service

932

Chapter 26 Introduction to RTI Persistence Service

933

Chapter 27 Configuring Persistence Service
27.1 How to Load the Persistence Service XML Configuration

935

27.2 XML Configuration File

936

27.2.1 Configuration File Syntax

937

27.2.2 XML Validation

938

xxxiv

27.2.2.1 Validation at Run Time

938

27.2.2.2 Validation During Editing

938

27.3 QoS Configuration

939

27.4 Configuring the Persistence Service Application

940

27.5 Configuring Remote Administration

942

27.6 Configuring Persistent Storage

943

27.7 Configuring Participants

946

27.8 Creating Persistence Groups

947

27.8.1 QoSs

952

27.8.2 DurabilityService QoS Policy

953

27.8.3 Sharing a Publisher/Subscriber

953

27.8.4 Sharing a Database Connection

954

27.8.5 Memory Management

954

27.9 Configuring Durable Subscriptions in Persistence Service
27.9.1 DDS Sample Memory Management With Durable Subscriptions

955
956

27.10 Synchronizing of Persistence Service Instances

956

27.11 Enabling RTI Distributed Logger in Persistence Service

957

27.12 Enabling RTI Monitoring Library in Persistence Service

958

27.13 Support for Extensible Types

959

27.13.1 Type Version Discrimination
27.14 TCP Transport Support in Persistence Service

960
960

Chapter 28 Running RTI Persistence Service
28.1 Starting Persistence Service

962

28.2 Stopping Persistence Service

965

Chapter 29 Administering Persistence Service from a Remote Location
29.1 Enabling Remote Administration

966

29.2 Remote Commands

967

29.2.1 start

967

29.2.2 stop

967

29.2.3 shutdown

968

29.2.4 status

968

29.3 Accessing Persistence Service from a Connext DDS Application

968

Chapter 30 Advanced Persistence Service Scenarios
30.1 Scenario: Load-balanced Persistence Services

972

30.2 Scenario: Delegated Reliability

974

30.3 Scenario: Slow Consumer

975

xxxv

Part 7: RTI CORBA Compatibility Kit

979

Chapter 31 Introduction to RTI CORBA Compatibility Kit

980

Chapter 32 Generating CORBA-Compatible Code
32.1 Generating C++ Code

983

32.2 Generating Java Code

984

Chapter 33 Supported IDL Types

985

Part 8: RTI TCP Transport

987

Chapter 34 TCP Communication Scenarios
34.1 Communication Within a Single LAN

988

34.2 Symmetric Communication Across NATs

989

34.3 Asymmetric Communication Across NATs

990

35.1 Configuring the TCP Transport

993

35.1.1 Choosing a Transport Mode

993

35.1.2 Explicitly Instantiating the TCP Transport Plugin

994

35.1.2.1 Additional Header Files and Include Directories

995

35.1.2.2 Additional Libraries and Compiler Flags

995

35.1.3 Configuring the TCP Transport with the Property QosPolicy

996

35.1.3.1 Configuring the TCP Transport to be Loaded Statically

998

35.1.3.2 Loading TLS Support Libraries Statically

999

35.1.4 Setting the Initial Peers
35.1.5 Support for External Hardware Load Balancers in TCP Transport Plugin
35.1.5.1 Session-ID Messages

999
1000
1002

35.1.6 TCP/TLS Transport Properties

1002

35.1.6.1 Connection Liveliness

1020

Part 9: RTI Monitoring Library

1022

Chapter 36 Using Monitoring Library in Your Application
36.1 Enabling Monitoring

1024

36.1.1 Method 1—Change the Participant QoS to Automatically Load the Dynamic Monitoring Library 1025
36.1.2 Method 2—Change the Participant QoS to Specify the Monitoring Library Create Function
Pointer and Explicitly Load the Monitoring Library

1025

36.1.2.1 Method 2-A: Change the Participant QoS by Specifying the Monitoring Library Create
Function Pointer in Source Code

1026

36.1.2.2 Method 2-B: Change the Participant QoS by Specifying the Monitoring Library Create
Function Pointer in an Environment Variable

1029

36.2 How does Monitoring Library Work?

1031

36.3 What Monitoring Topics are Published?

1031

36.4 Enabling Support for Large Type-Code (Optional)

1032

xxxvi

36.5 Troubleshooting Monitoring
36.5.1 Buffer Allocation Error

1033
1033

Chapter 37 Configuring Monitoring Library

1034

Part 10: RTI Distributed Logger

1039

Chapter 38 Using Distributed Logger in a Connext DDS Application
38.1 Using the API Directly

1041

38.2 Examples

1042

38.3 Data Type Resource

1043

38.4 Distributed Logger Topics

1044

38.5 Distributed Logger IDL

1044

38.6 Viewing Log Messages

1045

38.7 Logging Levels

1045

38.8 Distributed Logger Quality of Service Settings

1046

Chapter 39 Enabling Distributed Logger in RTI Services
39.1 Relationship Between Service Verbosity and Filter Level

1052

xxxvii

About this Document
Paths Mentioned in Documentation
The documentation refers to:
l


This refers to the installation directory for Connext DDS. The default installation paths are:
l Mac OS X systems:
/Applications/rti_connext_dds-5.2.3
l

l

l

l

UNIX-based systems, non-root user:
/home/your user name/rti_connext_dds-5.2.3
UNIX-based systems, root user:
/opt/rti_connext_dds-5.2.3
Windows systems, user without Administrator privileges:
\rti_connext_dds-5.2.3
Windows systems, user with Administrator privileges:
C:\Program Files\rti_connext_dds-5.2.3 (64-bit machines)
C:\Program Files (x86)\rti_connext_dds-5.2.3 (32-bit machines)

You may also see $NDDSHOME or %NDDSHOME%, which refers to an environment
variable set to the installation path.
Wherever you see  used in a path, replace it with your installation path.
Note for Windows Users: When using a command prompt to enter a command that
includes the path C:\Program Files (or any directory name that has a space), enclose the
path in quotation marks. For example:
“C:\Program Files\rti_connext_dds-5.2.3\bin\rtiddsgen”

Or if you have defined the NDDSHOME environment variable:
“%NDDSHOME%\bin\rtiddsgen”

l



By default, examples are copied into your home directory the first time you run RTI
Launcher or any script in /bin. This document refers to the location of the
copied examples as .
Wherever you see , replace it with the appropriate path.
Default path to the examples:
l Mac OS X systems: /Users/your user name/rti_workspace/5.2.3/examples
l

UNIX-based systems: /home/your user name/rti_workspace/5.2.3/examples

l

Windows systems: your Windows documents folder\rti_workspace\5.2.3\examples
Where 'your Windows documents folder' depends on your version of Windows. For
example, on Windows 7, the folder is C:\Users\your user name\Documents; on Windows Server 2003, the folder is C:\Documents and Settings\your user name\Documents.

Note: You can specify a different location for rti_workspace. You can also specify that you
do not want the examples copied to the workspace. For details, see Controlling Location for
RTI Workspace and Copying of Examples in the Connext DDS Core Libraries Getting Started Guide.

Programming Language Conventions
The terminology and example code in this manual assume you are using Traditional C++ without
namespace support.
C, Modern C++, C++/CLI, C#, and Java APIs are also available; they are fully described in the
API Reference HTML documentation. (Note: the Modern C++ API is not available for all
platforms, check the RTI Connext DDS Core Libraries Platform Notes to see if it is available for
your platform.)
Namespace support in Traditional C++, C++/CLI, and C# is also available; see the API Reference
HTML documentation (from the Modules page, select Using DDS:: Namespace) for details. In
the Modern C++ API all types, constants and functions are always in namespaces.

Traditional vs. Modern C++
Connext DDS provides two different C++ APIs, which we refer to as the "Traditional C++" and
"Modern C++" APIs. They provide substantially different programming paradigms and patterns.
The Traditional API could be considered as simply "C with classes," while the Modern API incorporates modern C++ techniques, most notably:

l

Generic programming

l

Integration with the standard library

l

Automatic object lifecycle management, providing full value types and reference types

l

C++11 support, such as move operations, initializer lists, and support for range for-loops.

These different programming styles make the Modern C++ API differ significantly with respect to
the other language APIs in several aspects; to name a few:
l

Creating and Deleting DDS Entities (Section 4.1.1 on page 153)

l

Creating User Data Types with IDL (Section 3.3 on page 69)

l

Interacting Dynamically with User Data Types (Section 3.8 on page 141)

l

Working with DDS Data Samples (Section 3.9 on page 145)

l

Using DataReaders to Access Data (Read & Take) (Section 7.4 on page 491)

l

QoS policies and QoS management

l

Naming conventions

This manual points out these kinds of differences whenever they are substantial.

Extensions to the DDS Standard
Connext DDS implements the DDS Standard published by the OMG. It also includes features that
are extensions to DDS. These include additional Quality of Service parameters, function calls,
structure fields, etc.
Extensions also include product-specific APIs that complement the DDS API. These include APIs
to create and use transport plug-ins, and APIs to control the verbosity and logging capabilities.
These APIs are prefixed with NDDS, such as NDDSTransportSupport::register_transport().

Environment Variables
Connext DDS documentation refers to path names that have been customized during installation.
NDDSHOME refers to the installation directory of Connext DDS.
Names of Supported Platforms
Connext DDS runs on several different target platforms. To support this vast array of platforms,
Connext DDS separates the executable, library, and object files for each platform into individual
directories.

Each platform name has four parts: hardware architecture, operating system, operating system version and compiler. For example, i86Linux2.4gcc3.2 is the directory that contains files specific to
Linux® version 2.4 for the Intel processor, compiled with gcc version 3.2.
For a full list of supported platforms, see the RTI Connext DDS Core Libraries Platform Notes.

Additional Resources
The details of each API (such as function parameters, return values, etc.) and examples are in the
API Reference HTML documentation. In case of discrepancies between the information in this document and the API Reference HTML documentation, the latter should be considered more up-todate.

Part 1: Welcome to RTI Connext DDS
RTI Connext DDS solutions provide a flexible data distribution infrastructure for integrating data
sources of all types. At its core is the world's leading ultra-high performance, distributed networking DataBus™. It connects data within applications as well as across devices, systems and networks. Connext DDS also delivers large data sets with microsecond performance and granular
quality-of-service control. Connext DDS is a standards-based, open architecture that connects
devices from deeply embedded real-time platforms to enterprise servers across a variety of networks.
Part 1 introduces the general concepts behind data-centric publish-subscribe communications and
provides a brief tour of Connext DDS.
l

Overview (Section Chapter 1 on page 2)

l

Data-Centric Publish-Subscribe Communications (Section Chapter 2 on page 10)

1

Chapter 1 Overview
RTI Connext DDS is network middleware for distributed real-time applications. Connext DDS simplifies application development, deployment and maintenance and provides fast, predictable distribution of time-critical data over a variety of transport networks.
Connext DDS solutions provide a flexible data distribution infrastructure for integrating data
sources of all types. At its core is the world's leading ultra-high performance, distributed networking DataBus™. It connects data within applications as well as across devices, systems and networks. Connext DDS also delivers large data sets with microsecond performance and granular
quality-of-service control. Connext DDS is a standards-based, open architecture that connects
devices from deeply embedded real-time platforms to enterprise servers across a variety of networks.
With Connext DDS, you can:
l
l

Perform complex one-to-many and many-to-many network communications.
Customize application operation to meet various real-time, reliability, and quality-of-service
goals.

l

Provide application-transparent fault tolerance and application robustness.

l

Use a variety of transports.

This section introduces basic concepts of middleware and common communication models, and
describes how Connext DDS’s feature-set addresses the needs of real-time systems.

1.1 What is Connext DDS?
Connext DDS is network middleware for real-time distributed applications. It provides the communications service programmers need to distribute time-critical data between embedded and/or
enterprise devices or nodes. Connext DDS uses the publish-subscribe communications model to
make data distribution efficient and robust.

2

1.2 Network Communications Models

Connext DDS implements the Data-Centric Publish-Subscribe (DCPS) API within the OMG’s Data Distribution Service (DDS) for Real-Time Systems. DDS is the first standard developed for the needs of realtime systems. DCPS provides an efficient way to transfer data in a distributed system.
With Connext DDS, systems designers and programmers start with a fault-tolerant and flexible communications infrastructure that will work over a wide variety of computer hardware, operating systems, languages, and networking transport protocols. Connext DDS is highly configurable so programmers can
adapt it to meet the application’s specific communication requirements.

1.2 Network Communications Models
The communications model underlying the network middleware is the most important factor in how applications communicate. The communications model impacts the performance, the ease to accomplish different
communication transactions, the nature of detecting errors, and the robustness to different error conditions.
Unfortunately, there is no “one size fits all” approach to distributed applications. Different communications
models are better suited to handle different classes of application domains.
This section describes three main types of network communications models:
l

Point-to-point

l

Client-server

l

Publish-subscribe

Point-to-point model:
Point-to-point is the simplest form of communication, as illustrated in Figure 1.1 Point-to-Point on the
facing page. The telephone is an example of an everyday point-to-point communications device. To use a
telephone, you must know the address (phone number) of the other party. Once a connection is established, you can have a reasonably high-bandwidth conversation. However, the telephone does not work as
well if you have to talk to many people at the same time. The telephone is essentially one-to-one communication.
TCP is a point-to-point network protocol designed in the 1970s. While it provides reliable, high-bandwidth
communication, TCP is cumbersome for systems with many communicating nodes.

3

1.2 Network Communications Models

Figure 1.1 Point-to-Point

Point-to-point is one-to-one communication.

Client-server model:
To address the scalability issues of the Point-to-Point model, developers turned to the Client-Server model.
Client-server networks designate one special server node that connects simultaneously to many client
nodes, as illustrated in Figure 1.2 Client-Server below.
Figure 1.2 Client-Server

Client-server is many-to-one communications.

4

1.2 Network Communications Models

Client-server is a "many-to-one" architecture. Ordering pizza over the phone is an example of client-server
communication. Clients must know the phone number of the pizza parlor to place an order. The parlor can
handle many orders without knowing ahead of time where people (clients) are located. After the order
(request), the parlor asks the client where the response (pizza) should be sent. In the client-server model,
each response is tied to a prior request. As a result, the response can be tailored to each request. In other
words, each client makes a request (order) and each reply (pizza) is made for one specific client in mind.
The client-server network architecture works best when information is centralized, such as in databases,
transaction processing systems, and file servers. However, if information is being generated at multiple
nodes, a client-server architecture requires that all information are sent to the server for later redistribution
to the clients. This approach is inefficient and precludes deterministic communications, since the client
does not know when new information is available. The time between when the information is available on
the server, and when the client asks and receives it adds a variable latency to the system.
Publish-subscribe model: In the publish-subscribe communications model (Figure 1.3 Publish-Subscribe
on the facing page), computer applications (nodes) “subscribe” to data they need and “publish” data they
want to share. Messages pass directly between the publisher and the subscribers, rather than moving into
and out of a centralized server. Most time-sensitive information intended to reach many people is sent by a
publish-subscribe system. Examples of publish-subscribe systems in everyday life include television,
magazines, and newspapers.
Publish-subscribe communication architectures are good for distributing large quantities of time-sensitive
information efficiently, even in the presence of unreliable delivery mechanisms. This direct and simultaneous communication among a variety of nodes makes publish-subscribe network architecture the best
choice for systems with complex time-critical data flows.
While the publish-subscribe model provides system architects with many advantages, it may not be the
best choice for all types of communications, including:
l

File-based transfers (alternate solution: FTP)

l

Remote Method Invocation (alternate solutions: CORBA, COM, SOAP)

l

Connection-based architectures (alternate solution: TCP/IP)

l

Synchronous transfers (alternate solution: CORBA)

5

1.3 What is Middleware?

Figure 1.3 Publish-Subscribe

Publish-subscribe is many-to-many communications.

1.3 What is Middleware?
Middleware is a software layer between an application and the operating system. Network middleware isolates the application from the details of the underlying computer architecture, operating system and network
stack (see Figure 1.4 Network Middleware on the next page). Network middleware simplifies the development of distributed systems by allowing applications to send and receive information without having to
program using lower-level protocols such as sockets and TCP or UDP/IP.

6

1.4 Features of Connext DDS

Figure 1.4 Network Middleware

Connext DDS is middleware that insulates applications from the raw operating-system network stack.

Publish-subscribe middleware:Connext DDS is based on a publish-subscribe communications model.
Publish-subscribe (PS) middleware provides a simple and intuitive way to distribute data. It decouples the
software that creates and sends data—the data publishers—from the software that receives and uses the
data—the data subscribers. Publishers simply declare their intent to send and then publish the data. Subscribers declare their intent to receive, then the data is automatically delivered by the middleware.
Despite the simplicity of the model, PS middleware can handle complex patterns of information flow. The
use of PS middleware results in simpler, more modular distributed applications. Perhaps most importantly,
PS middleware can automatically handle all network chores, including connections, failures, and network
changes, eliminating the need for user applications to program of all those special cases. What experienced
network middleware developers know is that handling special cases accounts for over 80% of the effort
and code.

1.4 Features of Connext DDS
Connext DDS supports mechanisms that go beyond the basic publish-subscribe model. The key benefit is
that applications that use Connext DDS for their communications are entirely decoupled. Very little of
their design time has to be spent on how to handle their mutual interactions. In particular, the applications
never need information about the other participating applications, including their existence or locations.
Connext DDS automatically handles all aspects of message delivery, without requiring any intervention
from the user applications, including:

7

1.4 Features of Connext DDS

l

determining who should receive the messages,

l

where recipients are located,

l

what happens if messages cannot be delivered.

This is made possible by how Connext DDS allows the user to specify Quality of Service (QoS) parameters as a way to configure automatic-discovery mechanisms and specify the behavior used when sending and receiving messages. The mechanisms are configured up-front and require no further effort on the
user's part. By exchanging messages in a completely anonymous manner, Connext DDS greatly simplifies
distributed application design and encourages modular, well-structured programs.
Furthermore, Connext DDS includes the following features, which are designed to meet the needs of distributed real-time applications:
l

l

l
l

l

Data-centric publish-subscribe communications: Simplifies distributed application programming
and provides time-critical data flow with minimal latency.
l Clear semantics for managing multiple sources of the same data.
l

Efficient data transfer, customizable Quality of Service, and error notification.

l

Guaranteed periodic samples, with maximum rate set by subscriptions.

l

Notification by a callback routine on data arrival to minimize latency.

l

Notification when data does not arrive by an expected deadline.

l

Ability to send the same message to multiple computers efficiently.

User-definable data types: Enables you to tailor the format of the information being sent to each
application.
Reliable messaging: Enables subscribing applications to specify reliable delivery of samples.
Multiple Communication Networks: Multiple independent communication networks (DDS
domains), each using Connext DDS, can be used over the same physical network. Applications are
only able to participate in the DDS domains to which they belong. Individual applications can be
configured to participate in multiple DDS domains.
Symmetric architecture: Makes your application robust:
l No central server or privileged nodes, so the system is robust to node failures.
l

l

Subscriptions and publications can be dynamically added and removed from the system at any
time.

Pluggable Transports Framework: Includes the ability to define new transport plug-ins and run
over them. Connext DDS comes with a standard UDP/IP pluggable transport and a shared memory
transport. It can be configured to operate over a variety of transport mechanisms, including backplanes, switched fabrics, and new networking technologies.

8

1.4 Features of Connext DDS

l
l

l

l

Multiple Built-in Transports: Includes UDP/IP and shared memory transports.
Multi-language support: Includes APIs for the C, C++ (Traditional and Modern APIs), C++/CLI,
C#, and Java™ programming languages.
Multi-platform support: Includes support for flavors of UNIX®, real-time operating systems, and
Windows®. (Consult the RTI Connext DDS Core Libraries Platform Notes to see which platforms
are supported in this release.)
Compliance with Standards:
l API complies with the DCPS layer of the OMG’s DDS specification.
l
l

Data types comply with OMG Interface Definition Language™ (IDL).
Data packet format complies with the International Engineering Consortium’s (IEC’s) publicly available specification for the RTPS wire protocol.

9

Chapter 2 Data-Centric Publish-Subscribe
Communications
This section describes the formal communications model used by Connext DDS: the Data-Centric
Publish-Subscribe (DCPS) standard. DCPS is a formalization (through a standardized API) and
extension of the publish-subscribe communications model presented in Network Communications
Models (Section 1.2 on page 3).
This section includes:

2.1 What is DCPS?
DCPS is the portion of the OMG DDS (Data Distribution Service) Standard that addresses datacentric publish-subscribe communications. The DDS standard defines a language-independent
model of publish-subscribe communications that has standardized mappings into various implementation languages. Connext DDS offers C, Traditional C++, Modern C++, C++/CLI, C#, and
Java versions of the DCPS API.
The publish-subscribe approach to distributed communications is a generic mechanism that can be
employed by many different types of applications. The DCPS model described in this chapter
extends the publish-subscribe model to address the specific needs of real-time, data-critical applications. As you’ll see, it provides several mechanisms that allow application developers to control
how communications works and how the middleware handles resource limitations and error conditions.
The “data-centric” portion of the term DCPS describes the fundamental concept supported by the
design of the API. In data-centric communications, the focus is on the distribution of data between
communicating applications. A data-centric system is comprised of data publishers and data subscribers. The communications are based on passing data of known types in named streams from
publishers to subscribers.

10

2.1.1 DCPS for Real-Time Requirements

In contrast, in object-centric communications the fundamental concept is the interface between the applications. An interface is comprised of a set of methods of known types (number and types of method arguments). An object-centric system is comprised of interface servers and interface clients, and
communications are based on clients invoking methods on named interfaces that are serviced by the corresponding server.
Data and object-centric communications are complementary paradigms in a distributed system. Applications may require both. However, real-time communications often fit a data-centric model more naturally.

2.1.1 DCPS for Real-Time Requirements
DCPS, and specifically the Connext DDS implementation, is well suited for real-time applications. For
instance, real-time applications often require the following features:
l

Efficiency

l

Real-time systems require efficient data collection and delivery. Only minimal delays should be introduced into the critical data-transfer path. Publish-subscribe is more efficient than client-server in both
latency and bandwidth for periodic data exchange.
Publish-subscribe greatly reduces the overhead required to send data over the network compared to
a client-server architecture. Occasional subscription requests, at low bandwidth, replace numerous
high-bandwidth client requests. Latency is also reduced, since the outgoing request message time is
eliminated. As soon as a new DDS sample becomes available, it is sent to the corresponding subscriptions.
Determinism

l

Real-time applications often care about the determinism of delivering periodic data as well as the
latency of delivering event data. Once buffers are introduced into a data stream to support reliable
connections, new data may be held undelivered for a unpredictable amount of time while waiting for
confirmation that old data was received.
Since publish-subscribe does not inherently require reliable connections, implementations, like Connext DDS, can provide configurable trade-offs between the deterministic delivery of new data and
the reliable delivery of all data.
Flexible delivery bandwidth
Typical real-time systems include both real-time and non-real-time nodes. The bandwidth requirements for these nodes—even for the same data—are quite different. For example, an application
may be sending DDS samples faster than a non-real-time application is capable of handling.
However, a real-time application may want the same data as fast as it is produced.
DCPS allows subscribers to the same data to set individual limits on how fast data should be
delivered to each subscriber. This is similar to how some people get a newspaper every day while
others can subscribe to only the Sunday paper.

11

2.2 DDS Data Types, Topics, Keys, Instances, and Samples

l

l

l

l

Thread awareness
Real-time communications must work without slowing the thread that sends DDS samples. On the
receiving side, some data streams should have higher priority so that new data for those streams are
processed before lower priority streams.
Connext DDS provides user-level configuration of its internal threads that process incoming data.
Users may configure Connext DDS so that different threads are created with different priorities to
process received data of different data streams.
Real-time communications must work without slowing the thread that sends DDS samples. On the
receiving side, some data streams should have higher priority so that new data for those streams are
processed before lower priority streams.
Connext DDS provides user-level configuration of its internal threads that process incoming data.
Users may configure Connext DDS so that different threads are created with different priorities to
process received data of different data streams.
Fault-tolerant operation
Real-time applications are often in control of systems that are required to run in the presence of component failures. Often, those systems are safety critical or carry financial penalties for loss of service.
The applications running those systems are usually designed to be fault-tolerant using redundant
hardware and software. Backup applications are often “hot” and interconnected to primary systems
so that they can take over as soon as a failure is detected.
Publish-subscribe is capable of supporting many-to-many connectivity with redundant DataWriters
and DataReaders. This feature is ideal for constructing fault-tolerant or high-availability applications
with redundant nodes and robust fault detection and handling services.
DCPS, and thus Connext DDS, was designed and implemented specifically to address the requirements above through configuration parameters known as QosPolicies defined by the DCPS standard
(see QosPolicies (Section 4.2 on page 162)). DDS Data Types, Topics, Keys, Instances, and
Samples (Section 2.2 below) introduces basic DCPS terminology and concepts.

2.2 DDS Data Types, Topics, Keys, Instances, and Samples
In data-centric communications, the applications participating in the communication need to share a common view of the types of data being passed around.
Within different programming languages there are several ‘primitive’ data types that all users of that language naturally share (integers, floating point numbers, characters, booleans, etc.). However, in any nontrivial software system, specialized data types are constructed out of the language primitives. So the data to
be shared between applications in the communication system could be structurally simple, using the primitive language types mentioned above, or it could be more complicated, using, for example, C and C++
structs, like this:
struct Time {
long year;
short day;

12

2.3 Data Topics — What is the Data Called?

short hour;
short minute;
short second;
};
struct StockPrice {
float price;
Time timeStamp;
};

Within a set of applications using DCPS, the different applications do not automatically know the structure
of the data being sent, nor do they necessarily interpret it in the same way (if, for instance, they use different operating systems, were written with different languages, or were compiled with different compilers). There must be a way to share not only the data, but also information about how the data is
structured.
In DCPS, data definitions are shared among applications using OMG IDL, a language-independent means
of describing data. For more information on data types and IDL, see Data Types and DDS Data Samples
(Section Chapter 3 on page 23).

2.3 Data Topics — What is the Data Called?
Shared knowledge of the data types is a requirement for different applications to communicate with DCPS.
The applications must also share a way to identify which data is to be shared. Data (of any data type) is
uniquely distinguished by using a name called a Topic. By definition, a Topic corresponds to a single data
type. However, several Topics may refer to the same data type.
Topics interconnect DataWriters and DataReaders. A DataWriter is an object in an application that tells
Connext DDS (and indirectly, other applications) that it has some values of a certain Topic. A corresponding DataReader is an object in an application that tells Connext DDS that it wants to receive values for the same Topic. And the data that is passed from the DataWriter to the DataReader is of the data
type associated with the Topic. DataWriters and DataReaders are described more in DataWriters/Publishers and DataReaders/Subscribers (Section 2.4 on page 15).
For a concrete example, consider a system that distributes stock quotes between applications. The applications could use a data type called StockPrice. There could be multiple Topics of the StockPrice data type,
one for each company’s stock, such as IBM, MSFT, GE, etc. Each Topic uses the same data type.
Data Type: StockPrice
struct StockPrice {
float price;
Time timeStamp;
};

Topic: “IBM”
Topic: “MSFT”
Topic: “GE”
13

2.3.1 DDS Samples, Instances, and Keys

Now, an application that keeps track of the current value of a client’s portfolio would subscribe to all of
the topics of the stocks owned by the client. As the value of each stock changes, the new price for the corresponding topic is published and sent to the application.

2.3.1 DDS Samples, Instances, and Keys
The value of data associated with a Topic can change over time. The different values of the Topic passed
between applications are called DDS samples. In our stock-price example, DDS samples show the price of
a stock at a certain point in time. So each DDS sample may show a different price.
For a data type, you can select one or more fields within the data type to form a key. A key is something
that can be used to uniquely identify one instance of a Topic from another instance of the same Topic.
Think of a key as a way to sub-categorize or group related data values for the same Topic. Note that not all
data types are defined to have keys, and thus, not all topics have keys. For topics without keys, there is
only a single instance of that topic.
However, for Topics with keys, a unique value for the key identifies a unique instance of the Topic. DDS
samples are then updates to particular instances of a Topic. Applications can subscribe to a Topic and
receive DDS samples for many different instances. Applications can publish DDS samples of one, all, or
any number of instances of a Topic. Many quality of service parameters actually apply on a per instance
basis. Keys are also useful for subscribing to a group of related data streams (instances) without pre-knowledge of which data streams (instances) exist at runtime.
For example, let’s change the StockPrice data type to include the symbol of the stock. Then instead of having a Topic for every stock, which would result in hundreds or thousands of Topics and related
DataWriters and DataReaders, each application would only have to publish or subscribe to a single Topic,
say “StockPrices.” Successive values of a stock would be presented as successive DDS samples of an
instance of “StockPrices”, with each instance corresponding to a single stock symbol.
Data Type: StockPrice
struct StockPrice {
float price;
Time timeStamp;
char *symbol;
//@key
};

Instance 1 = (Topic: “StockPrices”) + (Key: “MSFT”)
sample a, price = $28.00
sample b, price = $27.88
Instance 2 = (Topic: “StockPrices”) + (Key: “IBM”)
sample a, price = $74.02
sample b, price = $73.50

14

2.4 DataWriters/Publishers and DataReaders/Subscribers

Etc.
Just by subscribing to “StockPrices,” an application can get values for all of the stocks through a single
topic. In addition, the application does not have to subscribe explicitly to any particular stock, so that if a
new stock is added, the application will immediately start receiving values for that stock as well.
To summarize, the unique values of data being passed using DCPS are called DDS samples. A DDS
sample is a combination of a Topic (distinguished by a Topic name), an instance (distinguished by a key),
and the actual user data of a certain data type. As seen in Figure 2.1 Relationship of Topics, Keys, and
Instances below, a Topic identifies data of a single type, ranging from one single instance to a whole collection of instances of that given topic for keyed data types. For more information, see Data Types and
DDS Data Samples (Section Chapter 3 on page 23) and Topics (Section Chapter 5 on page 200).
Figure 2.1 Relationship of Topics, Keys, and Instances

By using keys, a Topic can identify a collection of data-object instances.

2.4 DataWriters/Publishers and DataReaders/Subscribers
In DCPS, applications must use APIs to create entities (objects) in order to establish publish-subscribe communications between each other. The entities and terminology associated with the data itself have been discussed already—Topics, keys, instances, DDS samples. This section will introduce the DCPS entities that
user code must create to send and receive the data. Note that Entity is actually a basic DCPS concept. In
object-oriented terms, Entity is the base class from which other DCPS classes—Topic, DataWriter,
DataReader, Publisher, Subscriber, DomainParticipants—derive. For general information on Entities, see
DDS Entities (Section Chapter 4 on page 151).

15

2.4 DataWriters/Publishers and DataReaders/Subscribers

The sending side uses objects called Publishers and DataWriters. The receiving side uses objects called
Subscribers and DataReaders. Figure 2.2 Overview below illustrates the relationship of these objects.
Figure 2.2 Overview

l

l

l

An application uses DataWriters to send data. A DataWriter is associated with a single Topic. You
can have multiple DataWriters and Topics in a single application. In addition, you can have more
than one DataWriter for a particular Topic in a single application.
A Publisher is the DCPS object responsible for the actual sending of data. Publishers own and manage DataWriters. A DataWriter can only be owned by a single Publisher while a Publisher can
own many DataWriters. Thus the same Publisher may be sending data for many different Topics of
different data types. When user code calls the write() method on a DataWriter, the DDS data
sample is passed to the Publisher object which does the actual dissemination of data on the network.
For more information, see Sending Data (Section Chapter 6 on page 242).
A Publisher is the DCPS object responsible for the actual sending of data. Publishers own and manage DataWriters. A DataWriter can only be owned by a single Publisher while a Publisher can
own many DataWriters. Thus the same Publisher may be sending data for many different Topics of
different data types. When user code calls the write() method on a DataWriter, the DDS data

16

2.4 DataWriters/Publishers and DataReaders/Subscribers

sample is passed to the Publisher object which does the actual dissemination of data on the network.
For more information, see Sending Data (Section Chapter 6 on page 242).
l

l

l

l

The association between a DataWriter and a Publisher is often referred to as a publication although
you never create a DCPS object known as a publication.
An application uses DataReaders to access data received over DCPS. A DataReader is associated
with a single Topic. You can have multiple DataReaders and Topics in a single application. In addition, you can have more than one DataReader for a particular Topic in a single application.
A Subscriber is the DCPS object responsible for the actual receipt of published data. Subscribers
own and manage DataReaders. A DataReader can only be owned by a single Subscriber while a
Subscriber can own many DataReaders. Thus the same Subscriber may receive data for many different Topics of different data types. When data is sent to an application, it is first processed by a
Subscriber; the DDS data sample is then stored in the appropriate DataReader. User code can either
register a listener to be called when new data arrives or actively poll the DataReader for new data
using its read() and take() methods. For more information, see Receiving Data (Section Chapter 7
on page 437).
The association between a DataReader and a Subscriber is often referred to as a subscription
although you never create a DCPS object known as a subscription.

Example:
The publish-subscribe communications model is analogous to that of magazine publications and subscriptions. Think of a publication as a weekly periodical such as Newsweek®. The Topic is the name of
the periodical (in this case the string "Newsweek"). The type specifies the format of the information, e.g., a
printed magazine. The user data is the contents (text and graphics) of each DDS sample (weekly issue).
The middleware is the distribution service (usually the US Postal service) that delivers the magazine from
where it is created (a printing house) to the individual subscribers (people’s homes). This analogy is illustrated in Figure 2.3 An Example of Publish-Subscribe on the facing page. Note that by subscribing to a
publication, subscribers are requesting current and future DDS samples of that publication (such as once a
week in the case of Newsweek), so that as new DDS samples are published, they are delivered without having to submit another request for data.

17

2.5 DDS Domains and DomainParticipants

Figure 2.3 An Example of Publish-Subscribe

The publish-subscribe model is analogous to publishing magazines. The Publisher sends DDS samples of a particular
Topic to all Subscribers of that Topic. With Newsweek® magazine, the Topic would be "Newsweek." The DDS sample
consists of the data (articles and pictures) sent to all Subscribers every week. The middleware (Connext DDS) is the distribution channel: all of the planes, trucks, and people who distribute the weekly issues to the Subscribers.

By default, each DDS sample is propagated individually, independently, and uncorrelated with other DDS
samples. However, an application may request that several DDS samples be sent as a coherent set, so that
they may be interpreted as such on the receiving side.

2.5 DDS Domains and DomainParticipants
You may have several independent DCPS applications all running on the same set of computers. You may
want to isolate one (or more) of those applications so that it isn’t affected by the others. To address this
issue, DCPS has a concept called DDS domains.
DDS domains represent logical, isolated, communication networks. Multiple applications running on the
same set of hosts on different DDS domains are completely isolated from each other (even if they are on
the same machine). DataWriters and DataReaders belonging to different DDS domains will never
exchange data.
Applications that want to exchange data using DCPS must belong to the same DDS domain. To belong to
a DDS domain, DCPS APIs are used to configure and create a DomainParticipant with a specific
Domain Index. DDS domains are differentiated by the domain index (an integer value). Applications that
have created DomainParticipants with the same domain index belong to the same DDS domain.
DomainParticipants own Topics, Publishers, and Subscribers, which in turn owns DataWriters and
DataReaders. Thus all DCPS Entities belong to a specific DDS domain.
An application may belong to multiple DDS domains simultaneously by creating multiple DomainParticipants with different domain indices. However, Publishers/DataWriters and Subscribers/DataReaders
only belong to the DDS domain in which they were created.

18

2.6 Quality of Service (QoS)

As mentioned before, multiple DDS domains may be used for application isolation, which is useful when
you are testing applications using computers on the same network or even the same computers. By assigning each user different domains, one can guarantee that the data produced by one user’s application won’t
accidentally be received by another. In addition, DDS domains may be a way to scale and construct larger
systems that are composed of multi-node subsystems. Each subsystem would use an internal DDS domain
for intra-system communications and an external DDS domain to connect to other subsystems.
For more information, see Working with DDS Domains (Section Chapter 8 on page 536).

2.6 Quality of Service (QoS)
The publish-subscribe approach to distributed communications is a generic mechanism that can be
employed by many different types of systems. The DCPS model described here extends the publish-subscribe model to address the needs of real-time, data-critical applications. It provides standardized mechanisms, known as Quality of Service Policies, that allow application developers to configure how
communications occur, to limit resources used by the middleware, to detect system incompatibilities and
setup error handling routines.

2.6.1 Controlling Behavior with Quality of Service (QoS) Policies
QosPolicies control many aspects of how and when data is distributed between applications. The overall
QoS of the DCPS system is made up of the individual QosPolicies for each DCPS Entity. There are
QosPolicies for Topics, DataWriters, Publishers, DataReaders, Subscribers, and DomainParticipants.
On the publishing side, the QoS of each Topic, the Topic’s DataWriter, and the DataWriter’s Publisher all
play a part in controlling how and when DDS samples are sent to the middleware. Similarly, the QoS of
the Topic, the Topic’s DataReader, and the DataReader’s Subscriber control behavior on the subscribing
side.
Users will employ QosPolicies to control a variety of behaviors. For example, the DEADLINE policy sets
up expectations of how often a DataReader expects to see DDS samples. The OWNERSHIP and
OWNERSHIP_STRENGTH policy are used together to configure and arbitrate whose data is passed to
the DataReader when there are multiple DataWriters for the same instance of a Topic. The HISTORY
policy specifies whether a DataWriter should save old data to send to new subscriptions that join the network later. Many other policies exist and they are presented in QosPolicies (Section 4.2 on page 162).
Some QosPolicies represent “contracts” between publications and subscriptions. For communications to
take place properly, the QosPolicies set on the DataWriter side must be compatible with corresponding
policies set on the DataReader side.
For example, the RELIABILITY policy is set by the DataWriter to state whether it is configured to send
data reliably to DataReaders. Because it takes additional resources to send data reliably, some DataWriters
may only support a best-effort level of reliability. This implies that for those DataWriters, Connext DDS
will not spend additional effort to make sure that the data sent is received by DataReaders or resend any
lost data. However, for certain applications, it could be imperative that their DataReaders receive every

19

2.7 Application Discovery

piece of data with total reliability. Running a system where the DataWriters have not been configured to
support the DataReaders could lead to erratic failures.
To address this issue, and yet keep the publications and subscriptions as decoupled as possible, DCPS
provides a way to detect and notify when QosPolicies set by DataWriters and DataReaders are incompatible. DCPS employs a pattern known as RxO (Requested versus Offered). The DataReader sets a
“requested” value for a particular QosPolicy. The DataWriter sets an “offered” value for that QosPolicy.
When Connext DDS matches a DataReader to a DataWriter, QosPolicies are checked to make sure that
all requested values can be supported by the offered values.
Note that not all QosPolicies are constrained by the RxO pattern. For example, it does not make sense to
compare policies that affect only the DataWriter but not the DataReader or vice versa.
If the DataWriter cannot satisfy the requested QosPolicies of a DataReader, Connext DDS will not connect the two DDS entities and will notify the applications on each side of the incompatibility if so configured.
For example, a DataReader sets its DEADLINE QoS to 4 seconds—that is, the DataReader is requesting
that it receive new data at least every 4 seconds.
In one application, the DataWriter sets its DEADLINE QoS to 2 seconds—that is, the DataWriter is committing to sending data at least every 2 seconds. This writer can satisfy the request of the reader, and thus,
Connext DDS will pass the data sent from the writer to the reader.
In another application, the DataWriter sets its DEADLINE QoS to 5 seconds. It only commits to sending
data at 5 second intervals. This will not satisfy the request of the DataReader. Connext DDS will flag this
incompatibility by calling user-installed listeners in both DataWriter and DataReader applications and not
pass data from the writer to the reader.
For a summary of the QosPolicies supported by Connext DDS, see QosPolicies (Section 4.2 on page
162).

2.7 Application Discovery
The DCPS model provides anonymous, transparent, many-to-many communications. Each time an application sends a DDS sample of a particular Topic, the middleware distributes the DDS sample to all the
applications that want that Topic. The publishing application does not need to specify how many applications receive the Topic, nor where those applications are located. Similarly, subscribing applications do
not specify the location of the publications. In addition, new publications and subscriptions of the Topic
can appear at any time, and the middleware will automatically interconnect them.
So how is this all done? Ultimately, in each application for each publication, Connext DDS must keep a
list of applications that have subscribed to the same Topic, nodes on which they are located, and some additional QoS parameters that control how the data is sent. Also, Connext DDS must keep a list of applications and publications for each of the Topics to which the application has subscribed.

20

2.7 Application Discovery

This propagation of this information (the existence of publications and subscriptions and associated QoS)
between applications by Connext DDS is known as the discovery process. While the DDS (DCPS) standard does not specify how discovery occurs, Connext DDS uses a standard protocol RTPS for both discovery and formatting on-the-wire packets.
When a DomainParticipant is created, Connext DDS sends out packets on the network to announce its
existence. When an application finds out that another application belongs to the same DDS domain, then it
will exchange information about its existing publications and subscriptions and associated QoS with the
other application. As new DataWriters and DataReaders are created, this information is sent to known
applications.
The Discovery process is entirely configurable by the user and is discussed extensively in Discovery (Section Chapter 14 on page 709).

21

Part 2: Core Concepts
This section includes:
l

Data Types and DDS Data Samples (Section Chapter 3 on page 23)

l

DDS Entities (Section Chapter 4 on page 151)

l

Topics (Section Chapter 5 on page 200)

l

Sending Data (Section Chapter 6 on page 242)

l

Receiving Data (Section Chapter 7 on page 437)

l

Working with DDS Domains (Section Chapter 8 on page 536)

l

Building Applications (Section Chapter 9 on page 622)

22

Chapter 3 Data Types and DDS Data
Samples
How data is stored or laid out in memory can vary from language to language, compiler to compiler, operating system to operating system, and processor to processor. This combination of language/compiler/operating system/processor is called a platform. Any modern middleware must be
able to take data from one specific platform (say C/gcc.3.2.2/Solaris/Sparc) and transparently
deliver it to another (for example, Java/JDK 1.6/Windows/Pentium). This process is commonly
called serialization/deserialization, or marshalling/demarshalling.
Messaging products have typically taken one of two approaches to this problem:
1. Do nothing. Messages consist only of opaque streams of bytes. The JMS BytesMessage is
an example of this approach.
2. Send everything, every time. Self-describing messages are at the opposite extreme, embedding full reflective information, including data types and field names, with each message.
The JMS MapMessage and the messages in TIBCO Rendezvous are examples of this
approach.
The “do nothing” approach is lightweight on its surface but forces you, the user of the middleware
API, to consider all data encoding, alignment, and padding issues. The “send everything” alternative results in large amounts of redundant information being sent with every packet, impacting performance.
Connext DDS takes an intermediate approach. Just as objects in your application program belong
to some data type, DDS data samples sent on the same Connext DDS topic share a data type. This
type defines the fields that exist in the DDS data samples and what their constituent types are. The
middleware stores and propagates this meta-information separately from the individual DDS data
samples, allowing it to propagate DDS samples efficiently while handling byte ordering and alignment issues for you.
To publish and/or subscribe to data with Connext DDS, you will carry out the following steps:

23

Chapter 3 Data Types and DDS Data Samples

1. Select a type to describe your data.
You have a number of choices. You can choose one of these options, or you can mix and match
them.
l Use a built-in type provided by the middleware.

l

This option may be sufficient if your data typing needs are very simple. If your data is highly
structured, or you need to be able to examine fields within that data for filtering or other purposes, this option may not be appropriate. The built-in types are described in Built-in Data
Types (Section 3.2 on page 30).
Use the RTI Code Generator to define a type at compile-time using a language-independent
description language.
Code generation offers two strong benefits not available with dynamic type definition: (1) it
allows you to share type definitions across programming languages, and (2) because the structure of the type is known at compile time, it provides rigorous static type safety.
The RTI Code Generator accepts input the following formats:
l

l

l

OMG IDL. This format is a standard component of both the DDS and CORBA specifications. It describes data types with a C++-like syntax. This format is described in
Creating User Data Types with IDL (Section 3.3 on page 69).
XML in a DDS-specific format. This XML format is terser, and therefore easier to
read and write by hand, than an XSD file. It offers the general benefits of XML-extensibility and ease of integration, while fully supporting DDS-specific data types and concepts. This format is described in Creating User Data Types with Extensible Markup
Language (XML) (Section 3.4 on page 121).

Define a type programmatically at run time.

This method may be appropriate for applications with dynamic data description needs: applications for which types change frequently or cannot be known ahead of time. It is described in
Defining New Types (Section 3.8.2 on page 141).
2. Register your type with a logical name.
If you've chosen to use a built-in type instead of defining your own, you can omit this step; the middleware pre-registers the built-in types for you.
This step is described in the Defining New Types (Section 3.8.2 on page 141).
3. Create a Topic using the type name you previously registered.
If you've chosen to use a built-in type instead of defining your own, you will use the API constant
corresponding to that type's name.
Creating and working with Topics is discussed in Topics (Section Chapter 5 on page 200).

24

3.1 Introduction to the Type System

4. Create one or more DataWriters to publish your data and one or more DataReaders to subscribe to
it.
The concrete types of these objects depend on the concrete data type you've selected, in order to
provide you with a measure of type safety.
Creating and working with DataWriters and DataReaders are described in Sending Data (Section
Chapter 6 on page 242) and Receiving Data (Section Chapter 7 on page 437), respectively.
Whether publishing or subscribing to data, you will need to know how to create and delete DDS data
samples and how to get and set their fields. These tasks are described in Working with DDS Data Samples
(Section 3.9 on page 145).
This section describes:

3.1 Introduction to the Type System
A user data type is any custom type that your application defines for use with Connext DDS. It may be a
structure, a union, a value type, an enumeration, or a typedef (or language equivalents).
Your application can have any number of user data types. They can be composed of any of the primitive
data types listed below or of other user data types.
Only structures, unions, and value types may be read and written directly by Connext DDS; enums,
typedefs, and primitive types must be contained within a structure, union, or value type. In order for a
DataReader and DataWriter to communicate with each other, the data types associated with their respective Topic definitions must be identical.
l

octet, char, wchar

l

short, unsigned short

l

long, unsigned long

l

long long, unsigned long long

l

float

l

double, long double

l

boolean

l

enum (with or without explicit values)

l

bounded and unbounded string and wstring

25

3.1.1 Sequences

The following type-building constructs are also supported:
l

module (also called a package or namespace)

l

pointer

l

array of primitive or user type elements

l

bounded/unbounded sequence of elements1—a sequence is a variable-length ordered collection,
such as a vector or list

l

typedef

l

bitfield2

l

union

l

struct

l

value type, a complex type that supports inheritance and other object-oriented features

To use a data type with Connext DDS, you must define that type in a way the middleware understands
and then register the type with the middleware. These steps allow Connext DDS to serialize, deserialize,
and otherwise operate on specific types. They will be described in detail in the following sections.

3.1.1 Sequences
A sequence contains an ordered collection of elements that are all of the same type. The operations supported in the sequence are documented in the API Reference HTML documentation, which is available for
all supported programming languages (select Modules, RTI Connext DDS API Reference, Infrastructure Module, Sequence Support).
Java sequences implement the java.util.List interface from the standard Collections framework.
In the Modern C++ API a sequences of type T maps to the type dds::core::vector. This type is similar
to std::vector.
Elements in a sequence are accessed with their index, just like elements in an array. Indices start at zero in
all APIs except Ada. In Ada, indices start at 1. Unlike arrays, however, sequences can grow in size. A
sequence has two sizes associated with it: a physical size (the "maximum") and a logical size (the
"length"). The physical size indicates how many elements are currently allocated by the sequence to hold;
the logical size indicates how many valid elements the sequence actually holds. The length can vary from
zero up to the maximum. Elements cannot be accessed at indices beyond the current length.

1Sequences of sequences are not supported directly. To work around this constraint, typedef the inner sequence and form a

sequence of that new type.
2Data types containing bitfield members are not supported by DynamicData. [RTI Bug # 12638]

26

3.1.1 Sequences

A sequence may be declared as bounded or unbounded. A sequence's "bound" is the maximum number of
elements that the sequence can contain at any one time. A finite bound is very important because it allows
Connext DDS to preallocate buffers to hold serialized and deserialized samples of your types; these buffers
are used when communicating with other nodes in your distributed system. If a sequence has no bound,
Connext DDS will not know how large to allocate its buffers and will therefore have to allocate them on
the fly as individual samples are read and written—impacting the latency and determinism of your application.
By default, any unbounded sequences found in an IDL file will be given a default bound of 100 elements.
This default value can be overwritten using the RTI Code Generator‘s -sequenceSize command-line argument (see the RTI Code Generator User’s Manual).
When using C, C++, or .NET, you can change the default behavior and used truly unbounded sequences
by using RTI Code Generator‘s -unboundedSupport command-line argument. When using this option,
the generated code will deserialize incoming samples by dynamically allocating and deallocating memory
to accommodate the actual size of the sequences.
Unbounded built-in types are only supported in the C, C++, Java and .NET APIs
To configure unbounded support for code generated with rtiddsgen -unboundedSupport:
1. Use these threshold QoS properties:
l dds.data_writer.history.memory_manager.fast_pool.pool_buffer_max_size on the
DataWriter
l

dds.data_reader.history.memory_manager.fast_pool.pool_buffer_max_size on the
DataReader (only if keyed)

2. Set the QoS value reader_resource_limits.dynamically_allocate_fragmented_samples on the
DataReader to true.
3. For the Java API, also set these properties accordingly for the Java serialization buffer:
l dds.data_writer.history.memory_manager.java_stream.min_size
l

dds.data_writer.history.memory_manager.java_stream.trim_to_size

l

dds.data_reader.history.memory_manager.java_stream.min_size

l

dds.data_reader.history.memory_manager.java_stream.trim_to_size

See also:
l

Unbounded Built-in Types (Section 3.2.7.2 on page 67)

l

Writer-Side Memory Management when Using Java (Section 20.1.3 on page 851)

l

Reader-Side Memory Management when Using Java (Section 20.2.2 on page 856)

27

3.1.2 Strings and Wide Strings

3.1.2 Strings and Wide Strings
Connext DDS supports both strings consisting of single-byte characters (the IDL string type) and strings
consisting of wide characters (IDL wstring). The wide characters supported by Connext DDS are four
bytes long, large enough to store not only two-byte Unicode/UTF16 characters but also UTF32 characters.
Like sequences, strings may be bounded or unbounded. A string's "bound" is its maximum length (not
counting the trailing NULL character in C and C++).
In the Modern C++ API strings map to the type dds::core::string, similar to std::string.
By default, any unbounded string found in an IDL file will be given a default bound of 255 elements. This
default value can be overwritten using the RTI Code Generator‘s -stringSize command-line argument (see
the RTI Code Generator User’s Manual).
In C, C++, and .NET, you can change the default behavior and used truly unbounded string by using RTI
Code Generator‘s -unboundedSupport command-line argument. When using this option, the generated
code will deserialize incoming samples by dynamically allocating and deallocating memory to accommodate the actual size of the strings.
Unbounded built-in types are only supported in the C, C++, Java and .NET APIs
To configure unbounded support for built-in types:
1. Set the properties dds.builtin_type.*.max_size and dds.builtin_type.*.alloc_size to
2,147,483,647.
2. Use these threshold QoS properties:
l dds.data_writer.history.memory_manager.fast_pool.pool_buffer_max_size on the
DataWriter
l

dds.data_reader.history.memory_manager.fast_pool.pool_buffer_max_size on the
DataReader (only if keyed)

3. Set the QoS value reader_resource_limits.dynamically_allocate_fragmented_samples on the
DataReader to true.
4. For the Java API, also set these properties accordingly for the Java serialization buffer:
l dds.data_writer.history.memory_manager.java_stream.min_size
l

dds.data_writer.history.memory_manager.java_stream.trim_to_size

l

dds.data_reader.history.memory_manager.java_stream.min_size

l

dds.data_reader.history.memory_manager.java_stream.trim_to_size

28

3.1.3 Introduction to TypeCode

See also:
l

Unbounded Built-in Types (Section 3.2.7.2 on page 67)

l

Writer-Side Memory Management when Using Java (Section 20.1.3 on page 851)

l

Reader-Side Memory Management when Using Java (Section 20.2.2 on page 856)

3.1.3 Introduction to TypeCode
Type schemas—the names and definitions of a type and its fields—are represented by TypeCode objects
(known as DynamicType in the Modern C++ API). A type code value consists of a type code kind (see
the TCKind enumeration below) and a list of members. For compound types like structs and arrays, this
list will recursively include one or more type code values.
enum TCKind {
TK_NULL,
TK_SHORT,
TK_LONG,
TK_USHORT,
TK_ULONG,
TK_FLOAT,
TK_DOUBLE,
TK_BOOLEAN,
TK_CHAR,
TK_OCTET,
TK_STRUCT,
TK_UNION
TK_ENUM,
TK_STRING,
TK_SEQUENCE,
TK_ARRAY,
TK_ALIAS,
TK_LONGLONG,
TK_ULONGLONG,
TK_LONGDOUBLE,
TK_WCHAR,
TK_WSTRING,
TK_VALUE
}

Type codes unambiguously match type representations and provide a more reliable test than comparing the
string type names.
The TypeCode class, modeled after the corresponding CORBA API, provides access to type-code information. For details on the available operations for the TypeCode class, see the API Reference HTML documentation, which is available for all supported programming languages (select Modules, RTI Connext
DDS API Reference, Topic Module, Type Code Support or, for the Modern C++ API select Modules,
RTI Connext DDS API Reference, Infrastructure Module, DynamicType and DynamicData).

29

3.1.3.1 Sending TypeCodes on the Network

Note: Type-code support must be enabled if you are going to use ContentFilteredTopics (Section 5.4 on
page 212) with the default SQL filter. You may disable type codes and use a custom filter, as described in
Creating ContentFilteredTopics (Section 5.4.3 on page 214).
3.1.3.1 Sending TypeCodes on the Network
In addition to being used locally, serialized type codes are typically published automatically during discovery as part of the built-in topics for publications and subscriptions. See Built-in DataReaders (Section
16.2 on page 773). This allows applications to publish or subscribe to topics of arbitrary types. This functionality is useful for generic system monitoring tools like the rtiddsspy debug tool (see the API Reference
HTML documentation).
Note: In the C, Traditional C++, Java and .NET APIs Type codes are not cached by Connext DDS upon
receipt and are therefore not available from the built-in data returned by the DataWriter's get_matched_
subscription_data() operation or the DataReader's get_matched_publication_data() operation; in the
Modern C++ API they are available.
If your data type has an especially complex type code, you may need to increase the value of the type_
code_max_serialized_length field in the DomainParticipant's DOMAIN_PARTICIPANT_
RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593). Or, to prevent the
propagation of type codes altogether, you can set this value to zero (0). Be aware that some features of
monitoring tools, as well as some features of the middleware itself (such as ContentFilteredTopics) will not
work correctly if you disable TypeCode propagation.

3.2 Built-in Data Types
Connext DDS provides a set of standard types that are built into the middleware. These types can be used
immediately; they do not require you to write IDL, use RTI Code Generator (rtiddsgen) (see Using RTI
Code Generator (rtiddsgen) (Section 3.6 on page 138)), or use the dynamic type API (see Managing
Memory for Built-in Types (Section 3.2.7 on page 62)).
The supported built-in types are String, KeyedString, Octets, and KeyedOctets. (The latter two types are
called Bytes and KeyedBytes, respectively, on Java and .NET platforms.)
The built-in type API is located under the DDS namespace in Traditional C++ and .NET. For Java, the
API is contained inside the package com.rti.dds.type.builtin. In the Modern C++ API they are located in
the dds::core namespace.
Built-in data types are discussed in the following sections:

3.2.1 Registering Built-in Types
By default, the built-in types are automatically registered when a DomainParticipant is created. You can
change this behavior by setting the DomainParticipant’s dds.builtin_type.auto_register property to 0

30

3.2.2 Creating Topics forBuilt-in Types

(false) using the PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394).

3.2.2 Creating Topics forBuilt-in Types
To create a topic for a built-in type, just use the standard DomainParticipant operations, create_topic() or
create_topic_with_profile() (see Creating Topics (Section 5.1.1 on page 202)); for the type_name parameter, use the value returned by the get_type_name() operation, listed below for each API.
Note: In the following examples, you will see the sentinel "."
For C and Traditional C++:  = String, KeyedString, Octets or KeyedOctets
For Java and .NET1:  = String, KeyedString, Bytes or KeyedBytes
C API:
const char* DDS_TypeSupport_get_type_name();

Traditional C++ API with namespace:
const char* DDS::TypeSupport::get_type_name();

Traditional C++ API without namespace:
const char* DDSTypeSupport::get_type_name();

C++/CLI API:
System::String^ DDS:TypeSupport::get_type_name();

C# API:
System.String DDS.TypeSupport.get_type_name();

Java API:
String
com.rti.dds.type.builtin.TypeSupport.get_type_name();

(This step is not required in the Modern C++ API)
3.2.2.1 Topic Creation Examples
For simplicity, error handling is not shown in the following examples.
C Example:
DDS_Topic * topic = NULL;
/* Create a builtin type Topic */

1RTI Connext DDS .NET language binding is currently supported for C# and C++/CLI.

31

3.2.2.1 Topic Creation Examples

topic = DDS_DomainParticipant_create_topic(
participant, "StringTopic",
DDS_StringTypeSupport_get_type_name(),
&DDS_TOPIC_QOS_DEFAULT, NULL,
DDS_STATUS_MASK_NONE);

Traditional C++ Example with namespaces:1
using namespace DDS;
...
/* Create a String builtin type Topic */
Topic * topic = participant->create_topic(
"StringTopic", StringTypeSupport::get_type_name(),
DDS_TOPIC_QOS_DEFAULT, NULL, DDS_STATUS_MASK_NONE);

Modern C++ Example:
dds::topic::Topic topic(participant, "StringTopic");

C++/CLI Example:
using namespace DDS;
...
/* Create a builtin type Topic */
Topic^ topic = participant->create_topic(
"StringTopic", StringTypeSupport::get_type_name(),
DomainParticipant::TOPIC_QOS_DEFAULT,
nullptr, StatusMask::STATUS_MASK_NONE);

C# Example:
using namespace DDS;
... /*
Create a builtin type Topic */
Topic topic = participant.create_topic(
"StringTopic", StringTypeSupport.get_type_name(),
DomainParticipant.TOPIC_QOS_DEFAULT,
null, StatusMask.STATUS_MASK_NONE);

Java Example:
import com.rti.dds.type.builtin.*;
...
/* Create a builtin type Topic */
Topic topic = participant.create_topic(
"StringTopic", StringTypeSupport.get_type_name(),
DomainParticipant.TOPIC_QOS_DEFAULT,
null, StatusKind.STATUS_MASK_NONE);

1This example uses C++ namespaces. If you're not using namespaces in your own code, prefix the name of each DDS class

with 'DDS.' For example, DDS::StringDataWriter becomes DDSStringDataWriter.

32

3.2.3 String Built-in Type

3.2.3 String Built-in Type
The String built-in type is represented by a NULL-terminated character array (char *) in C and C++ and
an immutable String object in Java and .NET1. This type can be used to publish and subscribe to a single
string.
3.2.3.1 Creating and Deleting Strings
In C and C++, Connext DDS provides a set of operations to create (DDS::String_alloc()), destroy
(DDS::String_free()), and clone strings (DDS::String_dup()). Select Modules, RTI Connext DDS
API Reference, Infrastructure Module, String support in the API Reference HTML documentation,
which is available for all supported programming languages.
Memory Considerations in Copy Operations:
When the read/take operations that take a sequence of strings as a parameter are used in copy mode,
Connext DDS allocates the memory for the string elements in the sequence if they are initialized to
NULL.
If the elements are not initialized to NULL, the behavior depends on the language:
l

l

In Java and .NET, the memory associated with the elements is reallocated with every DDS sample,
because strings are immutable objects.
In C and C++, the memory associated with the elements must be large enough to hold the received
data. Insufficient memory may result in crashes.

When take_next_sample() and read_next_sample() are called in C and C++, you must make sure
that the input string has enough memory to hold the received data. Insufficient memory may result in
crashes.
3.2.3.2 String DataWriter
The string DataWriter API matches the standard DataWriter API (see Using a Type-Specific DataWriter
(FooDataWriter) (Section 6.3.7 on page 281)). There are no extensions.
The following examples show how to write simple strings with a string built-in type DataWriter. For simplicity, error handling is not shown.
C Example:
DDS_StringDataWriter * stringWriter = ... ;

1RTI Connext DDS .NET language binding is currently supported for C# and C++/CLI.

33

3.2.3.2 String DataWriter

DDS_ReturnCode_t retCode; char * str = NULL;
/* Write some data */
retCode = DDS_StringDataWriter_write(
stringWriter, "Hello World!", &DDS_HANDLE_NIL);
str = DDS_String_dup("Hello World!");
retCode = DDS_StringDataWriter_write(
stringWriter, str, &DDS_HANDLE_NIL);
DDS_String_free(str);

Traditional C++ Example with namespaces:1
#include "ndds/ndds_namespace_cpp.h"
using namespace DDS;
...
StringDataWriter * stringWriter = ... ;
/* Write some data */
ReturnCode_t retCode = stringWriter->write(
"Hello World!", HANDLE_NIL);
char * str = DDS::String_dup("Hello World!");
retCode = stringWriter->write(str, HANDLE_NIL);
DDS::String_free(str);

Modern C++ Example:
dds::pub::DataWriter string_writer(
participant, string_topic);
string_writer.write("Hello World!");
dds::core::string str = "Hello World!";
string_writer.write(str);

C++/CLI Example:
using namespace System;
using namespace DDS;
...
StringDataWriter^ stringWriter = ... ;
/* Write some data */
stringWriter->write(
"Hello World!", InstanceHandle_t::HANDLE_NIL);
String^ str = "Hello World!";
stringWriter->write(
str, InstanceHandle_t::HANDLE_NIL);

1This example uses C++ namespaces. If you're not using namespaces in your own code, prefix the name of each DDS class

with 'DDS.' For example, DDS::StringDataWriter becomes DDSStringDataWriter.

34

3.2.3.3 String DataReader

C# Example:
using System;
using DDS;
...
StringDataWriter stringWriter = ... ;
/* Write some data */
stringWriter.write(
"Hello World!", InstanceHandle_t.HANDLE_NIL);
String str = "Hello World!";
stringWriter.write(
str, InstanceHandle_t.HANDLE_NIL);

Java Example:
import com.rti.dds.publication.*;
import com.rti.dds.type.builtin.*;
import com.rti.dds.infrastructure.*;
...
StringDataWriter stringWriter = ... ;
/* Write some data */
stringWriter.write(
"Hello World!", InstanceHandle_t.HANDLE_NIL);
String str = "Hello World!";
stringWriter.write(
str, InstanceHandle_t.HANDLE_NIL);

3.2.3.3 String DataReader
The string DataReader API matches the standard DataReader API (see Using a Type-Specific
DataReader (FooDataReader) (Section 7.4.1 on page 491)). There are no extensions.
The following examples show how to read simple strings with a string built-in type DataReader. For simplicity, error handling is not shown.
C Example:
struct DDS_StringSeq dataSeq =
DDS_SEQUENCE_INITIALIZER;
struct DDS_SampleInfoSeq infoSeq =
DDS_SEQUENCE_INITIALIZER;
DDS_StringDataReader * stringReader = ... ;
DDS_ReturnCode_t retCode;
int i;
/* Take and print the data */
retCode = DDS_StringDataReader_take(
stringReader, &dataSeq,
&infoSeq, DDS_LENGTH_UNLIMITED,
DDS_ANY_SAMPLE_STATE,

35

3.2.3.3 String DataReader

DDS_ANY_VIEW_STATE,
DDS_ANY_INSTANCE_STATE);
for (i = 0; i < DDS_StringSeq_get_length(&data_seq);
++i) {
if (DDS_SampleInfoSeq_get_reference(
&info_seq, i)->valid_data) {
DDS_StringTypeSupport_print_data(
DDS_StringSeq_get(&data_seq, i));
}
}
/* Return loan */
retCode = DDS_StringDataReader_return_loan(
stringReader, &data_seq, &info_seq);

Traditional C++ Example with namespaces:1
#include "ndds/ndds_namespace_cpp.h"
using namespace DDS;
...
StringSeq dataSeq;
SampleInfoSeq infoSeq;
StringDataReader * stringReader = ... ;
/* Take a print the data */
ReturnCode_t retCode = stringReader->take(
dataSeq, infoSeq,
LENGTH_UNLIMITED,
ANY_SAMPLE_STATE,
ANY_VIEW_STATE,
ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (infoSeq[i].valid_data) {
StringTypeSupport::print_data(dataSeq[i]);
}
}
/* Return loan */
retCode = stringReader->return_loan(
dataSeq, infoSeq);

Modern C++ Example:
using namespace dds::core;
using namespace dds::sub;
DataReader string_reader(
participant, string_topic);

1This example uses C++ namespaces. If you're not using namespaces in your own code, prefix the name of each DDS class

with 'DDS.' For example, DDS::StringDataWriter becomes DDSStringDataWriter.

36

3.2.3.3 String DataReader

LoanedSamples samples =
string_reader.take();
for (auto sample : samples) {
if (sample.info().valid()) {
std::cout << sample.data() << std::endl;
}
}

C++/CLI Example:
using namespace System;
using namespace DDS;
...
StringSeq^ dataSeq = gcnew StringSeq();
SampleInfoSeq^ infoSeq = gcnew SampleInfoSeq();
StringDataReader^ stringReader = ... ;
/* Take and print the data */
stringReader->take(
dataSeq, infoSeq,
ResourceLimitsQosPolicy::LENGTH_UNLIMITED,
SampleStateKind::ANY_SAMPLE_STATE,
ViewStateKind::ANY_VIEW_STATE,
InstanceStateKind::ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (infoSeq->get_at(i)->valid_data) {
StringTypeSupport::print_data(
dataSeq->get_at(i));
}
}
/* Return loan */
stringReader->return_loan(dataSeq, infoSeq);

C# Example:
using System;
using DDS;
...
StringSeq dataSeq = new StringSeq();
SampleInfoSeq infoSeq = new SampleInfoSeq();
StringDataReader stringReader = ... ;
/* Take and print the data */
stringReader.take(
dataSeq, infoSeq,
ResourceLimitsQosPolicy.LENGTH_UNLIMITED,
SampleStateKind.ANY_SAMPLE_STATE,
ViewStateKind.ANY_VIEW_STATE,
InstanceStateKind.ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (infoSeq.get_at(i)).valid_data) {

37

3.2.4 KeyedString Built-in Type

StringTypeSupport.print_data(
dataSeq.get_at(i));
}
}

Java Example:
import com.rti.dds.infrastructure.*;
import com.rti.dds.subscription.*;
import com.rti.dds.type.builtin.*;
...
StringSeq dataSeq = new StringSeq();
SampleInfoSeq infoSeq = new SampleInfoSeq();
StringDataReader stringReader = ... ;
/* Take and print the data */
stringReader.take(
dataSeq, infoSeq,
ResourceLimitsQosPolicy.LENGTH_UNLIMITED,
SampleStateKind.ANY_SAMPLE_STATE,
ViewStateKind.ANY_VIEW_STATE,
InstanceStateKind.ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (((SampleInfo)infoSeq.get(i)).valid_data) {
System.out.println(
(String)dataSeq.get(i));
}
}
/* Return loan */
stringReader.return_loan(dataSeq, infoSeq);

3.2.4 KeyedString Built-in Type
The Keyed String built-in type is represented by a (key, value) pair, where key and value are strings. This
type can be used to publish and subscribe to keyed strings. The language specific representations of the
type are as follows:
C/Traditional C++ Representation (without namespaces):
struct DDS_KeyedString {
char * key;
char * value;
};

Modern C++ Representation:
class dds::core::KeyedStringTopicType {
public:

38

3.2.4.1 Creating and Deleting Keyed Strings

dds::core::string& key();
dds::core::string& value();
// ... see API documentation for full definition
};

C++/CLI Representation:
namespace DDS {
public ref struct KeyedString: {
public:
System::String^ key;
System::String^ value;
...
};
};

C# Representation:
namespace DDS {
public class KeyedString {
public System.String key;
public System.String value;
};
};

Java Representation:
namespace DDS {
public class KeyedString {
public System.String key;
public System.String value;
};
};

3.2.4.1 Creating and Deleting Keyed Strings
Connext DDS provides a set of constructors/destructors to create/destroy Keyed Strings. For details, see
the API Reference HTML documentation, which is available for all supported programming languages
(select Modules, RTI Connext DDS API Reference, Topic Module, Built-in Types).
If you want to manipulate the memory of the fields 'value' and 'key' in the KeyedString struct in C/C++,
use the operations DDS::String_alloc(), DDS::String_dup(), and DDS::String_free(), as described in
the API Reference HTML documentation (select Modules, RTI Connext DDS API Reference, Infrastructure Module, String Support).

39

3.2.4.2 Keyed String DataWriter

3.2.4.2 Keyed String DataWriter
The keyed string DataWriter API is extended with the following methods (in addition to the standard methods described in Using a Type-Specific DataWriter (FooDataWriter) (Section 6.3.7 on page 281)):
DDS::ReturnCode_t
DDS::KeyedStringDataWriter::dispose(
const char* key,
const DDS::InstanceHandle_t* instance_handle);
DDS::ReturnCode_t
DDS::KeyedStringDataWriter::dispose_w_timestamp(
const char* key,
const DDS::InstanceHandle_t* instance_handle,
const struct DDS::Time_t* source_timestamp);
DDS::ReturnCode_t
DDS::KeyedStringDataWriter::get_key_value(
char * key,
const DDS::InstanceHandle_t* handle);
DDS::InstanceHandle_t
DDS::KeyedStringDataWriter::lookup_instance(
const char * key);
DDS::InstanceHandle_t
DDS::KeyedStringDataWriter::register_instance(
const char* key);
DDS::InstanceHandle_t
DDS_KeyedStringDataWriter::register_instance_w_timestamp(
const char * key,
const struct DDS_Time_t* source_timestamp);
DDS::ReturnCode_t
DDS::KeyedStringDataWriter::unregister_instance(
const char * key,
const DDS::InstanceHandle_t* handle);
DDS::ReturnCode_t
DDS::KeyedStringDataWriter::unregister_instance_w_timestamp(
const char* key,
const DDS::InstanceHandle_t* handle,
const struct DDS::Time_t* source_timestamp);
DDS::ReturnCode_t
DDS::KeyedStringDataWriter::write (
const char * key,
const char * str,
const DDS::InstanceHandle_t* handle);
DDS::ReturnCode_t
DDS::KeyedStringDataWriter::write_w_timestamp(
const char * key,
const char * str,
const DDS::InstanceHandle_t* handle,
const struct DDS::Time_t* source_timestamp);

40

3.2.4.2 Keyed String DataWriter

These operations are introduced to provide maximum flexibility in the format of the input parameters for
the write and instance management operations. For additional information and a complete description of
the operations, see the API Reference HTML documentation, which is available for all supported programming languages.
The following examples show how to write keyed strings using a keyed string built-in type DataWriter
and some of the extended APIs. For simplicity, error handling is not shown.
C Example:
DDS_KeyedStringDataWriter * stringWriter = ... ;
DDS_ReturnCode_t retCode;
struct DDS_KeyedString * keyedStr = NULL;
char * str = NULL;
/* Write some data using the KeyedString structure */
keyedStr = DDS_KeyedString_new(255, 255);
strcpy(keyedStr->key, "Key 1");
strcpy(keyedStr->value, "Value 1");
retCode = DDS_KeyedStringDataWriter_write_string_w_key(
stringWriter, keyedStr,
&DDS_HANDLE_NIL);
DDS_KeyedString_delete(keyedStr);
/* Write some data using individual strings */
retCode = DDS_KeyedStringDataWriter_write_string_w_key(
stringWriter, "Key 1",
"Value 1", &DDS_HANDLE_NIL);
str = DDS_String_dup("Value 2");
retCode = DDS_KeyedStringDataWriter_write_string_w_key(
stringWriter, "Key 1",
str, &DDS_HANDLE_NIL);
DDS_String_free(str);

C++ Example with Namespaces:1
#include "ndds/ndds_namespace_cpp.h"
using namespace DDS;
...
KeyedStringDataWriter * stringWriter = ... ;
/* Write some data using the KeyedString */
KeyedString * keyedStr = new KeyedString(255, 255);
strcpy(keyedStr->key, "Key 1");
strcpy(keyedStr->value, "Value 1");
ReturnCode_t retCode = stringWriter->write(
keyedStr, HANDLE_NIL);
delete keyedStr;

1This example uses C++ namespaces. If you're not using namespaces in your own code, prefix the name of each DDS class

with 'DDS.' For example, DDS::StringDataWriter becomes DDSStringDataWriter.

41

3.2.4.2 Keyed String DataWriter

C++/CLI Example:
using namespace System;
using namespace DDS;
...
KeyedStringDataWriter^ stringWriter = ... ;
/* Write some data using the KeyedString */
KeyedString^ keyedStr = gcnew KeyedString();
keyedStr->key = "Key 1";
keyedStr->value = "Value 1";
stringWriter->write(
keyedStr, InstanceHandle_t::HANDLE_NIL);
/* Write some data using individual strings */
stringWriter->write
"Key 1","Value 1",
InstanceHandle_t::HANDLE_NIL);
String^ str = "Value 2";
stringWriter->write(
"Key 1", str,
InstanceHandle_t::HANDLE_NIL);

C# Example:
using System;
using DDS;
...
KeyedStringDataWriter stringWriter = ... ;
/* Write some data using the KeyedString */
KeyedString keyedStr = new KeyedString();
keyedStr.key = "Key 1";
keyedStr.value = "Value 1";
stringWriter.write(
keyedStr, InstanceHandle_t.HANDLE_NIL);
/* Write some data using individual strings */
stringWriter.write(
"Key 1", "Value 1",
InstanceHandle_t.HANDLE_NIL);
String str = "Value 2";
stringWriter.write(
"Key 1", str,
InstanceHandle_t.HANDLE_NIL);

Java Example:
import com.rti.dds.publication.*;
import com.rti.dds.type.builtin.*;
import com.rti.dds.infrastructure.*;
...
KeyedStringDataWriter stringWriter = ... ;

42

3.2.4.3 Keyed String DataReader

/* Write some data using the KeyedString */
KeyedString keyedStr = new KeyedString();
keyedStr.key = "Key 1";
keyedStr.value = "Value 1";
stringWriter.write(
keyedStr, InstanceHandle_t.HANDLE_NIL);
/* Write some data using individual strings */
stringWriter.write(
"Key 1", "Value 1",
InstanceHandle_t.HANDLE_NIL);
String str = "Value 2";
stringWriter.write(
"Key 1", str,
InstanceHandle_t.HANDLE_NIL);

3.2.4.3 Keyed String DataReader
The KeyedString DataReader API is extended with the following operations (in addition to the standard
methods described in Using a Type-Specific DataReader (FooDataReader) (Section 7.4.1 on page 491)):
DDS::ReturnCode_t
DDS::KeyedStringDataReader::get_key_value(
char * key,
const DDS::InstanceHandle_t* handle);
DDS::InstanceHandle_t
DDS::KeyedStringDataReader::lookup_instance(
const char * key);

For additional information and a complete description of these operations in all supported languages, see
the API Reference HTML documentation, which is available for all supported programming languages.
Memory considerations in copy operations:
For read/take operations with copy semantics, such as read_next_sample() and take_next_sample(),
Connext DDS allocates memory for the fields 'value' and 'key' if they are initialized to NULL.
If the fields are not initialized to NULL, the behavior depends on the language:
l

l

In Java and .NET, the memory associated to the fields 'value' and 'key' will be reallocated with
every DDS sample.
In C and C++, the memory associated with the fields 'value' and 'key' must be large enough to
hold the received data. Insufficient memory may result in crashes.

The following examples show how to read keyed strings with a keyed string built-in type DataReader.
For simplicity, error handling is not shown.

43

3.2.4.3 Keyed String DataReader

C Example:
struct DDS_KeyedStringSeq dataSeq =
DDS_SEQUENCE_INITIALIZER;
struct DDS_SampleInfoSeq infoSeq =
DDS_SEQUENCE_INITIALIZER;
DDS_KeyedKeyedStringDataReader * stringReader = ... ;
DDS_ReturnCode_t retCode;
int i;
/* Take and print the data */
retCode = DDS_KeyedStringDataReader_take(
stringReader, &dataSeq,
&infoSeq,
DDS_LENGTH_UNLIMITED,
DDS_ANY_SAMPLE_STATE,
DDS_ANY_VIEW_STATE,
DDS_ANY_INSTANCE_STATE);
for (i = 0;
i < DDS_KeyedStringSeq_get_length(&data_seq);
++i) {
if (DDS_SampleInfoSeq_get_reference(
&info_seq, i)->valid_data) {
DDS_KeyedStringTypeSupport_print_data(
DDS_KeyedStringSeq_get_reference(&data_seq, i));
}
}
/* Return loan */
retCode = DDS_KeyedStringDataReader_return_loan(
stringReader, &data_seq, &info_seq);

C++ Example with Namespaces:1
#include "ndds/ndds_namespace_cpp.h"
using namespace DDS;
...
KeyedStringSeq dataSeq;
SampleInfoSeq infoSeq;
KeyedStringDataReader * stringReader = ... ;
/* Take a print the data */
ReturnCode_t retCode = stringReader->take(
dataSeq, infoSeq,
LENGTH_UNLIMITED,
ANY_SAMPLE_STATE,
ANY_VIEW_STATE,
ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {

1This example uses C++ namespaces. If you're not using namespaces in your own code, prefix the name of each DDS class

with 'DDS.' For example, DDS::StringDataWriter becomes DDSStringDataWriter.

44

3.2.4.3 Keyed String DataReader

if (infoSeq[i].valid_data) {
KeyedStringTypeSupport::print_data(&dataSeq[i]);
}
}
/* Return loan */
retCode = stringReader->return_loan(dataSeq, infoSeq);

C++/CLI Example:
using namespace System;
using namespace DDS;
...
KeyedStringSeq^ dataSeq = gcnew KeyedStringSeq();
SampleInfoSeq^ infoSeq = gcnew SampleInfoSeq();
KeyedStringDataReader^ stringReader = ... ;
/* Take and print the data */
stringReader->take(
dataSeq, infoSeq,
ResourceLimitsQosPolicy::LENGTH_UNLIMITED,
SampleStateKind::ANY_SAMPLE_STATE,
ViewStateKind::ANY_VIEW_STATE,
InstanceStateKind::ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (infoSeq->get_at(i)->valid_data) {
KeyedStringTypeSupport::print_data(
dataSeq->get_at(i));
}
}
/* Return loan */
stringReader->return_loan(dataSeq, infoSeq);

C# Example:
using System;
using DDS;
...
KeyedStringSeq dataSeq = new KeyedStringSeq();
SampleInfoSeq infoSeq = new SampleInfoSeq();
KeyedStringDataReader stringReader = ... ;
/* Take and print the data */
stringReader.take(dataSeq, infoSeq,
ResourceLimitsQosPolicy.LENGTH_UNLIMITED,
SampleStateKind.ANY_SAMPLE_STATE,
ViewStateKind.ANY_VIEW_STATE,
InstanceStateKind.ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (infoSeq.get_at(i)).valid_data) {
KeyedStringTypeSupport.print_data(
dataSeq.get_at(i));

45

3.2.5 Octets Built-in Type

}
}
/* Return loan */
stringReader.return_loan(dataSeq, infoSeq);

Java Example:
import com.rti.dds.infrastructure.*;
import com.rti.dds.subscription.*;
import com.rti.dds.type.builtin.*;
...
KeyedStringSeq dataSeq = new KeyedStringSeq();
SampleInfoSeq infoSeq = new SampleInfoSeq();
KeyedStringDataReader stringReader = ... ;
/* Take and print the data */
stringReader.take(dataSeq, infoSeq,
ResourceLimitsQosPolicy.LENGTH_UNLIMITED,
SampleStateKind.ANY_SAMPLE_STATE,
ViewStateKind.ANY_VIEW_STATE,
InstanceStateKind.ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (((SampleInfo)infoSeq.get(i)).valid_data) {
System.out.println((
(KeyedString)dataSeq.get(i)).toString());
}
}
/* Return loan */
stringReader.return_loan(dataSeq, infoSeq);

3.2.5 Octets Built-in Type
The octets built-in type is used to send sequences of octets. The language-specific representations are as follows:
C/Traditional C++ Representation (without Namespaces):
struct DDS_Octets {
int length;
unsigned char * value;
};

Modern C++ Representation:
class dds::core::BytesTopicType {
public:
uint8_t& operator [](uint32_t index);
// ... see API documentation for full definition

46

3.2.5.1 Creating and Deleting Octets

};

C++/CLI Representation:
namespace DDS {
public ref struct Bytes: {
public:
System::Int32 length;
System::Int32 offset;
array^ value;
...
};
};

C# Representation:
namespace DDS {
public class Bytes {
public System.Int32 length;
public System.Int32 offset;
public System.Byte[] value;
...
};
};

Java Representation:
package com.rti.dds.type.builtin;
public class Bytes implements Copyable {
public int length;
public int offset;
public byte[] value;
...
};

3.2.5.1 Creating and Deleting Octets
Connext DDS provides a set of constructors/destructors to create and destroy Octet objects. For details, see
the API Reference HTML documentation, which is available for all supported programming languages
(select Modules, RTI Connext DDS API Reference, Topic Module, Built-in Types).
If you want to manipulate the memory of the value field inside the Octets struct in C/Traditional C++, use
the operations DDS::OctetBuffer_alloc(), DDS::OctetBuffer_dup(), and DDS::OctetBuffer_free(),
described in the API Reference HTML documentation (select Modules, RTI Connext DDS API Reference, Infrastructure Module, Octet Buffer Support).

47

3.2.5.2 Octets DataWriter

3.2.5.2 Octets DataWriter
(Note: for Modern C++ API, refer to the API documentation)
In addition to the standard methods (see Using a Type-Specific DataWriter (FooDataWriter) (Section 6.3.7
on page 281)), the octets DataWriter API is extended with the following methods:
DDS::ReturnCode_t DDS::OctetsDataWriter::write(
const DDS::OctetSeq & octets,
const DDS::InstanceHandle_t & handle);
DDS::ReturnCode_t DDS::OctetsDataWriter::write(
const unsigned char * octets,
int length,
const DDS::InstanceHandle_t& handle);
DDS::ReturnCode_t DDS::OctetsDataWriter::write_w_timestamp(
const DDS::OctetSeq & octets,
const DDS::InstanceHandle_t & handle,
const DDS::Time_t & source_timestamp);
DDS::ReturnCode_t DDS::OctetsDataWriter::write_w_timestamp(
const unsigned char * octets,
int length,
const DDS::InstanceHandle_t& handle,
const DDS::Time_t& source_timestamp);

These methods are introduced to provide maximum flexibility in the format of the input parameters for the
write operations. For additional information and a complete description of these operations in all supported
languages, see the API Reference HTML documentation.
The following examples show how to write an array of octets using an octets built-in type DataWriter and
some of the extended APIs. For simplicity, error handling is not shown.
C Example:
DDS_OctetsDataWriter * octetsWriter = ... ;
DDS_ReturnCode_t retCode;
struct DDS_Octets * octets = NULL;
char * octetArray = NULL;
/* Write some data using the Octets structure */
octets = DDS_Octets_new_w_size(1024);
octets->length = 2;
octets->value[0] = 46;
octets->value[1] = 47;
retCode = DDS_OctetsDataWriter_write(
octetsWriter, octets, &DDS_HANDLE_NIL);
DDS_Octets_delete(octets);
/* Write some data using an octets array */
octetArray = (unsigned char *)malloc(1024);
octetArray[0] = 46;
octetArray[1] = 47;

48

3.2.5.2 Octets DataWriter

retCode = DDS_OctetsDataWriter_write_octets (
octetsWriter, octetArray, 2,
&DDS_HANDLE_NIL);
free(octetArray);

C++ Example with Namespaces:1
#include "ndds/ndds_namespace_cpp.h"
using namespace DDS;
...
OctetsDataWriter * octetsWriter = ... ;
/* Write some data using the Octets structure */
Octets * octets = new Octets(1024);
octets->length = 2;
octets->value[0] = 46;
octets->value[1] = 47;
ReturnCode_t retCode = octetsWriter->write(octets, HANDLE_NIL);
delete octets;
/* Write some data using an octet array */
unsigned char * octetArray = new unsigned char[1024];
octetArray[0] = 46;
octetArray[1] = 47;
retCode = octetsWriter->write(octetArray, 2, HANDLE_NIL);
delete []octetArray;

C++/CLI Example:
using namespace System;
using namespace DDS;
...
BytesDataWriter^ octetsWriter = ...;
/* Write some data using Bytes */
Bytes^ octets = gcnew Bytes(1024);
octets->value[0] =46;
octets->value[1] =47;
octets.length = 2;
octets.offset = 0;
octetWriter->write(octets, InstanceHandle_t::HANDLE_NIL);
/* Write some data using individual strings */
array^ octetAray = gcnew array(1024);
octetArray[0] = 46;
octetArray[1] = 47;
octetsWriter->write(octetArray, 0, 2, InstanceHandle_t::HANDLE_NIL);

C# Example:

1This example uses C++ namespaces. If you're not using namespaces in your own code, prefix the name of each DDS class

with 'DDS.' For example, DDS::StringDataWriter becomes DDSStringDataWriter.

49

3.2.5.3 Octets DataReader

using System;
using DDS;
...
BytesDataWriter stringWriter = ...;
/* Write some data using the Bytes */
Bytes octets = new Bytes(1024);
octets.value[0] = 46;
octets.value[1] = 47;
octets.length = 2;
octets.offset = 0;
octetWriter.write(octets, InstanceHandle_t.HANDLE_NIL);
/* Write some data using individual strings */
byte[] octetArray = new byte[1024];
octetArray[0] = 46;
octetArray[1] = 47;
octetsWriter.write(octetArray, 0, 2, InstanceHandle_t.HANDLE_NIL);

Java Example:
import com.rti.dds.publication.*;
import com.rti.dds.type.builtin.*;
import com.rti.dds.infrastructure.*;
...
BytesDataWriter octetsWriter = ... ;
/* Write some data using the Bytes class*/
Bytes octets = new Bytes(1024);
octets.length = 2;
octets.offset = 0;
octets.value[0] = 46;
octets.value[1] = 47;
octetsWriter.write(octets, InstanceHandle_t.HANDLE_NIL);
/* Write some data using a byte array */
byte[] octetArray = new byte[1024];
octetArray[0] = 46;
octetArray[1] = 47;
octetsWriter.write(octetArray, 0, 2, InstanceHandle_t.HANDLE_NIL);

3.2.5.3 Octets DataReader
(Note: for the Modern C++ API, refer to the API Reference HTML documentation)
The octets DataReader API matches the standard DataReader API (see Using a Type-Specific
DataReader (FooDataReader) (Section 7.4.1 on page 491)). There are no extensions.

50

3.2.5.3 Octets DataReader

Memory considerations in copy operations:
For read/take operations with copy semantics, such as read_next_sample() and take_next_sample(),
Connext DDS allocates memory for the field 'value' if it is initialized to NULL.
If the field 'value' is not initialized to NULL, the behavior depends on the language:
l

l

In Java and .NET, the memory for the field 'value' will be reallocated if the current size is not
large enough to hold the received data.
In C and C++, the memory associated with the field 'value' must be big enough to hold the
received data. Insufficient memory may result in crashes.

The following examples show how to read octets with an octets built-in type DataReader. For simplicity,
error handling is not shown.
C Example:
struct DDS_OctetsSeq dataSeq = DDS_SEQUENCE_INITIALIZER;
struct DDS_SampleInfoSeq infoSeq = DDS_SEQUENCE_INITIALIZER;
DDS_OctetsDataReader * octetsReader = ... ;
DDS_ReturnCode_t retCode;
int i;
/* Take and print the data */
retCode = DDS_OctetsDataReader_take(
octetsReader, &dataSeq,
&infoSeq, DDS_LENGTH_UNLIMITED,
DDS_ANY_SAMPLE_STATE,
DDS_ANY_VIEW_STATE,
DDS_ANY_INSTANCE_STATE);
for (i = 0; i < DDS_OctetsSeq_get_length(&dataSeq); ++i) {
if (DDS_SampleInfoSeq_get_reference(
&infoSeq, i)->valid_data) {
DDS_OctetsTypeSupport_print_data(
DDS_OctetsSeq_get_reference(&dataSeq, i));
}
}
/* Return loan */
retCode = DDS_OctetsDataReader_return_loan(
octetsReader, &dataSeq, &infoSeq);

C++ Example with Namespaces:1
#include "ndds/ndds_namespace_cpp.h"

1This example uses C++ namespaces. If you're not using namespaces in your own code, prefix the name of each DDS class

with 'DDS.' For example, DDS::StringDataWriter becomes DDSStringDataWriter.

51

3.2.5.3 Octets DataReader

using namespace DDS;
...
OctetsSeq dataSeq;
SampleInfoSeq infoSeq;
OctetsDataReader * octetsReader = ... ;
/* Take a print the data */
ReturnCode_t retCode = octetsReader->take(
dataSeq, infoSeq,
LENGTH_UNLIMITED, ANY_SAMPLE_STATE,
ANY_VIEW_STATE, ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (infoSeq[i].valid_data) {
OctetsTypeSupport::print_data(&dataSeq[i]);
}
}
/* Return loan */
retCode = octetsReader->return_loan(dataSeq, infoSeq);

C++/CLI Example:
using namespace System;
using namespace DDS;
...
BytesSeq^ dataSeq = gcnew BytesSeq();
SampleInfoSeq^ infoSeq = gcnew SampleInfoSeq();
BytesDataReader^ octetsReader = ... ;
/* Take and print the data */
octetsReader->take(
dataSeq, infoSeq,
ResourceLimitsQosPolicy::LENGTH_UNLIMITED,
SampleStateKind::ANY_SAMPLE_STATE,
ViewStateKind::ANY_VIEW_STATE,
InstanceStateKind::ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (infoSeq->get_at(i)->valid_data) {
BytesTypeSupport::print_data(dataSeq->get_at(i));
}
}
/* Return loan */
octetsReader->return_loan(dataSeq, infoSeq);

C# Example:
using System;
using DDS;
...
BytesSeq dataSeq = new BytesSeq();
SampleInfoSeq infoSeq = new SampleInfoSeq();
BytesDataReader octetsReader = ... ;

52

3.2.6 KeyedOctets Built-in Type

/* Take and print the data */
octetsReader.take(
dataSeq, infoSeq,
ResourceLimitsQosPolicy.LENGTH_UNLIMITED,
SampleStateKind.ANY_SAMPLE_STATE,
ViewStateKind.ANY_VIEW_STATE,
InstanceStateKind.ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (infoSeq.get_at(i)).valid_data) {
BytesTypeSupport.print_data(dataSeq.get_at(i));
}
}
/* Return loan */
octetsReader.return_loan(dataSeq, infoSeq);

Java Example:
import com.rti.dds.infrastructure.*;
import com.rti.dds.subscription.*;
import com.rti.dds.type.builtin.*;
...
BytesSeq dataSeq = new BytesSeq();
SampleInfoSeq infoSeq = new SampleInfoSeq();
BytesDataReader octetsReader = ... ;
/* Take and print the data */
octetsReader.take(dataSeq, infoSeq,
ResourceLimitsQosPolicy.LENGTH_UNLIMITED,
SampleStateKind.ANY_SAMPLE_STATE,
ViewStateKind.ANY_VIEW_STATE,
InstanceStateKind.ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (((SampleInfo)infoSeq.get(i)).valid_data) {
System.out.println(((Bytes)dataSeq.get(i)).toString());
}
}
/* Return loan */
octetsReader.return_loan(dataSeq, infoSeq);

3.2.6 KeyedOctets Built-in Type
The keyed octets built-in type is used to send sequences of octets with a key. The language-specific representations of the type are as follows:
C/Traditional C++ Representation (without Namespaces):
struct DDS_KeyedOctets {
char * key;
int length;
unsigned char * value;

53

3.2.6 KeyedOctets Built-in Type

};

Modern C++ Representation:
class dds::core::KeyedStringTopicType {
public:
dds::core::string& key();
uint8_t& operator [](uint32_t index);
// ... see API documentation for full definition
};

C++/CLI Representation:
namespace DDS {
public ref struct KeyedBytes {
public:
System::String^ key;
System::Int32 length;
System::Int32 offset;
array^ value;
...
};
};

C# Representation:
namespace DDS {
public class KeyedBytes {
public System.String key;
public System.Int32 length;
public System.Int32 offset;
public System.Byte[] value;
...
};
};

Java Representation:
package com.rti.dds.type.builtin;
public class KeyedBytes {
public String key;
public int length;
public int offset;
public byte[] value;
...
};

54

3.2.6.1 Creating and Deleting KeyedOctets

3.2.6.1 Creating and Deleting KeyedOctets
Connext DDS provides a set of constructors/destructors to create/destroy KeyedOctets objects. For details,
see the API Reference HTML documentation, which is available for all supported programming languages
(select Modules, RTI Connext DDS API Reference, Topic Module, Built-in Types).
To manipulate the memory of the value field in the KeyedOctets struct in C/C++: use DDS::OctetBuffer_alloc(), DDS::OctetBuffer_dup(), and DDS::OctetBuffer_free(). See the API Reference
HTML documentation (select Modules, RTI Connext DDS API Reference, Infrastructure Module,
Octet Buffer Support).
To manipulate the memory of the key field in the KeyedOctets struct in C/C++: use DDS::String_alloc(),
DDS::String_dup(), and DDS::String_free(). See the API Reference HTML documentation (select
Modules, RTI Connext DDS API Reference, Infrastructure Module, String Support).
3.2.6.2 Keyed Octets DataWriter
In addition to the standard methods (see Using a Type-Specific DataWriter (FooDataWriter) (Section 6.3.7
on page 281)), the keyed octets DataWriter API is extended with the following methods:
DDS::ReturnCode_t
DDS::KeyedOctetsDataWriter::dispose(
const char* key,
const DDS::InstanceHandle_t & instance_handle);
DDS::ReturnCode_t
DDS::KeyedOctetsDataWriter::dispose_w_timestamp(
const char* key,
const DDS::InstanceHandle_t & instance_handle,
const DDS::Time_t & source_timestamp);
DDS::ReturnCode_t
DDS::KeyedOctetsDataWriter::get_key_value(
char * key,
const DDS::InstanceHandle_t& handle);
DDS::InstanceHandle_t
DDS::KeyedOctetsDataWriter::lookup_instance(
const char * key);
DDS::InstanceHandle_t
DDS::KeyedOctetsDataWriter::register_instance(
const char* key);
DDS::InstanceHandle_t
DDS::KeyedOctetsDataWriter::
register_instance_w_timestamp(
const char * key,
const DDS::Time_t & source_timestamp);
DDS::ReturnCode_t
DDS::KeyedOctetsDataWriter::unregister_instance(
const char * key,
const DDS::InstanceHandle_t & handle);
DDS::ReturnCode_t
DDS::KeyedOctetsDataWriter::

55

3.2.6.2 Keyed Octets DataWriter

unregister_instance_w_timestamp(
const char* key,
const DDS::InstanceHandle_t & handle,
const DDS::Time_t & source_timestamp);
DDS::ReturnCode_t
DDS::KeyedOctetsDataWriter::write(
const char * key,
const unsigned char * octets,
int length,
const DDS::InstanceHandle_t& handle);
DDS::ReturnCode_t
DDS::KeyedOctetsDataWriter::write(
const char * key,
const DDS::OctetSeq & octets,
const DDS::InstanceHandle_t & handle);
DDS::ReturnCode_t
DDS::KeyedOctetsDataWriter::write_w_timestamp(
const char * key,
const unsigned char * octets,
int length,
const DDS::InstanceHandle_t& handle,
const DDS::Time_t& source_timestamp);
DDS::ReturnCode_t
DDS::KeyedOctetsDataWriter::write_w_timestamp(
const char * key,
const DDS::OctetSeq & octets,
const DDS::InstanceHandle_t & handle,
const DDS::Time_t & source_timestamp);

These methods are introduced to provide maximum flexibility in the format of the input parameters for the
write and instance management operations. For more information and a complete description of these operations in all supported languages, see the API Reference HTML documentation.
The following examples show how to write keyed octets using a keyed octets built-in type DataWriter and
some of the extended APIs. For simplicity, error handling is not shown.
C Example:
DDS_KeyedOctetsDataWriter * octetsWriter = ... ;
DDS_ReturnCode_t retCode;
struct DDS_KeyedOctets * octets = NULL;
char * octetArray = NULL;
/* Write some data using KeyedOctets structure */
octets = DDS_KeyedOctets_new_w_size(128,1024);
strcpy(octets->key, "Key 1");
octets->length = 2;
octets->value[0] = 46;
octets->value[1] = 47;
retCode = DDS_KeyedOctetsDataWriter_write(
octetsWriter, octets, &DDS_HANDLE_NIL);

56

3.2.6.2 Keyed Octets DataWriter

DDS_KeyedOctets_delete(octets);
/* Write some data using an octets array */
octetArray = (unsigned char *)malloc(1024);
octetArray[0] = 46;
octetArray[1] = 47;
retCode =
DDS_KeyedOctetsDataWriter_write_octets_w_key (
octetsWriter, "Key 1",
octetArray, 2, &DDS_HANDLE_NIL);
free(octetArray);

C++ Example with Namespaces:1
#include "ndds/ndds_namespace_cpp.h"
using namespace DDS;
...
KeyedOctetsDataWriter * octetsWriter = ...;
/* Write some data using KeyedOctets */
KeyedOctets * octets = new KeyedOctets(128,1024);
strcpy(octets->key, "Key 1");
octets->length = 2;
octets->value[0] = 46;
octets->value[1] = 47;
ReturnCode_t retCode =
octetsWriter->write(octets, HANDLE_NIL);
delete octets;
/* Write some data using an octet array */
unsigned char * octetArray = new unsigned char[1024];
octetArray[0] = 46;
octetArray[1] = 47;
retCode = octetsWriter->write(
"Key 1", octetArray, 2, HANDLE_NIL);
delete []octetArray;

C++/CLI Example:
using namespace System;
using namespace DDS;
...
KeyedOctetsDataWriter^ octetsWriter = ... ;
/* Write some data using KeyedBytes */
KeyedBytes^ octets = gcnew KeyedBytes(1024);
octets->key = "Key 1";
octets->value[0] =46;

1This example uses C++ namespaces. If you're not using namespaces in your own code, prefix the name of each DDS class

with 'DDS.' For example, DDS::StringDataWriter becomes DDSStringDataWriter.

57

3.2.6.2 Keyed Octets DataWriter

octets->value[1] =47;
octets.length = 2;
octets.offset = 0;
octetWriter->write(
octets, InstanceHandle_t::HANDLE_NIL);
/* Write some data using individual strings */
array^ octetAray = gcnew array(1024);
octetArray[0] = 46;
octetArray[1] = 47;
octetsWriter->write(
"Key 1", octetArray,
0, 2, InstanceHandle_t::HANDLE_NIL);

C# Example:
using System;
using DDS;
...
KeyedBytesDataWriter stringWriter = ... ;
/* Write some data using the KeyedBytes */
KeyedBytes octets = new KeyedBytes(1024);
octets.key = "Key 1";
octets.value[0] = 46;
octets.value[1] = 47;
octets.length = 2;
octets.offset = 0;
octetWriter.write(octets,
InstanceHandle_t.HANDLE_NIL);
/* Write some data using individual strings */
byte[] octetArray = new byte[1024];
octetArray[0] = 46;
octetArray[1] = 47;
octetsWriter.write(
"Key 1", octetArray,
0, 2, InstanceHandle_t.HANDLE_NIL);

Java Example:
import com.rti.dds.publication.*;
import com.rti.dds.type.builtin.*;
import com.rti.dds.infrastructure.*;
...
KeyedBytesDataWriter octetsWriter = ... ;
/* Write some data using KeyedBytes class */
KeyedBytes octets = new KeyedBytes(1024);
octets.key = "Key 1";
octets.length = 2;
octets.offset = 0;
octets.value[0] = 46;

58

3.2.6.3 Keyed Octets DataReader

octets.value[1] = 47;
octetsWriter.write(octets,
InstanceHandle_t.HANDLE_NIL);
/* Write some data using a byte array */
byte[] octetArray = new byte[1024];
octetArray[0] = 46;
octetArray[1] = 47;
octetsWriter.write(
"Key 1", octetArray,
0, 2, InstanceHandle_t.HANDLE_NIL);

3.2.6.3 Keyed Octets DataReader
The KeyedOctets DataReader API is extended with the following methods (in addition to the standard
methods described in Using a Type-Specific DataReader (FooDataReader) (Section 7.4.1 on page 491)):
DDS::ReturnCode_t
DDS::KeyedOctetsDataReader::get_key_value(
char * key,
const DDS::InstanceHandle_t* handle);
DDS::InstanceHandle_t
DDS::KeyedOctetsDataReader::lookup_instance(
const char * key);

For more information and a complete description of these operations in all supported languages, see the
API Reference HTML documentation.
Memory considerations in copy operations:
For read/take operations with copy semantics, such as read_next_sample() and take_next_sample(),
Connext DDS allocates memory for the fields 'value' and 'key' if they are initialized to NULL.
If the fields are not initialized to NULL, the behavior depends on the language:
l

l

In Java and .NET, the memory of the field 'value' will be reallocated if the current size is not
large enough to hold the received data. The memory associated with the field 'key' will be reallocated with every DDS sample (the key is an immutable object).
In C and C++, the memory associated with the fields 'value' and 'key' must be large enough to
hold the received data. Insufficient memory may result in crashes.

The following examples show how to read keyed octets with a keyed octets built-in type DataReader. For
simplicity, error handling is not shown.
C Example:

59

3.2.6.3 Keyed Octets DataReader

struct DDS_KeyedOctetsSeq dataSeq =
DDS_SEQUENCE_INITIALIZER;
struct DDS_SampleInfoSeq infoSeq =
DDS_SEQUENCE_INITIALIZER;
DDS_KeyedOctetsDataReader * octetsReader = ... ;
DDS_ReturnCode_t retCode;
int i;
/* Take and print the data */
retCode = DDS_KeyedOctetsDataReader_take(
octetsReader,
&dataSeq, &infoSeq, DDS_LENGTH_UNLIMITED,
DDS_ANY_SAMPLE_STATE, DDS_ANY_VIEW_STATE,
DDS_ANY_INSTANCE_STATE);
for (i = 0;
i < DDS_KeyedOctetsSeq_get_length(&data_seq);
++i) {
if (DDS_SampleInfoSeq_get_reference(
&info_seq, i)->valid_data) {
DDS_KeyedOctetsTypeSupport_print_data(
DDS_KeyedOctetsSeq_get_reference(
&data_seq, i));
}
}
/* Return loan */
retCode = DDS_KeyedOctetsDataReader_return_loan(
octetsReader, &data_seq, &info_seq);

C++ Example with Namespaces:1
#include "ndds/ndds_namespace_cpp.h"
using namespace DDS;
...
KeyedOctetsSeq dataSeq;
SampleInfoSeq infoSeq;
KeyedOctetsDataReader * octetsReader = ... ;
/* Take and print the data */
ReturnCode_t retCode = octetsReader->take(
dataSeq, infoSeq, LENGTH_UNLIMITED,
ANY_SAMPLE_STATE, ANY_VIEW_STATE,
ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (infoSeq[i].valid_data) {
KeyedOctetsTypeSupport::print_data(
&dataSeq[i]);
}

1This example uses C++ namespaces. If you're not using namespaces in your own code, prefix the name of each DDS class

with 'DDS.' For example, DDS::StringDataWriter becomes DDSStringDataWriter.

60

3.2.6.3 Keyed Octets DataReader

}
/* Return loan */
retCode = octetsReader->return_loan(
dataSeq, infoSeq);

C++/CLI Example:
using namespace System;
using namespace DDS;
...
KeyedBytesSeq^ dataSeq = gcnew KeyedBytesSeq();
SampleInfoSeq^ infoSeq = gcnew SampleInfoSeq();
KeyedBytesDataReader^ octetsReader = ... ;
/* Take and print the data */
octetsReader->take(dataSeq, infoSeq,
ResourceLimitsQosPolicy::LENGTH_UNLIMITED,
SampleStateKind::ANY_SAMPLE_STATE,
ViewStateKind::ANY_VIEW_STATE,
InstanceStateKind::ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i){
if (infoSeq->get_at(i)->valid_data){
KeyedBytesTypeSupport::print_data(
dataSeq->get_at(i));
}
}
/* Return loan */
octetsReader->return_loan(dataSeq, infoSeq);

C# Example:
using System;
using DDS;
...
KeyedBytesSeq dataSeq = new KeyedButesSeq();
SampleInfoSeq infoSeq = new SampleInfoSeq();
KeyedBytesDataReader octetsReader = ... ;
/* Take and print the data */
octetsReader.take(dataSeq, infoSeq,
ResourceLimitsQosPolicy.LENGTH_UNLIMITED,
SampleStateKind.ANY_SAMPLE_STATE,
ViewStateKind.ANY_VIEW_STATE,
InstanceStateKind.ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i) {
if (infoSeq.get_at(i)).valid_data) {
KeyedBytesTypeSupport.print_data(
dataSeq.get_at(i));
}
}
/* Return loan */

61

3.2.7 Managing Memory for Built-in Types

octetsReader.return_loan(dataSeq, infoSeq);

Java Example:
import com.rti.dds.infrastructure.*;
import com.rti.dds.subscription.*;
import com.rti.dds.type.builtin.*;
...
KeyedBytesSeq dataSeq = new KeyedBytesSeq();
SampleInfoSeq infoSeq = new SampleInfoSeq();
KeyedBytesDataReader octetsReader = ... ;
/* Take and print the data */
octetsReader.take(dataSeq, infoSeq,
ResourceLimitsQosPolicy.LENGTH_UNLIMITED,
SampleStateKind.ANY_SAMPLE_STATE,
ViewStateKind.ANY_VIEW_STATE,
InstanceStateKind.ANY_INSTANCE_STATE);
for (int i = 0; i < data_seq.length(); ++i){
if (((SampleInfo)infoSeq.get(i)).valid_data){
System.out.println(
((KeyedBytes)dataSeq.get(i)).toString());
}
}
/* Return loan */
octetsReader.return_loan(dataSeq, infoSeq);

3.2.7 Managing Memory for Built-in Types
When a DDS sample is written, the DataWriter serializes it and stores the result in a buffer obtained from a
pool of preallocated buffers. In the same way, when a DDS sample is received, the DataReader deserializes it and stores the result in a DDS sample coming from a pool of preallocated DDS samples.
By default, the buffers on the DataWriter and the samples on the DataReader are preallocated with their
maximum size. For example:
struct MyString
Unknown macro: { string<128> value; }

This IDL-defined type has a maximum serialized size of 133 bytes (4 bytes for length + 128 characters + 1
NULL terminating character). So the serialization buffers will have a size of 133 bytes. The buffer can
hold samples with 128 characters strings. Consequently, the preallocated samples will be sized to keep this
length.
However, for built-in types, the maximum size of the buffers/DDS samples is unknown and depends on
the nature of the application using the built-in type.

62

3.2.7 Managing Memory for Built-in Types

For example, a video surveillance application that is using the keyed octets built-in type to publish a stream
of images will require bigger buffers than a market-data application that uses the same built-in type to publish market-data values.
To accommodate both kinds of applications and optimize memory usage, you can configure the maximum
size of the built-in types on a per-DataWriter or per-Datareader basis using the PROPERTY QosPolicy
(DDS Extension) (Section 6.5.17 on page 394). Table 3.1 Properties for Allocating Size of Built-in
Types, per DataWriter and DataReader lists the supported built-in type properties. When the properties are
defined in the DomainParticipant, they are applicable to all DataWriters and DataReaders belonging to
the DomainParticipant, unless they are overwritten in the DataWriters and DataReaders.
These properties must be set consistently with respect to the corresponding *.max_size properties
in the DomainParticipant (see Table 3.14 Properties for Allocating Size of Built-in Types, per
DomainParticipant). The value of the alloc_size property must be less than or equal to the max_
size property with the same name prefix in the DomainParticipant.
Examples—Setting the Maximum Size for a String Programmatically (Section 3.2.7.1 on the next page)
includes examples of how to set the maximum size of a string built-in type for a DataWriter programmatically, for each API. You can also set the maximum size of the built-in types using XML QoS Profiles. For example, the following XML shows how to set the maximum size of a string built-in type for a
DataWriter.







dds.builtin_type.string.alloc_size
2048








dds.builtin_type.string.alloc_size
2048






63

3.2.7.1 Examples—Setting the Maximum Size for a String Programmatically




Table 3.1 Properties for Allocating Size of Built-in Types, per DataWriter and DataReader
Built-in
Type

string

Property

dds.builtin_
type.string.alloc_
size

dds.builtin_
type.keyed_string.
alloc_key_size

Description
Maximum size of the strings published by the DataWriter or received by the DataReader (includes the
NULL-terminated character).
Default: dds.builtin_type.string.max_size if defined (see Table 3.14 Properties for Allocating Size of
Built-in Types, per DomainParticipant). Otherwise, 1024.
Maximum size of the keys used by the DataWriter or DataReader (includes the NULL-terminated
character).
Default: dds.builtin_type.keyed_string.max_key_size if defined (see Table 3.14 Properties for Allocating
Size of Built-in Types, per DomainParticipant). Otherwise, 1024.

keyedstring
dds.builtin_
type.keyed_string.
alloc_size

octets

dds.builtin_
type.octets.alloc_
size

dds.builtin_
type.keyed_octets.
alloc_key_size
keyedoctets
dds.builtin_
type.keyed_octets.
alloc_size

Maximum size of the strings published by the DataWriter or received by the DataReader (includes the
NULL-terminated character).
Default: dds.builtin_type.keyed_string.max_size if defined (see Table 3.14 Properties for Allocating Size
of Built-in Types, per DomainParticipant). Otherwise, 1024.
Maximum size of the octet sequences published by the DataWriter or DataReader.
Default: dds.builtin_type.octets.max_size if defined (see Table 3.14 Properties for Allocating Size of
Built-in Types, per DomainParticipant). Otherwise, 2048.
Maximum size of the key published by the DataWriter or received by the DataReader (includes the
NULL-terminated character).
Default: dds.builtin_type.keyed_octets.max_key_size if defined (see Table 3.14 Properties for Allocating
Size of Built-in Types, per DomainParticipant). Otherwise, 1024.
Maximum size of the octet sequences published by the DataWriter or DataReader.
Default: dds.builtin_type.keyed_octets.max_size if defined (see Table 3.14 Properties for Allocating Size
of Built-in Types, per DomainParticipant). Otherwise, 2048.

3.2.7.1 Examples—Setting the Maximum Size for a String Programmatically
For simplicity, error handling is not shown in the following examples.
C Example:
DDS_DataWriter * writer = NULL;
DDS_StringDataWriter * stringWriter = NULL;
DDS_Publisher * publisher = ... ;
DDS_Topic * stringTopic = ... ;
struct DDS_DataWriterQos writerQos =

64

3.2.7.1 Examples—Setting the Maximum Size for a String Programmatically

DDS_DataWriterQos_INITIALIZER;
DDS_ReturnCode_t retCode;
retCode = DDS_DomainParticipant_get_default_datawriter_qos (
participant, &writerQos);
retCode = DDS_PropertyQosPolicyHelper_add_property (
&writerQos.property,
"dds.builtin_type.string.alloc_size", "1000",
DDS_BOOLEAN_FALSE);
writer = DDS_Publisher_create_datawriter(
publisher, stringTopic, &writerQos,
NULL, DDS_STATUS_MASK_NONE);
stringWriter = DDS_StringDataWriter_narrow(writer);
DDS_DataWriterQos_finalize(&writerQos);

Traditional C++ Example with Namespaces: 1
#include "ndds/ndds_namespace_cpp.h"
using namespace DDS;
...
Publisher * publisher = ... ;
Topic * stringTopic = ... ;
DataWriterQos writerQos;
ReturnCode_t retCode =
participant->get_default_datawriter_qos(writerQos);
retCode = PropertyQosPolicyHelper::add_property (
&writerQos.property,
"dds.builtin_type.string.alloc_size",
"1000", BOOLEAN_FALSE);
DataWriter * writer = publisher->create_datawriter(
stringTopic, writerQos,
NULL, STATUS_MASK_NONE);
StringDataWriter * stringWriter =
StringDataWriter::narrow(writer);

Modern C++ Example:
dds::pub::qos::DataWriterQos writer_qos =
participant.default_datawriter_qos();
writer_qos.policy().set({
"dds.builtin_type.string.alloc_size", "1000"});
dds::pub::DataWriter writer(
publisher, string_topic, writer_qos);

C++/CLI Example:

1This example uses C++ namespaces. If you're not using namespaces in your own code, prefix the name of each DDS class

with 'DDS.' For example, DDS::StringDataWriter becomes DDSStringDataWriter.

65

3.2.7.1 Examples—Setting the Maximum Size for a String Programmatically

using namespace DDS;
...
Topic^ stringTopic = ... ;
Publisher^ publisher = ... ;
DataWriterQos^ writerQos = gcnew DataWriterQos();
participant->get_default_datawriter_qos(writerQos);
PropertyQosPolicyHelper::add_property(
writerQos->property_qos,
"dds.builtin_type.string.alloc_size",
"1000", false);
DataWriter^ writer = publisher->create_datawriter(
stringTopic, writerQos,
nullptr, StatusMask::STATUS_MASK_NONE);
StringDataWriter^ stringWriter =
safe_cast(writer);

C# Example:
using DDS;
...
Topic stringTopic = ... ;
Publisher publisher = ... ;
DataWriterQos writerQos = new DataWriterQos();
participant.get_default_datawriter_qos(writerQos);
PropertyQosPolicyHelper.add_property (
writerQos.property_qos,
"dds.builtin_type.string.alloc_size",
"1000", false);
StringDataWriter stringWriter =
(StringDataWriter) publisher.create_datawriter(
stringTopic, writerQos, null,
StatusMask.STATUS_MASK_NONE);

Java Example:
import com.rti.dds.publication.*;
import com.rti.dds.type.builtin.*;
import com.rti.dds.infrastructure.*;
...
Topic stringTopic = ... ;
Publisher publisher = ... ;
DataWriterQos writerQos = new DataWriterQos();
participant.get_default_datawriter_qos(writerQos);
PropertyQosPolicyHelper.add_property (
writerQos.property,
"dds.builtin_type.string.alloc_size",
"1000", false);
StringDataWriter stringWriter =
(StringDataWriter) publisher.create_datawriter(

66

3.2.7.2 Unbounded Built-in Types

stringTopic, writerQos,
null, StatusKind.STATUS_MASK_NONE);

3.2.7.2 Unbounded Built-in Types
In some scenarios, the maximum size of a built-in type is not known in advance and there is no a reasonable maximum size. For example, this could occur in a file transfer application using the built-in type
Octets. Setting a large value for the dds.builtin_type.*.alloc_size property would involve high memory
usage.
For the above use case, you can configure the built-in type to be unbounded by setting the property
dds.builtin_type.*.alloc_size to the maximum value of a 32-bit signed integer: 2,147,483,647. Then the
middleware will not preallocate the DataReader queue's samples to their maximum size. Instead, it will
deserialize incoming samples by dynamically allocating and deallocating memory to accommodate the
actual size of the sample value.
To configure unbounded support for built-in types:
1. Set the properties dds.builtin_type.*.max_size and dds.builtin_type.*.alloc_size to
2,147,483,647.
2. Use these threshold QoS properties:
l dds.data_writer.history.memory_manager.fast_pool.pool_buffer_max_size on the
DataWriter
l

dds.data_reader.history.memory_manager.fast_pool.pool_buffer_max_size on the
DataReader (only if keyed)

3. Set the QoS value reader_resource_limits.dynamically_allocate_fragmented_samples on the
DataReader to true.
4. For the Java API, also set these properties accordingly for the Java serialization buffer:
l dds.data_writer.history.memory_manager.java_stream.min_size
l

dds.data_writer.history.memory_manager.java_stream.trim_to_size

l

dds.data_reader.history.memory_manager.java_stream.min_size

l

dds.data_reader.history.memory_manager.java_stream.trim_to_size

See these sections in the RTI Connext DDS Core Libraries User's Manual:
l

Section 20.1.3, Writer-Side Memory Management when Using Java

l

Section 20.2.2, Reader-Side Memory Management when Using Java

Unbounded built-in types are only supported in the C, C++, .NET, and Java APIs.

67

3.2.8 Type Codes for Built-in Types

3.2.8 Type Codes for Built-in Types
The type codes associated with the built-in types are generated from the following IDL type definitions:
module DDS {
/* String */
struct String {
string value;
};
/* KeyedString */
struct KeyedString {
string key; //@key
string value;
};
/* Octets */
struct Octets {
sequence value;
};
/* KeyedOctets */
struct KeyedOctets {
string key; //@key
sequence value;
};
};

The maximum size (max_size) of the strings and sequences that will be included in the type code definitions can be configured on a per-DomainParticipant-basis by using the properties in Table 3.2 Properties
for Allocating Size of Built-in Types, per DomainParticipant.
Table 3.2 Properties for Allocating Size of Built-in Types, per DomainParticipant
Built-in
Type

String

Property

Description

Maximum size of the strings published by the DataWriters and received by the DataReaders belonging to
dds.builtin_
type.string.max_ a DomainParticipant (includes the NULL-terminated character).
size
Default: 1024
dds.builtin_
type.keyed_
string.
max_key_size

Maximum size of the keys used by the DataWriters and DataReaders belonging to a DomainParticipant
(includes the NULL-terminated character).

dds.builtin_
type.keyed_
string.
max_size

Maximum size of the strings published by the DataWriters and received by the DataReaders belonging to
a DomainParticipant using the built-in type (includes the NULL-terminated character).

Default: 1024

KeyedString

Default: 1024

68

3.3 Creating User Data Types with IDL

Table 3.2 Properties for Allocating Size of Built-in Types, per DomainParticipant
Built-in
Type

Octets

KeyedOctets

Property

Description

Maximum size of the octet sequences published by the DataWriters and DataReaders belonging to a
dds.builtin_
type.octets.max_ DomainParticipant.
size
Default: 2048
dds.builtin_
type.keyed_
octets.
max_key_size

Maximum size of the key published by the DataWriter and received by the DataReaders belonging to the
DomainParticipant (includes the NULL-terminated character).

dds.builtin_
type.keyed_
octets.
max_size

Maximum size of the octet sequences published by the DataWriters and DataReaders belonging to a
DomainParticipant.

Default:1024.

Default: 2048

3.3 Creating User Data Types with IDL
You can create user data types in a text file using IDL (Interface Description Language). IDL is programming-language independent, so the same file can be used to generate code in C, Traditional C++,
Modern C++, C++/CLI, and Java (the languages supported by RTI Code Generator (rtiddsgen)). RTI
Code Generator parses the IDL file and automatically generates all the necessary routines and wrapper
functions to bind the types for use by Connext DDS at run time. You will end up with a set of required
routines and structures that your application and Connext DDS will use to manipulate the data.
Connext DDS only uses a subset of the IDL syntax. IDL was originally defined by the OMG for the use
of CORBA client/server applications in an enterprise setting. Not all of the constructs that can be described
by the language are as useful in the context of high-performance data-centric embedded applications.
These include the constructs that define method and function prototypes like “interface.”
RTI Code Generator will parse any file that follows version 3.0.3 of the IDL specification. It will quietly
ignore all syntax that is not recognized by Connext DDS. In addition, even though “anonymous
sequences” (sequences of sequences with no intervening typedef) are currently legal in IDL, they have
been deprecated by the specification; thus RTI Code Generator does not support them.
Certain keywords are considered reserved by the IDL specification; see Table 3.3 Reserved IDL Keywords.
Table 3.3 Reserved IDL Keywords
abstract

emits

local

pseudo

typeid

alias

enum

long

public

typename

69

3.3.1 Variable-Length Types

Table 3.3 Reserved IDL Keywords
any

eventtype

mirrorport

publishes

typeprefix

attribute

exception

module

raises

union

boolean

factory

multiple

readonly

unsigned

case

FALSE

native

sequence

uses

char

finder

object

setraises

valuebase

component

fixed

octet

short

valuetype

connector

float

oneway

string

void

const

getraises

out

struct

wchar

consumes

home

port

supports

wstring

context

import

porttype

switch

custom

in

primarykey

TRUE

default

inout

private

truncatable

double

interface

provides

typedef

The IDL constructs supported by RTI Code Generator are described in Table 3.5 Specifying Data Types
in IDL for C through Table 3.9 Specifying Data Types in IDL for Java. Use these tables to map primitive
types to their equivalent IDL syntax, and vice versa.
For C and Traditional C++, RTI Code Generator uses typedefs instead of the language keywords for primitive types. For example, DDS_Long instead of long or DDS_Double instead of double. This ensures that
the types are of the same size regardless of the platform.1
The remainder of this section includes:

3.3.1 Variable-Length Types
When RTI Code Generator generates code for data structures with variable-length types—strings and
sequences—it includes functions that create, initialize and finalize (destroy) those objects. These support

1The number of bytes sent on the wire for each data type is determined by the Common Data Representation (CDR) stand-

ard. For details on CDR, please see the Common Object Request Broker Architecture (CORBA) Specification, Version
3.1, Part 2: CORBA Interoperability, Section 9.3, CDR Transfer Syntax (http://www.omg.org/spec/CORBA/3.3/ ).

70

3.3.1.1 Sequences

functions will properly initialize pointers and allocate and deallocate the memory used for variable-length
types. All Connext DDS APIs assume that the data structures passed to them are properly initialized.
For variable-length types, the actual length (instead of the maximum length) of data is transmitted on the
wire when the DDS sample is written (regardless of whether the type has hard-coded bounds).
3.3.1.1 Sequences
C, Traditional C++, C++/CLI, and C# users can allocate memory from a number of sources: from the
heap, the stack, or from a custom allocator of some kind. In those languages, sequences provide the
concept of memory "ownership." A sequence may own the memory allocated to it or be loaned memory
from another source. If a sequence owns its memory, it will manage its underlying memory storage buffer
itself. When a sequence's maximum size is changed, the sequence will free and reallocate its buffer as
needed. However, if a sequence was created with loaned memory by user code, then its memory is not its
own to free or reallocate. Therefore, you cannot set the maximum size of a sequence whose memory is
loaned. See the API Reference HTML documentation, which is available for all supported programming
languages (select Modules, RTI Connext DDS API Reference, Infrastructure Module, Sequence Support)
for more information about how to loan and unloan memory for sequence.
In IDL, as described above, a sequence may be declared as bounded or unbounded. A sequence's "bound"
is the greatest value its maximum may take. If you use the initializer functions RTI Code Generator
provides for your types, all sequences will have their maximums set to their declared bounds. However,
the amount of data transmitted on the wire when the DDS sample is written will vary.
In the Modern C++ API, sequences (dds::core::vector) always own the memory.
3.3.1.2 Strings and Wide Strings
(Note: this section doesn't apply to the Modern C++ API, where dds::core::string behaves similarly to
std::string)
The initialization functions that RTI Code Generator provides for your types will allocate all of the
memory for strings in a type to their declared bounds. Take care—if you assign a string pointer (char *) in
a data structure allocated or initialized by a Connext DDS-generated function, you should release (free) the
memory originally allocated for the string, otherwise the memory will be leaked.
To Java and .NET users, an IDL string is a String object: it is immutable and knows its own length. C and
C++ users must take care, however, as there is no way to determine how much memory is allocated to a
character pointer "string"; all that can be determined is the string's current logical length. In some cases,
Connext DDS may need to copy a string into a structure that user code has provided. Connext DDS does
not free the memory of the string provided to it, as it cannot know from where that memory was allocated.
In the C and C++ APIs, Connext DDS therefore uses the following conventions:

71

3.3.2 Value Types

l

l

l

A string's memory is "owned" by the structure that contains that string. Calling the finalization function provided for a type will free all recursively contained strings. If you have allocated a contained
string in a special way, you must be careful to clean up your own memory and assign the pointer to
NULL before calling the type’s finalize() method, so that Connext DDS will skip over that string.
You must provide a non-NULL string pointer for Connext DDS to copy into. Otherwise, Connext
DDS will log an error.
When you provide a non-NULL string pointer in your data structure, Connext DDS will copy into
the provided memory without performing any additional memory allocations. Be careful—if you
provide Connext DDS with an uninitialized pointer or allocate a string that is too short, you may corrupt the memory or cause a program crash. Connext DDS will never try to copy a string that is
longer than the bound of the destination string. However, your application must insure that any
string that it allocates is long enough.

Connext DDS provides a small set of C functions for dealing with strings. These functions simplify common tasks, avoid some platform-specific issues (such as the lack of a strdup() function on some platforms), and provide facilities for dealing with wide strings, for which no standard C library exists. Connext
DDS always uses these functions internally for managing string memory; you are recommended—but not
required—to use them as well. See the API Reference HTML documentation, which is available for all
supported programming languages (select Modules, RTI DDS API Reference, Infrastructure Module,
String Support) for more information about strings.

3.3.2 Value Types
A value type is like a structure, but with support for additional object-oriented features such as inheritance.
It is similar to what is sometimes referred to in Java as a POJO—a Plain Old Java Object.
Readers familiar with value types in the context of CORBA should consult Table 3.4 Value Type Support
to see which value type-related IDL keywords are supported and what their behavior is in the context of
Connext DDS.
Table 3.4 Value Type Support
Aspect

Level of Support in RTI Code Generator

Inheritance

Single inheritance from other value types

Public state members

Supported

Private state members

Become public when code is generated

Custom keyword

Ignored (the value type is parsed without the keyword and code is generated to work with it)

Abstract value types

No code generated (the value type is parsed, but no code is generated)

72

3.3.3 Type Codes

Table 3.4 Value Type Support
Aspect

Level of Support in RTI Code Generator

Operations

No code generated (the value type is parsed, but no code is generated)

Truncatable keyword

Ignored (the value type is parsed without the keyword and code is generated to work with it)

3.3.3 Type Codes
Type codes are enabled by default when you run RTI Code Generator. The -notypecode option disables generation of type code information. Type-code support does increase the amount of memory used,
so if you need to save on memory, you may consider disabling type codes. (The
-notypecode option is described in the RTI Code Generator User’s Manual.)
Locally, your application can access the type code for a generated type "Foo" by calling the FooTypeSupport::get_typecode() (Traditional C++ Notation) operation in the code for the type generated by RTI
Code Generator (unless type-code support is disabled with the -notypecode option).
Note: Type-code support must be enabled if you are going to use ContentFilteredTopics (Section 5.4 on
page 212) with the default SQL filter. You may disable type codes and use a custom filter, as described in
Creating ContentFilteredTopics (Section 5.4.3 on page 214).

3.3.4 Translations for IDL Types
This section describes how to specify your data types in an IDL file. RTI Code Generator supports all the
types listed in the following tables:
l

Table 3.5 Specifying Data Types in IDL for C

l

Table 3.6 Specifying Data Types in IDL for Traditional C++

l

Table 3.8 Specifying Data Types in IDL for the Modern C++ API

l

Table 3.7 Specifying Data Types in IDL for C++/CLI

l

Table 3.9 Specifying Data Types in IDL for Java

l

Table 3.10 Specifying Data Types in IDL for Ada

In each table, the middle column shows the IDL syntax for a data type in an IDL file. The rightmost
column shows the corresponding language mapping created by RTI Code Generator.

73

3.3.4 Translations for IDL Types

Table 3.5 Specifying Data Types in IDL for C
IDL
Type
char

Example Entry in IDL File

Example Output Generated by
RTI Code Generator (rtiddsgen)

struct PrimitiveStruct {
char char_member;
};

typedef struct PrimitiveStruct
{
DDS_Char char_member;
} PrimitiveStruct;

wchar

struct PrimitiveStruct {
wchar wchar_member;
};

typedef struct PrimitiveStruct
{
DDS_Wchar wchar_member;
} PrimitiveStruct;

octet

struct PrimitiveStruct {
octet octet_member;
};

typedef struct PrimitiveStruct
{
DDS_Octet octect_member;
} PrimitiveStruct;

short

struct PrimitiveStruct {
short short_member;
};

typedef struct PrimitiveStruct
{
DDS_Short short_member;
} PrimitiveStruct;

unsigned
short

struct PrimitiveStruct {
unsigned short unsigned_short_member;
};

typedef struct PrimitiveStruct
{
DDS_UnsignedShort unsigned_short_member;
} PrimitiveStruct;

long

struct PrimitiveStruct {
long long_member;
};

typedef struct PrimitiveStruct
{
DDS_Long long_member;
} PrimitiveStruct;

unsigned
long

struct PrimitiveStruct {
unsigned long unsigned_long_member;
};

typedef struct PrimitiveStruct
{
DDS_UnsignedLong unsigned_long_member;
} PrimitiveStruct;

long long

struct PrimitiveStruct {
long long long_long_member;
};

typedef struct PrimitiveStruct
{
DDS_LongLong long_long_member;
} PrimitiveStruct;

(see
Note: 1
below)

74

3.3.4 Translations for IDL Types

Table 3.5 Specifying Data Types in IDL for C
IDL
Type

Example Entry in IDL File

Example Output Generated by
RTI Code Generator (rtiddsgen)

unsigned
long long

struct PrimitiveStruct {
unsigned long long unsigned_long_long_
member;
};

typedef struct PrimitiveStruct
{
DDS_UnsignedLongLong
unsigned_long_long_member;
} PrimitiveStruct;

float

struct PrimitiveStruct {
float float_member;
};

typedef struct PrimitiveStruct
{
DDS_Float float_member;
} PrimitiveStruct;

double

struct PrimitiveStruct {
double double_member;
};

typedef struct PrimitiveStruct
{
DDS_Double double_member;
} PrimitiveStruct;

struct PrimitiveStruct {
long double
long_double_member;
};

typedef struct PrimitiveStruct
{
DDS_LongDouble long_double_member;
} PrimitiveStruct;

struct MyStruct {
long * member;
};

typedef struct MyStruct {
DDS_Long * member;
} MyStruct;

struct PrimitiveStruct {
boolean boolean_member;
};

typedef struct PrimitiveStruct
{
DDS_Boolean boolean_member;
} PrimitiveStruct;

long
double
(see
Note: 2
below)
pointer
(see
Note: 9
below)

boolean

enum PrimitiveEnum {
ENUM1,
ENUM2,
ENUM3
};

enum

enum PrimitiveEnum {
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
};

typedef enum PrimitiveEnum
{
ENUM1,
ENUM2,
ENUM3
} PrimitiveEnum;
typedef enum PrimitiveEnum
{
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
} PrimitiveEnum;

75

3.3.4 Translations for IDL Types

Table 3.5 Specifying Data Types in IDL for C
IDL
Type
constant

Example Entry in IDL File

Example Output Generated by
RTI Code Generator (rtiddsgen)

const short SIZE = 5;

#define SIZE 5

struct PrimitiveStruct {
char char_member;
};

typedef struct PrimitiveStruct
{
char char_member;
} PrimitiveStruct;

union PrimitiveUnion switch (long){
case 1:
short short_member;
default:
long long_member;
};

typedef struct PrimitiveUnion
{
DDS_Long _d;
struct {
DDS_Short short_member;
DDS_Long long_member;
} _u;
} PrimitiveUnion;

typedef short TypedefShort;

typedef DDS_Short TypedefShort;

struct OneDArrayStruct {
short short_array[2];
};

typedef struct OneDArrayStruct
{
DDS_Short short_array[2];
} OneDArrayStruct;

struct

(see
Note: 10
below)
union

(see
Note: 3
and
Note: 10
below)
typedef

array of
above
types

bounded
sequence
of above
types

(see
Note: 11
and
Note: 15
below)

struct TwoDArrayStruct {
short short_array[1][2];
};

struct SequenceStruct {
sequence
short_sequence;
};

typedef struct TwoDArrayStruct
{
DDS_Short short_array[1][2];
} TwoDArrayStruct;

typedef struct SequenceStruct
{
DDSShortSeq short_sequence;
} SequenceStruct;

Note: Sequences of primitive types have been predefined by
Connext DDS.

76

3.3.4 Translations for IDL Types

Table 3.5 Specifying Data Types in IDL for C
IDL
Type
unbounde
d
sequence
of above
types

Example Entry in IDL File

struct SequenceStruct {
sequence short_sequence;
};

(see
Note: 11
and
Note: 15
below)

array of
sequences

Example Output Generated by
RTI Code Generator (rtiddsgen)

typedef struct SequenceStruct
{
DDSShortSeq short_sequence;
} SequenceStruct;

See Note: 12 below.

struct ArraysOfSequences{
sequence
sequences_array[2];
};

typedef struct ArraysOfSequences
{
DDS_ShortSeq sequences_array[2];
} ArraysOfSequences;

typedef DDS_Short ShortArray[2];

sequence
of arrays

(see
Note: 11
below)

sequence
of
sequences

(see
Note: 4
and
Note: 11
below)

DDS_SEQUENCE_NO_GET(ShortArraySeq,ShortArray);
typedef short ShortArray[2];
struct SequenceofArrays {
sequence
arrays_sequence;
};

typedef struct SequenceOfArrays
{
ShortArraySeq arrays_sequence;
} SequenceOfArrays;

DDS_SEQUENCE_NO_GET is a Connext DDS macro that
defines a new sequence type for a user data type. In this case, the
user data type is ShortArray.
typedef sequence
ShortSequence;
typedef DDS_ShortSeq ShortSequence;
struct SequencesOfSequences{
sequence
sequences_sequence;
};

struct PrimitiveStruct {

bounded
string

string<20> string_member;
};

DDS_SEQUENCE(ShortSequenceSeq, ShortSequence);
typedef struct SequencesOfSequences{
ShortSequenceSeq sequences_sequence;
} SequencesOfSequences;

typedef struct PrimitiveStruct {
char* string_member; /* maximum length =
(20) */
} PrimitiveStruct;

77

3.3.4 Translations for IDL Types

Table 3.5 Specifying Data Types in IDL for C
IDL
Type

unbounde
d string

Example Output Generated by
RTI Code Generator (rtiddsgen)

Example Entry in IDL File

struct PrimitiveStruct {
string string_member;
};

typedef struct PrimitiveStruct {
char* string_member; /* maximum length =
(255) */
} PrimitiveStruct;

See Note: 12 below.
typedef struct PrimitiveStruct {

bounded
wstring

DDS_Wchar * wstring_member;

struct PrimitiveStruct {
wstring<20> wstring_member;
};

/* maximum length = (20) */
} PrimitiveStruct;

unbounde
d wstring

struct PrimitiveStruct {
wstring wstring_member;
};

typedef struct PrimitiveStruct {
DDS_Wchar * wstring_member;
/* maximum length = (255) */
} PrimitiveStruct;

See Note: 12 below.
With the -namespace option (only available for C++):
namespace PackageName{

module

module PackageName {
struct Foo {
long field;
};
};

typedef struct Foo {
DDS_Long field;
} Foo;
};
Without the -namespace option:
typedef struct PackageName_Foo {
DDS_Long field;
} PackageName_Foo;

78

3.3.4 Translations for IDL Types

Table 3.5 Specifying Data Types in IDL for C
IDL
Type

valuetype

(see
Note: 9
and
Note: 10
below)

Example Entry in IDL File

valuetype MyValueType {
public MyValueType2 * member;
};
valuetype MyValueType {
public MyValueType2 member;
};
valuetype MyValueType:
MyBaseValueType {
public MyValueType2 * member;
};

Example Output Generated by
RTI Code Generator (rtiddsgen)
typedef struct MyValueType {
MyValueType2 * member;
} MyValueType;
typedef struct MyValueType {
MyValueType2 member;
} MyValueType;
typedef struct MyValueType
{
MyBaseValueType parent;
MyValueType2 * member;
} MyValueType;

Table 3.6 Specifying Data Types in IDL for Traditional C++
IDL
Type
char

Example Entry in IDL File

Example Output Generated by
RTI Code Generator (rtiddsgen)

struct PrimitiveStruct {
char char_member;
};

class PrimitiveStruct
{
DDS_Char char_member;
} PrimitiveStruct;

wchar

struct PrimitiveStruct {
wchar wchar_member;
};

class PrimitiveStruct
{
DDS_Wchar wchar_member;
} PrimitiveStruct;

octet

struct PrimitiveStruct {
octet octet_member;
};

class PrimitiveStruct
{
DDS_Octet octect_member;
} PrimitiveStruct;

short

struct PrimitiveStruct {
short short_member;
};

class PrimitiveStruct
{
DDS_Short short_member;
} PrimitiveStruct;

(see
Note: 1
below)

79

3.3.4 Translations for IDL Types

Table 3.6 Specifying Data Types in IDL for Traditional C++
IDL
Type

Example Entry in IDL File

Example Output Generated by
RTI Code Generator (rtiddsgen)

unsigned
short

struct PrimitiveStruct {
unsigned short unsigned_short_member;
};

class PrimitiveStruct
{
DDS_UnsignedShort unsigned_short_member;
} PrimitiveStruct;

long

struct PrimitiveStruct {
long long_member;
};

class PrimitiveStruct
{
DDS_Long long_member;
} PrimitiveStruct;

unsigned
long

struct PrimitiveStruct {
unsigned long unsigned_long_member;
};

class PrimitiveStruct
{
DDS_UnsignedLong unsigned_long_member;
} PrimitiveStruct;

long long

struct PrimitiveStruct {
long long long_long_member;
};

class PrimitiveStruct
{
DDS_LongLong long_long_member;
} PrimitiveStruct;

unsigned
long long

struct PrimitiveStruct {
unsigned long long unsigned_long_long_
member;
};

class PrimitiveStruct
{
DDS_UnsignedLongLong
unsigned_long_long_member;
} PrimitiveStruct;

float

struct PrimitiveStruct {
float float_member;
};

typedef struct PrimitiveStruct
{
DDS_Float float_member;
} PrimitiveStruct;

double

struct PrimitiveStruct {
double double_member;
};

class PrimitiveStruct
{
DDS_Double double_member;
} PrimitiveStruct;

struct PrimitiveStruct {
long double
long_double_member;
};

class PrimitiveStruct
{
DDS_LongDouble long_double_member;
} PrimitiveStruct;

long
double
(see
Note: 2
below)

80

3.3.4 Translations for IDL Types

Table 3.6 Specifying Data Types in IDL for Traditional C++
IDL
Type
pointer
(see
Note: 9
below)

boolean

Example Entry in IDL File

struct MyStruct {
long * member;
};

class MyStruct {
DDS_Long * member;
} MyStruct;

struct PrimitiveStruct {
boolean boolean_member;
};

class PrimitiveStruct
{
DDS_Boolean boolean_member;
} PrimitiveStruct;

enum PrimitiveEnum {
ENUM1,
ENUM2,
ENUM3
};

enum

constant

Example Output Generated by
RTI Code Generator (rtiddsgen)

typedef enum PrimitiveEnum
{
ENUM1,
ENUM2,
ENUM3
} PrimitiveEnum;

enum PrimitiveEnum {
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
};

typedef enum PrimitiveEnum
{
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
} PrimitiveEnum;

const short SIZE = 5;

static const DDS_Short size = 5;

struct PrimitiveStruct {
char char_member;
};

typedef struct PrimitiveStruct
{
char char_member;
} PrimitiveStruct;

union PrimitiveUnion switch (long){
case 1:
short short_member;
default:
long long_member;
};

class PrimitiveUnion
{
DDS_Long _d;
class{
DDS_Short short_member;
DDS_Long long_member;
} _u;
} PrimitiveUnion;

typedef short TypedefShort;

typedef DDS_Short TypedefShort;

struct

(see
Note: 10
below)
union

(see
Note: 3
and
Note: 10
below)
typedef

81

3.3.4 Translations for IDL Types

Table 3.6 Specifying Data Types in IDL for Traditional C++
IDL
Type

array of
above
types

bounded
sequence
of above
types

(see
Note: 11
and
Note: 15
below)
unbounde
d
sequence
of above
types

Example Entry in IDL File

struct OneDArrayStruct {
short short_array[2];
};
struct TwoDArrayStruct {
short short_array[1][2];
};

struct SequenceStruct {
sequence
short_sequence;
};

class OneDArrayStruct
{
DDS_Short short_array[2];
} OneDArrayStruct;
class TwoDArrayStruct
{
DDS_Short short_array[1][2];
} TwoDArrayStruct;

class SequenceStruct
{
DDSShortSeq short_sequence;
} SequenceStruct;

Note: Sequences of primitive types have been predefined by
Connext DDS.

struct SequenceStruct {
sequence short_sequence;
};

(see
Note: 11
and
Note: 15
below)

array of
sequences

Example Output Generated by
RTI Code Generator (rtiddsgen)

typedef struct SequenceStruct
{
DDSShortSeq short_sequence;
} SequenceStruct;

See Note: 12 below.

struct ArraysOfSequences{
sequence
sequences_array[2];
};

class ArraysOfSequences
{
DDS_ShortSeq sequences_array[2];
} ArraysOfSequences;

82

3.3.4 Translations for IDL Types

Table 3.6 Specifying Data Types in IDL for Traditional C++
IDL
Type

Example Entry in IDL File

Example Output Generated by
RTI Code Generator (rtiddsgen)
typedef DDS_Short ShortArray[2];
DDS_SEQUENCE_NO_GET(ShortArraySeq,
ShortArray);

sequence
of arrays

(see
Note: 11
below)

typedef short ShortArray[2];
struct SequenceofArrays {
sequence
arrays_sequence;
};

class SequenceOfArrays
{
ShortArraySeq arrays_sequence;
} SequenceOfArrays;

DDS_SEQUENCE_NO_GET is a Connext DDS macro that
defines a new sequence type for a user data type. In this case, the
user data type is ShortArray.
sequence
of
sequences

(see
Note: 4
and
Note: 11
below)

bounded
string

unbounde
d string

typedef sequence
ShortSequence;
struct SequencesOfSequences{
sequence
sequences_sequence;
};

struct PrimitiveStruct {
string<20> string_member;
};

struct PrimitiveStruct {
string string_member;
};

typedef DDS_ShortSeq ShortSequence;
DDS_SEQUENCE(ShortSequenceSeq, ShortSequence);
class SequencesOfSequences{
ShortSequenceSeq sequences_sequence;
} SequencesOfSequences;

class PrimitiveStruct {
char* string_member; /* maximum length =
(20) */
} PrimitiveStruct;

class PrimitiveStruct {
char* string_member; /* maximum length =
(255) */
} PrimitiveStruct;

See Note: 12 below.

bounded
wstring

struct PrimitiveStruct {
wstring<20> wstring_member;
};

class PrimitiveStruct {
DDS_Wchar * wstring_member;
/* maximum length = (20) */
} PrimitiveStruct;

83

3.3.4 Translations for IDL Types

Table 3.6 Specifying Data Types in IDL for Traditional C++
IDL
Type

unbounde
d wstring

Example Entry in IDL File

struct PrimitiveStruct {
wstring wstring_member;
};

Example Output Generated by
RTI Code Generator (rtiddsgen)
class PrimitiveStruct {
DDS_Wchar * wstring_member;
/* maximum length = (255) */
} PrimitiveStruct;

See Note: 12 below.
With the -namespace option (only available for C++):

module

module PackageName {
struct Foo {
long field;
};
};

namespace PackageName{
typedef struct Foo {
DDS_Long field;
} Foo;
};
Without the -namespace option:
class PackageName_Foo {
DDS_Long field;
} PackageName_Foo;

valuetype

(see
Note: 9
and
Note: 10
below)

valuetype MyValueType {
public MyValueType2 * member;
};
valuetype MyValueType {
public MyValueType2 member;
};
valuetype MyValueType:
MyBaseValueType {
public MyValueType2 * member;
};

class MyValueType {
public:
MyValueType2 * member;
};
class MyValueType {
public:
MyValueType2 member;
};
class MyValueType : public MyBaseValueType
{
public:
MyValueType2 * member;
};

84

3.3.4 Translations for IDL Types

Table 3.7 Specifying Data Types in IDL for C++/CLI
IDL Type

Example Entry in IDL File

Example Output Generated by
RTI Code Generator (rtiddsgen)

struct PrimitiveStruct {
char char_member;
};

public ref class PrimitiveStruct {
System::Char char_member;
};

wchar

struct PrimitiveStruct {
wchar wchar_member;
};

public ref class PrimitiveStruct {
System::Char wchar_member;
};

octet

struct PrimitiveStruct {
octet octet_member;
};

public ref class PrimitiveStruct {
System::Byte octet_member;
};

short

struct PrimitiveStruct {
short short_member;
};

public ref class PrimitiveStruct {
System::Int16 short_member;
};

unsigned short

struct PrimitiveStruct {
unsigned short
unsigned_short_member;
};

public ref class PrimitiveStruct {
System::UInt16
unsigned_short_member;
};

long

struct PrimitiveStruct {
long long_member;
};

public ref class PrimitiveStruct {
System::Int32 long_member;
};

unsigned long

struct PrimitiveStruct {
unsigned long
unsigned_long_member;
};

public ref class PrimitiveStruct {
System::UInt32
unsigned_long_member;
};

long long

struct PrimitiveStruct {
long long long_
long_member;
};

public ref class PrimitiveStruct {
System::Int64
long_long_member;
};

unsigned long long

struct PrimitiveStruct {
unsigned long long
unsigned_long_long_member;
};

public ref class PrimitiveStruct {
System::UInt64
unsigned_long_long_member;
};

char
(see Note: 1 below)

85

3.3.4 Translations for IDL Types

Table 3.7 Specifying Data Types in IDL for C++/CLI
IDL Type

Example Entry in IDL File

Example Output Generated by
RTI Code Generator (rtiddsgen)

float

struct PrimitiveStruct {
float float_member;
};

public ref class PrimitiveStruct {
System::Single
float_member;
};

double

struct PrimitiveStruct {
double double_member;
};

public ref class PrimitiveStruct {
System::Double
double_member;
} PrimitiveStruct;

struct PrimitiveStruct {
long double
long_double_member;
};

public ref class PrimitiveStruct {
DDS::LongDouble
long_double_member;
} PrimitiveStruct;

struct PrimitiveStruct {
boolean boolean_member;
};

public ref class PrimitiveStruct {
System::Boolean
boolean_member;
};

long double

(see Note: 2 below)

boolean

enum PrimitiveEnum {
ENUM1,
ENUM2,
ENUM3

enum

public enum class
PrimitiveEnum : System::Int32 {
ENUM1,
ENUM2,
ENUM3
};

};
enum PrimitiveEnum {
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
};

public enum class
PrimitiveEnum : System::Int32 {
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
};

constant

const short SIZE = 5;

public ref class SIZE {
public:
static System::Int16
VALUE = 5;
};

struct

struct PrimitiveStruct {
char char_member;
};

public ref class PrimitiveStruct {
System::Char char_member;
};

(see Note: 10 below)

86

3.3.4 Translations for IDL Types

Table 3.7 Specifying Data Types in IDL for C++/CLI
IDL Type

union

(see Note: 3 and Note: 10 below)

Example Entry in IDL File

union PrimitiveUnion switch (long)
{
case 1:
short short_member;
default:
long long_member;
};

Example Output Generated by
RTI Code Generator (rtiddsgen)
public ref class PrimitiveUnion
{
System::Int32 _d;
struct PrimitiveUnion_u {
System::Int16 short_member;
System::Int32 long_member;
} _u;
};

array of above types

bounded sequence of above types

(see Note: 11 and Note: 15 below)

unbounded sequence of above types

(see Note: 11 and Note: 15 below)

struct OneDArrayStruct {
short short_array[2];
};

struct SequenceStruct {
sequence
short_sequence;
};

struct SequenceStruct {
sequence
short_sequence;
};

public ref class OneDArrayStruct {
array^
short_array; /*length == 2*/
};

public ref class SequenceStruct {
ShortSeq^ short_sequence;
/*max = 4*/
};

Note: Sequences of primitive types
have been predefined by
Connext DDS
public ref class SequenceStruct {
ShortSeq^ short_sequence;
/*max = */
};

See Note: 12 below.

array of sequences

struct ArraysOfSequences{
sequence
sequences_array[2];
};

public ref class ArraysOfSequences
{
array^
sequences_array;
// maximum length = (2)
};

bounded string

struct PrimitiveStruct {
string<20> string_member;
};

public ref class PrimitiveStruct {
System::String^ string_member;
// maximum length = (20)
};

87

3.3.4 Translations for IDL Types

Table 3.7 Specifying Data Types in IDL for C++/CLI
IDL Type

unbounded string

Example Entry in IDL File

struct PrimitiveStruct {
string string_member;
};

Example Output Generated by
RTI Code Generator (rtiddsgen)
public ref class PrimitiveStruct {
System::String^ string_member;
// maximum length = (255)
};

See Note: 12 below.

bounded wstring

unbounded wstring

struct PrimitiveStruct {
wstring<20> wstring_member;
};

struct PrimitiveStruct {
wstring wstring_member;
};

public ref class PrimitiveStruct {
System::String^ string_member;
// maximum length = (20)
};

public ref class PrimitiveStruct {
System::String^ string_member;
// maximum length = (255)
};

See Note: 12 below.
module PackageName {
struct Foo {
long field;
};
};

module

namespace PackageName {
public ref class Foo {
System::Int32 field;
};
};

Table 3.8 Specifying Data Types in IDL for the Modern C++ API
IDL
Type
char
(see
Note: 1
below)

wchar

Example Entry in IDL
File

Example Output Generated by
RTI Code Generator (rtiddsgen)

struct PrimitiveStruct {
char char_member;
};

class PrimitiveStruct {
public:
char char_member() const OMG_NOEXCEPT;
void char_member(char value);
}

struct PrimitiveStruct {
wchar wchar_member;
};

class PrimitiveStruct {
public:
DDS_Wchar wchar_member() const OMG_NOEXCEPT;
void wchar_member(DDS_Wchar value);
};

88

3.3.4 Translations for IDL Types

Table 3.8 Specifying Data Types in IDL for the Modern C++ API
IDL
Type

Example Entry in IDL
File

octet

struct PrimitiveStruct {
octet octet_member;
};

class PrimitiveStruct {
public:
uint8_t octet_member() const OMG_NOEXCEPT;
void octet_member(uint8_t value);
};

short

struct PrimitiveStruct {
short short_member;
};

class PrimitiveStruct {
public:
int16_t short_member() const OMG_NOEXCEPT;
void short_member(int16_t value);
};

unsigne
d short

struct PrimitiveStruct {
unsigned short
unsigned_short_
member;
};

class PrimitiveStruct {
public:
uint16_t unsigned_short_member() const OMG_NOEXCEPT;
void unsigned_short_member(uint16_t value);
};

long

struct PrimitiveStruct {
long long_member;
};

class PrimitiveStruct {
public:
int32_t long_member() const OMG_NOEXCEPT;
void long_member(int32_t value);
};

unsigne
d long

struct PrimitiveStruct {
unsigned long
unsigned_long_
member;
};

class PrimitiveStruct {
public:
uint32_t long_member() const OMG_NOEXCEPT;
void unsigned_long_member(uint32_t value);
};

long
long

struct PrimitiveStruct {
long long
long_long_member;
};

class PrimitiveStruct {
public:
rti::core::int64 long_long_member() const OMG_NOEXCEPT;
void long_long_member(rti::core::int64 value);
};

unsigne
d long
long

struct PrimitiveStruct {
unsigned long long
unsigned_long_long_
member;
};

class PrimitiveStruct {
public:
rti::core::uint64 unsigned_long_long_member);
rti::core::uint64 unsigned_long_long_member() const OMG_NOEXCEPT;
};

Example Output Generated by
RTI Code Generator (rtiddsgen)

89

3.3.4 Translations for IDL Types

Table 3.8 Specifying Data Types in IDL for the Modern C++ API
IDL
Type

Example Entry in IDL
File

float

struct PrimitiveStruct {
float float_member;
};

class PrimitiveStruct {
public:
float float_member() const OMG_NOEXCEPT;
void float_member(float value);
};

double

struct PrimitiveStruct {
double double_
member;
};

class PrimitiveStruct {
public:
double double_member() const OMG_NOEXCEPT;
void double_member(double value);
};

struct PrimitiveStruct {

long
double
(see
Note: 2
below)

pointer
(see
Note: 9
below)

boolean

long double long_
double_member;
};

Example Output Generated by
RTI Code Generator (rtiddsgen)

class PrimitiveStruct {
public:
rti::core::LongDouble& long_double_member() OMG_NOEXCEPT;
const rti::core::LongDouble& long_double_member() const OMG_
NOEXCEPT;
void long_double_member(const rti::core::LongDouble& value);
}

struct MyStruct {
long * member;
};

class PrimitiveStruct {
int32_t * member() const OMG_NOEXCEPT;
void member(int32_t * value);
};

struct PrimitiveStruct {
boolean boolean_
member;
};

class PrimitiveStruct {
public:
bool boolean_member() const OMG_NOEXCEPT;
void boolean_member(bool value);
};

90

3.3.4 Translations for IDL Types

Table 3.8 Specifying Data Types in IDL for the Modern C++ API
IDL
Type

Example Entry in IDL
File

enum PrimitiveEnum {
ENUM1,
ENUM2,
ENUM3
};

enum

enum PrimitiveEnum {
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
};

Example Output Generated by
RTI Code Generator (rtiddsgen)
struct PrimitiveEnum_def {
enum type {
ENUM1,
ENUM2,
ENUM3
};
};
typedef dds::core::safe_enum PrimitiveEnum;
struct PrimitiveEnum_def {
enum type {
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
};
};
typedef dds::core::safe_enum PrimitiveEnum;

constant

const short SIZE = 5;

static const int16_t SIZE = 5;

struct PrimitiveStruct {
char char_member;
};

class PrimitiveStruct {
public:
....
char char_member() const OMG_NOEXCEPT;
void char_member(char value);
}

struct

(see
Note: 1
0 and
Note: 1
4
below)

91

3.3.4 Translations for IDL Types

Table 3.8 Specifying Data Types in IDL for the Modern C++ API
IDL
Type

union

(see
Note: 3
and
Note: 1
0
below)

typedef

Example Entry in IDL
File

union PrimitiveUnion
switch (long){
case 1:
short short_
member;
default:
long long_
member;
};

typedef short
TypedefShort;

struct OneDArrayStruct {
short short_array[2];
};

array of
above
types

struct TwoDArrayStruct {
short short_array[1]
[2];
};

Example Output Generated by
RTI Code Generator (rtiddsgen)
class PrimitiveUnion {
public:
int32_t _d() const ;
void _d(int32_t value);
int16_t short_member() const ;
void short_member(int16_t value);
int32_t long_member() const ;
void long_member(int32_t value);
static int32_t default_discriminator();
private:
int32_t m_d_;
struct Union_ {
int16_t m_short_member_;
int32_t m_long_member_;
Union_();
Union_(
int16_t short_member,
int32_t long_member);
};
Union_ m_u_;
};

typedef int16_t TypedefShort;
struct TypedefShort_AliasTag_t {};

class OneDArrayStruct {
public:
dds::core::array& short_array() OMG_NOEXCEPT;
const dds::core::array& short_array() const OMG_
NOEXCEPT;
void short_array(const dds::core::array& value);
};
class TwoDArrayStruct {
public:
dds::core::array, 1>& short_array()
OMG_NOEXCEPT;
const dds::core::array, 1>& short_
array() const OMG_NOEXCEPT;
void short_array(const dds::core::array, 1>& value);
};

92

3.3.4 Translations for IDL Types

Table 3.8 Specifying Data Types in IDL for the Modern C++ API
IDL
Type
bounde
d
sequenc
e of
above
types

Example Entry in IDL
File

struct SequenceStruct {
sequence
short_sequence;
};

(see
Note: 1
1
below)
unboun
ded
sequenc
e of
above
types

(see
Note: 1
1 and
Note: 1
5
below)

array of
sequenc
es

struct SequenceStruct {
sequence
short_sequence;
};

Example Output Generated by
RTI Code Generator (rtiddsgen)

class SequenceStruct {
public:
dds::core::vector& short_sequence() OMG_NOEXCEPT;
const dds::core::vector& short_sequence() const OMG_
NOEXCEPT;
void short_sequence(const dds::core::vector& value);
};

class SequenceStruct {
public:
dds::core::vector& short_sequence() OMG_NOEXCEPT;
const dds::core::vector& short_sequence() const OMG_
NOEXCEPT;
void short_sequence(const dds::core::vector& value);
};

See Note: 12 below.

struct ArraysOfSequences
{
sequence
sequences_array
[2];
};

class ArraysOfSequences {
public:
dds::core::array, 2>& sequences_array()
OMG_NOEXCEPT;
const dds::core::array, 2>& sequences_
array() const OMG_NOEXCEPT;
void sequences_array(const
dds::core::array, 2>& value);
};

93

3.3.4 Translations for IDL Types

Table 3.8 Specifying Data Types in IDL for the Modern C++ API
IDL
Type

Example Entry in IDL
File

sequenc
e of
arrays

typedef short ShortArray
[2];

(see
Note: 1
1 and
Note: 1
5
below)

sequenc
e of
sequenc
es

(see
Note: 4
and
Note: 1
1
below)

bounde
d string

unboun
ded
string

struct SequenceofArrays
{
sequence
arrays_sequence;

Example Output Generated by
RTI Code Generator (rtiddsgen)

typedef dds::core::array ShortArray;
class SequenceofArrays {
public:
dds::core::vector& arrays_sequence() OMG_NOEXCEPT;
const dds::core::vector& arrays_sequence() const OMG_
NOEXCEPT;
void arrays_sequence(const dds::core::vector& value);
};

};

typedef
sequence
ShortSequence;
struct
SequencesOfSequences{
sequence
sequences_
sequence;
};

struct PrimitiveStruct {
string<20> string_
member;
};

struct PrimitiveStruct {
string string_
member;
};

typedef dds::core::vector ShortSequence;
class SequencesOfSequences {
public:
dds::core::vector& sequences_sequence() OMG_
NOEXCEPT;
const dds::core::vector& sequences_sequence() const
OMG_NOEXCEPT;
void sequences_sequence(const dds::core::vector&
value);
};

class PrimitiveStruct {
public:
dds::core::string& string_member() OMG_NOEXCEPT;
const dds::core::string& string_member() const OMG_NOEXCEPT;
void string_member(const dds::core::string& value);
};

class PrimitiveStruct {
public:
dds::core::string& string_member() OMG_NOEXCEPT;
const dds::core::string& string_member() const OMG_NOEXCEPT;
void string_member(const dds::core::string& value);
};

See Note: 12 below.

94

3.3.4 Translations for IDL Types

Table 3.8 Specifying Data Types in IDL for the Modern C++ API
IDL
Type

Example Entry in IDL
File

bounde
d
wstring

struct PrimitiveStruct {
wstring<20> wstring_
member;
};

unboun
ded
wstring

struct PrimitiveStruct {
wstring wstring_
member;
};

Example Output Generated by
RTI Code Generator (rtiddsgen)
class PrimitiveStruct {
public:
dds::core::wstring& wstring_member() OMG_NOEXCEPT;
const dds::core::wstring& wstring_member() const OMG_NOEXCEPT;
void wstring_member(const dds::core::wstring& value);
};

class PrimitiveStruct {
public:
dds::core::wstring& wstring_member() OMG_NOEXCEPT;
const dds::core::wstring& wstring_member() const OMG_NOEXCEPT;
void wstring_member(const dds::core::wstring& value);
};

See Note: 12 below.

module

valuety
pe

(see
Note: 9
and
Note: 1
0
below)

module PackageName {
struct Foo {
long field;
};
};

namespace PackageName {
class Foo {
public:
int32_t field() const OMG_NOEXCEPT;
void field(int32_t value);
};
};

valuetype
MyBaseValueType
public long
};

class MyBaseValueType {
public:
int32_t member() const OMG_NOEXCEPT;
void member(int32_t value);
};

{
member;

valuetype MyValueType:
MyBaseValueType {
public short *
member2;
};

class MyValueType : public MyBaseValueType {
public:
int16_t * member2() const OMG_NOEXCEPT;
void member2(int16_t * value);
};

95

3.3.4 Translations for IDL Types

Table 3.9 Specifying Data Types in IDL for Java
IDL
Type

Example Entry in IDL file

Example Java Output Generated by
RTI Code Generator (rtiddsgen)

struct PrimitiveStruct {
char char_member;
};

public class PrimitiveStruct
{
public char char_member;
...
}

struct PrimitiveStruct {
wchar wchar_member;
};

public class PrimitiveStruct
{
public char wchar_member;
...
}

octet

struct PrimitiveStruct {
octet octet_member;
};

public class PrimitiveStruct
{
public byte byte_member;
...
}

short

struct PrimitiveStruct {
short short_member;
};

public class PrimitiveStruct
{
public short short_member;
...
}

struct PrimitiveStruct {
unsigned short
unsigned_short_member;
};

public class PrimitiveStruct
{
public short
unsigned_short_member;
...
}

struct PrimitiveStruct {
long long_member;
};

public class PrimitiveStruct
{
public int long_member;
...
}

struct PrimitiveStruct {
unsigned long
unsigned_long_member;
};

public class PrimitiveStruct
{
public int
unsigned_long_member;
...
}

char

(see Note: 5
below)
wchar

(see Note: 5
below)

unsigned
short

(see Note: 6
below)

long

unsigned
long

(see Note: 6
below)

96

3.3.4 Translations for IDL Types

Table 3.9 Specifying Data Types in IDL for Java
IDL
Type

Example Entry in IDL file

Example Java Output Generated by
RTI Code Generator (rtiddsgen)

struct PrimitiveStruct {
long long
long_long_member;
};

public class PrimitiveStruct
{
public long
long_long_member;
...
}

struct PrimitiveStruct {
unsigned long long
unsigned_long_long_member;
};

public class PrimitiveStruct
{
public long
unsigned_long_long_member;
...
}

float

struct PrimitiveStruct {
float float_member;
};

public class PrimitiveStruct
{
public float float_member;
...
}

double

struct PrimitiveStruct {
double double_member;
};

public class PrimitiveStruct
{
public double double_member;
...
}

struct PrimitiveStruct {
long double long_double_member;
};

public class PrimitiveStruct
{
public double long_double_member;
...
}

struct MyStruct {
long * member;
};

public class MyStruct {
public int member;
...
};

struct PrimitiveStruct {
boolean boolean_member;
};

public class PrimitiveStruct
{
public boolean boolean_member;
...
}

long long

unsigned
long long

(see Note: 7
below)

long double

(see Note: 7
below)

pointer
(see Note: 9
below)

boolean

97

3.3.4 Translations for IDL Types

Table 3.9 Specifying Data Types in IDL for Java
IDL
Type

Example Entry in IDL file

Example Java Output Generated by
RTI Code Generator (rtiddsgen)

enum PrimitiveEnum {
ENUM1,
ENUM2,
ENUM3
};

public class PrimitiveEnum extends Enum
{
public static PrimitiveEnum ENUM1 =
new PrimitiveEnum ("ENUM1", 0);
public static PrimitiveEnum ENUM2 =
new PrimitiveEnum ("ENUM2", 1);
public static PrimitiveEnum ENUM3 =
new PrimitiveEnum ("ENUM3", 2);
public static PrimitiveEnum
valueOf(int ordinal);
...
}

enum PrimitiveEnum {
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
};

public class PrimitiveEnum extends Enum
{
public static PrimitiveEnum ENUM1 =
new PrimitiveEnum ("ENUM1", 10);
public static PrimitiveEnum ENUM2 =
new PrimitiveEnum ("ENUM2", 10);
public static PrimitiveEnum ENUM3 =
new PrimitiveEnum ("ENUM3", 20);
public static PrimitiveEnum
valueOf(int ordinal);
...
}

const short SIZE = 5;

public class SIZE {
public static final short VALUE = 5;
}

struct PrimitiveStruct {
char char_member;
};

public class PrimitiveStruct
{
public char char_member;
}

union PrimitiveUnion switch (long){
case 1:
short short_member;
default:
long long_member;
};

public class PrimitiveUnion {
public int _d;
public short short_member;
public int long_member;
...
}

enum

constant

struct

(see
Note: 10
below)
union

(see
Note: 10
below)

98

3.3.4 Translations for IDL Types

Table 3.9 Specifying Data Types in IDL for Java
IDL
Type
typedef of
primitives,
enums,
strings

Example Entry in IDL file

typedef short ShortType;
struct PrimitiveStruct {
ShortType short_member;
};

(see Note: 8
below)

Example Java Output Generated by
RTI Code Generator (rtiddsgen)
/* typedefs are unwounded to the original
type when used */
public class PrimitiveStruct
{
public short short_member;
...
}

typedef short ShortArray[2];

/* Wrapper class */
public class ShortArray
{
public short[] userData = new
short[2];
...
}

struct OneDArrayStruct {
short short_array[2];
};

public class OneDArrayStruct
{
public short[] short_array = new
short[2];
...
}

struct TwoDArrayStruct {
short short_array[1][2];
};

public class TwoDArrayStruct
{
public short[][] short_array = new
short[1][2];
...
}

typedef of
sequences
or arrays

(see Note: 8
below)

array

bounded
sequence

(see
Note: 11
and
Note: 15
below)

struct SequenceStruct {
sequence
short_sequence;
};

public class SequenceStruct
{
public ShortSeq short_sequence = new
ShortSeq((4));
...
}

Note: Sequences of primitive types have been predefined by Connext
DDS.

99

3.3.4 Translations for IDL Types

Table 3.9 Specifying Data Types in IDL for Java
IDL
Type

Example Entry in IDL file

unbounded
sequence

(see
Note: 11
and
Note: 15
below)

array of
sequences

struct SequenceStruct {
sequence short_sequence;
};

Example Java Output Generated by
RTI Code Generator (rtiddsgen)
public class SequenceStruct
{
public ShortSeq short_sequence = new
ShortSeq((100));
...
}

See Note: 12 below.

struct ArraysOfSequences{
sequence
sequences_array[2];
};

public class ArraysOfSequences
{
public ShortSeq[] sequences_array =
new ShortSeq[2];
...
}

/* Wrapper class */
public class ShortArray
{
public short[] userData = new
short[2];
...
}

sequence of
arrays

(see
Note: 11
below)

typedef short ShortArray[2];
struct SequenceOfArrays{
sequence
arrays_sequence;
};

/* Sequence of wrapper class objects */
public final class ShortArraySeq
extends ArraySequence
{
...
}
public class SequenceOfArrays
{
public ShortArraySeq arrays_sequence
= new ShortArraySeq((2));
...
}

100

3.3.4 Translations for IDL Types

Table 3.9 Specifying Data Types in IDL for Java
IDL
Type

Example Entry in IDL file

Example Java Output Generated by
RTI Code Generator (rtiddsgen)
/* Wrapper class */
public class ShortSequence
{
public ShortSeq userData = new
ShortSeq((4));
...
}

sequence of
sequences

(see Note: 4
and
Note: 11
below)

bounded
string

unbounded
string

typedef sequence
ShortSequence;
struct SequencesOfSequences{
sequence
sequences_sequence;
};

/* Sequence of wrapper class objects */
public final class ShortSequenceSeq
extends ArraySequence
{
...
}
public class SequencesOfSequences
{
public ShortSequenceSeq
sequences_sequence = new
ShortSequenceSeq((2));
...
}

struct PrimitiveStruct {
string<20> string_member;
};

struct PrimitiveStruct {
string string_member;
};

public class PrimitiveStruct
{
public String string_member = new
String();
/* maximum length = (20) */
...
}

public class PrimitiveStruct
{
public String string_member = new String();
* maximum length = (255) */
...
}

See Note: 12 below.

bounded
wstring

struct PrimitiveStruct {
wstring<20> wstring_member;
};

public class PrimitiveStruct
{
public String wstring_member = new String();
/* maximum length = (20) */
...
}

101

3.3.4 Translations for IDL Types

Table 3.9 Specifying Data Types in IDL for Java
IDL
Type

unbounded
wstring

Example Java Output Generated by
RTI Code Generator (rtiddsgen)

Example Entry in IDL file

struct PrimitiveStruct {
wstring wstring_member;
};

public class PrimitiveStruct
{
public String wstring_member = new String();
/* maximum length = (255) */
...
}

See Note: 12 below.
package PackageName;

module

valuetype

(see Note: 9
and
Note: 10
below)

module PackageName {
struct Foo {
long field;
};
};

valuetype MyValueType {
public MyValueType2 * member;
};
valuetype MyValueType {
public MyValueType2 member;
};
valuetype MyValueType:
MyBaseValueType {
public MyValueType2 * member;
};

public class Foo
{
public int field;
...
}

public class MyValueType {
public MyValueType2 member;
...
};
public class MyValueType {
public MyValueType2 member;
...
};
public class MyValueType extends MyBaseValueType
{
public MyValueType2 member;
...
}

102

3.3.4 Translations for IDL Types

Table 3.10 Specifying Data Types in IDL for Ada
IDL
Type
char

Example Entry in IDL
File

Example Output Generated by
RTI Code Generator (rtiddsgen)

struct PrimitiveStruct
{
char char_member;
};

type PrimitiveStruct is record
char_member : aliased Standard.DDS.Char;
end record;

wchar

struct PrimitiveStruct
{
wchar wchar_member;
};

type PrimitiveStruct is record
wchar_member : aliased Standard.DDS.Wchar;
end record;

octet

struct PrimitiveStruct
{
octet octet_
member;
};

type PrimitiveStruct is record
octet_member: aliased Standard.DDS.Octet;
end record;

short

struct PrimitiveStruct
{
short short_
member;
};

type PrimitiveStruct is record
short_member: aliased Standard.DDS.Short;
end record;

unsign
ed
short

struct PrimitiveStruct
{
unsigned short
unsigned_short_
member;
};

type PrimitiveStruct is record
unsigned_short_member: aliased Standard.DDS.Unsigned_Short;
end record;

long

struct PrimitiveStruct
{
long long_member;
};

type PrimitiveStruct is record
long_member: aliased Standard.DDS.Long;
end record;

unsign
ed long

struct PrimitiveStruct
{
unsigned long
unsigned_long_
member;
};

type PrimitiveStruct is record
unsigned_long_member: aliased Standard.DDS.Unsigned_Long;
end record;

(see
Note: 1
3
below)

103

3.3.4 Translations for IDL Types

Table 3.10 Specifying Data Types in IDL for Ada
IDL
Type

Example Entry in IDL
File

long
long

struct PrimitiveStruct
{
long long
long_long_
member;
};

type PrimitiveStruct is record
long_long_member: aliased Standard.DDS.Long_Long;
end record;

unsign
ed long
long

struct PrimitiveStruct
{
unsigned long long
unsigned_long_
long_member;
};

type PrimitiveStruct is record
unsigned_long_long_member: aliased Standard.DDS.Unsigned_Long_Long;
end record;

float

struct PrimitiveStruct
{
float float_
member;
};

type PrimitiveStruct is record
float_member: aliased Standard.DDS.Float;
end record;

double

struct PrimitiveStruct
{
double double_
member;
};

type PrimitiveStruct is record
double_member: aliased Standard.DDS.Double;
end record;

struct PrimitiveStruct
{
long double
long_double_
member;
};

type PrimitiveStruct is record
long_double_member: aliased Standard.DDS.Long_Double;
end record;

struct MyStruct {
long * member;
};

type MyStruct is record
member : access Standard.DDS.Long;
end record;

struct PrimitiveStruct
{
boolean boolean_
member;
};

type PrimitiveStruct is record
boolean_member: aliased Standard.DDS.Boolean;
end record;

long
double
(see
Note: 2
below)
pointer
(see
Note: 9
below)

boolea
n

Example Output Generated by
RTI Code Generator (rtiddsgen)

104

3.3.4 Translations for IDL Types

Table 3.10 Specifying Data Types in IDL for Ada
IDL
Type

Example Entry in IDL
File
enum PrimitiveEnum {
ENUM1,
ENUM2,
ENUM3
};

enum

constan
t

enum PrimitiveEnum {
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
};

Example Output Generated by
RTI Code Generator (rtiddsgen)

type PrimitiveEnum is (ENUM1,

ENUM2,

ENUM3 );

type PrimitiveEnum is (ENUM1, ENUM2,
...
for PrimitiveEnum use ( ENUM1 => 10 ,

ENUM3 );
ENUM2 => 20 , ENUM3 => 30

const short SIZE = 5;

SIZE : constant Standard.DDS.Short := 5;

struct PrimitiveStruct
{
char char_member;
};

type PrimitiveStruct is record
char_member : aliased Standard.DDS.Char;
end record;

);

struct

(see
Note: 1
0
below)
union

(see
Note: 3
and
Note: 1
0
below)

typedef

array
of
above
types

union PrimitiveUnion
switch (long){
case 1:
short short_
member;
default:
long long_member;
};

type U_PrimitiveUnion is record
short_member : aliased Standard.DDS.Short;
long_member : aliased Standard.DDS.Long;
end record;
type PrimitiveUnion is record
d : Standard.DDS.Long;
u : U_PrimitiveUnion;
end record;

typedef short
TypedefShort;

type TypedefShort is new Standard.DDS.Short;

struct OneDArrayStruct
{
short short_array
[2];
};

type OneDArrayStruct is record
short_array : aliased Standard.DDS.Short_Array(1..2);
end record;

struct TwoDArrayStruct
{
short short_array
[1][2];
};

type TwoDArrayStruct_short_array_Array is array (1..1, 1..2) of aliased
Standard.DDS.Short;
type TwoDArrayStruct is record
short_array : aliased TwoDArrayStruct_short_array_Array;
end record;

105

3.3.4 Translations for IDL Types

Table 3.10 Specifying Data Types in IDL for Ada
IDL
Type
bounde
d
sequen
ce of
above
types

(see
Note: 1
1 and
Note: 1
5
below)
unboun
ded
sequen
ce of
above
types

(see
Note: 1
1 and
Note: 1
5
below)

array
of
sequen
ces

Example Entry in IDL
File

struct SequenceStruct
{
sequence
short_sequence;
};

struct SequenceStruct
{
sequence
short_sequence;
};

struct
ArraysOfSequences{
sequence
sequences_array
[2];
};

Example Output Generated by
RTI Code Generator (rtiddsgen)

type SequenceStruct is record
short_sequence : aliased Standard.DDS.Short_Seq.Sequence;
end record;

type SequenceStruct is record
short_sequence : aliased Standard.DDS.Short_Seq.Sequence;
end record;

See Note: 13 below.

type ArraysOfSequences_sequences_array_Array is array (1..2) of aliased
Standard.DDS.Short_Seq.Sequence;
type ArraysOfSequences is record
sequences_array : aliased ArraysOfSequences_sequences_array_Array;
end record;

106

3.3.4 Translations for IDL Types

Table 3.10 Specifying Data Types in IDL for Ada
IDL
Type

sequen
ce of
arrays

Example Entry in IDL
File

Example Output Generated by
RTI Code Generator (rtiddsgen)

typedef short
ShortArray[2];
struct
SequenceofArrays {

(see
Note: 1
1
below)

sequence

sequen
ce of
sequen
ces

typedef
sequence
ShortSequence;

arrays_
sequence;
};

struct
SequencesOfSequences{

type ShortArray is array (1..2) of Standard.DDS.Short;
...
type SequenceofArrays is record
arrays_sequence : aliased ADA_IDL_File.ShortArray_Seq.Sequence;
end record;

Note: ADA_IDL_File.ShortArray_Seq.Sequence is an instantiation of
Standard.DDS.Sequences_Generic for the user's data type

type ShortSequence is new Standard.DDS.Short_Seq.Sequence;
...
type SequencesOfSequences is record
sequences_sequence : aliased ADA_IDL_File.ShortSequence_Seq.Sequence;
end record;

(see
Note: 4
and
Note: 1
1
below)

sequence
sequences_
sequence;
};

bounde
d string

struct PrimitiveStruct
{
string<20> string_
member;
};

type PrimitiveStruct is record
string_member : aliased Standard.DDS.String;
-- maximum length = (20)
end record;

unboun
ded
string

struct PrimitiveStruct
{
string string_
member;
};

type PrimitiveStruct is record
string_member : aliased Standard.DDS.String; -end record;

bounde
d
wstring

struct PrimitiveStruct
{
wstring<20>
wstring_member;
};

type PrimitiveStruct is record
wstring_member : aliased Standard.DDS.Wide_String; -(20)
end record;

Note: ADA_IDL_File.ShortSequence_Seq.Sequence is an instantiation of
Standard.DDS.Sequences_Generic for the user's data type

maximum length = (255)

maximum length =

107

3.3.4 Translations for IDL Types

Table 3.10 Specifying Data Types in IDL for Ada
IDL
Type

Example Entry in IDL
File

unboun
ded
wstring

struct PrimitiveStruct
{
wstring wstring_
member;
};

type PrimitiveStruct is record
wstring_member : aliased Standard.DDS.Wide_String;
= (255)
end record;

module

module PackageName {
struct Foo {
long field;
};
};

package PackageName is
type Foo is record
field : aliased Standard.DDS.Long;
end record;
end PackageName;

valuety
pe

(see
Note: 9
and
Note: 1
0
below)

valuetype
MyBaseValueType
valuetype
MyBaseValueType
public long
member;
};

Example Output Generated by
RTI Code Generator (rtiddsgen)

--

maximum length

{
{

valuetype MyValueType:
MyBaseValueType {
public short *
member2;
};

type MyBaseValueType is record
member : aliased Standard.DDS.Long;
end record;
type MyValueType is record
parent : ADA_IDL_File.MyBaseValueType;
member2 : access Standard.DDS.Short;
end record;

Notes for Table 3.5 Specifying Data Types in IDL for C through Table 3.9 Specifying Data Types
in IDL for Java:
Note: 1: In C and C++, primitive types are not represented as native language types (e.g. long, char, etc.)
but as custom types in the DDS namespace (DDS_Long, DDS_Char, etc.). These typedefs are
used to ensure that a field’s size is the same across platforms.
Note: 2: Some platforms do not support long double or have different sizes for that type than defined by
IDL (16 bytes). On such platforms, DDS_LongDouble (as well as the unsigned version) is
mapped to a character array that matches the expected size of that type by default. If you are
using a platform whose native mapping has exactly the expected size, you can instruct Connext
DDS to use the native type instead. That is, if sizeof(long double) == 16, you can tell Connext
DDS to map DDS_LongDouble to long double by defining the following macro either in code
or on the compile line:

108

3.3.4 Translations for IDL Types

-DRTI_CDR_SIZEOF_LONG_DOUBLE=16

Note: 3: Unions in IDL are mapped to structs in C, C++ and records in ADA, so that Connext DDS will
not have to dynamically allocate memory for unions containing variable-length fields such as
strings or sequences. To be efficient, the entire struct (or class in C++/CLI) is not sent when the
union is published. Instead, Connext DDS uses the discriminator field of the struct to decide
what field in the struct is actually sent on the wire.
Note: 4: So-called "anonymous sequences" —sequences of sequences in which the sequence element
has no type name of its own—are not supported. Such sequences are deprecated in CORBA
and may be removed from future versions of IDL. For example, this is not supported:
sequence,4> MySequence;

Sequences of typedef’ed types, where the typedef is really a sequence, are supported. For example,
this is supported:
typedef sequence MyShortSequence;
sequence MySequence;

Note: 5: IDL wchar and char are mapped to Java char, 16-bit unsigned quantities representing Unicode
characters as specified in the standard OMG IDL to Java mapping. In C++/CLI, char and
wchar are mapped to System::Char.
Note: 6: The unsigned version for integer types is mapped to its signed version as specified in the standard OMG IDL to Java mapping.
Note: 7: There is no current support in Java for the IDL long double type. This type is mapped to double
as specified in the standard OMG IDL to Java mapping.
Note: 8: Java does not have a typedef construct, nor does C++/CLI. Typedefs for types that are neither
arrays nor sequences (struct, unions, strings, wstrings, primitive types and enums) are
"unwound" to their original type until a simple IDL type or user-defined IDL type (of the nontypedef variety) is encountered. For typedefs of sequences or arrays, RTI Code Generator will
generate wrapper classes if -corba is not used; no wrapper classes are generated if -corba is
used.
Note: 9: In C, C++ and ADA, all the members in a value type, structure or union that are declared with
the pointer symbol (‘*’) will be mapped to references (pointers). In C++/CLI and Java, the
pointer symbol is ignored because the members are always mapped as references.
Note: 10: In-line nested types are not supported inside structures, unions or valuetypes. For example, this
is not supported:

109

3.3.4 Translations for IDL Types

struct Outer {
short outer_short;
struct Inner {
char inner_char;
short inner_short;
} outer_nested_inner;
};

Note: 11: The sequence Seq is implicitly declared in the IDL file and therefore it cannot be
declared explicitly by the user. For example, this is not supported:
typedef sequence FooSeq; //error

Note: 12: RTI Code Generator will supply a default bound for sequences and strings. You can specify
that bound with the -sequenceSize or -stringSize command-line option, respectively. See the
RTI Code Generator User’s Manual.
Note: 13: In ADA, primitive types are not represented as native language types (e.g. , Character, etc.) but
as custom types in the DDS namespace (Standard.DDS.Long, Standard.DDS.Char, etc.).
These typedefs are used to ensure that a field’s size is the same across platforms.
Note: 14: Every type provides a default constructor, a copy constructor, a move constructor (C++11), a
constructor with parameters to set all the type's members, a destructor, a copy-assignment operator, and a move-assignment operator (C++11). Types also include equality operators, the operator << and a namespace-level swap function.
PrimitiveStruct();
explicit PrimitiveStruct(char char_member);
PrimitiveStruct(PrimitiveStruct&& other_) OMG_NOEXCEPT;
PrimitiveStruct& operator=(PrimitiveStruct&& other_) OMG_NOEXCEPT;
bool operator == (const PrimitiveStruct& other_) const;
bool operator != (const PrimitiveStruct& other_) const;
void swap(PrimitiveStruct& other_) OMG_NOEXCEPT ;
std::ostream& operator << (std::ostream& o,const PrimitiveStruct&
sample);

Note: 15: Sequences of pointers are not supported. For example, this is NOT supported:

110

3.3.5 Escaped Identifiers

sequence;

Sequences of typedef'ed types, where the typedef is really a pointer, are supported. For example, this is
supported:
typedef long* pointerToLong;
sequence;

3.3.5 Escaped Identifiers
To use an IDL keyword as an identifier, the keyword must be “escaped” by prepending an underscore,
‘_’. In addition, you must run RTI Code Generator with the -enableEscapeChar option. For example:
struct MyStruct {
octet _octet; // octet is a keyword. To use the type
// as a member name we add ‘_’
};

The use of ‘_’ is a purely lexical convention that turns off keyword checking. The generated code will not
contain ‘_’. For example, the mapping to C would be as follows:
struct MyStruct {
unsigned char octet;
};

Note: If you generate code from an IDL file to a language ‘X’ (for example, C++), the keywords of this
language cannot be used as IDL identifiers, even if they are escaped. For example:
struct MyStruct {
long int; // error
long _int; // error
};

3.3.6 Namespaces In IDL Files
In IDL, the module keyword is used to create namespaces for the declaration of types defined within the
file.
Here is an example IDL definition:
module PackageName {
struct Foo {
long field;
};
};

C Mapping:

111

3.3.6 Namespaces In IDL Files

The name of the module is concatenated to the name of the structure to create the namespace. The resulting code looks like this:
typedef struct PackageName_Foo {
DDS_Long field;
} PackageName_Foo;

C++ Mapping:
In the Traditional C++ API, when using the -namespace command-line option, RTI Code Generator generates a namespace, such as the following:
namespace PackageName{
class Foo {
public:
DDS_Long field;
}
}

Without the -namespace option, the mapping adds the module to the name of the class:
class PackageName_Foo {
public:
DDS_Long field;
}

In the Modern C++ API, namespaces are always used.
C++/CLI Mapping:
Independently of the usage of the -namespace command-line option, RTI Code Generator generates a
namespace, such as the following:
namespace PackageName{
public ref struct Foo: public DDS::ICopyable {
public:
System::Int32 field;
};
}

Java Mapping:
A Foo.java file will be created in a directory called PackageName to use the equivalent concept as
defined by Java. The file PackageName/Foo.java will contain a declaration of Foo class:

112

3.3.6 Namespaces In IDL Files

package PackageName;
public class Foo {
public int field;
};

In a more complex example, consider the following IDL definition:
module PackageName {
struct Bar {
long field;
};
struct Foo {
Bar barField;
};
};

When RTI Code Generator generates code for the above definition, it will resolve the Bar type to be
within the scope of the PackageName module and automatically generate fully qualified type names.
C Mapping:
typedef struct PackageName_Bar {
DDS_Long field;
} PackageName_Bar;
typedef struct PackageName_Foo {
PackageName_Bar barField;
} PackageName_Foo;

C++ Mapping:
With -namespace:
namespace PackageName {
class Bar {
public:
DDS_Long field;
};
class Foo {
public:
PackageName::Bar barField;
};
};

Without -namespace:

113

3.3.7 Referring to Other IDL Files

class PackageName_Bar {
public:
DDS_Long field;
};
class PackageName_Foo {
public:
PackageName_Bar barField;
};

C++/CLI Mapping:
namespace PackageName{
public ref struct Bar: public DDS::ICopyable {
public:
System::Int32 field;
};
public ref struct Foo: public DDS::ICopyable {
public:
PackageName::Bar^ barField;
};
};

Java Mapping:
PackageName/Bar.java and PackageName/Foo.java would be created with the following code, respectively:
package PackageName;
public class Bar {
public
int field;
};
package PackageName;
public class Foo {
public
PackageName.Bar barField = PackageName.Bar.create();
};

3.3.7 Referring to Other IDL Files
IDL files may refer to other IDL files using a syntax borrowed from C, C+, and C+/CLI preprocessors:
#include "Bar.idl"

If RTI Code Generator encounters such a statement in an IDL file Foo.idl and runs with the preprocessor
enabled (default), it will look in Bar.idl to resolve the types referenced in Foo.idl. For example:

114

3.3.8 Preprocessor Directives

Bar.idl
struct Bar {
};
Foo.idl
struct Foo {
Bar m1;
};

The parsing of Foo in the previous scenario will be successful as Bar can be found in Bar.idl. If Bar was
not declared in Bar.idl, RTI Code Generator will report an error indicating that the symbol could not be
found.
If the preprocessor is not enabled when running RTI Code Generator (see command-line option -ppDisable), the parsing of the previous IDL file will fail because RTI Code Generator will not be able to find a
reference to Bar within Bar.idl.
To prevent RTI Code Generator from resolving a type, use the //@resolve-name directive (see The @resolve-name Directive (Section 3.3.9.3 on page 119)).

3.3.8 Preprocessor Directives
RTI Code Generator supports the standard preprocessor directives defined by the IDL specification, such
as #if, #endif, #include, and #define.
To support these directives, RTI Code Generator calls an external C preprocessor before parsing the IDL
file. On Windows systems, the preprocessor is ‘cl.exe.’ On other architectures, the preprocessor is ‘cpp.’
You can change the default preprocessor with the –ppPath option. If you do not want to run the preprocessor, use the –ppDisable option (see the RTI Code Generator User’s Manual).

3.3.9 Using Custom Directives
The following RTI Code Generator-specific directives can be used in your IDL file:
//@key (see The @key Directive (Section 3.3.9.1 on the next page))
//@copy (see The @copy and Related Directives (Section 3.3.9.2 on page 117))
//@copy-c
//@copy-cppcli
//@copy-java
//@copy-java-begin
//@copy-declaration
//@copy-c-declaration
//@copy-cppcli-declaration
//@copy-java-declaration
//@copy-java-declaration-begin
//@resolve-name [true | false] (see The @resolve-name Directive (Section 3.3.9.3 on page 119))
//@top-level [true | false] (see The @top-level Directive (Section 3.3.9.4 on page 120))

115

3.3.9.1 The @key Directive

Notes:
l

To apply multiple directives to the same member or structure in an IDL file, put each additional directive on a new line, as shown below:
struct A {
long a; //@key
//@ID 20
long b;
}; //@Extensibility FINAL_EXTENSIBILITY
//@top-level false

l

Custom directives start with “//@”. Do not put a space between the slashes and the @, or the directive will not be recognized by RTI Code Generator.

The directives are case-sensitive. For instance, you must use //@key (not //@Key).
3.3.9.1 The @key Directive
To declare a key for your data type, insert the @key directive in the IDL file after one or more fields of the
data type.
With each key, Connext DDS associates an internal 16-byte representation, called a key-hash.
If the maximum size of the serialized key is greater than 16 bytes, to generate the key-hash, Connext DDS
computes the MD5 key-hash of the serialized key in network-byte order. Otherwise (if the maximum size
of the serialized key is <= 16 bytes), the key-hash is the serialized key in network-byte order.
Only struct definitions in IDL may have key fields. When RTI Code Generator encounters //@key, it considers the previously declared field in the enclosing structure to be part of the key. Table 3.11 Example
Keys shows some examples of keys.
Table 3.11 Example Keys
Type

Key Fields

struct NoKey {
long member1;
long member2;
}

struct SimpleKey {
long member1; //@key
long member2;
}

member1

116

3.3.9.2 The @copy and Related Directives

Table 3.11 Example Keys
Type

Key Fields

struct NestedNoKey {
SimpleKey member1;
long member2;
}

struct NestedKey {
SimpleKey member1; //@key
long member2;
}

struct NestedKey2 {
NoKey member1; //@key
long member2;
}

valuetype BaseValueKey {
public long member1; //@key
}

valuetype DerivedValueKey :BaseValueKey {
public long member2; //@key
}

valuetype DerivedValue : BaseValueKey {
public long member2;
}

struct ArrayKey {
long member1[3]; //@key
}

member1.member1

member1.member1
member1.member2

member1

member1
member2

member1

member1[0]
member1[1]
member1[2]

3.3.9.2 The @copy and Related Directives
To copy a line of text verbatim into the generated code files, use the @copy directive in the IDL file. This
feature is particularly useful when you want your generated code to contain text that is valid in the target
programming language but is not valid IDL. It is often used to add user comments or headers or preprocessor commands into the generated code.
//@copy
//@copy
//@copy
//@copy
//@copy

// Modification History
// -------------------// 17Jul05aaa, Created.
// #include “MyTypes.h”

117

3.3.9.2 The @copy and Related Directives

These variations allow you to use the same IDL file for multiple languages:
@copy-c

Copies code if the language is C or C++

@copy-cppcli

Copies code if the language is C++/CLI

@copy-java

Copies code if the language is Java.

@copy-ada

Copies code if the language is Ada.

For example, to add import statements to generated Java code:
//@copy-java import java.util.*;

The above line would be ignored if the same IDL file was used to generate non-Java code.
In C, C++, and C++/CLI, the lines are copied into all of the foo*.[h, c, cxx, cpp] files generated from
foo.idl. For Java, the lines are copied into all of the *.java files that were generated from the original “.idl”
file. The lines will not be copied into any additional files that are generated using the -example command
line option.
@copy-java-begin copies a line of text at the beginning of all the Java files generated for a type. The directive only applies to the first type that is immediately below in the IDL file. A similar directive for Ada
files is also available, @copy-ada-begin.
If you want RTI Code Generator to copy lines only into the files that declare the data types—foo.h for C,
C++, and C++/CLI, foo.java for Java—use the //@copy*declaration forms of this directive.
Note that the first whitespace character to follow //@copy is considered a delimiter and will not be copied
into generated files. All subsequent text found on the line, including any leading whitespaces will be
copied.
//@copy-declaration

Copies the text into the file where the type is declared (.h for C and C++, or .java for Java)

//@copy-c-declaration

Same as //@copy-declaration, but for C and C++ code

//@copy-cppcli-declaration

Same as //@copy-declaration, but for C++/CLI code

//@copy-java-declaration

Same as //@copy-declaration, but for Java-only code

//@copy-ada-declaration

Same as //@copy-declaration, but for Ada-only code

//@copy-java-declaration-begin

Same as //@copy-java-declaration, but only copies the text into the file where the type is declared

//@copy-ada-declaration-begin

Same as //@copy-java-declaration-begin, but only for Ada-only code

118

3.3.9.3 The @resolve-name Directive

3.3.9.3 The @resolve-name Directive
By default, the RTI Code Generator tries to resolve all the references to types and constants in an IDL file.
For example:
module PackageName {
struct Foo {
Bar barField;
};
};

The compilation of the previous IDL file will report an error like the following:
ERROR com.rti.ndds.nddsgen.Main Foo.idl line x:x member type 'Bar' not found

In most cases, this is the expected behavior. However, in some cases, you may want to skip the resolution
step. For example, assume that the Bar type is defined in a separate IDL file and thatyou arerunning RTI
Code Generator without an external preprocessor by using the command-line option -ppDisable (maybe
because the preprocessor is not available in their host platform, see Preprocessor Directives (Section 3.3.8
on page 115)):
Bar.idl
module PackageName {
struct Bar {
long field;
};
};

Foo.idl
#include "Bar.idl"
module PackageName {
struct Foo {
Bar barField;
};
};

In this case, compiling Foo.idl would generate the 'not found' error. However, Bar is defined in Bar.idl.
To specify that RTI Code Generator should not resolve a type reference, use the //@resolve-name false
directive. For example:
#include "Bar.idl"
module PackageName {
struct Foo {

119

3.3.9.4 The @top-level Directive

Bar barField; //@resolve-name false
};
};

When this directive is used, then for the field preceding the directive, RTI Code Generator will assume that
the type is a unkeyed 'structure' and it will use the type name unmodified in the generated code.
Java mapping:
package PackageName;
public class Foo {
public Bar barField = Bar.create();
};

C++ mapping:
namespace PackageName {
class Foo {
public:
Bar barField;
};
};

It is up to you to include the correct header files (or if using Java, to import the correct packages) so that
the compiler resolves the ‘Bar’ type correctly. If needed, this can be done using the copy directives (see
The @copy and Related Directives (Section 3.3.9.2 on page 117)).
When used at the end of the declaration of a structure in IDL, then the directive applies to all types within
the structure, including the base type if defined. For example:
struct MyStructure: MyBaseStructure
{
Foo member1;
Bar member2;
};
//@resolve-name false

3.3.9.4 The @top-level Directive
By default, RTI Code Generator generates user-level type-specific methods for all structures/unions found
in an IDL file. These methods include the methods used by DataWriters and DataReaders to send and
receive data of a given type. General methods for writing and reading that take a void pointer are not
offered by Connext DDS because they are not type safe. Instead, type-specific methods must be created to
support a particular data type.
We use the term ‘top-level type’ to refer to the data type for which you intend to create a DCPS Topic that
can be published or subscribed to. For top-level types, RTI Code Generator must create all of the type120

3.4 Creating User Data Types with Extensible Markup Language (XML)

specific methods previously described in addition to the code to serialize/deserialize those types. However,
some of structures/unions defined in the IDL file are only embedded within higher-level structures and are
not meant to be published or subscribed to individually. For non-top-level types, the DataWriters and
DataReaders methods to send or receive data of those types are superfluous and do not need to be created.
Although the existence of these methods is not a problem in and of itself, code space can be saved if these
methods are not generated in the first place.
You can mark non-top-level types in an IDL file with the directive ‘//@top-level false’ to tell RTI Code
Generator not to generate type-specific methods. Code will still be generated to serialize and deserialize
those types, since they may be embedded in top-level types.
In this example, RTI Code Generator will generate DataWriter/DataReader code for TopLevelStruct
only:
struct EmbeddedStruct{
short member;
}; //@top-level false
struct TopLevelStruct{
EmbeddedStruct member;
};

3.4 Creating User Data Types with Extensible Markup Language (XML)
You can describe user data types with Extensible Markup Language (XML) notation. Connext DDS
provides DTD and XSD files that describe the XML format; see /resource/app/app_
support/rtiddsgen/schema/rti_dds_topic_types.dtd and /resource/app/app_support/rtiddsgen/schema/rti_dds_topic_types.xsd, respectively (in 5.x.y, the x and y stand for the version
numbers of the current release). ( is described in Paths Mentioned in Documentation (Section on page xxxviii).)
The XML validation performed by RTI Code Generator always uses the DTD definition. If the
 tag is not in the XML file, RTI Code Generator will look for the default DTD document
in /resource/schema. Otherwise, it will use the location specified in .
We recommend including a reference to the XSD/DTD files in the XML documents. This provides helpful features in code editors such as Visual Studio® and Eclipse™, including validation and auto-completion while you are editing the XML. We recommend including the reference to the XSD document in
the XML files because it provides stricter validation and better auto-completion than the DTD document.
To include a reference to the XSD document in your XML file, use the attribute
xsi:noNamespaceSchemaLocation in the  tag. For example :


...


121

3.4 Creating User Data Types with Extensible Markup Language (XML)

To include a reference to the DTD document in your XML file, use the  tag. For example:

/resource/app/app_support/rtiddsgen/schema/rti_dds_topic_types.dtd">

...


Table 3.12 Mapping Type System Constructs to XML shows how to map the type system constructs into
XML.
Table 3.12 Mapping Type System Constructs to XML
Type/Construct
IDL

char

wchar

octet

short

unsigned
short

long

unsigned
long

XML

Example
IDL

XML

char

struct PrimitiveStruct {
char char_member;
};





wchar

struct PrimitiveStruct {
wchar wchar_member;
};





octet

struct PrimitiveStruct {
octet octet_member;
};





short

struct PrimitiveStruct {
short short_member;
};





unsignedShort

struct PrimitiveStruct {
unsigned short
unsigned_short_member;
};





long

struct PrimitiveStruct {
long long_member;
};





unsignedLong

struct PrimitiveStruct {
unsigned long
unsigned_long_member;
};





122

3.4 Creating User Data Types with Extensible Markup Language (XML)

Table 3.12 Mapping Type System Constructs to XML
Type/Construct
IDL

long long

unsigned
long long

float

double

long double

boolean

Example

XML

IDL
struct PrimitiveStruct {
long long
long_long_member;

XML

};





unsignedLongLong

struct PrimitiveStruct {
unsigned long long
unsigned_long_long_
member;
};





float

struct PrimitiveStruct {
float float_member;
};





double

struct PrimitiveStruct {
double double_member;
};





longDouble

struct PrimitiveStruct {
long double
long_double_member;
};





boolean

struct PrimitiveStruct {
boolean boolean_member;
};





longLong





unbounded
string

bounded
string

string without stringMaxLength attribute
or with stringMaxLength set to -1

string with stringMaxLength attribute

struct PrimitiveStruct {
string string_member;
};

struct PrimitiveStruct {
string<20> string_member;
};

or







123

3.4 Creating User Data Types with Extensible Markup Language (XML)

Table 3.12 Mapping Type System Constructs to XML
Type/Construct
IDL

XML

Example
IDL

XML




wstring without stringMaxLength
attribute or with stringMaxLength set to 1

struct PrimitiveStruct {
wstring wstring_member;
};

bounded
wstring

wstring with stringMaxLength attribute

struct PrimitiveStruct {
wstring<20> wstring_
member;
};





pointer

pointer attribute with values true,false,0
or 1
Default (if not present): 0

struct PrimitiveStruct {
long * long_member;
};





bitfield attribute with the bitfield length

struct BitfieldStruct {
short short_member: 1;
unsigned short
unsignedShort_member: 1;
short short_nmember_2: 0;
long long_member : 5;
};








struct
KeyedPrimitiveStruct {
short short_member;
//@key
};





unbounded
wstring

bitfield1

key attribute with values
key directive
true, false, 0 or 1
2
Default (if not present): 0

or




1Data types containing bitfield members are not supported by DynamicData (Interacting Dynamically with User Data

Types (Section 3.8 on page 141)).
2Directives are RTI extensions to the standard IDL grammar. For additional information about directives see Using Custom

Directives (Section 3.3.9 on page 115).

124

3.4 Creating User Data Types with Extensible Markup Language (XML)

Table 3.12 Mapping Type System Constructs to XML
Type/Construct
IDL

resolvename
directive1

top-level
directive 2

Other
directives 3

enum

constant

XML

Example
IDL

XML

resolveName attribute with values true,
false, 0 or 1
Default (if not present): 1

struct
UnresolvedPrimitiveStruct
{
PrimitiveStruct
primitive_member;
//@resolve-name false
};





topLevel attribute with values true, false,
0 or 1
Default (if not present): 1

struct
TopLevelPrimitiveStruct {
short short_member;
}; //@top-level false





directive tag

//@copy This text will be
copied in the generated
files


This text will be copied in the
generated files


enum PrimitiveEnum {
ENUM1,
ENUM2,
ENUM3
};







enum PrimitiveEnum {
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
};







const double PI = 3.1415;



enum tag

const tag

1Directives are RTI extensions to the standard IDL grammar. For additional information about directives see Using Custom

Directives (Section 3.3.9 on page 115).
2Directives are RTI extensions to the standard IDL grammar. For additional information about directives see Using Custom

Directives (Section 3.3.9 on page 115).
3Directives are RTI extensions to the standard IDL grammar. For additional information about directives see Using Custom

Directives (Section 3.3.9 on page 115).

125

3.4 Creating User Data Types with Extensible Markup Language (XML)

Table 3.12 Mapping Type System Constructs to XML
Type/Construct
IDL

struct

union

XML

struct tag

union tag

Example
IDL
struct PrimitiveStruct {
short short_member;
};

union PrimitiveUnion
switch
(long) {
case 1:
short short_member;
case 2:
case 3:
float float_member;
default:
long long_member;
};

valuetype BaseValueType {
public long long_member;
};

valuetype

valuetype tag

valuetype
DerivedValueType:
BaseValueType {
public long
long_member_2;
};

typedef short ShortType;

typedef

typedef tag

struct PrimitiveStruct {
short short_member;
};
typedef PrimitiveStruct
PrimitiveStructType;

XML































126

3.4 Creating User Data Types with Extensible Markup Language (XML)

Table 3.12 Mapping Type System Constructs to XML
Type/Construct
IDL

arrays

bounded
sequence

unbounded
sequence

array of
sequences

XML

Example
IDL
struct OneArrayStruct {
short short_array[2];
};





struct TwoArrayStruct {
short short_array[1][2];
};





Attribute sequenceMaxLength > 0

struct SequenceStruct {
sequence
short_sequence;
};





Attribute sequenceMaxLength set to -1

struct SequenceStruct {
sequence
short_sequence;
};





Attributes sequenceMaxLength and
arrayDimensions

struct
ArrayOfSequencesStruct {
sequence
short_sequence_array[2];
};





Attribute
arrayDimensions

typedef short
ShortArray[2];

sequence of
arrays

XML

Must be implemented with a typedef tag

struct
SequenceOfArraysStruct {
sequence
short_array_sequence;
};






127

3.4.1 Primitive Types

Table 3.12 Mapping Type System Constructs to XML
Type/Construct
IDL

XML

Example
IDL

XML

ShortSequence;

type="short"sequenceMaxLength="4"/>

struct
SequenceOfSequencesStruct
{
sequence
short_sequence_sequence;
};












module

module tag

module PackageName {
struct PrimitiveStruct {
long long_member;
};
};

include

include tag

#include
"PrimitiveTypes.idl"

3.4.1 Primitive Types
The primitive types char, wchar, long double, and wstring are not supported natively in XSD. Connext
DDS provides definitions for these types in the file /resource/app/app_support/rtiddsgen/schema. All files that use the primitive types char, wchar, long double and wstring must
reference rti_dds_topic_types_common.xsd. For example:










3.5 Creating User Data Types with XML Schemas (XSD)
You can describe data types with XML schemas (XSD). The format is based on the standard IDL-toWSDL mapping described in the OMG document "CORBA to WSDL/SOAP Interworking
128

3.5 Creating User Data Types with XML Schemas (XSD)

Specification."
Example Header for XSD:



...


Mapping Type System Constructs to XSD (Section Table 3.13 below) describes how to map IDL types to
XSD. The Connext DDS code generator, rtiddsgen, will only accept XSD files that follow this mapping.
Table 3.13 Mapping Type System Constructs to XSD
Type/Construct
IDL

char

wchar

octet

XSD

dds:chara

dds:wcharb

xsd:
unsignedByte

Example
IDL

XSD

struct PrimitiveStruct {
char char_member; };







struct PrimitiveStruct {
wchar wchar_member;
};







struct PrimitiveStruct {
octet octet_member;
};







a All files that use the primitive types char, wchar, long double and wstring must reference rti_dds_topic_

types_common.xsd. See Primitive Types (Section 3.4.1 on the previous page).
bAll files that use the primitive types char, wchar, long double and wstring must reference rti_dds_topic_

types_common.xsd. See Primitive Types (Section 3.4.1 on the previous page)
129

3.5 Creating User Data Types with XML Schemas (XSD)

Table 3.13 Mapping Type System Constructs to XSD
Type/Construct
IDL

short

unsigned
short

long

unsigned
long

long long

unsigned
long long

Example
IDL

XSD

struct PrimitiveStruct {
short short_member;
};







struct PrimitiveStruct {
unsigned short unsigned_
short_member; };



 


xsd:int

struct PrimitiveStruct {
long long_member;
};







xsd:
unsignedInt

struct PrimitiveStruct {
unsigned long unsigned_
long_member;
};



 


xsd:long

struct PrimitiveStruct {
long long long_long_
member;
};

 




xsd:
unsignedLong

struct PrimitiveStruct {
unsigned long long
unsigned_long_long_
member;
};







XSD

xsd:short

xsd:
unsignedShort

130

3.5 Creating User Data Types with XML Schemas (XSD)

Table 3.13 Mapping Type System Constructs to XSD
Type/Construct
IDL

float

double

long
double

boolean

unbounded
string

bounded
string

Example
IDL

XSD

xsd:float

struct PrimitiveStruct {
float float_member;
};



 


xsd:double

struct PrimitiveStruct {
double double_member;
};



  

dds:
longDouble



struct PrimitiveStruct {

};



xsd:boolean

struct PrimitiveStruct {
boolean boolean_member;
};



  

struct PrimitiveStruct {
string string_member;
};







struct PrimitiveStruct {
string<20> string_
member;
};

   
  
  
    

XSD

xsd:string

xsd:string with
restriction
to specify the
maximum length

131

3.5 Creating User Data Types with XML Schemas (XSD)

Table 3.13 Mapping Type System Constructs to XSD
Type/Construct
IDL

unbounded
wstring

XSD

dds:wstring a

Example
IDL

XSD

struct PrimitiveStruct {
wstring wstring_member;
};



  

struct PrimitiveStruct {
wstring<20> wstring_
member;
};

   
  
  
   
 

bounded
wstring

xsd:wstring with
restriction
to specify the
maximum
length

pointer


-->
Default (if not
specified): false

struct PrimitiveStruct {
long * long_member;
};








key
directiveb


-->
Default (if not
specified): false

struct
KeyedPrimitiveStruct {
long long_member; //@key
};



  


aAll files that use the primitive types char, wchar, long double and wstring must reference rti_dds_topic_

types_common.xsd. See Primitive Types (Section 3.4.1 on page 128)
bDirectives are RTI extensions to the standard IDL grammar. For additional information about directives,

see Using Custom Directives.
132

3.5 Creating User Data Types with XML Schemas (XSD)

Table 3.13 Mapping Type System Constructs to XSD
Type/Construct
IDL

XSD


directivea
-->
Default (if not
specified): true

Example
IDL

XSD

struct

UnresolvedPrimitiveStruct 
{

//@resolve-name false
 
};


top-level
directiveb


-->
Default (if not
specified): true


struct
 
}; //@top-level false

 

other
directives



-->

//@copy This text will be

files

enum

constant

xsd:simpleType
with
enumeration

enum PrimitiveEnum {
ENUM1,
ENUM2,
ENUM3
};
enum PrimitiveEnum {
ENUM1 = 10,
ENUM2 = 20,
ENUM3 = 30
};

  
   
  
  10
  
 
 20 
    
30  
  

IDL constants are mapped by substituting their value directly in the generated file

a Directives are RTI extensions to the standard IDL grammar. For additional information about directives,

see Using Custom Directives.
bDirectives are RTI extensions to the standard IDL grammar. For additional information about directives,

see Using Custom Directives.
133

3.5 Creating User Data Types with XML Schemas (XSD)

Table 3.13 Mapping Type System Constructs to XSD
Type/Construct
IDL

XSD

Example
IDL

XSD

xsd:complexType
struct PrimitiveStruct {
with
short short_member; };
xsd:sequence







union

union PrimitiveUnion
switch (long) {
xsd:complexType case 1:
short short_member;
with xsd:choice
default:
long long_member; };

 

 a   
1  
    
default  
  


valuetype

 
 
  
long long_member2; public








struct

aThe discriminant values can be described using comments (as specified by the standard) or xsd:annotation

tags. We recommend using annotations because comments may be removed by XSD/XML parsers.
134

3.5 Creating User Data Types with XML Schemas (XSD)

Table 3.13 Mapping Type System Constructs to XSD
Type/Construct
IDL

typedef

arrays

XSD

Type definitions
are
mapped to XML
schema
type restrictions

Example
IDL

XSD

typedef short ShortType;
struct PrimitiveStruct {
short short_member; };
typedef PrimitiveType =
PrimitiveStructType;

    

 
 






 


n
xsd:complexType
with
sequence
containing one
element with
struct OneArrayStruct {
min & max
short short_array[2]; };
occurs
There is one
xsd:complexType
per array
dimension

 
  
     
 

135

3.5 Creating User Data Types with XML Schemas (XSD)

Table 3.13 Mapping Type System Constructs to XSD
Type/Construct
IDL

arrays
(cont’d)

XSD

Example
IDL

n
xsd:complexType
with
sequence
containing one
element with min struct TwoArrayStruct {
short short_array[2][1];
& max
};
occurs
There is one
xsd:complexType
per array
dimension

bounded
sequence

unbounded
sequence

XSD
 
  
    

    
  


xsd:complexType
with
struct SequenceStruct {
sequence
sequence
containing one
short_sequence; };
element
with min & max
occurs







  
  


xsd:complexType
with sequence
struct SequenceStruct {
containing one
sequence short_
element with
sequence; };
min & max
occurs

 
   

 

 

136

3.5 Creating User Data Types with XML Schemas (XSD)

Table 3.13 Mapping Type System Constructs to XSD
Type/Construct
IDL

array of
sequences

XSD

Example
IDL

n+1
xsd:complexType
with
sequence
containing one
element
with min & max struct
ArrayOfSequencesStruct {
occurrences.
sequence

There is one
sequence_sequence[2]; };
xsd:complexType
per
array dimension
and one
xsd:complexType
for the sequence.

sequence of
arrays

Sequences of
arrays must be
implemented
using an explicit
type definition
(typedef) for
the array

typedef short ShortArray
[2]; struct
SequenceOfArraysStruct {
sequence
arrays_sequence; };

XSD








  
     
  


   
 
 
  








137

3.6 Using RTI Code Generator (rtiddsgen)

Table 3.13 Mapping Type System Constructs to XSD
Type/Construct
IDL

XSD

Example
IDL

XSD

sequence of
sequences

Sequences of
sequences must
be implemented
using an
explicit type
definition
(typedef)
for the second
sequence

     
 

 
 

module

Modules are
mapped adding
the
name of the
module before
the
name of each
type inside the
module

module PackageName {
struct PrimitiveStruct {
long long_member; };
};

 

 

include

xsd:include

#include
"PrimitiveType.idl"



3.6 Using RTI Code Generator (rtiddsgen)
RTI Code Generator creates the code needed to define and register a user-data type with Connext DDS.
Using this tool is optional if:
l

l

You are using dynamic types (see Managing Memory for Built-in Types (Section 3.2.7 on page
62))
You are using one of the built-in types (see Built-in Data Types (Section 3.2 on page 30))

See the RTI Code Generator User’s Manual for more information.

138

3.7 Using Generated Types without Connext DDS (Standalone)

3.7 Using Generated Types without Connext DDS (Standalone)
You can use the generated type-specific source and header files without linking the Connext DDS libraries
or even including the Connext DDS header files. That is, the files generated by RTI Code Generator for
your data types can be used standalone.
The directory /resource/app/app_support/rtiddsgen/standalone contains the required
helper files:
l

include: header and templates files for C and C++.

l

src: source files for C and C++.

l

class: Java jar file.

Note: You must use RTI Code Generator’s -notypecode option to generate code for standalone use. See
the RTI Code Generator User’s Manual for more information.

3.7.1 Using Standalone Types in C
The generated files that can be used standalone are:
l

.c: Types source file

l

.h: Types header file

The type plug-in code (Plugin.[c,h]) and type-support code (Support.[c,h]) cannot be
used standalone.
To use the generated types in a standalone manner:
1. Make sure you use rtiddsgen’s -notypecode option to generate the code.
2. Include the directory /resource/app/app_support/rtiddsgen/standalone/include
in the list of directories to be searched for header files.
3. Add the source files, ndds_standalone_type.c and .c, to your project.
4. Include the file .h in the source files that will use the generated types in a standalone
manner.
5. Compile the project using the following two preprocessor definitions:
l NDDS_STANDALONE_TYPE
l

The definition for your platform (RTI_VXWORKS, RTI_QNX, RTI_WIN32, RTI_INTY,
RTI_LYNX or RTI_UNIX)

139

3.7.2 Using Standalone Types in C++

3.7.2 Using Standalone Types in C++
(This section applies to the Traditional C++ API only)
The generated files that can be used standalone are:
l

.cxx: Types source file

l

.h: Types header file

The type-plugin code (Plugin.[cxx,h]) and type-support code (Support.[cxx,h]) cannot
be used standalone.
To use the generated types in a standalone manner:
1. Make sure you use RTI Code Generator’s -notypecode option to generate the code.
2. Include the directory /resource/app/app_support/rtiddsgen/standalone/include
in the list of directories to be searched for header files.
3. Add the source files, ndds_standalone_type.cxx and .cxx, to your project.
4. Include the file .h in the source files that will use the RTI Code Generator types in a
standalone manner.
5. Compile the project using the following two preprocessor definitions:
l NDDS_STANDALONE_TYPE
l

The definition for your platform (such as RTI_VXWORKS, RTI_QNX, RTI_WIN32, RTI_
INTY, RTI_LYNX or RTI_UNIX)

3.7.3 Standalone Types in Java
The generated files that can be used standalone are:
l

.java

l

Seq.java

The type code (TypeCode.java), type-support code (TypeSupport.java),
DataReader code (DataReader.java) and DataWriter code (DataWriter.java) cannot
be used standalone.
To use the generated types in a standalone manner:
1. Make sure you use RTI Code Generator’s -notypecode option to generate the code.
2. Include the file ndds_standalone_type.jar in the classpath of your project.

140

3.8 Interacting Dynamically with User Data Types

3. Compile the project using the standalone types files (.java and Seq.java).

3.8 Interacting Dynamically with User Data Types
3.8.1 Type Schemas and TypeCode Objects
Type schemas—the names and definitions of a type and its fields—are represented by TypeCode objects,
described in Introduction to TypeCode (Section 3.1.3 on page 29).

3.8.2 Defining New Types
This section does not apply when using the separate add-on product, Ada Language Support,
which does not support Dynamic Types.
Locally, your application can access the type code for a generated type "Foo" by calling the FooTypeSupport::get_typecode() (Traditional C++ Notation) operation in the code for the type generated by RTI
Code Generator (unless type-code support is disabled with the -notypecode option). But you can also create TypeCodes at run time without any code generation.
Creating a TypeCode is parallel to the way you would define the type statically: you define the type itself
with some name, then you add members to it, each with its own name and type.
For example, consider the following statically defined type. It might be in C, C++, or IDL; the syntax is
largely the same.
struct MyType {
long my_integer;
float my_float;
bool my_bool;
string<128> my_string; // @key
};

This is how you would define the same type at run time in the Traditional C++ API:
DDS_ExceptionCode_t ex = DDS_NO_EXCEPTION_CODE;
DDS_StructMemberSeq structMembers; // ignore for now
DDS_TypeCodeFactory* factory =
DDS_TypeCodeFactory::get_instance();
DDS_TypeCode* structTc = factory->create_struct_tc(
"MyType", structMembers, ex);
// If structTc is NULL, check 'ex' for more information.
structTc->add_member(
"my_integer", DDS_TYPECODE_MEMBER_ID_INVALID,
factory->get_primitive_tc(DDS_TK_LONG)
DDS_TYPECODE_NONKEY_REQUIRED_MEMBER, ex);
structTc->add_member(
"my_float", DDS_TYPECODE_MEMBER_ID_INVALID,

141

3.8.2 Defining New Types

factory->get_primitive_tc(DDS_TK_FLOAT),
DDS_TYPECODE_NONKEY_REQUIRED_MEMBER, ex);
structTc->add_member(
"my_bool", DDS_TYPECODE_MEMBER_ID_INVALID,
factory->get_primitive_tc(DDS_TK_BOOLEAN),
DDS_TYPECODE_NONKEY_REQUIRED_MEMBER, ex);
structTc->add_member(
"my_string", DDS_TYPECODE_MEMBER_ID_INVALID,
factory->create_string_tc(128),
DDS_TYPECODE_KEY_MEMBER, ex);

More detailed documentation for the methods and constants you see above, including example code, can
be found in the API Reference HTML documentation, which is available for all supported programming
languages.
If, as in the example above, you know all of the fields that will exist in the type at the time of its construction, you can use the StructMemberSeq to simplify the code:
DDS_StructMemberSeq structMembers;
structMembers.ensure_length(4, 4);
DDS_TypeCodeFactory* factory = DDS_TypeCodeFactory::get_instance();
structMembers[0].name = DDS_String_dup("my_integer");
structMembers[0].type = factory->get_primitive_tc(DDS_TK_LONG);
structMembers[1].name = DDS_String_dup("my_float");
structMembers[1].type = factory->get_primitive_tc(DDS_TK_FLOAT);
structMembers[2].name = DDS_String_dup("my_bool");
structMembers[2].type = factory->get_primitive_tc(DDS_TK_BOOLEAN);
structMembers[3].name = DDS_String_dup("my_string");
structMembers[3].type = factory->create_string_tc(128);
structMembers[3].is_key = DDS_BOOLEAN_TRUE;
DDS_ExceptionCode_t ex = DDS_NO_EXCEPTION_CODE;
DDS_TypeCode* structTc =
factory->create_struct_tc(
"MyType", structMembers, ex);

After you have defined the TypeCode, you will register it with a DomainParticipant using a logical name
(note: this step is not required in the Modern C++ API). You will use this logical name later when you create a Topic.
DDSDynamicDataTypeSupport* type_support =
new DDSDynamicDataTypeSupport(structTc,
DDS_DYNAMIC_DATA_TYPE_PROPERTY_DEFAULT);
DDS_ReturnCode_t retcode =
type_support->register_type(participant,
"My Logical Type Name");

For code examples for the Modern C++ API, please refer to the API Reference HTML documentation:
Modules, Programming How-To's, DynamicType and DynamicData Use Cases.

142

3.8.3 Sending Only a Few Fields

Now that you have created a type, you will need to know how to interact with objects of that type. See
Sending Only a Few Fields (Section 3.8.3 below) for more information.

3.8.3 Sending Only a Few Fields
In some cases, your data model may contain a large number of potential fields, but it may not be desirable
or appropriate to include a value for every one of them with every DDS data sample.
l

l

It may use too much bandwidth. You may have a very large data structure, parts of which are
updated very frequently. Rather than resending the entire data structure with every change, you may
wish to send only those fields that have changed and rely on the recipients to reassemble the complete state themselves.
It may not make sense. Some fields may only have meaning in the presence of other fields. For
example, you may have an event stream in which certain fields are only relevant for certain kinds of
events.

To support these and similar cases, Connext DDS supports mutable types and optional members (see the
RTI Connext DDS Core Libraries Getting Started Guide Addendum for Extensible Types).

3.8.4 Sending Type Codes on the Network
In addition to being used locally, serialized type codes are typically published automatically during discovery as part of the built-in topics for publications and subscriptions. See Built-in DataReaders (Section
16.2 on page 773). This allows applications to publish or subscribe to topics of arbitrary types. This functionality is useful for generic system monitoring tools like the rtiddsspy debug tool. For details on using
rtiddsspy, see the API Reference HTML documentation (select Modules, Programming Tools).
Note: Type

codes are not cached by Connext DDS upon receipt and are therefore not available from the
built-in data returned by the DataWriter's get_matched_subscription_data() operation or the
DataReader's get_matched_publication_data() operation.
If your data type has an especially complex type code, you may need to increase the value of the type_
code_max_serialized_length field in the DomainParticipant's DOMAIN_PARTICIPANT_
RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593). Or, to prevent the
propagation of type codes altogether, you can set this value to zero (0). Be aware that some features of
monitoring tools, as well as some features of the middleware itself (such as ContentFilteredTopics) will not
work correctly if you disable TypeCode propagation.
3.8.4.1 Type Codes for Built-in Types
The type codes associated with the built-in types are generated from the following IDL type definitions:

143

3.8.4.1 Type Codes for Built-in Types

module DDS {
/* String */
struct String {
string value;
};
/* KeyedString */
struct KeyedString {
string key; //@key
string value;
};
/* Octets */
struct Octets {
sequence value;
};
/* KeyedOctets */
struct KeyedOctets {
string key; //@key
sequence value;
};
};

The maximum size (max_size) of the strings and sequences that will be included in the type code definitions can be configured on a per-DomainParticipant-basis by using the properties in Table 3.14 Properties
for Allocating Size of Built-in Types, per DomainParticipant.
Table 3.14 Properties for Allocating Size of Built-in Types, per DomainParticipant
Built-in
Type

String

Property

Description

Maximum size of the strings published by the DataWriters and received by the DataReaders belonging to
dds.builtin_
type.string.max_ a DomainParticipant (includes the NULL-terminated character).
size
Default: 1024
dds.builtin_
type.keyed_
string.
max_key_size

Maximum size of the keys used by the DataWriters and DataReaders belonging to a DomainParticipant
(includes the NULL-terminated character).

dds.builtin_
type.keyed_
string.
max_size

Maximum size of the strings published by the DataWriters and received by the DataReaders belonging to
a DomainParticipant using the built-in type (includes the NULL-terminated character).

Default: 1024

KeyedString

Octets

Default: 1024

Maximum size of the octet sequences published by the DataWriters and DataReaders belonging to a
dds.builtin_
type.octets.max_ DomainParticipant.
size
Default: 2048

144

3.9 Working with DDS Data Samples

Table 3.14 Properties for Allocating Size of Built-in Types, per DomainParticipant
Built-in
Type

KeyedOctets

Property

Description

dds.builtin_
type.keyed_
octets.
max_key_size

Maximum size of the key published by the DataWriter and received by the DataReaders belonging to the
DomainParticipant (includes the NULL-terminated character).

dds.builtin_
type.keyed_
octets.
max_size

Maximum size of the octet sequences published by the DataWriters and DataReaders belonging to a
DomainParticipant.

Default: 1024.

Default: 2048

3.9 Working with DDS Data Samples
You should now understand how to define and work with data types, whether you're using the simple data
types built into the middleware (see Built-in Data Types (Section 3.2 on page 30)), dynamically defined
types (see Managing Memory for Built-in Types (Section 3.2.7 on page 62)), or code generated from IDL
or XML files (see Creating User Data Types with IDL (Section 3.3 on page 69) and Creating User Data
Types with Extensible Markup Language (XML) (Section 3.4 on page 121)).
Now that you have chosen one or more data types to work with, this section will help you understand how
to create and manipulate objects of those types.

3.9.1 Objects of Concrete Types
If you use one of the built-in types or decide to generate custom types from an IDL or XML file, your Connext DDS data type is like any other data type in your application: a class or structure with fields, methods,
and other members that you interact with directly.
In C and Traditional C++:
You create and delete your own objects from factories, just as you create Connext DDS objects from
factories. In the case of user data types, the factory is a singleton object called the type support. Objects
allocated from these factories are deeply allocated and fully initialized.
/* In the generated header file: */
struct MyData {
char* myString;
};
/* In your code: */
MyData* sample = MyDataTypeSupport_create_data();
char* str = sample->myString; /*empty, non-NULL string*/
/* ... */
MyDataTypeSupport_delete_data(sample);

145

3.9.1 Objects of Concrete Types

In Traditional C++:
You create and delete objects using the TypeSupport factories.
MyData* sample = MyDataTypeSupport::create_data(); char* str = sample>myString; // empty, non-NULL string // ... MyDataTypeSupport::delete_data
(sample);

In Modern C++:
Generated types have value-type semantics and provide a default constructor, a constructor with parameters to initialize all the members, a copy constructor and assignment operator ,a move constructor and
move-assignment operator (C++11 only), a destructor, equality operators, a swap function and an overloaded operator<<. Data members are accessed using getters and setters.
// In the generated header file
class MyData {
public:
MyData();
explicit MyData(const dds::core::string& myString);
// Note: the implicit destructor, copy and
// move constructors, and assignment operators
// are available
dds::core::string& myString() OMG_NOEXCEPT;
const dds::core::string& myString() const OMG_NOEXCEPT;
void myString(const dds::core::string& value);
bool operator == (const MyData& other_) const;
bool operator != (const MyData& other_) const;
private:
// ...
};
void swap(MyData& a, MyData& b) OMG_NOEXCEPT
std::ostream& operator <<
(std::ostream& o,const MyData& sample);
// In your code:
MyData sample("Hello");
sample.myString("Bye");

In C# and C++/CLI:
You can use a no-argument constructor to allocate objects. Those objects will be deallocated by the
garbage collector as appropriate.
// In the generated code (C++/CLI):
public ref struct MyData {
public: System::String^ myString;

146

3.9.2 Objects of Dynamically Defined Types

};
// In your code, if you are using C#:
MyData sample = new MyData();
System.String str = sample.myString;
// empty, non-null string
// In your code, if you are using C++/CLI:
MyData^ sample = gcnew MyData();
System::String^ str = sample->myString;
// empty, non-nullptr string

In Java:
You can use a no-argument constructor to allocate objects. Those objects will be deallocated by the
garbage collector as appropriate.
// In the generated code:
public class MyData {
public String myString = "";
}
// In your code:
MyData sample = new MyData();
String str = sample->myString;
// empty, non-null string

3.9.2 Objects of Dynamically Defined Types
If you are working with a data type that was discovered or defined at run time, you will use the reflective
API provided by the DynamicData class to get and set the fields of your object.
Consider the following type definition:
struct MyData {
long myInteger;
};

As with a statically defined type, you will create objects from a TypeSupport factory. How to create or otherwise obtain a TypeCode, and how to subsequently create from it a DynamicDataTypeSupport, is
described in Defining New Types (Section 3.8.2 on page 141). In the Modern C++ API you will use the
DynamicData constructor, which receives a DynamicType.
For more information about the DynamicData and DynamicDataTypeSupport classes, consult the API
Reference HTML documentation, which is available for all supported programming languages (select
Modules, RTI Connext DDS API Reference, Topic Module, Dynamic Data).
In C:
DDS_DynamicDataTypeSupport* support = ...;
DDS_DynamicData* sample = DDS_DynamicDataTypeSupport_create_data(support);
DDS_Long theInteger = 0;

147

3.9.2 Objects of Dynamically Defined Types

DDS_ReturnCode_t success = DDS_DynamicData_set_long(sample,
"myInteger", DDS_DYNAMIC_DATA_MEMBER_ID_UNSPECIFIED, 5);
/* Error handling omitted. */
success = DDS_DynamicData_get_long(
sample, &theInteger,
"myInteger", DDS_DYNAMIC_DATA_MEMBER_ID_UNSPECIFIED);
/* Error handling omitted. "theInteger" now contains the value 5
if no error occurred.
*/

In Traditional C++:
DDSDynamicDataTypeSupport* support = ...;
DDS_DynamicData* sample = support->create_data();
DDS_ReturnCode_t success = sample->set_long("myInteger",
DDS_DYNAMIC_DATA_MEMBER_ID_UNSPECIFIED, 5);
// Error handling omitted.
DDS_Long theInteger = 0;
success = sample->get_long(
&theInteger, "myInteger",
DDS_DYNAMIC_DATA_MEMBER_ID_UNSPECIFIED);
// Error handling omitted.
// "theInteger" now contains the value 5 if no error occurred.

In Modern C++:
using namespace dds::core::xtypes;
StructType type(
"MyData", {
Member("myInteger", primitive_type())
}
);
DynamicData sample(type);
sample.value("myInteger", 5);
int32_t the_int = sample.value("myInteger");
// "the_int" now contains the value 5 if no exception was thrown

In C++/CLI:
using DDS;
DynamicDataTypeSupport^ support = ...;
DynamicData^ sample = support->create_data();
sample->set_long("myInteger",
DynamicData::MEMBER_ID_UNSPECIFIED, 5);
int theInteger = sample->get_long("myInteger",
0 /*redundant w/ field name*/);
/* Exception handling omitted.
* "theInteger" now contains the value 5 if no error occurred.
*/

In C#:
using namespace DDS;
DynamicDataTypeSupport support = ...;
DynamicData sample = support.create_data();

148

3.9.3 Serializing and Deserializing Data Samples

sample.set_long("myInteger", DynamicData.MEMBER_ID_UNSPECIFIED, 5);
int theInteger = sample.get_long("myInteger",
DynamicData.MEMBER_ID_UNSPECIFIED);
/* Exception handling omitted.
* "theInteger" now contains the value 5 if no error occurred.
*/

In Java:
import com.rti.dds.dynamicdata.*;
DynamicDataTypeSupport support = ...;
DynamicData sample = (DynamicData) support.create_data();
sample.set_int("myInteger", DynamicData.MEMBER_ID_UNSPECIFIED, 5);
int theInteger = sample.get_int("myInteger",
DynamicData.MEMBER_ID_UNSPECIFIED);
/* Exception handling omitted.
* "theInteger" now contains the value 5 if no error occurred.
*/

IThe Modern C++ API provides convenience functions to convert among DynamicData samples and
typed samples (such as MyData, from the previous example). For example:
#include "MyData.hpp"
// ...
MyData typed_sample(44);
DynamicData dynamic_sample = rti::core::xtypes::convert(typed_sample);
assert (dynamic_sample.value("myInteger") == 44);
dynamic_sample.value("myInteger", 33);
typed_sample = rti::core::xtypes::convert(dynamic_sample);
assert (typed_sample.myInteger() == 33);

3.9.3 Serializing and Deserializing Data Samples
There are two TypePlugin operations to serialize a sample into a buffer and deserialize a sample from a buffer. The sample serialization/deserialization uses CDR representation.
The feature is supported in the following languages: C, Modern and Traditional C++, Java, and .NET.
C:
#include "FooSupport.h"
FooTypeSupport_serialize_data_to_cdr_buffer(...)
FooTypeSupport_deserialize_data_from_cdr_buffer(...)

Traditional C++
#include "FooSupport.h"
FooTypeSupport::serialize_data_to_cdr_buffer(...)
FooTypeSupport::deserialize_data_from_cdr_buffer(...)

Modern C++

149

3.9.4 Accessing the Discriminator Value in a Union

#include "Foo.hpp"
dds::topic::topic_type_support::to_cdr_buffer(...)
dds::topic::topic_type_support::from_cdr_buffer(...)

Java:
FooTypeSupport.get_instance().serialize_to_cdr_buffer(...)
FooTypeSupport.get_instance().deserialize_from_cdr_buffer(...)

C++/CLI:
FooTypeSupport::serialize_data_to_cdr_buffer(...)
FooTypeSupport::deserialize_data_from_cdr_buffer(...)

C#:
FooTypeSupport.serialize_data_to_cdr_buffer(...)
FooTypeSupport.deserialize_data_from_cdr_buffer(...)

3.9.4 Accessing the Discriminator Value in a Union
A union type can only hold a single member. The member_id for this member is equal to the discriminator value. To get the value of the discriminator, use the operation get_member_info_by_index()
on the DynamicData using an index value of 0. This operation fills in a DynamicDataMemberInfo structure, which includes a member_id field that is the value of the discriminator.
Once you know the discriminator value, you can use the proper version of get_() (such as get_long
()) to access the member value.
For example:
DynamicDataMemberInfo memberInfo = new DynamicDataMemberInfo();
myDynamicData.get_member_info_by_index(memberInfo, 0);
int discriminatorValue = memberInfo.member_id;
int myMemberValue = myDynamicData.get_long(null, discriminatorValue);

The Modern C++ API provides the method discriminator_value() to achieve the same result:
int32_t my_member_value = my_dynamic_data.value(
my_dynamic_data.discriminator_value());

150

Chapter 4 DDS Entities
The main classes extend an abstract base class called a DDS Entity. Every DDS Entity has a set of
associated events known as statuses and a set of associated Quality of Service Policies
(QosPolicies). In addition, a Listener may be registered with the Entity to be called when status
changes occur. DDS Entities may also have attached DDS Conditions, which provide a way to
wait for status changes. Figure 4.1 Overview of DDS Entities on the next page presents an overview in a UML diagram.
This section describes the common operations and general designed patterns shared by all DDS
Entities including DomainParticipants, Topics, Publishers, DataWriters, Subscribers, and
DataReaders. In subsequent chapters, the specific statuses, Listeners, Conditions, and QosPolicies
for each class will be discussed in detail.

151

4.1 Common Operations for All DDS Entities

Figure 4.1 Overview of DDS Entities

4.1 Common Operations for All DDS Entities
All DDS Entities (DomainParticipants, Topics, Publishers, DataWriters, Subscribers, and DataReaders)
provide operations for:

152

4.1.1 Creating and Deleting DDS Entities

4.1.1 Creating and Deleting DDS Entities
l

C, Traditional C++, Java, and .NET:
The factory design pattern is used in creating and deleting DDS Entities. Instead of declaring and
constructing or destructing Entities directly, a factory object is used to create an Entity. Almost all
Entity factories are objects that are also Entities. The only exception is the factory for a DomainParticipant. See Table 4.1 Entity Factories.
Table 4.1 Entity Factories
Entity
DomainParticipant

Created by
DomainParticipantFactory (a static singleton object provided by Connext DDS)

Topic
Publisher
Subscriber

DomainParticipant

DataWriter
DataReader
DataWriter

Publisher

DataReader

Subscriber

All Entities that are factories have:
l Operations to create and delete child Entities. For example:
DDSPublisher::create_datawriter()

l

DDSDomainParticipant::delete_topic()
Operations to get and set the default QoS values used when creating child Entities. For
example:
DDSSubscriber::get_default_datareader_qos()

l

DDSDomainParticipantFactory::set_default_participant_qos()
And ENTITYFACTORY QosPolicy (Section 6.4.2 on page 315) to specify whether or not
the newly created child Entity should be automatically enabled upon creation.

DataWriters may be created by a DomainParticipant or a Publisher. Similarly, DataReaders may
be created by a DomainParticipant or a Subscriber.

153

4.1.2 Enabling DDS Entities

An entity that is a factory cannot be deleted until all the child Entities created by it have been
deleted.
Each Entity obtained through create_() must eventually be deleted by calling delete_
(), or by calling delete_contained_entities().
l

Modern C++:
In the Modern C++ API the factory pattern is not explicit. Entities have constructors and destructors.
The first argument to an Entity's constructor is its "factory" (except for the DomainParticipant). For
example:
// Note: this example shows the simplest version of each Entity's constructor:
dds::domain::DomainParticipant participant(MY_DOMAIN_ID);
dds::topic::Topic topic(participant, "Example Foo");
dds::sub::Subscriber subscriber(participant);
dds::sub::DataReader reader(subscriber, topic);
dds::pub::Publisher publisher(participant);
dds::pub::DataWriter writer(publisher, topic);

Entities are reference types. In a reference type copy operations, such as copy-construction and
copy-assignment are shallow. The reference types are modeled after shared pointers. Similar to pointers, it is important to distinguish between an entity and a reference (or handle) to it. A single entity
may have multiple references. Copying a reference does not copy the entity it is referring to—creating additional references from the existing reference(s) is a relatively inexpensive operation.
The lifecycle of references and the entity they are referring to is not the same. In general, the entity
lives as long as there is at least one reference to it. When the last reference to the entity ceases to
exists, the entity it is referring to is destroyed.
Applications can override the automatic destruction of Entities. An Entity can be explicitly closed
(by calling the method close()) or retained (by calling retain())
Closing an Entity destroys the underlying object and invalidates all references to it.
Retaining an Entity disables the automatic destruction when it loses all its reference. A retained
Entity can be looked up (see Looking Up DomainParticipants (Section 8.2.4 on page 546)) and has
to be explicitly destroyed with close().

4.1.2 Enabling DDS Entities
The enable() operation changes an Entity from a non-operational to an operational state. Entity objects can
be created disabled or enabled. This is controlled by the value of the ENTITYFACTORY QosPolicy (Section 6.4.2 on page 315) on the corresponding factory for the Entity (not on the Entity itself).
By default, all Entities are automatically created in the enabled state. This means that as soon as the Entity
is created, it is ready to be used. In some cases, you may want to create the Entity in a ‘disabled’ state. For
example, by default, as soon as you create a DataReader, the DataReader will start receiving new DDS
154

4.1.2.1 Rules for Calling enable()

samples for its Topic if they are being sent. However, your application may still be initializing other components and may not be ready to process the data at that time. In that case, you can tell the Subscriber to
create the DataReader in a disabled state. After all of the other parts of the application have been created
and initialized, then the DataReader can be enabled to actually receive messages.
To create a particular entity in a disabled state, modify the EntityFactory QosPolicy of its corresponding
factory entity before calling create_(). For example, to create a disabled DataReader, modify the
Subscriber’s QoS as follows:
DDS_SubscriberQos subscriber_qos;
subscriber->get_qos(subscriber_qos);
subscriber_qos.entity_factory.autoenable_created_entities = DDS_BOOLEAN_FALSE;
subscriber->set_qos(subscriber_qos);
DDSDataReader* datareader =
subscriber->create_datareader(topic, DDS_DATAREADER_QOS_DEFAULT, listener);

When the application is ready to process received data, it can enable the DataReader:
datareader->enable();

4.1.2.1 Rules for Calling enable()
In the following, a ‘Factory’ refers to a DomainParticipant, Publisher, or Subscriber; a ‘child’ refers to an
entity created by the factory:
l

l

l

l

l

l

l
l

If the factory is disabled, its children are always created disabled, regardless of the setting in the factory's EntityFactoryQoS.
If the factory is enabled, its children will be created either enabled or disabled, according to the setting in the factory's EntityFactory Qos.
Calling enable() on a child whose factory object is still disabled will fail and return DDS_
RECODE_RECONDITION_NOT_MET.
Calling enable() on a factory with EntityFactoryQoS set to DDS_BOOLEAN_TRUE will recursively enable all of the factory’s children. If the factory’s EntityFactoryQoS is set to DDS_
BOOLEAN_FALSE, only the factory itself will be enabled.
Calling enable() on an entity that is already enabled returns DDS_RETCODE_OK and has no
effect.
There is no complementary “disable” operation. You cannot disable an entity after it is enabled. Disabled Entities must have been created in that state.
An entity’s Listener will only be invoked if the entity is enabled.
The existence of an entity is not propagated to other DomainParticipants until the entity is enabled
(see Discovery (Section Chapter 14 on page 709)).

155

4.1.2.1 Rules for Calling enable()

l

l

If a DataWriter/DataReader is to be created in an enabled state, then the associated Topic must
already be enabled. The enabled state of the Topic does not matter, if the Publisher/Subscriber has
its EntityFactory QosPolicy to create children in a disabled state.
When calling enable() for a DataWriter/DataReader, both the Publisher/Subscriber and the Topic
must be enabled, or the operation will fail and return DDS_RETCODE_PRECONDITION_NOT_
MET.

The following operations may be invoked on disabled Entities:
l

get_qos() and set_qos()Some DDS-specified QosPolicies are immutable—they cannot be changed
after an Entity is enabled. This means that for those policies, if the entity was created in the disabled
state, get/set_qos() can be used to change the values of those policies until enabled() is called on the
Entity. After the Entity is enabled, changing the values of those policies will not affect the Entity.
However, there are mutable QosPolicies whose values can be changed at anytime–even after the
Entity has been enabled.
Finally, there are extended QosPolicies that are not a part of the DDS specification but offered by
Connext DDS to control extended features for an Entity. Some of those extended QosPolicies cannot be changed after the Entity has been created—regardless of whether the Entity is enabled or disabled.

l

l

l
l

l

Into which exact categories a QosPolicy falls—mutable at any time, immutable after enable, immutable after creation—is described in the documentation for the specific policy.
get_status_changes() and get_*_status()The status of an Entity can be retrieved at any time (but
the status of a disabled Entity never changes). (Note: get_*_status() resets the related status so it no
longer considered “changed.”)
get_statuscondition()An Entity’s StatusCondition can be checked at any time (although the status
of a disabled Entity never changes).
get_listener() and set_listener()An Entity’s Listener can be changed at any time.
create_*() and delete_*()A factory Entity can still be used to create or delete any child Entity that it
can produce. Note: following the rules discussed previously, a disabled Entity will always create its
children in a disabled state, no matter what the value of the EntityFactory QosPolicy is.
lookup_*()An Entity can always look up children it has previously created.

Most other operations are not allowed on disabled Entities. Executing one of those operations when an
Entity is disabled will result in a return code of DDS_RETCODE_NOT_ENABLED. The documentation
for a particular operation will explicitly state if it is not allowed to be used if the Entity is disabled.

156

4.1.3 Getting an Entity’s Instance Handle

The builtin transports are implicitly registered when (a) the DomainParticipant is enabled, (b) the
first DataWriter/DataReader is created, or (c) you look up a builtin data reader, whichever
happens first. Any changes to the builtin transport properties that are made after the builtin
transports have been registered will have no affect on any DataWriters/DataReaders.

4.1.3 Getting an Entity’s Instance Handle
The Entity class provides an operation to retrieve an instance handle for the object. The operation is
simply:
InstanceHandle_t get_instance_handle()

An instance handle is a global ID for the entity that can be used in methods that allow user applications to
determine if the entity was locally created, if an entity is owned (created) by another entity, etc.

4.1.4 Getting Status and Status Changes
The get_status_changes() operation retrieves the set of events, also known in DDS terminology as communication statuses, in the Entity that have changed since the last time get_status_changes() was called.
This method actually returns a value that must be bitwise AND’ed with an enumerated bit mask to test
whether or not a specific status has changed. The operation can be used in a polling mechanism to see if
any statuses related to the Entity have changed. If an entity is disabled, all communication statuses are in
the “unchanged” state so the list returned by the get_status_changes() operation will be empty.
A set of statuses is defined for each class of Entities. For each status, there is a corresponding operation,
get__status(), that can be used to get its current value. For example, a DataWriter has a
DDS_OFFERED_DEADLINE_MISSED status; it also has a get_offered_deadline_missed_status()
operation:
DDS_StatusMask statuses;
DDS_OfferedDeadlineMissedStatus deadline_stat;
statuses = datawriter->get_status_changes();
if (statuses & DDS_OFFERED_DEADLINE_MISSED_STATUS) {
datawriter->get_offered_deadline_missed_status(
&deadline_stat);
printf(“Deadline missed %d times.\n”,
deadline_stat.total_count);
}

To reset a status (so that it is no longer considered “changed”), call get__status(). Or, in
the case of the DDS_DATA_AVAILABLE status, call read(), take(), or one of their variants.
If you use a StatusCondition to be notified that a particular status has changed, the
StatusCondition’s trigger_value will remain true unless you call get_*_status() to reset the status.
See also: Statuses (Section 4.3 on page 169) and StatusConditions (Section 4.6.8 on page 197).

157

4.1.5 Getting and Setting Listeners

4.1.5 Getting and Setting Listeners
Each type of Entity has an associated Listener, see Listeners (Section 4.4 on page 177). A Listener represents a set of functions that users may install to be called asynchronously when the state of communication statuses change.
The get_listener() operation returns the current Listener attached to the Entity.
The set_listener() operation installs a Listener on an Entity. The Listener will only be invoked on the
changes of statuses specified by the accompanying mask. Only one listener can be attached to each Entity.
If a Listener was already attached, set_listener() will replace it with the new one.
The get_listener() and set_listener() operations are directly provided by the DomainParticipant, Topic,
Publisher, DataWriter, Subscriber, and DataReader classes so that listeners and masks used in the argument list are specific to each Entity.
Note: The set_listener() operation is not synchronized with the listener callbacks, so it is possible to set a
new listener on an participant while the old listener is in a callback. Therefore you should be careful not to
delete any listener that has been set on an enabled participant unless some application-specific means are
available of ensuring that the old listener cannot still be in use.
See Listeners (Section 4.4 on page 177) for more information about Listeners.

4.1.6 Getting the StatusCondition
Each type of Entity may have an attached StatusCondition, which can be accessed through the get_
statuscondition() operation. You can attach the StatusCondition to a WaitSet, to cause your application to
wait for specific status changes that affect the Entity.
See Conditions and WaitSets (Section 4.6 on page 187) for more information about StatusConditions and
WaitSets.

4.1.7 Getting, Setting, and Comparing QosPolicies
Each type of Entity has an associated set of QosPolicies (see QosPolicies (Section 4.2 on page 162)).
QosPolicies allow you to configure and set properties for the Entity.
While most QosPolicies are defined by the DDS specification, some are offered by Connext DDS as extensions to control parameters specific to the implementation.
There are two ways to specify a QoS policy:
l
l

Programmatically, as described in this section.
QosPolicies can also be configured from XML resources (files, strings)—with this approach, you
can change the QoS without recompiling the application. The QoS settings are automatically loaded

158

4.1.7 Getting, Setting, and Comparing QosPolicies

by the DomainParticipantFactory when the first DomainParticipant is created. See Configuring
QoS with XML (Section Chapter 17 on page 791).
The get_qos() operation retrieves the current values for the set of QosPolicies defined for the Entity.
QosPolicies can be set programmatically when an Entity is created, or modified with the Entity's set_qos()
operation.
The set_qos() operation sets the QosPolicies of the entity. Note: not all QosPolicy changes will take effect
instantaneously; there may be a delay since some QosPolicies set for one entity, for example, a
DataReader, may actually affect the operation of a matched entity in another application, for example, a
DataWriter.
The get_qos() and set_qos() operations are passed QoS structures that are specific to each derived entity
class, since the set of QosPolicies that effect each class of Entities is different.
The equals() operation compares two Entity’s QoS structures for equality. It takes two parameters for the
two Entities’ QoS structures to be compared, then returns TRUE is they are equal (all values are the same)
or FALSE if they are not equal.
Each QosPolicy has default values (listed in the API Reference HTML documentation). If you want to use
custom values, there are three ways to change QosPolicy settings:
l

l

l

Before Entity creation (if custom values should be used for multiple Entities). See Changing the
QoS Defaults Used to Create DDS Entities: set_default_*_qos() (Section 4.1.7.1 on the next page).
During Entity creation (if custom values are only needed for a particular Entity). See Setting QoS
During Entity Creation (Section 4.1.7.2 on the next page).
After Entity creation (if the values initially specified for a particular Entity are no longer appropriate). See Changing the QoS for an Existing Entity (Section 4.1.7.3 on page 161).

Regardless of when or how you make QoS changes, there are some rules to follow:
l

l

Some QosPolicies interact with each other and thus must be set in a consistent manner. For instance,
the maximum value of the HISTORY QosPolicy’s depth parameter is limited by values set in the
RESOURCE_LIMITS QosPolicy. If the values within a QosPolicy structure are inconsistent, then
set_qos() will return the error INCONSISTENT_POLICY, and the operation will have no effect.
Some policies can only be set when the Entity is created, or before the Entity is enabled. Others can
be changed at any time. In general, all standard DDS QosPolicies can be changed before the Entity
is enabled. A subset can be changed after the Entity is enabled. Connext DDS-specific QosPolicies
either cannot be changed after creation or can be changed at any time. The changeability of each
QosPolicy is documented in the API Reference HTML documentation as well as in Table 4.2
QosPolicies. If you attempt to change a policy after it cannot be changed, set_qos() will fail with a
return IMMUTABLE_POLICY.

159

4.1.7.1 Changing the QoS Defaults Used to Create DDS Entities: set_default_*_qos()

4.1.7.1 Changing the QoS Defaults Used to Create DDS Entities: set_default_*_qos()
Each parent factory has a set of default QoS settings that are used when the child entity is created. The
DomainParticipantFactory has default QoS values for creating DomainParticipants. A DomainParticipant has a set of default QoS for each type of entity that can be created from the DomainParticipant
(Topic, Publisher, Subscriber, DataWriter, and DataReader). Likewise, a Publisher has a set of default
QoS values used when creating DataWriters, and a Subscriber has a set of default QoS values used when
creating DataReaders.
An entity’s QoS are set when it is created. Once an entity is created, all of its QoS—for itself and its child
Entities—are fixed unless you call set_qos() or set_qos_with_profile() on that entity. Calling set_
default__qos() on a parent entity will have no effect on child Entities that have already been created.
You can change these default values so that they are automatically applied when new child Entities are created. For example, suppose you want all DataWriters for a particular Publisher to have their
RELIABILITY QosPolicy set to RELIABLE. Instead of making this change for each DataWriter when it
is created, you can change the default used when any DataWriter is created from the Publisher by using
the Publisher’s set_default_datawriter_qos() operation.
DDS_DataWriterQos default_datawriter_qos;
// get the current default values
publisher->get_default_datawriter_qos(default_datawriter_qos);
// change to desired default values
default_datawriter_qos.reliability.kind =
DDS_RELIABLE_RELIABILITY_QOS;
// set the new default values
publisher->set_default_datawriter_qos(default_datawriter_qos);
// created datawriters will use new default values
datawriter =
publisher->create_datawriter(topic, NULL, NULL, NULL);

It is not safe to get or set the default QoS values for an entity while another thread may be
simultaneously calling get_default__qos(), set_default__qos(), or create_
() with DDS__QOS_DEFAULT as the qos parameter (for the same entity).
Another way to make QoS changes is by using XML resources (files, strings). For more information, see
Configuring QoS with XML (Section Chapter 17 on page 791).
4.1.7.2 Setting QoS During Entity Creation
If you only want to change a QosPolicy for a particular entity, you can pass in the desired QosPolicies for
an entity in its creation routine.
To customize an entity's QoS before creating it:

160

4.1.7.3 Changing the QoS for an Existing Entity

1. (C API Only) Initialize a QoS object with the appropriate INITIALIZER constructor.
2. Call the relevant get__default_qos() method.
3. Modify the QoS values as desired.
4. Create the entity.
For example, to change the RELIABLE QosPolicy for a DataWriter before creating it:
// Initialize the QoS object
DDS_DataWriterQos datawriter_qos;
// Get the default values
publisher->get_default_datawriter_qos(datawriter_qos);
// Modify the QoS values as desired
datawriter_qos.reliability.kind = DDS_BEST_EFFORT_RELIABILITY_QOS;
// Create the DataWriter with new values
datawriter = publisher->create_datawriter(
topic, datawriter_qos, NULL, NULL);

Another way to set QoS during entity creation is by using a QoS profile. For more information, see Configuring QoS with XML (Section Chapter 17 on page 791).
4.1.7.3 Changing the QoS for an Existing Entity
Some policies can also be changed after the entity has been created. To change such a policy after the
entity has been created, use the entity’s set_qos() operation.
For example, suppose you want to tweak the DEADLINE QoS for an existing DataWriter:
DDS_DataWriterQos datawriter_qos;
// get the current values
datawriter->get_qos(datawriter_qos);
// make desired changes
datawriter_qos.deadline.period.sec = 3;
datawriter_qos.deadline.period.nanosec = 0;
// set new values
datawriter->set_qos(datawriter_qos);

Another way to make QoS changes is by using a QoS profile. For more information, see Configuring QoS
with XML (Section Chapter 17 on page 791).
In the code examples presented in this section, we are not testing for the return code for the set_qos(),
set_default_*_qos() functions. If the values used in the QosPolicy structures are inconsistent then the functions will fail and return INCONSISTENT_POLICY. In addition, set_qos() may return IMMUTABLE_
POLICY if you try to change a QosPolicy on an Entity after that policy has become immutable. User code
should test for and address those anomalous conditions.
Note:

161

4.1.7.4 Default QoS Values

4.1.7.4 Default QoS Values
Connext DDS provides special constants for each Entity type that can be used in set_qos() and set_
default_*_qos() to reset the QosPolicy values to the original DDS default values:
l

DDS_PARTICIPANT_QOS_DEFAULT

l

DDS_PUBLISHER_QOS_DEFAULT

l

DDS_SUBSCRIBER_QOS_DEFAULT

l

DDS_DATAWRITER_QOS_DEFAULT

l

DDS_DATAREADER_QOS_DEFAULT

l

DDS_TOPIC_QOS_DEFAULT

For example, if you want to set a DataWriter’s QoS back to their DDS-specified default values:
datawriter->set_qos(DDS_DATAWRITER_QOS_DEFAULT);

Or if you want to reset the default QosPolicies used by a Publisher to create DataWriters back to their
DDS-specified default values:
publisher->set_default_datawriter_qos(DDS_DATAWRITER_QOS_DEFAULT);

These defaults cannot be used to initialize a QoS structure for an entity. For example, the following is
NOT allowed:
DataWriterQos dataWriterQos = DATAWRITER_QOS_DEFAULT;
// modify QoS...
create_datawriter(dataWriterQos);

4.2 QosPolicies
Connext DDS’s behavior is controlled by the Quality of Service (QoS) policies of the data communication
Entities (DomainParticipant, Topic, Publisher, Subscriber, DataWriter, and DataReader) used in your
applications. This section summarizes each of the QosPolicies that you can set for the various Entities.
The QosPolicy class is the abstract base class for all the QosPolicies. It provides the basic mechanism for
an application to specify quality of service parameters. Table 4.2 QosPolicies lists each supported
QosPolicy (in alphabetical order), provides a summary, and points to a section in the manual that provides
further details.
The detailed description of a QosPolicy that applies to multiple Entities is provided in the first chapter that
discusses an Entity whose behavior the QoS affects. Otherwise, the discussion of a QosPolicy can be
found in the chapter of the particular Entity to which the policy applies. As you will see in the detailed

162

4.2 QosPolicies

description sections, all QosPolicies have one or more parameters that are used to configure the policy.
The how’s and why’s of tuning the parameters are also discussed in those sections.
As first discussed in Controlling Behavior with Quality of Service (QoS) Policies (Section 2.6.1 on page
19), QosPolicies may interact with each other, and certain values of QosPolicies can be incompatible with
the values set for other policies.
The set_qos() operation will fail if you attempt to specify a set of values would result in an inconsistent set
of policies. To indicate a failure, set_qos() will return INCONSISTENT_POLICY. QoS Requested vs.
Offered Compatibility—the RxO Property (Section 4.2.1 on page 167) provides further information on
QoS compatibility within an Entity as well as across matching Entities, as does the discussion/reference section for each QosPolicy listed in Table 4.2 QosPolicies.
The values of some QosPolicies cannot be changed after the Entity is created or after the Entity is enabled.
Others may be changed at any time. The detailed section on each QosPolicy states when each policy can
be changed. If you attempt to change a QosPolicy after it becomes immutable (because the associated
Entity has been created or enabled, depending on the policy), set_qos() will fail with a return code of
IMMUTABLE_POLICY.
Table 4.2 QosPolicies
QosPolicy
AsynchronousPublisher

Summary
Configures the mechanism that sends user data in an external middleware thread. See
ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension) (Section 6.4.1 on page 313).
This QoS policy is used in the context of two features:

Availability

For a Collaborative DataWriter, specifies the group of DataWriters expected to collaboratively provide
data and the timeouts that control when to allow data to be available that may skip DDS samples.
For a Durable Subscription, configures a set of Durable Subscriptions on a DataWriter.
See AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on page 337).

Batch

Specifies and configures the mechanism that allows Connext DDS to collect multiple DDS data samples
to be sent in a single network packet, to take advantage of the efficiency of sending larger packets and thus
increase effective throughput. See BATCH QosPolicy (DDS Extension) (Section 6.5.2 on page 341).

Database

Various settings and resource limits used by Connext DDS to control its internal database. See
DATABASE QosPolicy (DDS Extension) (Section 8.5.1 on page 577).

DataReaderProtocol

This QosPolicy configures the Connext DDS on-the-network protocol, RTPS. See DATA_READER_
PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511).

DataReaderResourceLimits

Various settings that configure how DataReaders allocate and use physical memory for internal resources.
See DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 7.6.2 on page 517).

DataWriterProtocol

This QosPolicy configures the Connext DDS on-the-network protocol, RTPS. See DATA_WRITER_
PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).

163

4.2 QosPolicies

Table 4.2 QosPolicies
QosPolicy

DataWriterResourceLimits

Summary
Controls how many threads can concurrently block on a write() call of this DataWriter. Also controls the
number of batches managed by the DataWriter and the instance-replacement kind used by the DataWriter.
See DATA_WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 6.5.4 on page 359).
For a DataReader, specifies the maximum expected elapsed time between arriving DDS data samples.

Deadline

For a DataWriter, specifies a commitment to publish DDS samples with no greater elapsed time between
them.
See DEADLINE QosPolicy (Section 6.5.5 on page 363).

DestinationOrder

Controls how Connext DDS will deal with data sent by multiple DataWriters for the same topic. Can be set
to "by reception timestamp" or to "by source timestamp." See DESTINATION_ORDER QosPolicy
(Section 6.5.6 on page 365).

Discovery

Configures the mechanism used by Connext DDS to automatically discover and connect with new remote
applications. See DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580).

DiscoveryConfig

Controls the amount of delay in discovering Entities in the system and the amount of discovery traffic in the
network. See DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585).

DomainParticipantResourceLimits

Various settings that configure how DomainParticipants allocate and use physical memory for internal
resources, including the maximum sizes of various properties. See DOMAIN_PARTICIPANT_
RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593).

Durability

Specifies whether or not Connext DDS will store and deliver data that were previously published to new
DataReaders. See DURABILITY QosPolicy (Section 6.5.7 on page 368).

DurabilityService

Various settings to configure the external Persistence Service used by Connext DDS for DataWriters with
a Durability QoS setting of Persistent Durability. See DURABILITY SERVICE QosPolicy (Section 6.5.8
on page 372).

EntityFactory

Controls whether or not child Entities are created in the enabled state. See ENTITYFACTORY QosPolicy
(Section 6.4.2 on page 315).

EntityName

Assigns a name and role_name to an Entity. See ENTITY_NAME QosPolicy (DDS Extension) (Section
6.5.9 on page 374).

Event

Configures the DomainParticipant’s internal thread that handles timed events. See EVENT QosPolicy
(DDS Extension) (Section 8.5.5 on page 602).

ExclusiveArea

Configures multi-thread concurrency and deadlock prevention capabilities. See EXCLUSIVE_AREA
QosPolicy (DDS Extension) (Section 6.4.3 on page 318).

GroupData

Along with TOPIC_DATA QosPolicy (Section 5.2.1 on page 209) and USER_DATA QosPolicy
(Section 6.5.26 on page 417), this QosPolicy is used to attach a buffer of bytes to Connext DDS's
discovery meta-data. See GROUP_DATA QosPolicy (Section 6.4.4 on page 320).

164

4.2 QosPolicies

Table 4.2 QosPolicies
QosPolicy

Summary

History

Specifies how much data must be stored by Connext DDS for the DataWriter or DataReader. This
QosPolicy affects the RELIABILITY QosPolicy (Section 6.5.19 on page 400) as well as the
DURABILITY QosPolicy (Section 6.5.7 on page 368). See HISTORY QosPolicy (Section 6.5.10 on
page 376).

LatencyBudget

Suggestion to Connext DDS on how much time is allowed to deliver data. See LATENCYBUDGET QoS
Policy (Section 6.5.11 on page 380).

Lifespan

Specifies how long Connext DDS should consider data sent by an user application to be valid. See
LIFESPAN QoS Policy (Section 6.5.12 on page 381).

Liveliness

Specifies and configures the mechanism that allows DataReaders to detect when DataWriters become
disconnected or "dead." See LIVELINESS QosPolicy (Section 6.5.13 on page 382).

Logging

Configures the properties associated with Connext DDS logging. See LOGGING QosPolicy (DDS
Extension) (Section 8.4.1 on page 572).

MultiChannel

Configures a DataWriter’s ability to send data on different multicast groups (addresses) based on the value
of the data. See MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386).

Ownership

Along with Ownership Strength, specifies if DataReaders for a topic can receive data from multiple
DataWriters at the same time. See OWNERSHIP QosPolicy (Section 6.5.15 on page 389).

OwnershipStrength

Used to arbitrate among multiple DataWriters of the same instance of a Topic when Ownership QoSPolicy
is EXLUSIVE. See OWNERSHIP_STRENGTH QosPolicy (Section 6.5.16 on page 393).

Partition

Adds string identifiers that are used for matching DataReaders and DataWriters for the same Topic. See
PARTITION QosPolicy (Section 6.4.5 on page 323).

Presentation

Controls how Connext DDS presents data received by an application to the DataReaders of the data. See
PRESENTATION QosPolicy (Section 6.4.6 on page 330).

Profile

Configures the way that XML documents containing QoS profiles are loaded by RTI. See PROFILE
QosPolicy (DDS Extension) (Section 8.4.2 on page 573).

Property

Stores name/value(string) pairs that can be used to configure certain parameters of Connext DDS that are
not exposed through formal QoS policies. It can also be used to store and propagate application-specific
name/value pairs, which can be retrieved by user code during discovery. See PROPERTY QosPolicy
(DDS Extension) (Section 6.5.17 on page 394).

PublishMode

Specifies how Connext DDS sends application data on the network. By default, data is sent in the user
thread that calls the DataWriter’s write() operation. However, this QosPolicy can be used to tell Connext
DDS to use its own thread to send the data. See PUBLISH_MODE QosPolicy (DDS Extension) (Section
6.5.18 on page 397).

ReaderDataLifeCycle

Controls how a DataReader manages the lifecycle of the data that it has received. See READER_DATA_
LIFECYCLE QoS Policy (Section 7.6.3 on page 523).

165

4.2 QosPolicies

Table 4.2 QosPolicies
QosPolicy

Summary

ReceiverPool

Configures threads used by Connext DDS to receive and process data from transports (for example, UDP
sockets). See RECEIVER_POOL QosPolicy (DDS Extension) (Section 8.5.6 on page 604).

Reliability

Specifies whether or not Connext DDS will deliver data reliably. See RELIABILITY QosPolicy (Section
6.5.19 on page 400).

ResourceLimits

Controls the amount of physical memory allocated for Entities, if dynamic allocations are allowed, and how
they occur. Also controls memory usage among different instance values for keyed topics. See
RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405).

Service

Intended for use by RTI infrastructure services. User applications should not modify its value. See
SERVICE QosPolicy (DDS Extension) (Section 6.5.21 on page 408).

SystemResourceLimits

Configures DomainParticipant-independent resources used by Connext DDS. Mainly used to change the
maximum number of DomainParticipants that can be created within a single process (address space). See
SYSTEM_RESOURCE_LIMITS QoS Policy (DDS Extension) (Section 8.4.3 on page 575).

TimeBasedFilter

Set by a DataReader to limit the number of new data values received over a period of time. See TIME_
BASED_FILTER QosPolicy (Section 7.6.4 on page 526).

TopicData

Along with Group Data QosPolicy and User Data QosPolicy, used to attach a buffer of bytes to Connext
DDS's discovery meta-data. See TOPIC_DATA QosPolicy (Section 5.2.1 on page 209).

TransportBuiltin

Specifies which built-in transport plugins are used. See TRANSPORT_BUILTIN QosPolicy (DDS
Extension) (Section 8.5.7 on page 606).

TransportMulticast

Specifies the multicast address on which a DataReader wants to receive its data. Can specify a port number
as well as a subset of the available transports with which to receive the multicast data. See TRANSPORT_
MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on page 529).

TransportMulticastMapping

Specifies the automatic mapping between a list of topic expressions and multicast address that can be used
by a DataReader to receive data for a specific topic. See TRANSPORT_MULTICAST_MAPPING
QosPolicy (DDS Extension) (Section 8.5.8 on page 608).

TransportPriority

Set by a DataWriter or DataReader to tell Connext DDS that the data being sent is a different "priority"
than other data. See TRANSPORT_PRIORITY QosPolicy (Section 6.5.22 on page 409).

TransportSelection

Allows you to select which physical transports a DataWriter or DataReader may use to send or receive its
data. See TRANSPORT_SELECTION QosPolicy (DDS Extension) (Section 6.5.23 on page 411).

TransportUnicast

Specifies a subset of transports and port number that can be used by an Entity to receive data. See
TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412).

Defines rules that determine whether the type used to publish a given data stream is consistent with that
TypeConsistencyEnforcement used to subscribe to it. See TYPE_CONSISTENCY_ENFORCEMENT QosPolicy (Section 7.6.6 on page
532).

166

4.2.1 QoS Requested vs. Offered Compatibility—the RxO Property

Table 4.2 QosPolicies
QosPolicy

Summary

TypeSupport

Used to attach application-specific value(s) to a DataWriter or DataReader. These values are passed to the
serialization or deserialization routine of the associated data type. Also controls whether padding bytes are
set to 0 during serialization. See TYPESUPPORT QosPolicy (DDS Extension) (Section 6.5.25 on page
416).

UserData

Along with Topic Data QosPolicy and Group Data QosPolicy, used to attach a buffer of bytes to Connext
DDS's discovery meta-data. See USER_DATA QosPolicy (Section 6.5.26 on page 417).

WireProtocol

Specifies IDs used by the RTPS wire protocol to create globally unique identifiers. See WIRE_
PROTOCOL QosPolicy (DDS Extension) (Section 8.5.9 on page 610).

WriterDataLifeCycle

Controls how a DataWriter handles the lifecycle of the instances (keys) that the DataWriter is registered to
manage. See WRITER_DATA_LIFECYCLE QoS Policy (Section 6.5.27 on page 419).

4.2.1 QoS Requested vs. Offered Compatibility—the RxO Property
Some QosPolicies that apply to Entities on the sending and receiving sides must have their values set in a
compatible manner. This is known as the policy’s ‘requested vs. offered’ (RxO) property. Entities on the
publishing side ‘offer’ to provide a certain behavior. Entities on the subscribing side ‘request’ certain behavior. For Connext DDS to connect the sending entity to the receiving entity, the offered behavior must satisfy the requested behavior.
For some QosPolicies, the allowed values may be graduated in a way that the offered value will satisfy the
requested value if the offered value is either greater than or less than the requested value. For example, if a
DataWriter’s DEADLINE QosPolicy specifies a duration less than or equal to a DataReader’s
DEADLINE QosPolicy, then the DataWriter is promising to publish data at least as fast or faster than the
DataReader requires new data to be received. This is a compatible situation (see DEADLINE QosPolicy
(Section 6.5.5 on page 363)).
Other QosPolicies require the values on the sending side and the subscribing side to be exactly equal for
compatibility to be met. For example, if a DataWriter’s OWNERSHIP QosPolicy is set to SHARED, and
the matching DataReader’s value is set to EXCLUSIVE, then this is an incompatible situation since the
DataReader and DataWriter have different expectations of what will happen if more than one DataWriter
publishes an instance of the Topic (see OWNERSHIP QosPolicy (Section 6.5.15 on page 389)).
Finally there are QosPolicies that do not require compatibility between the sending entity and the receiving
entity, or that only apply to one side or the other. Whether or not related Entities on the publishing and subscribing sides must use compatible settings for a QosPolicy is indicated in the policy’s RxO property,
which is provided in the detailed section on each QosPolicy.
l

RxO = YES The policy is set at both the publishing and subscribing ends and the values must be set
in a compatible manner. What it means to be compatible is defined by the QosPolicy.

167

4.2.2 Special QosPolicy Handling Considerations for C

l

RxO = NO The policy is set only on one end or at both the publishing and subscribing ends, but the
two settings are independent. There the requested vs. offered semantics are not used for these
QosPolicies.

For those QosPolicies that follow the RxO semantics, Connext DDS will compare the values of those
policies for compatibility. If they are compatible, then Connext DDS will connect the sending entity to the
receiving entity allowing data to be sent between them. If they are found to be incompatible, then Connext
DDS will not interconnect the Entities preventing data to be sent between them.
In addition, Connext DDS will record this event by changing the associated communication status in both
the sending and receiving applications, see Types of Communication Status (Section 4.3.1 on page 170).
Also, if you have installed Listeners on the associated Entities, then Connext DDS will invoke the associated callback functions to notify user code that an incompatible QoS combination has been found, see
Types of Listeners (Section 4.4.1 on page 177).
For Publishers and DataWriters, the status corresponding to this situation is OFFERED_
INCOMPATIBLE_QOS_STATUS. For Subscribers and DataReaders, the corresponding status is
REQUESTED_INCOMPATIBLE_QOS_STATUS. The question of why a DataReader is not receiving data sent from a matching DataWriter can often be answered if you have instrumented the application
with Listeners for the statuses noted previously.

4.2.2 Special QosPolicy Handling Considerations for C
Many QosPolicy structures contain variable-length sequences to store their parameters. In the C++,
C++/CLI, C# and Java languages, the memory allocation related to sequences are handled automatically
through constructors/destructors and overloaded operators. However, the C language is limited in what it
provides to automatically handle memory management. Thus, Connext DDS provides functions and macros in C to initialize, copy, and finalize (free) QosPolicy structures defined for Entities.
In the C language, it is not safe to use an Entity’s QosPolicy structure declared in user code unless it has
been initialized first. In addition, user code should always finalize an Entity’s QosPolicy structure to
release any memory allocated for the sequences–even if the Entity’s QosPolicy structure was declared as a
local, stack variable.
Thus, for a general Entity’s QosPolicy, Connext DDS will provide:
l

DDS_Qos_INITIALIZER This is a macro that should be used when a DDS_
Qos structure is declared in a C application.
struct DDS_Qos qos = DDS_Qos_INITIALIZER;

l

DDS_Qos_initialize() This is a function that can be used to initialize a DDS_
Qos structure instead of the macro above.

168

4.3 Statuses

struct DDS_Qos qos;
DDS_QOS_initialize(&qos);

l

DDS_Qos_finalize() This is a function that should be used to finalize a DDS_
Qos structure when the structure is no longer needed. It will free any memory allocated for
sequences contained in the structure.
struct DDS_Qos qos = DDS_Qos_INITIALIZER;
...

...
// now done with qos
DDS_QoS_finalize(&qos);

l

DDSQos_copy() This is a function that can be used to copy one DDS_Qos
structure to another. It will copy the sequences contained in the source structure and allocate
memory for sequence elements if needed. In the code below, both dstQos and srcQos must have
been initialized at some point earlier in the code.
DDS_QOS_copy(&dstQos, &srcQos);

4.3 Statuses
This section describes the different statuses that exist for an entity. A status represents a state or an event
regarding the entity. For instance, maybe Connext DDS found a matching DataReader for a DataWriter,
or new data has arrived for a DataReader.
Your application can retrieve an Entity’s status by:
l

explicitly checking for any status changes with get_status_changes().

l

explicitly checking a specific status with get__status().

l

using a Listener, which provides asynchronous notification when a status changes.

l

using StatusConditions and WaitSets, which provide a way to wait for status changes.

If you want your application to be notified of status changes asynchronously: create and install a Listener
for the Entity. Then internal Connext DDS threads will call the listener methods when the status changes.
See Listeners (Section 4.4 on page 177).
If you want your application to wait for status changes: set up StatusConditions to indicate the statuses of
interest, attach the StatusConditions to a WaitSet, and then call the WaitSet’s wait() operation. The call to
wait() will block until statuses in the attached Conditions changes (or until a timeout period expires). See
Conditions and WaitSets (Section 4.6 on page 187).

169

4.3.1 Types of Communication Status

This section includes the following:

4.3.1 Types of Communication Status
Each Entity is associated with a set of Status objects representing the “communication status” of that
Entity. The list of statuses actively monitored by Connext DDS is provided in Table 4.3 Communication
Statuses. A status structure contains values that give you more information about the status; for example,
how many times the event has occurred since the last time the user checked the status, or how many time
the event has occurred in total.
Changes to status values cause activation of corresponding StatusCondition objects and trigger invocation
of the corresponding Listener functions to asynchronously inform the application that the status has
changed. For example, a change in a Topic’s INCONSISTENT_TOPIC_STATUS may trigger the TopicListener’s on_inconsistent_topic() callback routine (if such a Listener is installed).
Table 4.3 Communication Statuses
Related
Entity

Topic

Status (DDS_*_
STATUS)
INCONSISTENT_
TOPIC

Description

Another Topic exists with the same name but different characteristics—
for example, a different type.

This status indicates that a DataWriter has received an application-level
acknowledgment for a DDS sample. The listener provides the identities
APPLICATION_
of the DDS sample and acknowledging DataReader, as well as userACKNOWLEDGMENT
specified response data sent from the DataReader by the
acknowledgment message.

DataWriter

DATA_WRITER_
CACHE

DATA_WRITER_
PROTOCOL

The status of the DataWriter’s cache.
This status does not have a Listener.

The status of a DataWriter’s internal protocol related metrics (such as
the number of DDS samples pushed, pulled, filtered) and the status of
wire protocol traffic.
This status does not have a Listener.

Reference
INCONSISTENT_
TOPIC Status (Section
5.3.1 on page 211)
Application
Acknowledgment
(Section 6.3.12 on page
288)
DATA_WRITER_
CACHE_STATUS
(Section 6.3.6.2 on page
272)
DATA_WRITER_
PROTOCOL_STATUS
(Section 6.3.6.3 on page
273)

170

4.3.1 Types of Communication Status

Table 4.3 Communication Statuses
Related
Entity

Status (DDS_*_
STATUS)

Description

Reference

LIVELINESS_LOST

The liveliness that the DataWriter has committed to (through its
Liveliness QosPolicy) was not respected (assert_liveliness() or write()
not called in time), thus DataReaders may consider the DataWriter as
no longer active.

LIVELINESS_LOST
Status (Section 6.3.6.4
on page 276)

OFFERED_
DEADLINE_
MISSED

The deadline that the DataWriter has committed through its Deadline
QosPolicy was not respected for a specific instance of the Topic.

OFFERED_
DEADLINE_MISSED
Status (Section 6.3.6.5
on page 277)

OFFERED_
INCOMPATIBLE_
QOS

An offered QosPolicy value was incompatible with what was requested
by a DataReader of the same Topic.

OFFERED_
INCOMPATIBLE_
QOS Status (Section
6.3.6.6 on page 277)

PUBLICATION_
MATCHED

The DataWriter found a DataReader that matches the Topic, has
compatible QoSs and a common partition, or a previously matched
DataReader has been deleted.

PUBLICATION_
MATCHED Status
(Section 6.3.6.7 on page
278)

The number of unacknowledged DDS samples in a reliable
DataWriter's cache has reached one of the predefined trigger points.

RELIABLE_WRITER_
CACHE_CHANGED
Status (DDS Extension)
(Section 6.3.6.8 on page
279)

RELIABLE_READER_
ACTIVITY_
CHANGED

One or more reliable DataReaders has either been discovered, deleted,
or changed between active and inactive state as specified by the
LivelinessQosPolicy of the DataReader.

RELIABLE_READER_
ACTIVITY_
CHANGED Status
(DDS Extension)
(Section 6.3.6.9 on page
281)

Subscriber DATA_ON_READERS

New data is available for any of the readers that were created from the
Subscriber.

Statuses for Subscribers
(Section 7.2.9 on page
458)

DataWriter
cont'd

RELIABLE_WRITER_
CACHE_CHANGED

171

4.3.1 Types of Communication Status

Table 4.3 Communication Statuses
Related
Entity

Status (DDS_*_
STATUS)

DATA_AVAILABLE

DATA_READER_
CACHE
DataReader

DataReader
cont'd

Description

New data (one or more DDS samples) are available for the specific
DataReader.

The status of the reader's cache.
This status does not have a Listener.

Reference
DATA_AVAILABLE
Status (Section 7.3.7.1
on page 471)
DATA_READER_
CACHE_STATUS
(Section 7.3.7.2 on page
471)

DATA_READER_
PROTOCOL

The status of a DataReader’s internal protocol related metrics (such as
DATA_READER_
the number of DDS samples received, filtered, rejected) and the status of PROTOCOL_STATUS
wire protocol traffic.
(Section 7.3.7.3 on page
472)
This status does not have a Listener.

LIVELINESS_
CHANGED

The liveliness of one or more DataWriters that were writing instances
read by the DataReader has either been discovered, deleted, or changed
between active and inactive state as specified by the
LivelinessQosPolicy of the DataWriter.

LIVELINESS_
CHANGED Status
(Section 7.3.7.4 on page
475)

REQUESTED_
DEADLINE_
MISSED

New data was not received for an instance of the Topic within the time
period set by the DataReader’s Deadline QosPolicy.

REQUESTED_
DEADLINE_MISSED
Status (Section 7.3.7.5
on page 476)

REQUESTED_
A requested QosPolicy value was incompatible with what was offered
INCOMPATIBLE_QOS by a DataWriter of the same Topic.

REQUESTED_
INCOMPATIBLE_
QOS Status (Section
7.3.7.6 on page 477)

SAMPLE_LOST

A DDS sample sent by Connext DDS has been lost (never received).

SAMPLE_LOST Status
(Section 7.3.7.7 on page
478)

SAMPLE_REJECTED

A received DDS sample has been rejected due to a resource limit
(buffers filled).

SAMPLE_REJECTED
Status (Section 7.3.7.8
on page 479)

SUBSCRIPTION_
MATCHED

The DataReader has found a DataWriter that matches the Topic, has
compatible QoSs and a common partition, or an existing matched
DataWriter has been deleted.

SUBSCRIPTION_
MATCHED Status
(Section 7.3.7.9 on page
482)

Statuses can be grouped into two categories:

172

4.3.1.1 Changes in Plain Communication Status

l

Plain communication status:
In addition to a flag that indicates whether or not a status has changed, a plain communication status
also contains state and thus has a corresponding structure to hold its current value.

l

Read communication status:
A read communication status is more like an event and has no state other than whether or not it has
occurred. Only two statuses listed in Table 4.3 Communication Statuses are read communications
statuses: DATA_AVAILABLE and DATA_ON_READERS.

As mentioned in Getting Status and Status Changes (Section 4.1.4 on page 157), all Entities have a get_
status_changes() operation that can be used to explicitly poll for changes in any status related to the entity.
For plain statuses, each entry has operations to get the current value of the status; for example, the Topic class
has a get_inconsistent_topic_status() operation. For read statuses, your application should use the take()
operation on the DataReader to retrieve the newly arrived data that is indicated by DATA_AVAILABLE
and DATA_ON_READER.
Note that the two read communication statuses do not change independently. If data arrives for a DataReader,
then its DATA_AVAILABLE status changes. At the same time, the DATA_ON_READERS status
changes for the DataReader’s Subscriber.
Both types of status have a StatusChangedFlag. This flag indicates whether that particular communication status has changed since the last time the status was read by the application. The way the
StatusChangedFlag is maintained is slightly different for the plain communication status and the read communication status, as described in the following sections:
l

Changes in Plain Communication Status (Section 4.3.1.1 below)

l

Changes in Read Communication Status (Section 4.3.1.2 on the next page)

4.3.1.1 Changes in Plain Communication Status
As seen in Figure 4.2 Status Changes for Plain Communication Status on the next page, for the plain communication status, the StatusChangedFlag flag is initially set to FALSE. It becomes TRUE whenever the
plain communication status changes and is reset to FALSE each time the application accesses the plain
communication status via the proper get_*_status() operation.

173

4.3.1.2 Changes in Read Communication Status

Figure 4.2 Status Changes for Plain Communication Status

The communication status is also reset to FALSE whenever the associated listener operation is called, as
the listener implicitly accesses the status which is passed as a parameter to the operation.
The fact that the status is reset prior to calling the listener means that if the application calls the get_*_
status() operation from inside the listener, it will see the status already reset.
An exception to this rule is when the associated listener is the 'nil' listener. The 'nil' listener is treated as a
NO-OP and the act of calling the 'nil' listener does not reset the communication status. (See Types of
Listeners (Section 4.4.1 on page 177).)
For example, the value of the StatusChangedFlag associated with the REQUESTED_DEADLINE_
MISSED status will become TRUE each time new deadline occurs (which increases the RequestedDeadlineMissed status’ total_count field). The value changes to FALSE when the application accesses
the status via the corresponding get_requested_deadline_missed_status() operation on the proper Entity.
4.3.1.2 Changes in Read Communication Status
As seen in Figure 4.3 Status Changes for Read Communication Status on the facing page, for the read
communication status, the StatusChangedFlag flag is initially set to FALSE. The StatusChangedFlag
becomes TRUE when either a DDS data sample arrives or the ViewStateKind, SampleStateKind, or
InstanceStateKind of any existing DDS sample changes for any reason other than a call to one of the read/take operations. Specifically, any of the following events will cause the StatusChangedFlag to become
TRUE:
l
l

The arrival of new data.
A change in the InstanceStateKind of a contained instance. This can be caused by either:
l Notification that an instance has been disposed by:
l the DataWriter that owns it, if OWNERSHIP = EXCLUSIVE
l

l

or by any DataWriter, if OWNERSHIP = SHARED

The loss of liveliness of the DataWriter of an instance for which there is no other DataWriter.

174

4.3.1.2 Changes in Read Communication Status

l

The arrival of the notification that an instance has been unregistered by the only DataWriter
that is known to be writing the instance.

Depending on the kind of StatusChangedFlag, the flag transitions to FALSE (that is, the status is reset)
as follows:
l

l

The DATA_AVAILABLE StatusChangedFlag becomes FALSE when either on_data_available
() is called or the read/take operation (or their variants) is called on the associated DataReader.
The DATA_ON_READERS StatusChangedFlag becomes FALSE when any of the following
occurs:
l on_data_on_readers() is called.
l
l

on_data_available() is called on any DataReader belonging to the Subscriber.
read(), take(), or one of their variants is called on any DataReader belonging to the Subscriber.

Figure 4.3 Status Changes for Read Communication Status

175

4.3.2 Special Status-Handling Considerations for C

4.3.2 Special Status-Handling Considerations for C
Some status structures contain variable-length sequences to store their values. In the C++, C++/CLI, C#
and Java languages, the memory allocation related to sequences are handled automatically through constructors/destructors and overloaded operators. However, the C language is limited in what it provides to
automatically handle memory management. Thus, Connext DDS provides functions and macros in C to initialize, copy, and finalize (free) status structures.
In the C language, it is not safe to use a status structure that has internal sequences declared in user code
unless it has been initialized first. In addition, user code should always finalize a status structure to release
any memory allocated for the sequences–even if the status structure was declared as a local, stack variable.
Thus, for a general status structure, Connext DDS will provide:
l

DDS_STATUS_INITIALIZER This is a macro that should be used when a DDS_
Status structure is declared in a C application.
struct DDS_Status status =
DDS_Status_INITIALIZER;

176

4.4 Listeners

l

DDS_Status_initialize() This is a function that can be used to initialize a DDS_
Status structure instead of the macro above.
struct DDS_Status status;
DDS_Status_initialize(&Status);

l

DDS_Status_finalize() This is a function that should be used to finalize a DDS_
Status structure when the structure is no longer needed. It will free any memory allocated
for sequences contained in the structure.
struct DDS_Status status =
DDS_Status_INITIALIZER;
...

...
// now done with Status
DDS_Status_finalize(&status);

l

DDSStatus_copy() This is a function that can be used to copy one DDS_Status
structure to another. It will copy the sequences contained in the source structure and allocate
memory for sequence elements if needed. In the code below, both dstStatus and srcStatus must
have been initialized at some point earlier in the code.
DDS_Status_copy(&dstStatus, &srcStatus);

Note that many status structures do not have sequences internally. For those structures, you do not need to
use the macro and methods provided above. However, they have still been created for your convenience.

4.4 Listeners
Listeners are triggered by changes in an entity’s status. For instance, maybe Connext DDS found a matching DataReader for a DataWriter, or new data has arrived for a DataReader.
This section describes Listeners and how to use them:

4.4.1 Types of Listeners
The Listener class is the abstract base class for all listeners. Each entity class (DomainParticipant, Topic,
Publisher, DataWriter, Subscriber, and DataReader) has its own derived Listener class that add methods
for handling entity-specific statuses. The hierarchy of Listener classes is presented in Figure 4.4 Listener
Class Hierarchy on the next page. The methods are called by an internal Connext DDS thread when the
corresponding status for the Entity changes value.

177

4.4.1 Types of Listeners

Figure 4.4 Listener Class Hierarchy

You can choose which changes in status will trigger a callback by installing a listener with a bit-mask. Bits
in the mask correspond to different statuses. The bits that are true indicate that the listener will be called
back when there are changes in the corresponding status.
You can specify a listener and set its bit-mask before or after you create an Entity:
178

4.4.2 Creating and Deleting Listeners

During Entity creation:
DDS_StatusMask mask = DDS_REQUESTED_DEADLINE_MISSED_STATUS |
DDS_DATA_AVAILABLE_STATUS;
datareader = subscriber->create_datareader(topic,
DDS_DATAREADER_QOS_DEFAULT,
listener, mask);

or afterwards:
DDS_StatusMask mask = DDS_REQUESTED_DEADLINE_MISSED_STATUS |
DDS_DATA_AVAILABLE_STATUS;
datareader->set_listener(listener, mask);

As you can see in the above examples, there are two components involved when setting up listeners: the
listener itself and the mask. Both of these can be null. Table 4.4 Effect of Different Combinations of Listeners and Status Bit Masks describes what happens when a status change occurs. See Hierarchical Processing of Listeners (Section 4.4.4 on the next page) for more information.
Table 4.4 Effect of Different Combinations of Listeners and Status Bit Masks
No Bits Set in Mask
Listener is Connext DDS finds the next most relevant listener for the
Specified changed status.

Some/All Bits Set in Mask
For the statuses that are enabled in the mask, the most relevant
listener will be called.
The 'statusChangedFlag' for the relevant status is reset.

Listener is Connext DDS behaves as if the listener is not installed and Connext DDS behaves as if the listener callback is installed, but
NULL
finds the next most relevant listener for that status.
the callback is doing nothing. This is called a ‘nil’ listener.

4.4.2 Creating and Deleting Listeners
There is no factory for creating or deleting a Listener; use the natural means in each language binding (for
example, “new” or “delete” in C++ or Java). For example:
class HelloWorldListener : public DDSDataReaderListener {
virtual void on_data_available(DDSDataReader* reader);
};
void HelloWorldListener::on_data_available(DDSDataReader* reader)
{
printf("received data\n");
}
// Create a Listener
HelloWorldListener *reader_listener = NULL;
reader_listener = new HelloWorldListener();
// Delete a Listener
delete reader_listener;

179

4.4.3 Special Considerations for Listeners in C

A listener cannot be deleted until the entity it is attached to has been deleted. For example, you must delete
the DataReader before deleting the DataReader’s listener.
Note: Due to a thread-safety issue, the destruction of a DomainParticipantListener from an enabled
DomainParticipant should be avoided—even if the DomainParticipantListener has been removed from
the DomainParticipant. (This limitation does not affect the Java API.)

4.4.3 Special Considerations for Listeners in C
In C, a Listener is a structure with function pointers to the user callback routines. Often, you may only be
interested in a subset of the statuses that can be monitored with the Listener. In those cases, you may not
set all of the functions pointers in a listener structure to a valid function. In that situation, we recommend
that the unused, callback-function pointers are set to NULL. While setting the DDS_StatusMask to
enable only the callbacks for the statuses in which you are interested (and thus only enabling callbacks on
the functions that actually exist) is safe, we still recommend that you clear all of the unused callback pointers in the Listener structure.
To help, in the C language, we provide a macro that can be used to initialize a Listener structure so that all
of its callback pointers are set to NULL. For example
DDS_Listener listener = DDS_Listener_INITIALIZER;
// now only need to set the listener callback pointers for statuses // to be monitored

There is no need to do this in languages other than C.

4.4.4 Hierarchical Processing of Listeners
As seen in Listener Class Hierarchy (Section Figure 4.4 on page 178), Listeners for some Entities derive
from the Connext DDS Listeners for related Entities. This means that the derived Listener has all of the methods of its parent class. You can install Listeners at all levels of the object hierarchy. At the top is the
DomainParticipantListener; only one can be installed in a DomainParticipant. Then every Subscriber and
Publisher can have their own Listener. Finally, each Topic, DataReader and DataWriter can have their
own listeners. All are optional.
Suppose, however, that an Entity does not install a Listener, or installs a Listener that does not have particular communication status selected in the bitmask. In this case, if/when that particular status changes for
that Entity, the corresponding Listener for that Entity’s parent is called. Status changes are “propagated”
from child Entity to parent Entity until a Listener is found that is registered for that status. Connext DDS
will give up and drop the status-change event only if no Listeners have been installed in the object hierarchy to be called back for the specific status. This is true for plain communication statuses. Read communication statuses are handle somewhat differently, see Processing Read Communication Statuses
(Section 4.4.4.1 on the facing page).
For example, suppose that Connext DDS finds a matching DataWriter for a local DataReader. This event
will change the SUBSCRIPTION_MATCHED status. So the local DataReader object is checked to see

180

4.4.4.1 Processing Read Communication Statuses

if the application has installed a listener that handles the SUBSCRIPTION_MATCH status. If not, the
Subscriber that created the DataReader is checked to see if it has a listener installed that handles the same
event. If not, the DomainParticipant is checked. The DomainParticipantListener methods are called only
if none of the descendent Entities of the DomainParticipant have listeners that handle the particular status
that has changed. Again, all listeners are optional. Your application does not have to handle any communication statuses.
Table 4.5 Listener Callback Functions lists the callback functions that are available for each Entity’s status
listener.
Table 4.5 Listener Callback Functions
Entity Listener for:

Callback Functions
Topics

on_inconsistent_topic()
on_liveliness_lost()
on_offered_deadline_missed()
on_offered_incompatible_qos()

Publishers and DataWriters
on_publication_matched()
on_reliable_reader_activity_changed()
on_reliable_writer_cache_changed()
DomainParticipants

Subscribers

on_data_on_readers()
on_data_available
on_liveliness_changed()
on_requested_deadline_missed()

Subscribers and DataReaders

on_requested_incompatible_qos()
on_sample_lost()
on_sample_rejected()
on_subscription_matched()

4.4.4.1 Processing Read Communication Statuses
The processing of the DATA_ON_READERS and DATA_AVAILABLE read communication
statuses are handled slightly differently since, when new data arrives for a DataReader, both statuses
change simultaneously. However, only one, if any, Listener will be called to handle the event.
181

4.4.5 Operations Allowed within Listener Callbacks

If there is a Listener installed to handle the DATA_ON_READERS status in the DataReader’s Subscriber or in the DomainParticipant, then that Listener’s on_data_on_readers() function will be called
back. The DataReaderListener’s on_data_available() function is called only if the DATA_ON_
READERS status is not handle by any relevant listeners.
This can be useful if you have generic processing to do whenever new data arrives for any DataReader.
You can execute the generic code in the on_data_on_readers() method, and then dispatch the processing
of the actual data to the specific DataReaderListener’s on_data_available() function by calling the
notify_datareaders() method on the Subscriber.
For example:
void on_data_on_readers (DDSSubscriber *subscriber)
{
// Do some general processing that needs to be done
// whenever new data arrives, but is independent of
// any particular DataReader
< generic processing code here >
// Now dispatch the actual processing of the data
// to the specific DataReader for which the data
// was received
subscriber->notify_datareaders();
}

4.4.5 Operations Allowed within Listener Callbacks
Due to the potential for deadlock, some Connext DDS APIs should not be invoked within the functions of
listener callbacks. Exactly which Connext DDS APIs are restricted depends on the Entity upon which the
Listener is installed, as well as the configuration of ‘Exclusive Areas,’ as discussed in Exclusive Areas
(EAs) (Section 4.5 below).
Please read and understand Exclusive Areas (EAs) (Section 4.5 below) and Restricted Operations in
Listener Callbacks (Section 4.5.1 on page 185) to ensure that the calls made from your Listeners are
allowed and will not cause potential deadlock situations.

4.5 Exclusive Areas (EAs)
Listener callbacks are invoked by internal Connext DDS threads. To prevent undesirable, multi-threaded
interaction, the internal threads may take and hold semaphores (mutexes) used for mutual exclusion. In
your listener callbacks, you may want to invoke functions provided by the Connext DDS API. Internally,
those Connext DDS functions also may take mutexes to prevent errors due to multi-threaded access to critical data or operations.
Once there are multiple mutexes to protect different critical regions, the possibility for deadlock exists. Consider Figure 4.5 Multiple Mutexes Leading to a Deadlock Condition on the facing page’s scenario, in
which there are two threads and two mutexes.

182

4.5 Exclusive Areas (EAs)

Figure 4.5 Multiple Mutexes Leading to a Deadlock Condition

Thread1 takes MutexA while simultaneously Thread2 takes MutexB. Then, Thread1 takes MutexB and simultaneously
Thread2 takes MutexA. Now both threads are blocked since they hold a mutex that the other thread is trying to take.
This is a deadlock condition.

While the probability of entering the deadlock situation in Figure 4.5 Multiple Mutexes Leading to a Deadlock Condition above depends on execution timing, when there are multiple threads and multiple mutexes,
care must be taken in writing code to prevent those situations from existing in the first place. Connext
DDS has been carefully created and analyzed so that we know our threads internally are safe from deadlock interactions.
However, when Connext DDS threads that are holding mutexes call user code in listeners, it is possible for
user code to inadvertently cause the threads to deadlock if Connext DDS APIs that try to take other
mutexes are invoked. To help you avoid this situation, RTI has defined a concept known as Exclusive
Areas, some restrictions regarding the use of Connext DDS APIs within user callback code, and a QoS
policy that allows you to configure Exclusive Areas.
Connext DDS uses Exclusive Areas (EAs) to encapsulate mutexes and critical regions. Only one thread at
a time can be executing code within an EA. The formal definition of EAs and their implementation
ensures safety from deadlock and efficient entering and exiting of EAs. While every Entity created by Connext DDS has an associated EA, EAs may be shared among several Entities. A thread is automatically in
the entity's EA when it is calling the entity’s listener.
Connext DDS allows you to configure all the Entities within an application in a single DDS domain to
share a single Exclusive Area. This would greatly restrict the concurrency of thread execution within Connext DDS’s multi-threaded core. However, doing so would release all restrictions on using Connext DDS
APIs within your callback code.

183

4.5 Exclusive Areas (EAs)

You may also have the best of both worlds by configuring a set of Entities to share a global EA and others
to have their own. For the Entities that have their own EAs, the types of Connext DDS operations that you
can call from the Entity’s callback are restricted.
To understand why the general EA framework limits the operations that can be called in an EA, consider a
modification to the example previously presented in Figure 4.5 Multiple Mutexes Leading to a Deadlock
Condition on the previous page. Suppose we create a rule that is followed when we write our code. “For
all situations in which a thread has to take multiple mutexes, we write our code so that the mutexes are
always taken in the same order.” Following the rule will ensure us that the code we write cannot enter a
deadlock situation due to the taking of the mutexes, see Figure 4.6 Taking Multiple Mutexes in a Specific
Order to Eliminate Deadlock below.
Figure 4.6 Taking Multiple Mutexes in a Specific Order to Eliminate Deadlock

By creating an order in which multiple mutexes are taken, you can guarantee that no deadlock situation will arise. In
this case, if a thread must take both MutexA and MutexB, we write our code so that in those cases MutexA is always
taken before MutexB.

Connext DDS defines an ordering of the mutexes it creates. Generally speaking, there are three ordered
levels of Exclusive Areas:
l

ParticipantEA
There is only one ParticipantEA per participant. The creation and deletion of all Entities (create_
xxx(), delete_xxx()) take the ParticipantEA. In addition, the enable() method for an Entity and the
setting of the Entity’s QoS, set_qos(), also take the ParticipantEA. There are other functions that
take the ParticipantEA: get_discovered_participants(), get_publishers(), get_subscribers(), get_

184

4.5.1 Restricted Operations in Listener Callbacks

discovered_topics(), ignore_participant(), ignore_topic(), ignore_publication(), ignore_subscription(), remove_peer(), and register_type().
l

SubscriberEA
This EA is created on a per-Subscriber basis by default. You can assume that the methods of a Subscriber will take the SubscriberEA. In addition, the DataReaders created by a Subscriber share the
EA of its parent. This means that the methods of a DataReader (including take() and read()) will
take the EA of its Subscriber. Therefore, operations on DataReaders of the same Subscriber, will
be serialized, even when invoked from multiple concurrent application threads. As mentioned, the
enable() and set_qos() methods of both Subscribers and DataReaders will take the ParticipantEA.
The same is true for the create_datareader() and delete_datareader() methods of the Subscriber.

l

PublisherEA
This EA is created on a per-Publisher basis by default. You can assume that the methods of a Publisher will take the PublisherEA. In addition, the DataWriters created by a Publisher share the EA
of its parent. This means that the methods of a DataWriter including write() will take the EA of its
Publisher. Therefore, operations on DataWriters of the same Publisher will be serialized, even
when invoked from multiple concurrent application threads. As mentioned, the enable() and set_
qos() methods of both Publishers and DataWriters will take the ParticipantEA, as well as the create_datawriter() and delete_datawriter() methods of the Publisher.

In addition, you should also be aware that:
l

l

l

The three EA levels are ordered in the following manner:
ParticipantEA < SubscriberEA < PublisherEA
When executing user code in a listener callback of an Entity, the internal Connext DDS thread is
already in the EA of that Entity or used by that Entity.
If a thread is in an EA, it can call methods associated with either a higher EA level or that share the
same EA. It cannot call methods associated with a lower EA level nor ones that use a different EA
at the same level.

4.5.1 Restricted Operations in Listener Callbacks
Based on the background and rules provided in Exclusive Areas (EAs) (Section 4.5 on page 182), this section describes how EAs restrict you from using various Connext DDS APIs from within the Listener callbacks of different Entities. Reader callbacks take the SubscriberEA. Writer callbacks take the
PublisherEA. DomainParticipant callbacks take the ParticipantEA.
These restrictions do not apply to builtin topic listener callbacks.

185

4.5.1 Restricted Operations in Listener Callbacks

By default, each Publisher and Subscriber creates and uses its own EA, and shares it with its children
DataWriters and DataReaders, respectively. In that case:
Within a DataWriter/DataReader’s Listener callback, do not:
l

Create any Entities

l

Delete any Entities

l

Enable any Entities

l

Set QoS on any Entities

Within a Subscriber/DataReader’s Listener callback, do not call any operations on:
l

Other Subscribers

l

DataReaders that belong to other Subscribers

l

Publishers/DataWriters that have been configured to use the ParticipantEA (see below)

Within a Publisher/DataWriter Listener callback, do not call any operations on:
l

Other Publishers

l

DataWriters that belong to other Publishers

l

Any Subscribers

l

Any DataReaders

Connext DDS will enforce the rules to avoid deadlock, and any attempt to call an illegal method from
within a Listener callback will return DDS_RETCODE_ILLEGAL_OPERATION.
However, as previously mentioned, if you are willing to trade-off concurrency for flexibility, you may configure individual Publishers and Subscribers (and thus their DataWriters and DataReaders) to share the
EA of their participant. In the limit, only a single ParticipantEA is shared among all Entities. When doing
so, the restrictions above are lifted at a cost of greatly reduced concurrency. You may create/delete/enable/set_qos’s and generally call all of the methods of any other entity in the Listener callbacks
of Entities that share the ParticipantEA.
Use the EXCLUSIVE_AREA QosPolicy (DDS Extension) (Section 6.4.3 on page 318) of the Publisher
or Subscriber to set whether or not to use a shared exclusive area. By default, Publishers and Subscribers
will create and use their own individual EAs. You can configure a subset of the Publishers and Subscribers to share the ParticipantEA if you need the Listeners associated with those Entities or child Entities
to be able to call any of the restricted methods listed above.

186

4.6 Conditions and WaitSets

Regardless of how the EXCLUSIVE_AREA QosPolicy is set, the following operations are never allowed
in any Listener callback:
l

l

Destruction of the entity to which the Listener is attached. For instance, a DataWriter/DataReader
Listener callback must not destroy its DataWriter/DataReader.
Within the TopicListener callback, you cannot call any operations on DataReaders, DataWriters,
Publishers, Subscribers or DomainParticipants.

4.6 Conditions and WaitSets
Conditions and WaitSets provide another way for Connext DDS to communicate status changes (including
the arrival of data) to your application. While a Listener is used to provide a callback for asynchronous
access, Conditions and WaitSets provide synchronous data access. In other words, Listeners are notification-based and Conditions are wait-based.
A WaitSet allows an application to wait until one or more attached Conditions becomes true (or until a
timeout expires).
Briefly, your application can create a WaitSet, attach one or more Conditions to it, then call the WaitSet’s
wait() operation. The wait() blocks until one or more of the WaitSet’s attached Conditions becomes
TRUE.
A Condition has a trigger_value that can be TRUE or FALSE. You can retrieve the current value by calling the Condition’s only operation, get_trigger_value().
There are three kinds of Conditions. A Condition is a root class for all the conditions that may be attached
to a WaitSet. This basic class is specialized in three classes:
l

l

l

GuardConditions (Section 4.6.6 on page 194) are created by your application. Each GuardCondition has a single, user-settable, boolean trigger_value. Your application can manually trigger the
GuardCondition by calling set_trigger_value(). Connext DDS does not trigger or clear this type of
condition—it is completely controlled by your application.
ReadConditions and QueryConditions (Section 4.6.7 on page 195) are created by your application,
but triggered by Connext DDS. ReadConditions provide a way for you to specify the DDS data
samples that you want to wait for, by indicating the desired sample-states, view-states, and instancestates1.
StatusConditions (Section 4.6.8 on page 197) are created automatically by Connext DDS, one for
each Entity. A StatusCondition is triggered by Connext DDS when there is a change to any of that
Entity’s enabled statuses.

1These states are described in The SampleInfo Structure (Section 7.4.6 on page 504).

187

4.6.1 Creating and Deleting WaitSets

Figure 4.7 Conditions and WaitSets below shows the relationship between these objects and other Entities
in the system.
Figure 4.7 Conditions and WaitSets

A WaitSet can be associated with more than one Entity (including multiple DomainParticipants). It can be
used to wait on Conditions associated with different DomainParticipants. A WaitSet can only be in use by
one application thread at a time.

4.6.1 Creating and Deleting WaitSets
There is no factory for creating or deleting a WaitSet; use the natural means in each language binding (for
example, “new” or “delete” in C++ or Java).

188

4.6.2 WaitSet Operations

There are two ways to create a WaitSet—with or without specifying WaitSet properties (DDS_
WaitSetProperty_t, described in Table 4.6 WaitSet Properties (DDS_WaitSet_Property_t)). Waiting for
Conditions (Section 4.6.3 on the next page) describes how the properties are used.
Table 4.6 WaitSet Properties (DDS_WaitSet_Property_t)
Type

long

Field
Name
max_
event_
count

DDS_
max_
Duration_ event_
t
delay

Description

Maximum number of trigger events to cause a WaitSet to wake up.

Maximum delay from occurrence of first trigger event to cause a WaitSet to wake up.
This value should reflect the maximum acceptable latency increase (time delay from occurrence of the event to
waking up the WaitSet) incurred as a result of waiting for additional events before waking up the WaitSet.

To create a WaitSet with default behavior:
WaitSet* waitset = new WaitSet();

To create a WaitSet with properties:
DDS_WaitSetProperty_t prop;
Prop.max_event_count = 5;
DDSWaitSet* waitset = new DDSWaitSet(prop);

To delete a WaitSet:
delete waitset;

4.6.2 WaitSet Operations
WaitSets have only a few operations, as listed in Table 4.7 WaitSet Operations. For details, see the API
Reference HTML documentation.

189

4.6.3 Waiting for Conditions

Table 4.7 WaitSet Operations
Operation

Description
Attaches a Condition to this WaitSet.

attach_
condition

You may attach a Condition to a WaitSet that is currently being waited upon (via the wait() operation). In this case, if the
Condition has a trigger_value of TRUE, then attaching the Condition will unblock the WaitSet.
Adding a Condition that is already attached to the WaitSet has no effect. If the Condition cannot be attached, Connext DDS
will return an OUT_OF_RESOURCES error code.

detach_
condition

Detaches a Condition from the WaitSet. Attempting to detach a Condition that is not to attached the WaitSet will result in a
PRECONDITION_NOT_MET error code.

wait

Blocks execution of the thread until one or more attached Conditions becomes true, or until a user-specified timeout
expires. See Waiting for Conditions (Section 4.6.3 below).

dispatch

(Modern C++ API only) Blocks execution of the thread until one or more attached Conditions becomes true, or until a
user-specified timeout expires. Then it calls the handlers attached to the active conditions and returns. For more information
see the API Reference HTML documentation for the DDS Modern C++ API (Modules, Infrastructure Module, Conditions
and WaitSets).

get_
conditions

Retrieves a list of attached Conditions.

get_property Retrieves the DDS_WaitSetProperty_t structure of the associated WaitSet.
set_property

Sets the DDS_WaitSetProperty_t structure, to configure the associated WaitSet to return after one or more trigger events
have occurred.

4.6.3 Waiting for Conditions
The WaitSet’s wait() operation allows an application thread to wait for any of the attached Conditions to
trigger (become TRUE).
If any of the attached Conditions are already TRUE when wait() is called, it returns immediately.
If none of the attached Conditions are already TRUE, wait() blocks—suspending the calling thread. The
waiting behavior depends on whether or not properties were set when the WaitSet was created:
l

If properties are not specified when the WaitSet is created:
The WaitSet will wake up as soon as a trigger event occurs (that is, when an attached Condition
becomes true). This is the default behavior if properties are not specified.
This ‘immediate wake-up’ behavior is optimal if you want to minimize latency (to wake up and process the data or event as soon as possible). However, "waking up" involves a context switch—the
operating system must signal and schedule the thread that is waiting on the WaitSet. A context

190

4.6.3.1 How WaitSets Block

switch consumes significant CPU and therefore waking up on each data update is not optimal in situations where the application needs to maximize throughput (the number of messages processed per
second). This is especially true if the receiver is CPU limited.
l

If properties are specified when the WaitSet is created:
The properties configure the waiting behavior of a WaitSet. If no conditions are true at the time of
the call to wait, the WaitSet will wait for (a) max_event_count trigger events to occur, (b) up to
max_event_delay time from the occurrence of the first trigger event, or (c) up to the timeout maximum wait duration specified in the call to wait(). (Note: The resolution of the timeout period is constrained by the resolution of the system clock.)

If wait() does not timeout, it returns a list of the attached Conditions that became TRUE and therefore
unblocked the wait.
If wait() does timeout, it returns TIMEOUT and an empty list of Conditions.
Only one application thread can be waiting on the same WaitSet. If wait() is called on a WaitSet that
already has a thread blocking on it, the operation will immediately return PRECONDITION_NOT_MET.
If you detach a Condition from a Waitset that is currently in a wait state (that is, you are waiting on
it), wait() may return OK and an empty sequence of conditions.
4.6.3.1 How WaitSets Block
The blocking behavior of the WaitSet is illustrated in Figure 4.8 WaitSet Blocking Behavior on the next
page. The result of a wait() operation depends on the state of the WaitSet, which in turn depends on
whether at least one attached Condition has a trigger_value of TRUE.
If the wait() operation is called on a WaitSet with state BLOCKED, it will block the calling thread. If wait
() is called on a WaitSet with state UNBLOCKED, it will return immediately.
When the WaitSet transitions from BLOCKED to UNBLOCKED, it wakes up the thread (if there is one)
that had called wait() on it. There is no implied “event queuing” in the awakening of a WaitSet. That is, if
several Conditions attached to the WaitSet have their trigger_value transition to true in sequence, Connext
DDS will only unblock the WaitSet once.

191

4.6.4 Processing Triggered Conditions—What to do when Wait() Returns

Figure 4.8 WaitSet Blocking Behavior

4.6.4 Processing Triggered Conditions—What to do when Wait() Returns
When wait() returns, it provides a list of the attached Condition objects that have a trigger_value of true.
Your application can use this list to do the following for each Condition in the returned list:
l

If it is a StatusCondition:
l First, call get_status_changes() to see what status changed.
l

l

l

l

If the status changes refer to plain communication status: call get_()
on the relevant Entity.
If the status changes refer to DATA_ON_READERS1: call get_datareaders() on the relevant Subscriber.
If the status changes refer to DATA_AVAILABLE: call read() or take() on the relevant
DataReader.

If it is a ReadCondition or a QueryCondition: You may want to call read_w_condition() or take_
w_condition() on the DataReader, with the ReadCondition as a parameter (see read_w_condition
and take_w_condition (Section 7.4.3.6 on page 500)).

1And then read/take on the returned DataReader objects.

192

4.6.5 Conditions and WaitSet Example

l

Note that this is just a suggestion, you do not have to use the “w_condition” operations (or any read/take operations, for that matter) simply because you used a WaitSet. The “w_condition” operations
are just a convenient way to use the same status masks that were set on the ReadCondition or
QueryCondition.
If it is a GuardCondition: check to see which GuardCondition changed, then react accordingly.
Recall that GuardConditions are completely controlled by your application.

See Conditions and WaitSet Example (Section 4.6.5 below) to see how to determine which of the
attached Conditions is in the returned list.

4.6.5 Conditions and WaitSet Example
This example creates a WaitSet and then waits for one or more attached Conditions to become true.
// Create a WaitSet
WaitSet* waitset = new WaitSet();
// Attach Conditions
DDSCondition* cond1 = ...;
DDSCondition* cond2 = entity->get_statuscondition();
DDSCondition* cond3 = reader->create_readcondition(
DDS_NOT_READ_SAMPLE_STATE,
DDS_ANY_VIEW_STATE,
DDS_ANY_INSTANCE_STATE);
DDSCondition* cond4 = new DDSGuardCondition();
DDSCondition* cond5 = ...;
DDS_ReturnCode_t retcode;
retcode = waitset->attach_condition(cond1);
if (retcode != DDS_RETCODE_OK) {
// ... error
}
retcode = waitset->attach_condition(cond2);
if (retcode != DDS_RETCODE_OK) {
// ... error
}
retcode = waitset->attach_condition(cond3);
if (retcode != DDS_RETCODE_OK) {
// ... error
}
retcode = waitset->attach_condition(cond4);
if (retcode != DDS_RETCODE_OK) {
// ... error
}
retcode = waitset->attach_condition(cond5);
if (retcode != DDS_RETCODE_OK) {
// ... error
}
// Wait for a condition to trigger or timeout
DDS_Duration_t timeout = { 0, 1000000 }; // 1ms
DDSConditionSeq active_conditions; // holder for active conditions
bool is_cond1_triggered = false;

193

4.6.6 GuardConditions

bool is_cond2_triggered = false;
DDS_ReturnCode_t retcode;
retcode = waitset->wait(active_conditions, timeout);
if (retcode == DDS_RETCODE_TIMEOUT) {
// handle timeout
printf("Wait timed out. No conditions were triggered.\n");
}
else if (retcode != DDS_RETCODE_OK) {
// ... check for cause of failure
} else {
// success
if (active_conditions.length() == 0) {
printf("Wait timed out!! No conditions triggered.\n");
} else
// check if "cond1" or "cond2" are triggered:
for(i = 0; i < active_conditions.length(); ++i) {
if (active_conditions[i] == cond1) {
printf("Cond1 was triggered!");
is_cond1_triggered = true;
}
if (active_conditions[i] == cond2) {
printf("Cond2 was triggered!");
is_cond2_triggered = true;
}
if (is_cond1_triggered && is_cond2_triggered) {
break;
}
}
}
}
if (is_cond1_triggered) {
// ... do something because "cond1" was triggered ...
}
if (is_cond2_triggered) {
// ... do something because "cond2" was triggered ...
}
// Delete the waitset
delete waitset;
waitset = NULL;

4.6.6 GuardConditions
GuardConditions are created by your application. GuardConditions provide a way for your application to
manually awaken a WaitSet. Like all Conditions, it has a single boolean trigger_value. Your application
can manually trigger the GuardCondition by calling set_trigger_value().
Connext DDS does not trigger or clear this type of condition—it is completely controlled by your application.
A GuardCondition has no factory. It is created as an object directly by the natural means in each language
binding (e.g., using “new” in C++ or Java). For example:

194

4.6.7 ReadConditions and QueryConditions

// Create a Guard Condition
Condition* my_guard_condition = new GuardCondition();
// Delete a Guard Condition
delete my_guard_condition;

When first created, the trigger_value is FALSE.
A GuardCondition has only two operations, get_trigger_value() and set_trigger_value().
When your application calls set_trigger_value(DDS_BOOLEAN_TRUE), Connext DDS will awaken
any WaitSet to which the GuardCondition is attached.

4.6.7 ReadConditions and QueryConditions
ReadConditions are created by your application, but triggered by Connext DDS. ReadConditions provide
a way for you to specify the DDS data samples that you want to wait for, by indicating the desired samplestates, view-states, and instance-states1. Then Connext DDS will trigger the ReadCondition when suitable
DDS samples are available.
A QueryCondition is a special ReadCondition that allows you to specify a query expression and parameters, so you can filter on the locally available (already received) data. QueryConditions use the same
SQL-based filtering syntax as ContentFilteredTopics for query expressions, parameters, etc. Unlike ContentFilteredTopics, QueryConditions are applied to data already received, so they do not affect the reception of data.
Multiple mask combinations can be associated with a single content filter. This is important because the
maximum number of content filters that may be created per DataReader is 32, but more than 32
QueryConditions may be created per DataReader, if they are different mask-combinations of the same content filter.
ReadConditions and QueryConditions are created by using the DataReader’s create_readcondition() and
create_querycondition() operations. For example:
DDSReadCondition* my_read_condition = reader->create_readcondition(
DDS_NOT_READ_SAMPLE_STATE,
DDS_ANY_VIEW_STATE,
DDS_ANY_INSTANCE_STATE);
DDSQueryCondition* my_query_condition = reader->create_querycondition(
DDS_NOT_READ_SAMPLE_STATE,
DDS_ANY_VIEW_STATE,
DDS_ANY_INSTANCE_STATE
query_expression,
query_parameters);

1These states are described in The SampleInfo Structure (Section 7.4.6 on page 504).

195

4.6.7.1 How ReadConditions are Triggered

If you are using a ReadCondition to simply detect the presence of new data, consider using a
StatusCondition (StatusConditions (Section 4.6.8 on the facing page)) with the DATA_
AVAILABLE_STATUS instead, which will perform better in this situation.
A DataReader can have multiple attached ReadConditions and QueryConditions. A ReadCondition or
QueryCondition may only be attached to one DataReader.
To delete a ReadCondition or QueryCondition, use the DataReader’s delete_readcondition() operation:
DDS_ReturnCode_t

delete_readcondition (DDSReadCondition *condition)

After a ReadCondition is triggered, use the FooDataReader’s read/take “with condition” operations (see
read_w_condition and take_w_condition (Section 7.4.3.6 on page 500)) to access the DDS samples.
ReadCondition and QueryCondition Operations (Section Table 4.8 below) lists the operations available on
ReadConditions.
Table 4.8 ReadCondition and QueryCondition Operations
Operation

Description

get_
datareader

Returns the DataReader to which the ReadCondition or QueryCondition is attached.

get_
instance_
state_mask

Returns the instance states that were specified when the ReadCondition or QueryCondition was created. These are the DDS
sample’s instance states that Connext DDS checks to determine whether or not to trigger the ReadCondition or
QueryCondition .

get_sample_ Returns the sample-states that were specified when the ReadCondition or QueryCondition was created. These are the sample
state_mask
states that Connext DDS checks to determine whether or not to trigger the ReadCondition or QueryCondition.
get_view_
state_mask

Returns the view-states that were specified when the ReadCondition or QueryCondition was created. These are the view
states that Connext DDS checks to determine whether or not to trigger the ReadCondition or QueryCondition.

4.6.7.1 How ReadConditions are Triggered
A ReadCondition has a trigger_value that determines whether the attached WaitSet is BLOCKED or
UNBLOCKED. Unlike the StatusCondition, the trigger_value of the ReadCondition is tied to the presence of at least one DDS sample with a sample-state, view-state, and instance-state that matches those set
in the ReadCondition. Furthermore, for the QueryCondition to have a trigger_value==TRUE, the data
associated with the DDS sample must be such that the query_expression evaluates to TRUE.
The trigger_value of a ReadCondition depends on the presence of DDS samples on the associated
DataReader. This implies that a single ‘take’ operation can potentially change the trigger_value of several
ReadConditions or QueryConditions. For example, if all DDS samples are taken, any ReadConditions and
QueryConditions associated with the DataReader that had trigger_value==TRUE before will see the trigger_value change to FALSE. Note that this does not guarantee that WaitSet objects that were separately

196

4.6.7.2 QueryConditions

attached to those conditions will not be awakened. Once we have trigger_value==TRUE on a condition,
it may wake up the attached WaitSet, the condition transitioning to trigger_value==FALSE does not
necessarily 'unwakeup' the WaitSet, since 'unwakening' may not be possible. The consequence is that an
application blocked on a WaitSet may return from wait() with a list of conditions, some of which are no
longer “active.” This is unavoidable if multiple threads are concurrently waiting on separate WaitSet
objects and taking data associated with the same DataReader.
Consider the following example: A ReadCondition that has a sample_state_mask = {NOT_READ} will
have a trigger_value of TRUE whenever a new DDS sample arrives and will transition to FALSE as
soon as all the newly arrived DDS samples are either read (so their status changes to READ) or taken (so
they are no longer managed by Connext DDS). However, if the same ReadCondition had a sample_
state_mask = {READ, NOT_READ}, then the trigger_value would only become FALSE once all the
newly arrived DDS samples are taken (it is not sufficient to just read them, since that would only change
the SampleState to READ), which overlaps the mask on the ReadCondition.
4.6.7.2 QueryConditions
A QueryCondition is a special ReadCondition that allows your application to also specify a filter on the
locally available data.
The query expression is similar to a SQL WHERE clause and can be parameterized by arguments that are
dynamically changeable by the set_query_parameters() operation.
QueryConditions are triggered in the same manner as ReadConditions, with the additional requirement that
the DDS sample must also satisfy the conditions of the content filter associated with the QueryCondition.
Table 4.9 QueryCondition Operations
Operation

Description

get_query_
expression

Returns the query expression specified when the QueryCondition was created.

get_query_
parameters

Returns the query parameters associated with the QueryCondition. That is, the parameters specified on the last successful call
to set_query_parameters(), or if set_query_parameters() was never called, the arguments specified when the
QueryCondition was created.

set_query_
parameters

Changes the query parameters associated with the QueryCondition.

4.6.8 StatusConditions
StatusConditions are created automatically by Connext DDS, one for each Entity. Connext DDS will trigger the StatusCondition when there is a change to any of that Entity’s enabled statuses.

197

4.6.8 StatusConditions

By default, when Connext DDS creates a StatusCondition, all status bits are turned on, which means it
will check for all statuses to determine when to trigger the StatusCondition. If you only want Connext
DDS to check for specific statuses, you can use the StatusCondition’s set_enabled_statuses() operation
and set just the desired status bits.
The trigger_value of the StatusCondition depends on the communication status of the Entity (e.g., arrival
of data, loss of information, etc.), ‘filtered’ by the set of enabled statuses on the StatusCondition.
The set of enabled statuses and its relation to Listeners and WaitSets is detailed in How StatusConditions
are Triggered (Section 4.6.8.1 on the facing page).
Table 4.10 StatusCondition Operations lists the operations available on StatusConditions.
Table 4.10 StatusCondition Operations
Operation

Description
Defines the list of communication statuses that are taken into account to determine the trigger_value of the StatusCondition.
This operation may change the trigger_value of the StatusCondition.

set_enabled_
WaitSets behavior depend on the changes of the trigger_value of their attached conditions. Therefore, any WaitSet to which
statuses
the StatusCondition is attached is potentially affected by this operation.
If this function is not invoked, the default list of enabled statuses includes all the statuses.
Retrieves the list of communication statuses that are taken into account to determine the trigger_value of the
get_enabled_
StatusCondition. This operation returns the statuses that were explicitly set on the last call to set_enabled_statuses() or, if
statuses
set_enabled_statuses() was never called, the default list
get_entity

Returns the Entity associated with the StatusCondition. Note that there is exactly one Entity associated with each
StatusCondition.

Unlike other types of Conditions, StatusConditions are created by Connext DDS, not by your application.
To access an Entity’s StatusCondition, use the Entity’s get_statuscondition() operation. For example:
Condition* my_status_condition = entity->get_statuscondition();

In the Modern C++ API, use the StatusCondition constructor to obtain a reference to the Entity’s condition. For example:
dds::core::cond::StatusCondition my_status_condition(entity)

After a StatusCondition is triggered, call the Entity’s get_status_changes() operation to see which status
(es) changed.
Note: Not all statuses will activate the StatusCondition. Refer to the API Reference HTML documentation
of the individual statuses for that information.

198

4.6.8.1 How StatusConditions are Triggered

4.6.8.1 How StatusConditions are Triggered
The trigger_value of a StatusCondition is the boolean OR of the ChangedStatusFlag of all the communication statuses to which it is sensitive. That is, trigger_value is FALSE only if all the values of the
ChangedStatusFlags are FALSE.
The sensitivity of the StatusCondition to a particular communication status is controlled by the list of
enabled_statuses set on the Condition by means of the set_enabled_statuses() operation.
Once a StatusCondition’s trigger_value becomes true, it remains true until the status that changed is reset.
To reset a status, call the related get_*_status() operation. Or, in the case of the data available status, call
read(), take(), or one of their variants.
Therefore, if you are using a StatusCondition on a WaitSet to be notified of events, your thread will wake
up when one of the statuses associated with the StatusCondition becomes true. If you do not reset the
status, the StatusCondition’s trigger_value remains true and your WaitSet will not block again—it will
immediately wake up when you call wait().

4.6.9 Using Both Listeners and WaitSets
You can use Listeners and WaitSets in the same application. For example, you may want to use WaitSets
and Conditions to access the data, and Listeners to be warned asynchronously of erroneous communication statuses.
We recommend that you choose one or the other mechanism for each particular communication status (not
both). However, if both are enabled, the Listener mechanism is used first, then the WaitSet objects are
signaled.

199

Chapter 5 Topics
For a DataWriter and DataReader to communicate, they need to use the same Topic. A Topic
includes a name and an association with a user data type that has been registered with Connext
DDS. Topic names are how different parts of the communication system find each other. Topics
are named streams of data of the same data type. DataWriters publish DDS samples into the
stream; DataReaders subscribe to data from the stream. More than one Topic can use the same
user data type, but each Topic needs a unique name.
Topics, DataWriters, and DataReaders relate to each other as follows:
l

Multiple Topics (each with a unique name) can use the same user data type.

l

Applications may have multiple DataWriters for each Topic.

l

Applications may have multiple DataReaders for each Topic.

l

l

DataWriters and DataReaders must be associated with the same Topic in order for them to
be connected.
Topics are created and deleted by a DomainParticipant, and as such, are owned by that
DomainParticipant. When two applications (DomainParticipants) want to use the same
Topic, they must both create the Topic (even if the applications are on the same node).

Connext DDS uses ‘Builtin Topics’ to discover and keep track of remote entities, such as new participants in the DDS domain. Builtin Topics are discussed in Built-In Topics (Section Chapter 16
on page 772).
This section includes the following sections:

5.1 Topics
Before you can create a Topic, you need a user data type (see Data Types and DDS Data Samples
(Section Chapter 3 on page 23)) and a DomainParticipant (DomainParticipants (Section 8.3 on

200

5.1 Topics

page 547)). The user data type must be registered with the DomainParticipant (see Type Codes for Builtin Types (Section 3.8.4.1 on page 143)).
Once you have created a Topic, what do you do with it? Topics are primarily used as parameters in other
Entities’ operations. For instance, a Topic is required when a Publisher or Subscriber creates a DataWriter
or DataReader, respectively. Topics do have a few operations of their own, as listed in Table 5.1 Topic
Operations. For details on using these operations, see the reference section or the API Reference HTML
documentation.
Figure 5.1 Topic Module

201

5.1.1 Creating Topics

Table 5.1 Topic Operations
Purpose

Configuring
the Topic

Checking
Status

Navigating
Relationships

Operation

Description

enable

Enables the Topic.

get_qos

Gets the Topic’s current QosPolicy settings. This is most often used in
preparation for calling set_qos().

set_qos

Sets the Topic’s QoS. You can use this operation to change the values for
the Topic’s QosPolicies. Note, however, that not all QosPolicies can be
changed after the Topic has been created.

equals

Compares two Topic’s QoS structures for equality.

set_qos_
with_
profile

Sets the Topic’s QoS based on a specified QoS profile.

get_listener

Gets the currently installed Listener.

set_listener

Sets the Topic’s Listener. If you create the Topic without a Listener, you
can use this operation to add one later. Setting the listener to NULL will
remove the listener from the Topic.

narrow

A type-safe way to cast a pointer. This takes a DDSTopicDescription
pointer and ‘narrows’ it to a DDSTopic pointer.

Reference
Enabling DDS Entities (Section
4.1.2 on page 154)

Setting Topic QosPolicies
(Section 5.1.3 on page 204)

Comparing QoS Values
(Section 5.1.3.2 on page 207)

Setting Up TopicListeners
(Section 5.1.5 on page 208)

Using a Type-Specific
DataWriter (FooDataWriter)
(Section 6.3.7 on page 281)

get_
Allows an application to retrieve a Topic’s INCONSISTENT_TOPIC_
inconsistent_
STATUS status.
topic_status

INCONSISTENT_TOPIC
Status (Section 5.3.1 on page
211)

get_status_
changes

Gets a list of statuses that have changed since the last time the application
read the status or the listeners were called.

Getting Status and Status
Changes (Section 4.1.4 on
page 157)

get_name

Gets the topic_name string used to create the Topic.

get_type_
name

Gets the type_name used to create the Topic.

get_
participant

Gets the DomainParticipant to which this Topic belongs.

Creating Topics (Section 5.1.1
below)

Finding a Topic’s
DomainParticipant (Section
5.1.6.1 on page 209)

5.1.1 Creating Topics
Topics are created using the DomainParticipant’s create_topic() or create_topic_with_profile() operation.

202

5.1.1 Creating Topics

A QoS profile is way to use QoS settings from an XML file or string. With this approach, you can change
QoS settings without recompiling the application. For details, see Configuring QoS with XML (Section
Chapter 17 on page 791).
DDSTopic * create_topic (
const char *topic_name,
const char *type_name,
const DDS_TopicQos &qos,
DDSTopicListener *listener,
DDS_StatusMask mask)
DDSTopic * create_topic_with_profile (
const char *topic_name,
const char *type_name,
const char *library_name,
const char *profile_name,
DDSTopicListener *listener,
DDS_StatusMask mask)

Where:
topic_name

Name for the new Topic , must not exceed 255 characters.

type_name

Name for the user data type, must not exceed 255 characters. It must be the same name that was
used to register the DDS type, and the DDS type must be registered with the same
DomainParticipant used to create this Topic . See Using RTI Code Generator (rtiddsgen)
(Section 3.6 on page 138).

qos

If you want to use the default QoS settings (described in the API Reference HTML
documentation), use DDS_TOPIC_QOS_DEFAULT for this parameter (see Figure 5.2
Creating a Topic with Default QosPolicies on the facing page ). If you want to customize any
of the QosPolicies, supply a QoS structure (see Setting Topic QosPolicies (Section 5.1.3 on
the facing page)).
If you use DDS_TOPIC_QOS_DEFAULT, it is not safe to create the topic while another
thread may be simultaneously calling the DomainParticipant’s set_default_topic_qos()
operation.

listener

Listeners are callback routines. Connext DDS uses them to notify your application of specific
events (status changes) that may occur with respect to the Topic . The listener parameter may be
set to NULL if you do not want to install a Listener. If you use NULL, the Listener of the
DomainParticipant to which the Topic belongs will be used instead (if it is set). For more
information on TopicListeners, see Setting Up TopicListeners (Section 5.1.5 on page 208).

mask

This bit-mask indicates which status changes will cause the Listener to be invoked. The bits in
the mask that are set must have corresponding callbacks implemented in the Listener. If you use
NULL for the Listener, use DDS_STATUS_MASK_NONE for this parameter. If the Listener
implements all callbacks, use DDS_STATUS_MASK_ALL. For information on statuses, see
Listeners (Section 4.4 on page 177).

203

5.1.2 Deleting Topics

library_name A QoS Library is a named set of QoS profiles. See URL Groups (Section 17.8 on page 814). If
NULL is used for library_name , the DomainParticipant’s default library is assumed.
profile_name

A QoS profile groups a set of related QoS, usually one per entity. See URL Groups (Section
17.8 on page 814). If NULL is used for profile_name , the DomainParticipant’s default profile
is assumed and library_name is ignored.

It is not safe to create a topic while another thread is calling lookup_topicdescription() for that
same topic (see Looking up Topic Descriptions (Section 8.3.7 on page 568)).
Figure 5.2 Creating a Topic with Default QosPolicies

const char *type_name = NULL;
// register the DDS type
type_name = FooTypeSupport::get_type_name();
retcode = FooTypeSupport::register_type(
participant, type_name);
if (retcode != DDS_RETCODE_OK) {
// handle error
}
// create the topic
DDSTopic* topic = participant->create_topic(
"Example Foo", type_name,
DDS_TOPIC_QOS_DEFAULT,
NULL, DDS_STATUS_MASK_NONE);
if (topic == NULL) {
// process error here
};

For more examples, see Configuring QoS Settings when the Topic is Created (Section 5.1.3.1 on page
206).

5.1.2 Deleting Topics
To delete a Topic, use the DomainParticipant’s delete_topic() operation:
DDS_ReturnCode_t delete_topic (DDSTopic *

topic)

Note, however, that you cannot delete a Topic if there are any existing DataReaders or DataWriters
(belonging to the same DomainParticipant) that are still using it. All DataReaders and DataWriters associated with the Topic must be deleted first.
Note: in the Modern C++ API, Entities are automatically destroyed.

5.1.3 Setting Topic QosPolicies
A Topic’s QosPolicies control its behavior, or more specifically, the behavior of the DataWriters and
DataReaders of the Topic. You can think of the policies as the ‘properties’ for the Topic. The DDS_

204

5.1.3 Setting Topic QosPolicies

TopicQos structure has the following format:
DDS_TopicQos struct {
DDS_TopicDataQosPolicy
DDS_DurabilityQosPolicy
DDS_DurabilityServiceQosPolicy
DDS_DeadlineQosPolicy
DDS_LatencyBudgetQosPolicy
DDS_LivelinessQosPolicy
DDS_ReliabilityQosPolicy
DDS_DestinationOrderQosPolicy
DDS_HistoryQosPolicy
DDS_ResourceLimitsQosPolicy
DDS_TransportPriorityQosPolicy
DDS_LifespanQosPolicy
DDS_OwnershipQosPolicy
} DDS_TopicQos;

topic_data;
durability;
durability_service;
deadline;
latency_budget;
liveliness;
reliability;
destination_order;
history;
resource_limits;
transport_priority;
lifespan;
ownership;

Table 5.2 Topic QosPolicies summarizes the meaning of each policy (arranged alphabetically). For information on why you would want to change a particular QosPolicy, see the section noted in the Reference
column. For defaults and valid ranges, please refer to the API Reference HTML documentation for each
policy.
Table 5.2 Topic QosPolicies
QosPolicy

Description
For a DataReader, specifies the maximum expected elapsed time between arriving DDS data samples.

Deadline

For a DataWriter, specifies a commitment to publish DDS samples with no greater elapsed time between them.
See DEADLINE QosPolicy (Section 6.5.5 on page 363).

Controls how Connext DDS will deal with data sent by multiple DataWriters for the same topic. Can be set to "by
DestinationOrder reception timestamp" or to "by source timestamp". See DESTINATION_ORDER QosPolicy (Section 6.5.6 on page
365).
Durability

Specifies whether or not Connext DDS will store and deliver data that were previously published to new
DataReaders. See DURABILITY QosPolicy (Section 6.5.7 on page 368).

DurabilityService

Various settings to configure the external Persistence Service used by Connext DDS for DataWriters with a
Durability QoS setting of Persistent Durability. See DURABILITY SERVICE QosPolicy (Section 6.5.8 on page 372).

History

Specifies how much data must to stored by Connext DDS for the DataWriter or DataReader. This QosPolicy affects
the RELIABILITY QosPolicy (Section 6.5.19 on page 400) as well as the DURABILITY QosPolicy (Section 6.5.7
on page 368). See HISTORY QosPolicy (Section 6.5.10 on page 376).

LatencyBudget

Suggestion to Connext DDS on how much time is allowed to deliver data. See LATENCYBUDGET QoS Policy
(Section 6.5.11 on page 380).

205

5.1.3.1 Configuring QoS Settings when the Topic is Created

Table 5.2 Topic QosPolicies
QosPolicy

Description

Lifespan

Specifies how long Connext DDS should consider data sent by an user application to be valid. See LIFESPAN QoS
Policy (Section 6.5.12 on page 381).

Liveliness

Specifies and configures the mechanism that allows DataReaders to detect when DataWriters become disconnected or
"dead." See LIVELINESS QosPolicy (Section 6.5.13 on page 382).

Ownership

Along with Ownership Strength, specifies if DataReaders for a topic can receive data from multiple DataWriters at the
same time. See OWNERSHIP QosPolicy (Section 6.5.15 on page 389).

Reliability

Specifies whether or not Connext DDS will deliver data reliably. See RELIABILITY QosPolicy (Section 6.5.19 on
page 400).

ResourceLimits

Controls the amount of physical memory allocated for entities, if dynamic allocations are allowed, and how they occur.
Also controls memory usage among different instance values for keyed topics. See RESOURCE_LIMITS QosPolicy
(Section 6.5.20 on page 405).

TopicData

Along with Group Data QosPolicy and User Data QosPolicy, used to attach a buffer of bytes to Connext DDS's
discovery meta-data. See TOPIC_DATA QosPolicy (Section 5.2.1 on page 209).

TransportPriority

Set by a DataWriter to tell Connext DDS that the data being sent is a different "priority" than other data. See
TRANSPORT_PRIORITY QosPolicy (Section 6.5.22 on page 409).

5.1.3.1 Configuring QoS Settings when the Topic is Created
As described in Creating Topics (Section 5.1.1 on page 202), there are different ways to create a Topic,
depending on how you want to specify its QoS (with or without a QoS profile).
In Creating a Topic with Default QosPolicies (Section Figure 5.2 on page 204), we saw an example of
how to create a Topic with default QosPolicies by using the special constant, DDS_TOPIC_QOS_
DEFAULT, which indicates that the default QoS values for a Topic should be used. The default Topic
QoS values are configured in the DomainParticipant; you can change them with the DomainParticipant’s
set_default_topic_qos() or set_default_topic_qos_with_profile() operations (see Getting and Setting
Default QoS for Child Entities (Section 8.3.6.5 on page 568)).
To create a Topic with non-default QoS values, without using a QoS profile, use the DomainParticipant’s
get_default_topic_qos() operation to initialize a DDS_TopicQos structure. Then change the policies from
their default values before passing the QoS structure to create_topic().
You can also create a Topic and specify its QoS settings via a QoS profile. To do so, call create_topic_
with_profile().
If you want to use a QoS profile, but then make some changes to the QoS before creating the Topic, call
get_topic_qos_from_profile(), modify the QoS and use the modified QoS when calling create_topic().

206

5.1.3.2 Comparing QoS Values

5.1.3.2 Comparing QoS Values
The equals() operation compares two Topic’s DDS_TopicQoS structures for equality. It takes two parameters for the two Topics’ QoS structures to be compared, then returns TRUE is they are equal (all values
are the same) or FALSE if they are not equal.
5.1.3.3 Changing QoS Settings After the Topic Has Been Created
There are two ways to change an existing Topic’s QoS after it is has been created—again depending on
whether or not you are using a QoS Profile.
To change QoS programmatically (that is, without using a QoS Profile), see the example code in Figure
5.3 Changing the QoS of an Existing Topic (without a QoS Profile) below. It retrieves the current values
by calling the Topic’s get_qos() operation. Then it modifies the value and calls set_qos() to apply the new
value. Note, however, that some QosPolicies cannot be changed after the Topic has been enabled—this
restriction is noted in the descriptions of the individual QosPolicies.
You can also change a Topic’s (and all other Entities’) QoS by using a QoS Profile. For an example, see
Figure 5.4 Changing the QoS of an Existing Topic with a QoS Profile below. For more information, see
Configuring QoS with XML (Section Chapter 17 on page 791).
Figure 5.3 Changing the QoS of an Existing Topic (without a QoS Profile)
DDS_TopicQos topic_qos;1
// Get current QoS. topic points to an existing DDSTopic.
if (topic->get_qos(topic_qos) != DDS_RETCODE_OK) {
// handle error
}
// Next, make changes.
// New ownership kind will be Exclusive
topic_qos.ownership.kind = DDS_EXCLUSIVE_OWNERSHIP_QOS;
// Set the new QoS
if (topic->set_qos(topic_qos) != DDS_RETCODE_OK ) {
// handle error
}

Figure 5.4 Changing the QoS of an Existing Topic with a QoS Profile

retcode = topic->set_qos_with_profile(
“FooProfileLibrary”,”FooProfile”);
if (retcode != DDS_RETCODE_OK) {
// handle error
}

1For the C API, use DDS_TopicQos_INITIALIZER or DDS_TopicQos_initialize(). See Special

QosPolicy Handling Considerations for C (Section 4.2.2 on page 168)
207

5.1.4 Copying QoS From a Topic to a DataWriter or DataReader

5.1.4 Copying QoS From a Topic to a DataWriter or DataReader
Only the TOPIC_DATA QosPolicy strictly applies to Topics—it is described in this section, while the others are described in the sections noted Table 5.2 Topic QosPolicies. The rest of the QosPolicies for a Topic
can also be set on the corresponding DataWriters and/or DataReaders. Actually, the values that Connext
DDS uses for those policies are taken directly from those set on the DataWriters and DataReaders. The
values for those policies are stored only for reference in the DDS_TopicQos structure.
Because many QosPolicies affect the behavior of matching DataWriters and DataReaders, the DDS_TopicQos structure is provided as a convenient way to set the values for those policies in a single place in the
application. Otherwise, you would have to modify the individual QosPolicies within separate DataWriter
and DataReader QoS structures. And because some QosPolicies are compared between DataReaders and
DataWriters, you will need to make certain that the individual values that you set are compatible (see QoS
Requested vs. Offered Compatibility—the RxO Property (Section 4.2.1 on page 167)).
The use of the DDS_TopicQos structure to set the values of any QosPolicy except TOPIC_DATA—
which only applies to Topics—is really a way to share a single set of values with the associated
DataWriters and DataReaders, as well as to avoid creating those entities with inconsistent QosPolicies.
To cause a DataWriter to use its Topic’s QoS settings, either:
l

Pass DDS_DATAWRITER_QOS_USE_TOPIC_QOS to create_datawriter(), or

l

Call the Publisher’s copy_from_topic_qos() operation

To cause a DataReader to use its Topic’s QoS settings, either:
l

Pass DDS_DATAREADER_QOS_USE_TOPIC_QOS to create_datareader(), or

l

Call the Subscriber’s copy_from_topic_qos() operation

Please refer to the API Reference HTML documentation for the Publisher’s create_datawriter() and Subscriber’s create_datareader() methods for more information about using values from the Topic
QosPolicies when creating DataWriters and DataReaders.

5.1.5 Setting Up TopicListeners
When you create a Topic, you have the option of giving it a Listener. A TopicListener includes just one
callback routine, on_inconsistent_topic(). If you create a TopicListener (either as part of the Topic creation call, or later with the set_listener() operation), Connext DDS will invoke the TopicListener’s on_
inconsistent_topic() method whenever it detects that another application has created a Topic with same
name but associated with a different user data type. For more information, see INCONSISTENT_TOPIC
Status (Section 5.3.1 on page 211).
Note: Some operations cannot be used within a listener callback, see Restricted Operations in Listener
Callbacks (Section 4.5.1 on page 185).

208

5.1.6 Navigating Relationships Among Entities

If a Topic’s Listener has not been set and Connext DDS detects an inconsistent Topic, the DomainParticipantListener (if it exists) will be notified instead (see Setting Up DomainParticipantListeners (Section
8.3.5 on page 560)). So you only need to set up a TopicListener if you need to perform specific actions
when there is an error on that particular Topic. In most cases, you can set the TopicListener to NULL and
process inconsistent-topic errors in the DomainParticipantListener instead.

5.1.6 Navigating Relationships Among Entities
5.1.6.1 Finding a Topic’s DomainParticipant
To retrieve a handle to the Topic’s DomainParticipant, use the get_participant() operation:
DDSDomainParticipant* DDSTopicDescription::get_participant()

Notice that this method belongs to the DDSTopicDescription class, which is the base class for
DDSTopic.
5.1.6.2 Retrieving a Topic’s Name or DDS Type Name
If you want to retrieve the topic_name or type_name used in the create_topic() operation, use these methods:
const char* DDSTopicDescription::get_type_name();
const char* DDSTopicDescription::get_name();

Notice that these methods belong to the DDSTopicDescription class, which is the base class for
DDSTopic.

5.2 Topic QosPolicies
This section describes the only QosPolicy that strictly applies to Topics (and no other types of Entities)—
the TOPIC_DATA QosPolicy. For a complete list of the QosPolicies that can be set for Topics, see Table
5.2 Topic QosPolicies.
Most of the QosPolicies that can be set on a Topic can also be set on the corresponding DataWriter and/or
DataReader. The Topic’s QosPolicy is essentially just a place to store QoS settings that you plan to share
with multiple entities that use that Topic (see how in Setting Topic QosPolicies (Section 5.1.3 on page
204)); they are not used otherwise and are not propagated on the wire.

5.2.1 TOPIC_DATA QosPolicy
This QosPolicy provides an area where your application can store additional information related to the
Topic. This information is passed between applications during discovery (see Discovery (Section Chapter
14 on page 709)) using builtin-topics (see Built-In Topics (Section Chapter 16 on page 772)). How this
information is used will be up to user code. Connext DDS does not do anything with the information
209

5.2.1.1 Example

stored as TOPIC_DATA except to pass it to other applications. Use cases are usually application-to-application identification, authentication, authorization, and encryption purposes.
The value of the TOPIC_DATA QosPolicy is sent to remote applications when they are first discovered,
as well as when the Topic’s set_qos() method is called after changing the value of the TOPIC_DATA.
User code can set listeners on the builtin DataReaders of the builtin Topics used by Connext DDS to
propagate discovery information. Methods in the builtin topic listeners will be called whenever new applications, DataReaders, and DataWriters are found. Within the user callback, you will have access to the
TOPIC_DATA that was set for the associated Topic.
Currently, TOPIC_DATA of the associated Topic is only propagated with the information that declares a
DataWriter or DataReader. Thus, you will need to access the value of TOPIC_DATA through DDS_
PublicationBuiltinTopicData or DDS_SubscriptionBuiltinTopicData (see Built-In Topics (Section Chapter
16 on page 772)).
The structure for the TOPIC_DATA QosPolicy includes just one field, as seen in Table 5.3 DDS_TopicDataQosPolicy. The field is a sequence of octets that translates to a contiguous buffer of bytes whose
contents and length is set by the user. The maximum size for the data are set in the DOMAIN_
PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593).
Table 5.3 DDS_TopicDataQosPolicy
Type
DDS_OctetSeq

Field Name
value

Description
default: empty

This policy is similar to the GROUP_DATA (GROUP_DATA QosPolicy (Section 6.4.4 on page 320))
and USER_DATA (USER_DATA QosPolicy (Section 6.5.26 on page 417)) policies that apply to other
types of Entities.
5.2.1.1 Example
One possible use of TOPIC_DATA is to send an associated XML schema that can be used to process the
data stored in the associated user data structure of the Topic. The schema, which can be passed as a long
sequence of characters, could be used by an XML parser to take DDS samples of the data received for a
Topic and convert them for updating some graphical user interface, web application or database.
5.2.1.2 Properties
This QosPolicy can be modified at any time. A change in the QosPolicy will cause Connext DDS to send
packets containing the new TOPIC_DATA to all of the other applications in the DDS domain.
Because Topics are created independently by the applications that use the Topic, there may be different
instances of the same Topic (same topic name and DDS data type) in different applications. The TOPIC_
DATA for different instances of the same Topic may be set differently by different applications.

210

5.2.1.3 Related QosPolicies

5.2.1.3 Related QosPolicies
l

GROUP_DATA QosPolicy (Section 6.4.4 on page 320)

l

USER_DATA QosPolicy (Section 6.5.26 on page 417)

l

DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
on page 593)

5.2.1.4 Applicable DDS Entities
l

Topics (Section 5.1 on page 200)

5.2.1.5 System Resource Considerations
As mentioned earlier, the maximum size of the TOPIC_DATA is set in the topic_data_max_length field
of the DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
on page 593). Because Connext DDS will allocate memory based on this value, you should only increase
this value if you need to. If your system does not use TOPIC_DATA, then you can set this value to 0 to
save memory. Setting the value of the TOPIC_DATA QosPolicy to hold data longer than the value set in
the topic_data_max_length field will result in failure and an INCONSISTENT_QOS_POLICY return
code.
However, should you decide to change the maximum size of TOPIC_DATA, you must make certain that
all applications in the DDS domain have changed the value of topic_data_max_length to be the same. If
two applications have different limits on the size of TOPIC_DATA, and one application sets the TOPIC_
DATA QosPolicy to hold data that is greater than the maximum size set by another application, then the
DataWriters and DataReaders of that Topic between the two applications will not connect. This is also
true for the GROUP_DATA (GROUP_DATA QosPolicy (Section 6.4.4 on page 320)) and USER_
DATA (USER_DATA QosPolicy (Section 6.5.26 on page 417)) QosPolicies.

5.3 Status Indicator for Topics
There is only one communication status defined for a Topic, ON_INCONSISTENT_TOPIC. You can
use the get_inconsistent_topic_status() operation to access the current value of the status or use a TopicListener to catch the change in the status as it occurs. See Listeners (Section 4.4 on page 177) for a general discussion on Listeners and Statuses.

5.3.1 INCONSISTENT_TOPIC Status
In order for a DataReader and a DataWriter with the same Topic to communicate, their DDS types must
be consistent according to the DataReader’s type-consistency enforcement policy value, defined in its
TYPE_CONSISTENCY_ENFORCEMENT QosPolicy (Section 7.6.6 on page 532)). This status indicates that another DomainParticipant has created a Topic using the same name as the local Topic, but with
an inconsistent DDS type.

211

5.4 ContentFilteredTopics

The status is a structure of type DDS_InconsistentTopicStatus, see Table 5.4 DDS_InconsistentTopicStatus Structure. The total_count keeps track of the total number of (DataReader,
DataWriter) pairs with topic names that match the Topic to which this status is attached, but whose DDS
types are inconsistent. The TopicListener’s on_inconsistent_topic() operation is invoked when this status
changes (an inconsistent topic is found). You can also retrieve the current value by calling the Topic’s get_
inconsistent_topic_status() operation.
The value of total_count_change reflects the number of inconsistent topics that were found since the last
time get_inconsistent_topic_status() was called by user code or on_inconsistent_topic() was invoked by
Connext DDS.
Table 5.4 DDS_InconsistentTopicStatus Structure
Type

Field
Name

DDS_
Long

total_count

Total cumulative count of (DataReader, DataWriter) pairs whose topic names match the Topic to which this status
is attached, but whose DDS types are inconsistent.

DDS_
Long

total_count_
change

The change in total_count since the last time this status was read.

Description

5.4 ContentFilteredTopics
A ContentFilteredTopic is a Topic with filtering properties. It makes it possible to subscribe to topics and
at the same time specify that you are only interested in a subset of the Topic’s data.
For example, suppose you have a Topic that contains a temperature reading for a boiler, but you are only
interested in temperatures outside the normal operating range. A ContentFilteredTopic can be used to limit
the number of DDS data samples a DataReader has to process and may also reduce the amount of data
sent over the network.
This section includes the following:

5.4.1 Overview
A ContentFilteredTopic creates a relationship between a Topic, also called the related topic, and user-specified filtering properties. The filtering properties consist of an expression and a set of parameters.
l

l

The filter expression evaluates a logical expression on the Topic content. The filter expression is similar to the WHERE clause in a SQL expression.
The parameters are strings that give values to the 'parameters' in the filter expression. There must be
one parameter string for each parameter in the filter expression.

212

5.4.2 Where Filtering is Applied—Publishing vs. Subscribing Side

A ContentFilteredTopic is a type of topic description, and can be used to create DataReaders. However, a
ContentFilteredTopic is not an entity—it does not have QosPolicies or Listeners.
A ContentFilteredTopic relates to other entities in Connext DDS as follows:
l

ContentFilteredTopics are used when creating DataReaders, not DataWriters.

l

Multiple DataReaders can be created with the same ContentFilteredTopic.

l

A ContentFilteredTopic belongs to (is created/deleted by) a DomainParticipant.

l

A ContentFilteredTopic and Topic must be in the same DomainParticipant.

l

A ContentFilteredTopic can only be related to a single Topic.

l

A Topic can be related to multiple ContentFilteredTopics.

l

l
l

l

l

A ContentFilteredTopic can have the same name as a Topic, but ContentFilteredTopics must have
unique names within the same DomainParticipant.
A DataReader created with a ContentFilteredTopic will use the related Topic's QoS and Listeners.
Changing filter parameters on a ContentFilteredTopic causes all DataReaders using the same ContentFilteredTopic to see the change.
A Topic cannot be deleted as long as at least one ContentFilteredTopic that has been created with it
exists.
A ContentFilteredTopic cannot be deleted as long as at least one DataReader that has been created
with the ContentFilteredTopic exists.

5.4.2 Where Filtering is Applied—Publishing vs. Subscribing Side
Filtering may be performed on either side of the distributed application. (The DataWriter obtains the filter
expression and parameters from the DataReader during discovery.)
When batching is enabled, content filtering is always done on the reader side.
Connext DDS also supports network-switch filtering for multi-channel DataWriters (see Multi-channel
DataWriters (Section Chapter 18 on page 824)).
A DataWriter will automatically filter DDS data samples for a DataReader if all of the following are true;
otherwise filtering is performed by the DataReader.
1. The DataWriter is filtering for no more than writer_resource_limits.max_remote_reader_filters
DataReaders at the same time.
l

There is a resource-limit on the DataWriter called writer_resource_limits.max_remote_
reader_filters (see DATA_WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension)
(Section 6.5.4 on page 359)). This value can be from [0, (2^31)-2]. 0 means do not filter any
DataReader; 32 (default value) means filter up to 32 DataReaders.

213

5.4.3 Creating ContentFilteredTopics

l

If a DataWriter is filtering max_remote_reader_filters DataReaders at the same time and a
new filtered DataReader is created, then the newly created DataReader (max_remote_
reader_filters + 1) is not filtered. Even if one of the first (max_remote_reader_filters)
DataReaders is deleted, that already created DataReader (max_remote_reader_filters + 1)
will still not be filtered. However, any subsequently created DataReaders will be filtered as
long as the number of DataReaders currently being filtered is not more than writer_
resource_limits.max_remote_reader_filters.

2. The DataReader is not subscribing to data using multicast.
3. There are no more than 4 matching DataReaders in the same locator (see Peer Descriptor Format
(Section 14.2.1 on page 713)).
4. The DataWriter has infinite liveliness. (See LIVELINESS QosPolicy (Section 6.5.13 on page
382).)
5. The DataWriter is not using an Asynchronous Publisher. (That is, the DataWriter’s PUBLISH_
MODE QosPolicy (DDS Extension) (Section 6.5.18 on page 397) kind is set to DDS_
SYNCHRONOUS_PUBLISHER_MODE_QOS.) See Note below.
6. If you are using a custom filter (not the default one), it must be registered in the DomainParticipant
of the DataWriter and the DataReader.
7. The DataWriter is not configured to use batching.
Notes:
l

l

l

Connext DDS supports limited writer-side filtering if asynchronous publishing is enabled. The middleware will not send any DDS sample to a destination if the DDS sample is filtered out by all the
DataReaders on that destination. However, if there is one DataReader to which the DDS sample
has to be sent, all the DataReaders on the destination will do reader side filtering for the incoming
DDS sample.
In addition to filtering new DDS samples, a DataWriter can also be configured to filter previously
written DDS samples stored in the DataWriter’s queue for newly discovered DataReaders. To do
so, use the refilter field in the DataWriter’s HISTORY QosPolicy (Section 6.5.10 on page 376).
When batching is enabled, content filtering is always done on the reader side. See BATCH
QosPolicy (DDS Extension) (Section 6.5.2 on page 341).

5.4.3 Creating ContentFilteredTopics
To create a ContentFilteredTopic that uses the default SQL filter, use the DomainParticipant’s create_contentfilteredtopic() operation:
DDS_ContentFilteredTopic *create_contentfilteredtopic(
const char * name,
const DDS_Topic * related_topic,

214

5.4.3 Creating ContentFilteredTopics

const char * filter_expression,
const DDS_StringSeq & expression_parameters)

Or, to use a custom filter or the builtin STRINGMATCH filter (see STRINGMATCH Filter Expression
Notation (Section 5.4.7 on page 231)), use the create_contentfilteredtopic_with_filter() variation:
DDS_ContentFilteredTopic *create_contentfilteredtopic_with_filter(
const char * name,
DDSTopic * related_topic,
const char * filter_expression,
const DDS_StringSeq & expression_parameters,
const char * filter_name = DDS_SQLFILTER_NAME)

Where:
name

Name of the ContentFilteredTopic. Note that it is legal for a ContentFilteredTopic to have the
same name as a Topic in the same DomainParticipant, but a ContentFilteredTopic cannot have
the same name as another ContentFilteredTopic in the same DomainParticipant. This parameter
cannot be NULL.

related_topic

The related Topic to be filtered. The related topic must be in the same DomainParticipant as the
ContentFilteredTopic. This parameter cannot be NULL. The same related topic can be used in
many different ContentFilteredTopics.

filter_
expression

A logical expression on the contents on the Topic. If the expression evaluates to TRUE, a DDS
sample is received; otherwise it is discarded. This parameter cannot be NULL. The notation for
this expression depends on the filter that you are using (specified by the filter_name
parameter). See SQL Filter Expression Notation (Section 5.4.6 on page 222) and
STRINGMATCH Filter Expression Notation (Section 5.4.7 on page 231). The filter_
expression can be changed with set_expression() (Setting an Expression’s Filter and
Parameters (Section 5.4.5.2 on page 220)).

expression_
parameters

A string sequence of filter expression parameters. Each parameter corresponds to a positional
argument in the filter expression: element 0 corresponds to positional argument 0, element 1 to
positional argument 1, and so forth.
The expression_parameters can be changed with set_expression_parameters() or set_
expression() (Setting an Expression’s Filter and Parameters (Section 5.4.5.2 on page 220)),
append_to_expression_parameter() (Appending a String to an Expression Parameter
(Section 5.4.5.3 on page 220)) and remove_from_expression_parameter() (Removing a
String from an Expression Parameter (Section 5.4.5.4 on page 221)).

filter_name

215

5.4.3 Creating ContentFilteredTopics

Name of the content filter to use for filtering. The filter must have been previously registered
with the DomainParticipant (see Registering a Custom Filter (Section 5.4.8.2 on page 234)).
There are two builtin filters, DDS_SQLFILTER_NAME1 (the default filter) and DDS_
STRINGMATCHFILTER_NAME—these are automatically registered.
To use the STRINGMATCH filter, call create_contentfilteredtopic_with_filter() with
"DDS_STRINGMATCHFILTER_NAME" as the filter_name . STRINGMATCH filter
expressions have the syntax:
 MATCH  (see STRINGMATCH Filter Expression Notation
(Section 5.4.7 on page 231)).
2

If you run RTI Code Generator with -notypecode, you must use the "with_filter" version with a custom
filter instead—do not use the builtin SQL filter or the STRINGMATCH filter with the -notypecode
option because they require type codes.
To summarize:
l

l

To use the builtin default SQL filter:
l Do not use -notypecode when running RTI Code Generator
l

Call create_contentfilteredtopic()

l

See SQL Filter Expression Notation (Section 5.4.6 on page 222)

To use the builtin STRINGMATCH filter:
l Do not use -notypecode when running RTI Code Generator
l

l

l

l

Call create_contentfilteredtopic_with_filter(), setting the filter_name to DDS_
STRINGMATCHFILTER_NAME
See STRINGMATCH Filter Expression Notation (Section 5.4.7 on page 231)

To use a custom filter:
l Call create_contentfilteredtopic_with_filter(), setting the filter_name to a registered custom
filter
To use RTI Code Generator with -notypecode:
l Call create_contentfilteredtopic_with_filter(), setting the filter_name to a registered custom
filter

1 In the Java and C# APIs, you can access the names of the builtin filters by using

DomainParticipant.SQLFILTER_NAME and DomainParticipant.STRINGMATCHFILTER_NAME.
2 In the Java and C# APIs, you can access the names of the builtin filters by using

DomainParticipant.SQLFILTER_NAME and DomainParticipant.STRINGMATCHFILTER_NAME.

216

5.4.3.1 Creating ContentFilteredTopics for Built-in DDS Types

Be careful with memory management of the string sequence in some of the ContentFilteredTopic
APIs. See the String Support section in the API Reference HTML documentation (within the
Infrastructure module) for details on sequences.
5.4.3.1 Creating ContentFilteredTopics for Built-in DDS Types
To create a ContentFilteredTopic for a built-in DDS type (see Built-in Data Types (Section 3.2 on page
30)), use the standard DomainParticipant operations, create_contentfilteredtopic() or create_contentfilteredtopic_with_filter.
The field names used in the filter expressions for the built-in SQL (see SQL Filter Expression Notation
(Section 5.4.6 on page 222)) and StringMatch filters (see STRINGMATCH Filter Expression Notation
(Section 5.4.7 on page 231)) must correspond to the names provided in the IDL description of the built-in
DDS types.
ContentFilteredTopic Creation Examples:
For simplicity, error handling is not shown in the following examples.
C Example:
DDS_Topic * topic = NULL;
DDS_ContentFilteredTopic * contentFilteredTopic = NULL;
struct DDS_StringSeq parameters = DDS_SEQUENCE_INITIALIZER;
/* Create a string ContentFilteredTopic */
topic = DDS_DomainParticipant_create_topic(
participant, "StringTopic",
DDS_StringTypeSupport_get_type_name(),
&DDS_TOPIC_QOS_DEFAULT,NULL,
DDS_STATUS_MASK_NONE);
contentFilteredTopic =
DDS_DomainParticipant_create_contentfilteredtopic(
participant,
"StringContentFilteredTopic",
topic,
"value = 'Hello World!'", ¶meters);

C++ Example with Namespaces:
using namespace DDS;
...
/* Create a String ContentFilteredTopic */
Topic * topic = participant->create_topic(
"StringTopic",
StringTypeSupport::get_type_name(),
TOPIC_QOS_DEFAULT,
NULL, STATUS_MASK_NONE);
StringSeq parameters;
ContentFilteredTopic * contentFilteredTopic =
participant->create_contentfilteredtopic(
"StringContentFilteredTopic", topic,
"value = 'Hello World!'", parameters);

217

5.4.4 Deleting ContentFilteredTopics

C++/CLI Example:
using namespace DDS;
...
/* Create a String ContentFilteredTopic */
Topic^ topic = participant->create_topic(
"StringTopic", StringTypeSupport::get_type_name(),
DomainParticipant::TOPIC_QOS_DEFAULT,
nullptr, StatusMask::STATUS_MASK_NONE);
StringSeq^ parameters = gcnew StringSeq();
ContentFilteredTopic^ contentFilteredTopic =
participant->create_contentfilteredtopic(
"StringContentFilteredTopic", topic,
"value = 'Hello World!'", parameters);

C# Example:
using namespace DDS;
...
/* Create a String ContentFilteredTopic */
Topic topic = participant.create_topic(
"StringTopic", StringTypeSupport.get_type_name(),
DomainParticipant.TOPIC_QOS_DEFAULT,
null, StatusMask.STATUS_MASK_NONE);
StringSeq parameters = new StringSeq();
ContentFilteredTopic contentFilteredTopic =
participant.create_contentfilteredtopic(
"StringContentFilteredTopic", topic,
"value = 'Hello World!'", parameters);

Java Example:
import com.rti.dds.type.builtin.*;
...
/* Create a String ContentFilteredTopic */
Topic topic = participant.create_topic(
"StringTopic", StringTypeSupport.get_type_name(),
DomainParticipant.TOPIC_QOS_DEFAULT,
null, StatusKind.STATUS_MASK_NONE);
StringSeq parameters = new StringSeq();
ContentFilteredTopic contentFilteredTopic =
participant.create_contentfilteredtopic(
"StringContentFilteredTopic", topic,
"value = 'Hello World!'", parameters);

5.4.4 Deleting ContentFilteredTopics
To delete a ContentFilteredTopic, use the DomainParticipant’s delete_contentfilteredtopic() operation:
Make sure no DataReaders are using the ContentFilteredTopic. (If this is not true, the operation returns
PRECONDITION_NOT_MET.)
Delete the ContentFilteredTopic by using the DomainParticipant’s delete_contentfilteredtopic() operation.

218

5.4.5 Using a ContentFilteredTopic

DDS_ReturnCode_t delete_contentfilteredtopic
(DDSContentFilteredTopic * a_contentfilteredtopic)

5.4.5 Using a ContentFilteredTopic
Once you’ve created a ContentFilteredTopic, you can use the operations listed in Table 5.5 ContentFilteredTopic Operations.
Table 5.5 ContentFilteredTopic Operations
Operation

Description

Reference

append_to_expression_
parameter

Concatenates a string value to the input
expression parameter

Appending a String to an Expression Parameter (Section
5.4.5.3 on the facing page)

get_expression_
parameters

Gets the expression parameters.

Getting the Current Expression Parameters (Section 5.4.5.1
below)

get_filter_expression

Gets the expression.

Getting the Filter Expression (Section 5.4.5.5 on page 221)

get_related_topic

Gets the related Topic.

Getting the Related Topic (Section 5.4.5.6 on page 221)

narrow

Casts a DDS_TopicDescription pointer to a
ContentFilteredTopic pointer.

‘Narrowing’ a ContentFilteredTopic to a TopicDescription
(Section 5.4.5.7 on page 222)

remove_from_
expression_parameter

Removes a string value from the input expression Removing a String from an Expression Parameter (Section
parameter
5.4.5.4 on page 221)

set_expression

Changes the filter expression and parameters.

set_expression_
parameters

Changes the expression parameters.

Setting an Expression’s Filter and Parameters (Section
5.4.5.2 on the facing page)

5.4.5.1 Getting the Current Expression Parameters
To get the expression parameters, use the ContentFilteredTopic’s get_expression_parameters() operation:
DDS_ReturnCode_t get_expression_parameters(
struct DDS_StringSeq & parameters)

Where:
parameters

The filter expression parameters.

219

5.4.5.2 Setting an Expression’s Filter and Parameters

The memory for the strings in this sequence is managed as described in the String Support
section of the API Reference HTML documentation (within the Infrastructure module). In
particular, be careful to avoid a situation in which Connext DDS allocates a string on your behalf
and you then reuse that string in such a way that Connext DDS believes it to have more memory
allocated to it than it actually does. This parameter cannot be NULL.

This operation gives you the expression parameters that were specified on the last successful call to set_
expression_parameters() or set_expression(), or if they were never called, the parameters specified when
the ContentFilteredTopic was created.
5.4.5.2 Setting an Expression’s Filter and Parameters
To change the filter expression and expression parameters associated with a ContentFilteredTopic:
DDS_ReturnCode set_expression(
const char * expression,
const struct DDS_StringSeq & parameters)

To change just the expression parameters (not the filter expression):
DDS_ReturnCode_t set_expression_parameters(
const struct DDS_StringSeq & parameters)

Where:
expression

The new expression to be set in the ContentFilteredTopic.

parameters

The filter expression parameters. Each element in the parameter sequence corresponds to a
positional parameter in the filter expression. When using the default DDS_SQLFILTER_
NAME, parameter strings are automatically converted to the member type. For example, "4" is
converted to the integer 4. This parameter cannot be NULL.

The ContentFilteredTopic’s operations do not manage the sequences; you must ensure that the
parameter sequences are valid. Please refer to the String Support section in the API Reference
HTML documentation (within the Infrastructure module) for details on sequences.
5.4.5.3 Appending a String to an Expression Parameter
To concatenate a string to an expression parameter, use the ContentFilteredTopic's append_to_expression_parameter() operation:
DDS_ReturnCode_t append_to_expression_parameter(
const DDS_Long index,
const char* value);

When using the STRINGMATCH filter, index must be 0.

220

5.4.5.4 Removing a String from an Expression Parameter

This function is only intended to be used with the builtin SQL and STRINGMATCH filters. This function
can be used in expression parameters associated with MATCH operators (see SQL Extension: Regular
Expression Matching (Section 5.4.6.5 on page 228)) to add a pattern to the match pattern list. For
example, if filter_expression is:
symbol MATCH 'IBM'

Then append_to_expression_parameter(0, "MSFT") would generate the expression:
symbol MATCH 'IBM,MSFT'

5.4.5.4 Removing a String from an Expression Parameter
To remove a string from an expression parameter use the ContentFilteredTopic's remove_from_expression_parameter() operation:
DDS_ReturnCode_t remove_from_expression_parameter(
const DDS_Long index,
const char* value)

When using the STRINGMATCH filter, index must be 0.
This function is only intended to be used with the builtin SQL and STRINGMATCH filters. It can be
used in expression parameters associated with MATCH operators (see SQL Extension: Regular Expression Matching (Section 5.4.6.5 on page 228)) to remove a pattern from the match pattern list. For
example, if filter_expression is:
symbol MATCH 'IBM,MSFT'

Then remove_from_expression_parameter(0, "IBM") would generate the expression:
symbol MATCH 'MSFT'

5.4.5.5 Getting the Filter Expression
To get the filter expression that was specified when the ContentFilteredTopic was created or when set_
expression() was used:
const char* get_filter_expression ()

5.4.5.6 Getting the Related Topic
To get the related Topic that was specified when the ContentFilteredTopic was created:
DDS_Topic * get_related_topic ()

221

5.4.5.7 ‘Narrowing’ a ContentFilteredTopic to a TopicDescription

5.4.5.7 ‘Narrowing’ a ContentFilteredTopic to a TopicDescription
To safely cast a DDS_TopicDescription pointer to a ContentFilteredTopic pointer, use the ContentFilteredTopic’s narrow() operation:
DDS_TopicDescription* narrow ()

5.4.6 SQL Filter Expression Notation
A SQL filter expression is similar to the WHERE clause in SQL. The SQL expression format provided
by Connext DDS also supports the MATCH operator as an extended operator (see SQL Extension: Regular Expression Matching (Section 5.4.6.5 on page 228)).
The following sections provide more information:
l

Example SQL Filter Expressions (Section 5.4.6.1 below)

l

SQL Grammar (Section 5.4.6.2 on page 224)

l

Token Expressions (Section 5.4.6.3 on page 225)

l

Type Compatibility in the Predicate (Section 5.4.6.4 on page 227)

l

SQL Extension: Regular Expression Matching (Section 5.4.6.5 on page 228)

l

Composite Members (Section 5.4.6.6 on page 229)

l

Strings (Section 5.4.6.7 on page 229)

l

Enumerations (Section 5.4.6.8 on page 230)

l

Pointers (Section 5.4.6.9 on page 230)

l

Arrays (Section 5.4.6.10 on page 230)

l

Sequences (Section 5.4.6.11 on page 231)

5.4.6.1 Example SQL Filter Expressions
Assume that you have a Topic with two floats, X and Y, which are the coordinates of an object moving
inside a rectangle measuring 200 x 200 units. This object moves quite a bit, generating lots of DDS
samples that you are not interested in. Instead you only want to receive DDS samples outside the middle of
the rectangle, as seen in Filtering Example (Section Figure 5.5 on the next page). That is, you want to filter
out data points in the gray box.

222

5.4.6.1 Example SQL Filter Expressions

Figure 5.5 Filtering Example

The filter expression would look like this (remember the expression is written so that DDS samples that we
do want will pass):
"(X < 50 or X > 150) and (Y < 50 or Y > 150)"

While this filter works, it cannot be changed after the ContentFilteredTopic has been created. Suppose you
would like the ability to adjust the coordinates that are considered outside the acceptable range (changing
the size of the gray box). You can achieve this by using filter parameters. An more flexible way to write
the expression is this:
"(X < %0 or X > %1) and (Y < %2 or Y > %3)"

Recall that when you create a ContentFilteredTopic (see Creating ContentFilteredTopics (Section 5.4.3
on page 214)), you pass a expression_parameters string sequence as one of the parameters. Each element
in the string sequence corresponds to one argument.
See the String and Sequence Support sections of the API Reference HTML documentation (from the
Modules page, select RTI Connext DDS API Reference, Infrastructure Module).
In C++, the filter parameters could be assigned like this:
FilterParameter[0]
FilterParameter[1]
FilterParameter[2]
FilterParameter[3]

=
=
=
=

"50";
"150";
"50";
"150";

223

5.4.6.2 SQL Grammar

With these parameters, the filter expression is identical to the first approach. However, it is now possible to
change the parameters by calling set_expression_parameters(). For example, perhaps you decide that
you only want to see data points where X < 10 or X > 190. To make this change:
FilterParameter[0] = 10
FilterParameter[1] = 190
set_expression_parameters(....)

The new filter parameters will affect all DataReaders that have been created with this
ContentFilteredTopic.
5.4.6.2 SQL Grammar
This section describes the subset of SQL syntax, in Backus–Naur Form (BNF), that you can use to form
filter expressions.
The following notational conventions are used:
NonTerminals are typeset in italics.
'Terminals' are quoted and typeset in a fixed-width font. They are written in upper case in most cases in the
BNF-grammar below, but should be case insensitive.
TOKENS are typeset in bold.
The notation (element // ',') represents a non-empty, comma-separated list of elements.
Expression ::= FilterExpression
|
TopicExpression
|
QueryExpression
.
FilterExpression ::= Condition
TopicExpression ::= SelectFrom { Where } ';'
QueryExpression ::= { Condition }{ 'ORDER BY' (FIELDNAME // ',') }
.
SelectFrom
::= 'SELECT' Aggregation 'FROM' Selection
.
Aggregation
::= '*'
|
(SubjectFieldSpec // ',')
.
SubjectFieldSpec ::= FIELDNAME
|
FIELDNAME 'AS' IDENTIFIER
|
FIELDNAME IDENTIFIER
.
Selection
::= TOPICNAME
|
TOPICNAME NaturalJoin JoinItem
.
JoinItem
::= TOPICNAME
|
TOPICNAME NaturalJoin JoinItem
|
'(' TOPICNAME NaturalJoin JoinItem ')'
.
NaturalJoin ::= 'INNER JOIN'

224

5.4.6.3 Token Expressions

|
'INNER NATURAL JOIN'
|
'NATURAL JOIN'
|
'NATURAL INNER JOIN'
.
Where
::= 'WHERE' Condition
.
Condition
::= Predicate
|
Condition 'AND' Condition
|
Condition 'OR' Condition
|
'NOT' Condition
|
'(' Condition ')'
.
Predicate
::= ComparisonPredicate
|
BetweenPredicate
.
ComparisonPredicate ::= ComparisonTerm RelOp ComparisonTerm
.
ComparisonTerm
::= FieldIdentifier
| Parameter
.
BetweenPredicate
::= FieldIdentifier 'BETWEEN' Range
|
FieldIdentifier 'NOT BETWEEN' Range
.
FieldIdentifier
::= FIELDNAME
| IDENTIFIER
.
RelOp
::= '=' | '>' | '>=' | '<' | '<=' | '<>' | 'LIKE' | 'MATCH'
.
Range
::= Parameter 'AND' Parameter
.
Parameter ::= INTEGERVALUE
|
CHARVALUE
|
FLOATVALUE
|
STRING
|
ENUMERATEDVALUE
|
BOOLEANVALUE
|
PARAMETER

INNER JOIN, INNER NATURAL JOIN, NATURAL JOIN, and NATURAL INNER JOIN are all aliases, in the sense that they have the same semantics. They are all supported because they all are part of the
SQL standard.
5.4.6.3 Token Expressions
The syntax and meaning of the tokens used in SQL grammar is described as follows:
IDENTIFIER—An

identifier for a FIELDNAME, defined as any series of characters 'a', ..., 'z', 'A', ..., 'Z',
'0', ..., '9', '_' but may not start with a digit.
IDENTIFIER: LETTER (PART_LETTER)*

where LETTER: [ "A"-"Z","_","a"-"z" ] PART_LETTER: [ "A"-"Z","_","a"-"z","0"-"9" ]

225

5.4.6.3 Token Expressions

FIELDNAME—A reference to a field in the data structure. A dot '.' is used to navigate through nested
structures. The number of dots that may be used in a FIELDNAME is unlimited. The FIELDNAME can
refer to fields at any depth in the data structure. The names of the field are those specified in the IDL definition of the corresponding structure, which may or may not match the fieldnames that appear on the language-specific (e.g., C/C++, Java) mapping of the structure. To reference the n+1 element in an array or
sequence, use the notation '[n]', where n is a natural number (zero included). FIELDNAME must
resolve to a primitive IDL type; that is either boolean, octet, (unsigned) short, (unsigned) long, (unsigned)
long long, float double, char, wchar, string, wstring, or enum.
FIELDNAME: FieldNamePart ( "." FieldNamePart )*

where FieldNamePart : IDENTIFIER ( "[" Index "]" )* Index> : (["0"-"9"])+ | ["0x","0X"](["0"-"9",
"A"-"F", "a"-"f"])+
Primitive IDL types referenced by FIELDNAME are treated as different types in Predicate according to
the following table:
Predicate Data Type

IDL Type

BOOLEANVALUE

boolean

INTEGERVALUE

octet, (unsigned) short, (unsigned) long, (unsigned) long long

FLOATVALUE

float, double

CHARVALUE

char, wchar

STRING

string, wstring

ENUMERATEDVALUE

enum

TOPICNAME—An identifier for a topic, and is defined as any series of characters 'a', ..., 'z', 'A', ..., 'Z',
'0', ..., '9', '_' but may not start with a digit.
TOPICNAME : IDENTIFIER

INTEGERVALUE—Any series of digits, optionally preceded by a plus or minus sign, representing a
decimal integer value within the range of the system. 'L' or 'l' must be used for long long, otherwise long is
assumed. A hexadecimal number is preceded by 0x and must be a valid hexadecimal expression.
INTEGERVALUE : (["+","-"])? (["0"-"9"])+ [("L","l")]?
| (["+","-"])? ["0x","0X"](["0"-"9",
"A"-"F", "a"-"f"])+ [("L","l")]?

CHARVALUE—A single character enclosed between single quotes.
CHARVALUE : "'" (~["'"])? "'"

226

5.4.6.4 Type Compatibility in the Predicate

FLOATVALUE—Any series of digits, optionally preceded by a plus or minus sign and optionally including a floating point ('.'). 'F' or 'f' must be used for float, otherwise double is assumed. A power-of-ten
expression may be postfixed, which has the syntax en or En, where n is a number, optionally preceded by
a plus or minus sign.
FLOATVALUE : (["+","-"])? (["0"-"9"])* (".")? (["0"-"9"])+
(EXPONENT)?[("F",’f’)]?

where EXPONENT: ["e","E"] (["+","-"])? (["0"-"9"])+
STRING—Any series of characters encapsulated in single quotes, except the single quote itself.
STRING : "'" (~["'"])* "'"

ENUMERATEDVALUE—A reference to a value declared within an enumeration. Enumerated values
consist of the name of the enumeration label enclosed in single quotes. The name used for the enumeration
label must correspond to the label names specified in the IDL definition of the enumeration.
ENUMERATEDVALUE : "'" ["A" - "Z", "a" - "z"]
["A" - "Z", "a" - "z", "_", "0" - "9"]* "'"

BOOLEANVALUE—Can either be TRUE or FALSE, and is case insensitive.
BOOLEANVALUE : ["TRUE","FALSE"]

PARAMETER—Takes the form %n, where n represents a natural number (zero included) smaller than
100. It refers to the (n + 1)th argument in the given context. This argument can only be in primitive type
value format. It cannot be a FIELDNAME.
PARAMETER : "%" (["0"-"9"])+

5.4.6.4 Type Compatibility in the Predicate
As seen in Table 5.6 Valid Type Comparisons, only certain combinations of type comparisons are valid in
the Predicate.
Table 5.6 Valid Type Comparisons
BOOLEAN
VALUE
BOOLEAN

INTEGER
VALUE

FLOAT
VALUE

CHAR
VALUE

STRING

ENUMERATED
VALUE

YES

INTEGERVALUE

YES

YES

FLOATVALUE

YES

YES

227

5.4.6.5 SQL Extension: Regular Expression Matching

Table 5.6 Valid Type Comparisons
BOOLEAN
VALUE

INTEGER
VALUE

FLOAT
VALUE

CHAR
VALUE

STRING

ENUMERATED
VALUE

CHARVALUE

YES

YES

YES

STRING

YES

YES 1

YES

YES2

YES 3

YES 4

ENUMERATED
VALUE

YES

5.4.6.5 SQL Extension: Regular Expression Matching
The relational operator MATCH may only be used with string fields. The right-hand operator is a string
pattern. A string pattern specifies a template that the left-hand field must match.
MATCH is case-sensitive. These characters have special meaning: ,/?*[]-^!\%
The pattern allows limited "wild card" matching under the rules in Table 5.7 Wild Card Matching.
The syntax is similar to the POSIX® fnmatch syntax5. The MATCH syntax is also similar to the 'subject'
strings of TIBCO Rendezvous®. Some example expressions include:
"symbol MATCH 'NASDAQ/[A-G]*'"
"symbol MATCH 'NASDAQ/GOOG,NASDAQ/MSFT'"

Table 5.7 Wild Card Matching
Character
,

Meaning
A , separates a list of alternate patterns. The field string is matched if it matches one or more of the patterns.

aSee SQL Extension: Regular Expression Matching (Section 5.4.6.5 below).
2Because of the formal notation of the Enumeration values, they are compatible with string and char literals, but they are

not compatible with string or char variables, i.e., "MyEnum='EnumValue'" is correct, but "MyEnum=MyString" is not
allowed.
3Because of the formal notation of the Enumeration values, they are compatible with string and char literals, but they are

not compatible with string or char variables, i.e., "MyEnum='EnumValue'" is correct, but "MyEnum=MyString" is not
allowed.
4Only for same-type Enums.
5See http://www.opengroup.org/onlinepubs/000095399/functions/fnmatch.html.

228

5.4.6.6 Composite Members

Table 5.7 Wild Card Matching
Character

Meaning

/

A / in the pattern string matches a / in the field string. It separates a sequence of mandatory substrings.

?

A ? in the pattern string matches any single non-special characters in the field string.

*

A * in the pattern string matches 0 or more non-special characters in field string.

%

This special character is used to designate filter expression parameters.

\

(Not supported) Escape character for special characters.

[charlist]

Matches any one of the characters in charlist.

[!charlist] or [^charlist]

(Not supported) Matches any one of the characters not in charlist.

[s-e]

Matches any character from s to e, inclusive.

[!s-e] or [^s-e]

(Not supported) Matches any character not in the interval s to e.

5.4.6.6 Composite Members
Any member can be used in the filter expression, with the following exceptions:
l

128-bit floating point numbers (long doubles) are not supported

l

bitfields are not supported

l

LIKE is not supported

Composite members are accessed using the familiar dot notation, such as "x.y.z > 5". For unions, the notation is special due to the nature of the IDL union type.
On the publishing side, you can access the union discriminator with myunion._d and the actual member
with myunion._u.mymember. If you want to use a ContentFilteredTopic on the subscriber side and filter
a DDS sample with a top-level union, you can access the union discriminator directly with _d and the
actual member with mymember in the filter expression.
5.4.6.7 Strings
The filter expression and parameters can use IDL strings. String constants must appear between single quotation marks (').
For example:
" fish = 'salmon' "

229

5.4.6.8 Enumerations

Strings used as parameter values must contain the enclosing quotation marks (') within the parameter value;
do not place the quotation marks within the expression statement. For example, the expression " symbol
MATCH %0 " with parameter 0 set to " 'IBM' " is legal, whereas the expression " symbol MATCH '%0' "
with parameter 0 set to " IBM " will not compile.
5.4.6.8 Enumerations
A filter expression can use enumeration values, such as GREEN, instead of the numerical value. For
example, if x is an enumeration of GREEN, YELLOW and RED, the following expressions are valid:
"x = 'GREEN'"
"X < 'RED'"

5.4.6.9 Pointers
Pointers can be used in filter expressions and are automatically dereferenced to the correct value.
For example:
struct Point {
long x;
long y;
};
struct Rectangle {
Point *u_l;
Point *l_r;
};

The following expression is valid on a Topic of type Rectangle:
"u_l.x > l_r.x"

5.4.6.10 Arrays
Arrays are accessed with the familiar [] notation.
For example:
struct ArrayType {
long value[255][5];
};

The following expression is valid on a Topic of type ArrayType:
"value[244][2] = 5"

230

5.4.6.11 Sequences

In order to compare an array of bytes(octets in idl), instead of comparing each individual element of the
array using [] notation, Connext DDS provides a helper function, hex(). The hex() function can be used to
represent an array of bytes (octets in IDL). To use the hex() function, use the notation &hex() and pass the
byte array as a sequence of hexadecimal values.
For example:
&hex

(07 08 09 0A 0B 0c 0D 0E 0F 10 11 12 13 14 15 16)

Here the leftmost-pair represents the byte and index 0.
Note: If the length of the octet array represented by the hex() function does not match the length of the
field being compared, it will result in a compilation error.
For example:
struct ArrayType {
octet value[2];
};

The following expression is valid:
"value = &hex(12 0A)"

5.4.6.11 Sequences
Sequence elements can be accessed using the () or [] notation.
For example:
struct SequenceType {
sequence s;
};

The following expressions are valid on a Topic of type SequenceType:
"s(1) = 5"
"s[1] = 5"

5.4.7 STRINGMATCH Filter Expression Notation
The STRINGMATCH Filter is a subset of the SQL filter; it only supports the MATCH relational operator
on a single string field. It is introduced mainly for the use case of partitioning data according to channels in
the DataWriter's MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386) in Market Data applications.
A STRINGMATCH filter expression has the following syntax:
 MATCH 

231

5.4.7.1 Example STRINGMATCH Filter Expressions

The STRINGMATCH filter is provided to support the narrow use case of filtering a single string field of
the DDS sample against a comma-separated list of matching string values. It is intended to be used in conjunction with ContentFilteredTopic helper routines append_to_expression_parameter() (Appending a
String to an Expression Parameter (Section 5.4.5.3 on page 220)) and remove_from_expression_parameter() (Removing a String from an Expression Parameter (Section 5.4.5.4 on page 221)), which allow
you to easily append and remove individual string values from the comma-separated list of string values.
The STRINGMATCH filter must contain only one , and a single occurrence of the MATCH
operator. The  must be either the single parameter %0, or a single, comma-separated list of
strings without intervening spaces.
During creation of a STRINGMATCH filter, the  is automatically parameterized. That is,
during creation, if the  specified in the filter expression is not the parameter %0, then the
comma-separated list of strings is copied to the initial contents of parameter 0 and the  in
the filter expression is replaced with the parameter %0.
The initial matching string list is converted to an explicit parameter value so that subsequent additions and
deletions of string values to and from the list of matching strings may be performed with the append_to_
expression_parameter() and remove_from_expression_parameter() operations mentioned above.
5.4.7.1 Example STRINGMATCH Filter Expressions
This expression evaluates to TRUE if the value of symbol is equal to NASDAQ/MSFT:
symbol MATCH 'NASDAQ/MSFT'

This expression evaluates to TRUE if the value of symbol is equal to NASDAQ/IBM or
NASDAQ/MSFT:
symbol MATCH 'NASDAQ/IBM,NASDAQ/MSFT'

This expression evaluates to TRUE if the value of symbol corresponds to NASDAQ and starts with a letter between M and Y:
symbol MATCH 'NASDAQ/[M-Y]*'

5.4.7.2 STRINGMATCH Filter Expression Parameters
In the builtin STRINGMATCH filter, there is one, and only one, parameter: parameter 0. (If you want to
add more parameters, see Appending a String to an Expression Parameter (Section 5.4.5.3 on page 220).)
The parameter can be specified explicitly using the same syntax as the SQL filter or implicitly by using a
constant string pattern. For example:
symbol MATCH %0

(Explicit parameter)

symbol MATCH ‘IBM’ (Implicit parameter initialized to IBM)

232

5.4.8 Custom Content Filters

Strings used as parameter values must contain the enclosing quotation marks (') within the parameter value;
do not place the quotation marks within the expression statement. For example, the expression " symbol
MATCH %0 " with parameter 0 set to " 'IBM' " is legal, whereas the expression " symbol MATCH '%0' "
with parameter 0 set to " IBM " will not compile.

5.4.8 Custom Content Filters
By default, a ContentFilteredTopic will use a SQL-like content filter, DDS_SQLFILTER_NAME (see
SQL Filter Expression Notation (Section 5.4.6 on page 222)), which implements a superset of the content
filter. There is another builtin filter, DDS_STRINGMATCHFILTER_NAME (see STRINGMATCH
Filter Expression Notation (Section 5.4.7 on page 231)). Both of these are automatically registered.
If you want to use a different filter, you must register it first, then create the ContentFilteredTopic using
create_contentfilteredtopic_with_filter() (see Creating ContentFilteredTopics (Section 5.4.3 on page
214)).
One reason to use a custom filter is that the default filter can only filter based on relational operations
between topic members, not on a computation involving topic members. For example, if you want to filter
based on the sum of the members, you must create your own filter.
Notes:
l

The API for using a custom content filter is subject to change in a future release.

l

Custom content filters are not supported when using the .NET APIs

5.4.8.1 Filtering on the Writer Side with Custom Filters
There are two approaches for performing writer-side filtering. The first approach is to evaluate each written DDS sample against filters of all the readers that have content filter specified and identify the readers
whose filter passes the DDS sample.
The second approach is to evaluate the written DDS sample once for the writer and then rely on the filter
implementation to provide a set of readers whose filter passes the DDS sample. This approach allows the
filter implementation to cache the result of filtering, if possible. For example, consider a scenario where the
data is described by the struct shown below, where 10x > x ? DDS_BOOLEAN_FALSE : DDS_BOOLEAN_TRUE);
}

The function may use the following parameters:
compile_data The last return value from the compile function for this instance of the content filter. Can be
NULL.
sample

A pointer to a C structure with the data to filter. Note that the evaluate function always receives
deserialized data.

meta_data

A pointer to the meta data associated with the DDS sample.

238

5.4.8.7 Finalize Function

Note: Currently the meta_data field only supports related_sample_identity (described in Table 6.16
DDS_WriteParams_t).
5.4.8.7 Finalize Function
The finalize function specified in the ContentFilter will be called when an instance of the custom content
filter is no longer needed. When this function is called, it is safe to free all resources used by this particular
instance of the custom content filter.
For example:
void sample_finalize_function ( void* compile_data) {
/* free parameter string from compile function */
DDS_String_free((char *)compile_data);
}

The finalize function may use the following optional parameters:
system_key
See Compile Function (Section 5.4.8.5 on page 237).
handle
This is the opaque returned by the last call to the compile function.
5.4.8.8 Writer Attach Function
The writer_attach function specified in the WriterContentFilter will be used to create some state that can
be used by the filter to perform writer-side filtering more efficiently. It is entirely up to you, as the implementer of the filter, to decide if the filter requires this state.
The function has the following parameter:
writer_filter_data

A user-specified opaque pointer to some state created on the writer side that will help perform writer-side filtering efficiently.

5.4.8.9 Writer Detach Function
The writer_detach function specified in the WriterContentFilter will be used to free up any state that was
created using the writer_attach function.
The function has the following parameter:
writer_filter_data
A pointer to the state created using the writer_attach function.
5.4.8.10 Writer Compile Function
The writer_compile function specified in the WriterContentFilter will be used by a DataWriter to compile
a filter expression and parameters associated with a DataReader for which the DataWriter is performing fil-

239

5.4.8.11 Writer Evaluate Function

tering. The function will receive as input a DDS_Cookie_t that uniquely identifies the DataReader for
which the function was invoked.
The function has the following parameters:
writer_filter_ A pointer to the state created using the writer_attach function.
data
prop

A pointer to DDS_ExpressionProperty. This is an output parameter. It allows you to indicate to
Connext DDS if a filter expression can be optimized (as described in Filtering on the Writer
Side with Custom Filters (Section 5.4.8.1 on page 233)).

expression

An ASCIIZ string with the filter expression the ContentFilteredTopic was created with. Note
that the memory used by the parameter pointer is owned by Connext DDS. If you want to
manipulate this string, you must make a copy of it first. Do not free the memory for this string.

parameters

A string sequence of expression parameters used to create the ContentFilteredTopic. The string
sequence is equal (but not identical) to the string sequence passed to create_
contentfilteredtopic() (see expression_parameters in Creating ContentFilteredTopics
(Section 5.4.3 on page 214)).
The sequence passed to the compile function is owned by Connext DDS and must not be
referred to outside the writer_compile function.

type_code

A pointer to the type code of the related Topic. A type code is a description of the topic
members, such as their type (long, octet, etc.), but does not contain any information with respect
to the memory layout of the structures. The type code can be used to write filters that can be
used with any type. See Using Generated Types without Connext DDS (Standalone) (Section
3.7 on page 139). [Note: If you are using the Java API, this parameter will always be NULL.]

type_class_
name

The fully qualified class name of the related Topic.

cookie

A DDS_Cookie_t to uniquely identify the DataReader for which the writer_compile function
was called.

5.4.8.11 Writer Evaluate Function
The writer_evaluate function specified in the WriterContentFilter will be used by a DataWriter to retrieve
the list of DataReaders whose filter passed the DDS sample. The writer_evaluate function returns a
sequence of cookies which identifies the set of DataReaders whose filter passes the DDS sample.
The function has the following parameters:
writer_filter_
data

A pointer to the state created using the writer_attach function.

sample

A pointer to the data to be filtered. Note that the writer_evaluate function always receives
deserialized data.

240

5.4.8.12 Writer Return Loan Function

meta_data

A pointer to the meta-data associated with the DDS sample.

Note: Currently the meta_data field only supports related_sample_identity (described in Table 6.16
DDS_WriteParams_t).
5.4.8.12 Writer Return Loan Function
Connext DDS uses the writer_return_loan function specified in the WriterContentFilter to indicate to the
filter implementation that it has finished using the sequence of cookies returned by the filter’s writer_evaluate function. Your filter implementation should not free the memory associated with the cookie sequence
before the writer_return_loan function is called.
The function has the following parameters:
writer_filter_data A pointer to the state created using the writer_attach function.
cookies

The sequence of cookies for which the writer_return_loan function was called.

5.4.8.13 Writer Finalize Function
The writer_finalize function specified in the WriterContentFilter will be called when the DataWriter no
longer matches with a DataReader that was created with ContentFilteredTopic. This will allow the filter
implementation to delete any state it was maintaining for the DataReader.
The function has the following parameters:
writer_filter_
A pointer to the state created using the writer_attach function.
data
cookie

A DDS_Cookie_t to uniquely identify the DataReader for which the writer_finalize
was called.

241

Chapter 6 Sending Data
This section discusses how to create, configure, and use Publishers and DataWriters to send data.
It describes how these Entities interact, as well as the types of operations that are available for
them.
This section includes the following sections:
The goal of this section is to help you become familiar with the Entities you need for sending data.
For up-to-date details such as formal parameters and return codes on any mentioned operations,
please see the API Reference HTML documentation.

6.1 Preview: Steps to Sending Data
To send DDS samples of a data instance:
1. Create and configure the required Entities:
a. Create a DomainParticipant (see Creating a DomainParticipant (Section 8.3.1 on
page 556)).
b. Register user data types1 with the DomainParticipant. For example, the
‘FooDataType’. (This step is not necessary in the Modern C++ API--the Topic instantiation automatically registers the type)
c. Use the DomainParticipant to create a Topic with the registered data type.
d. Optionally2, use the DomainParticipant to create a Publisher.

1Type registration is not required for built-in types (see Registering Built-in Types (Section 3.2.1 on page 30)).
2You are not required to explicitly create a Publisher; instead, you can use the 'implicit Publisher' created from

the DomainParticipant. See Creating Publishers Explicitly vs. Implicitly (Section 6.2.1 on page 248).

242

6.2 Publishers

e. Use the Publisher or DomainParticipant to create a DataWriter for the Topic.
f. Use a type-safe method to cast the generic DataWriter created by the Publisher to a type-specific DataWriter. For example, ‘FooDataWriter’. (This step doesn't apply to the Modern
C++ API where you directly instantiate a type-safe ‘DataWriter.')
g. Optionally, register data instances with the DataWriter. If the Topic’s user data type contain
key fields, then registering a data instance (data with a specific key value) will improve performance when repeatedly sending data with the same key. You may register many different
data instances; each registration will return an instance handle corresponding to the specific
key value. For non-keyed data types, instance registration has no effect. See DDS Samples,
Instances, and Keys (Section 2.3.1 on page 14) for more information on keyed data types and
instances.
2. Every time there is changed data to be published:
a. Store the data in a variable of the correct data type (for instance, variable ‘Foo’ of the type
‘FooDataType’).
b. Call the FooDataWriter’s write() operation, passing it a reference to the variable ‘Foo’.
l For non-keyed data types or for non-registered instances, also pass in DDS_
HANDLE_NIL.
l

For keyed data types, pass in the instance handle corresponding to the instance stored in
‘Foo’, if you have registered the instance previously. This means that the data stored in
‘Foo’ has the same key value that was used to create instance handle.

c. The write() function will take a snapshot of the contents of ‘Foo’ and store it in Connext
DDS internal buffers from where the DDS data sample is sent under the criteria set by the
Publisher’s and DataWriter’s QosPolicies. If there are matched DataReaders, then the DDS
data sample will have been passed to the physical transport plug-in/device driver by the time
that write() returns.

6.2 Publishers
An application that intends to publish information needs the following Entities: DomainParticipant, Topic,
Publisher, and DataWriter. All Entities have a corresponding specialized Listener and a set of
QosPolicies. A Listener is how Connext DDS notifies your application of status changes relevant to the
Entity. The QosPolicies allow your application to configure the behavior and resources of the Entity.
l
l

l

A DomainParticipant defines the DDS domain in which the information will be made available.
A Topic defines the name under which the data will be published, as well as the type (format) of the
data itself.
An application writes data using a DataWriter. The DataWriter is bound at creation time to a Topic,
thus specifying the name under which the DataWriter will publish the data and the type associated

243

6.2 Publishers

with the data. The application uses the DataWriter’s write() operation to indicate that a new value
of the data is available for dissemination.
l

A Publisher manages the activities of several DataWriters. The Publisher determines when the data
is actually sent to other applications. Depending on the settings of various QosPolicies of the Publisher and DataWriter, data may be buffered to be sent with the data of other DataWriters or not
sent at all. By default, the data is sent as soon as the DataWriter’s write() function is called.
You may have multiple Publishers, each managing a different set of DataWriters, or you may
choose to use one Publisher for all your DataWriters.
For more information, see Creating Publishers Explicitly vs. Implicitly (Section 6.2.1 on page 248).

Figure 6.1 Publication Module below shows how these Entities are related, as well as the methods defined
for each Entity.
Figure 6.1 Publication Module

Publishers are used to perform the operations listed in Table 6.1 Publisher Operations on the next page.
You can find more information about the operations by looking in the section listed under the Reference

244

6.2 Publishers

column. For details such as formal parameters and return codes, please see the API Reference HTML documentation.
Some operations cannot be used within a listener callback, see Restricted Operations in Listener
Callbacks (Section 4.5.1 on page 185).
Table 6.1 Publisher Operations
Working
with ...

Operation

Description

begin_coherent_
changes

Indicates that the application will begin a coherent set of modifications.

create_datawriter

Creates a DataWriter that will belong to the Publisher.

DataWriters create_
datawriter_
with_profile

copy_from_
topic_qos

Sets the DataWriter’s QoS based on a specified QoS profile.

Copies relevant QosPolicies from a Topic into a DataWriterQoS structure.

delete_contained_
Deletes all of the DataWriters that were created by the Publisher.
entities

DataWriters delete_datawriter
cont'd

end_coherent_
changes

Reference
Writing Coherent Sets of
DDS Data Samples
(Section 6.3.10 on page
287)

Creating DataWriters
(Section 6.3.1 on page
266)

Other Publisher QoSRelated Operations
(Section 6.2.4.6 on page
257)
Deleting Contained
DataWriters (Section
6.2.3.1 on page 251)

Deletes a DataWriter that belongs to the Publisher.

Deleting DataWriters
(Section 6.3.3 on page
268)

Ends the coherent set initiated by begin_coherent_changes().

Writing Coherent Sets of
DDS Data Samples
(Section 6.3.10 on page
287)

245

6.2 Publishers

Table 6.1 Publisher Operations
Working
with ...

DataWriters
cont'd

Operation

Description

Reference

get_all_
datawriters

Retrieves all the DataWriters created from this Publisher.

Getting All DataWriters
(Section 6.3.2 on page
268)

get_default_
datawriter_qos

Copies the Publisher’s default DataWriterQoS values into a DataWriterQos
structure.

Setting DataWriter
QosPolicies (Section
6.3.15 on page 300)

get_status_
changes

Will always return 0 since there are no Statuses currently defined for
Publishers.

Getting Status and Status
Changes (Section 4.1.4 on
page 157)

lookup_
datawriter

Retrieves a DataWriter previously created for a specific Topic.

Finding a Publisher’s
Related DDS Entities
(Section 6.2.6 on page
259)

set_default_
datawriter_qos

Sets or changes the default DataWriterQos values.

set_default_
datawriter_
qos_with_profile

Sets or changes the default DataWriterQos values based on a QoS profile.

Blocks until all data written by the Publisher’s reliable DataWriters are
DataWriters wait_for_
acknowledged by all matched reliable DataReaders, or until the a specified
acknowledgments
cont'd
timeout duration, max_wait, elapses.

Getting and Setting
Default QoS for
DataWriters (Section
6.2.4.5 on page 256)

Waiting for
Acknowledgments in a
Publisher (Section 6.2.7
on page 260)

Indicates if a sample has been application-acknowledged by all the matching
DataReaders that were alive when the sample was written.
is_sample_app_
acknowledged

Application
If a DataReader does not enable application acknowledgment (by setting the Acknowledgment (Section
ReliabilityQosPolicy's acknowledgment_kind to a value other than DDS_
6.3.12 on page 288)
PROTOCOL_ACKNOWLEDGMENT_MODE), the sample is considered
application-acknowledged for that DataReader.

246

6.2 Publishers

Table 6.1 Publisher Operations
Working
with ...

Operation
get_default_
library

Gets the Publisher’s default QoS profile library.

get_default_
profile

Gets the Publisher’s default QoS profile.

get_default_
Libraries
profile_
and Profiles
library

Gets the library that contains the Publisher’s default QoS profile.

set_default_
library

Sets the default library for a Publisher.

set_default_
profile

Sets the default profile for a Publisher.

Reference

Getting and Setting the
Publisher’s Default QoS
Profile and Library
(Section 6.2.4.4 on page
255)

Gets the DomainParticipant that was used to create the Publisher.

Finding a Publisher’s
Related DDS Entities
(Section 6.2.6 on page
259)

enable

Enables the Publisher.

Enabling DDS Entities
(Section 4.1.2 on page
154)

equals

Compares two Publisher’s QoS structures for equality.

Comparing QoS Values
(Section 6.2.4.2 on page
254)

get_qos

Gets the Publisher’s current QosPolicy settings. This is most often used in
preparation for calling set_qos().

set_qos

Sets the Publisher’s QoS. You can use this operation to change the values
for the Publisher’s QosPolicies. Note, however, that not all QosPolicies can
be changed after the Publisher has been created.

set_qos_with_
profile

Sets the Publisher’s QoS based on a specified QoS profile.

Participants get_participant

Publishers

Description

Setting Publisher
QosPolicies (Section 6.2.4
on page 251)

247

6.2.1 Creating Publishers Explicitly vs. Implicitly

Table 6.1 Publisher Operations
Working
with ...

Publishers
cont'd

Operation

Description

get_listener

Gets the currently installed Listener.

set_listener

Sets the Publisher’s Listener. If you created the Publisher without a
Listener, you can use this operation to add one later.

suspend_
publications
resume_
publications

Reference
Setting Up
PublisherListeners
(Section 6.2.5 on page
257)

Provides a hint that multiple data-objects within the Publisher are about to be
written. Connext DDS does not currently use this hint.
Suspending and Resuming
Publications (Section 6.2.9
on page 261)
Reverses the action of suspend_publications().

6.2.1 Creating Publishers Explicitly vs. Implicitly
To send data, your application must have a Publisher. However, you are not required to explicitly
create one. If you do not create one, the middleware will implicitly create a Publisher the first time you
create a DataWriter using the DomainParticipant’s operations. It will be created with default QoS (DDS_
PUBLISHER_QOS_DEFAULT) and no Listener.
A Publisher (implicit or explicit) gets its own default QoS and the default QoS for its child DataWriters
from the DomainParticipant. These default QoS are set when the Publisher is created. (This is true for
Subscribers and DataReaders, too.)
The 'implicit Publisher' can be accessed using the DomainParticipant’s get_implicit_publisher() operation (see Getting the Implicit Publisher or Subscriber (Section 8.3.9 on page 569)). You can use this
‘implicit Publisher’ just like any other Publisher (it has the same operations, QosPolicies, etc.). So you can
change the mutable QoS and set a Listener if desired.
DataWriters are created by calling create_datawriter() or create_datawriter_with_profile()—these
operations exist for DomainParticipants and Publishers. If you use the DomainParticipant to create a
DataWriter, it will belong to the implicit Publisher. If you use a Publisher to create a DataWriter, it will
belong to that Publisher.
The middleware will use the same implicit Publisher for all DataWriters that are created using the
DomainParticipant’s operations.
Having the middleware implicitly create a Publisher allows you to skip the step of creating a Publisher.
However, having all your DataWriters belong to the same Publisher can reduce the concurrency of the system because all the write operations will be serialized.

248

6.2.2 Creating Publishers

6.2.2 Creating Publishers
Before you can explicitly create a Publisher, you need a DomainParticipant (see DomainParticipants (Section 8.3 on page 547)). To create a Publisher, use the DomainParticipant’s create_publisher() or create_
publisher_with_profile() operations.
A QoS profile is way to use QoS settings from an XML file or string. With this approach, you can change
QoS settings without recompiling the application. For details, see Configuring QoS with XML (Section
Chapter 17 on page 791).
Note: The Modern C++ API Publishers provide constructors whose first and only required argument is the
DomainParticipant.
DDSPublisher * create_publisher (
const DDS_PublisherQos &qos,
DDSPublisherListener *listener,
DDS_StatusMask mask)
DDSPublisher * create_publisher_with_profile (
const char *library_name,
const char *profile_name,
DDSPublisherListener *listener,
DDS_StatusMask mask)

Where:
qos

If you want the default QoS settings (described in the API Reference HTML documentation),
use DDS_PUBLISHER_QOS_DEFAULT for this parameter (see Creating a Publisher with
Default QosPolicies (Section Figure 6.2 on the facing page)).
If you want to customize any of the QosPolicies, supply a QoS structure (see Creating a
Publisher with Non-Default QosPolicies (not from a profile) (Section Figure 6.3 on page
253)). The QoS structure for a Publisher is described in Publisher/Subscriber QosPolicies
(Section 6.4 on page 312).

Note: If you use DDS_PUBLISHER_QOS_DEFAULT, it is not safe to create the Publisher
while another thread may be simultaneously calling set_default_publisher_qos().
listener

Listeners are callback routines. Connext DDS uses them to notify your application when specific
events (status changes) occur with respect to the Publisher or the DataWriters created by the
Publisher.
The listener parameter may be set to NULL if you do not want to install a Listener. If you use
NULL, the Listener of the DomainParticipant to which the Publisher belongs will be used
instead (if it is set). For more information on PublisherListeners, see Setting Up
PublisherListeners (Section 6.2.5 on page 257).

mask

This bit-mask indicates which status changes will cause the Publisher’s Listener to be invoked.
The bits set in the mask must have corresponding callbacks implemented in the Listener.

249

6.2.3 Deleting Publishers

If you use NULL for the Listener, use DDS_STATUS_MASK_NONE for this parameter. If
the Listener implements all callbacks, use DDS_STATUS_MASK_ALL. For information on
statuses, see Listeners (Section 4.4 on page 177).

library_name A QoS Library is a named set of QoS profiles. See URL Groups (Section 17.8 on page 814). If
NULL is used for library_name , the DomainParticipant’s default library is assumed (see
Getting and Setting the Publisher’s Default QoS Profile and Library (Section 6.2.4.4 on page
255)).
profile_name A QoS profile groups a set of related QoS, usually one per entity. See URL Groups (Section
17.8 on page 814). If NULL is used for profile_name , the DomainParticipant’s default profile
is assumed and library_name is ignored

Figure 6.2 Creating a Publisher with Default QosPolicies

// create the publisher
DDSPublisher* publisher =
participant->create_publisher(
DDS_PUBLISHER_QOS_DEFAULT,
NULL, DDS_STATUS_MASK_NONE);
if (publisher == NULL) {
// handle error
};

For more examples, see Configuring QoS Settings when the Publisher is Created (Section 6.2.4.1 on page
252).
After you create a Publisher, the next step is to use the Publisher to create a DataWriter for each Topic,
see Creating DataWriters (Section 6.3.1 on page 266). For a list of operations you can perform with a Publisher, see Table 6.1 Publisher Operations.

6.2.3 Deleting Publishers
(Note: in the Modern C++ API, Entities are automatically destroyed, see Creating and Deleting DDS Entities (Section 4.1.1 on page 153))
This section applies to both implicitly and explicitly created Publishers.
To delete a Publisher:
1. You must first delete all DataWriters that were created with the Publisher. Use the Publisher’s
delete_datawriter() operation to delete them one at a time, or use the delete_contained_entities()
operation (Deleting Contained DataWriters (Section 6.2.3.1 on the next page)) to delete them all at
the same time.
DDS_ReturnCode_t delete_datawriter (DDSDataWriter *a_datawriter)

250

6.2.3.1 Deleting Contained DataWriters

2. Delete the Publisher by using the DomainParticipant’s delete_publisher() operation.
DDS_ReturnCode_t delete_publisher (DDSPublisher *p)

Note: A Publisher cannot be deleted within a Listener callback, see Restricted Operations in Listener Callbacks (Section 4.5.1 on page 185).
6.2.3.1 Deleting Contained DataWriters
The Publisher’s delete_contained_entities() operation deletes all the DataWriters that were created by the
Publisher.
DDS_ReturnCode_t delete_contained_entities ()

After this operation returns successfully, the application may delete the Publisher (see Deleting Publishers
(Section 6.2.3 on the previous page)).

6.2.4 Setting Publisher QosPolicies
A Publisher’s QosPolicies control its behavior. Think of the policies as the configuration and behavior
‘properties’ of the Publisher. The DDS_PublisherQos structure has the following format:
DDS_PublisherQos struct {
DDS_PresentationQosPolicy presentation;
DDS_PartitionQosPolicy partition;
DDS_GroupDataQosPolicy group_data;
DDS_EntityFactoryQosPolicy entity_factory;
DDS_AsynchronousPublisherQosPolicy asynchronous_publisher;
DDS_ExclusiveAreaQosPolicy exclusive_area;
DDS_EntityNameQosPolicy publisher_name;
} DDS_PublisherQos;

Note: set_qos() cannot always be used in a listener callback; see Restricted Operations in Listener Callbacks (Section 4.5.1 on page 185).
Table 6.2 Publisher QosPolicies summarizes the meaning of each policy. (They appear alphabetically in
the table.) For information on why you would want to change a particular QosPolicy, see the referenced
section. For defaults and valid ranges, please refer to the API Reference HTML documentation for each
policy.

251

6.2.4.1 Configuring QoS Settings when the Publisher is Created

Table 6.2 Publisher QosPolicies
QosPolicy

Description

ASYNCHRONOUS_PUBLISHER
QosPolicy (DDS Extension) (Section
6.4.1 on page 313)

Configures the mechanism that sends user data in an external middleware thread.

ENTITYFACTORY QosPolicy
(Section 6.4.2 on page 315)

Controls whether or not child Entities are created in the enabled state.

ENTITY_NAME QosPolicy (DDS
Extension) (Section 6.5.9 on page 374)

Assigns a name and role_name to a Publisher.

EXCLUSIVE_AREA QosPolicy (DDS
Extension) (Section 6.4.3 on page 318)

Configures multi-thread concurrency and deadlock prevention capabilities.

GROUP_DATA QosPolicy (Section
6.4.4 on page 320)

Along with TOPIC_DATA QosPolicy (Section 5.2.1 on page 209) and USER_DATA
QosPolicy (Section 6.5.26 on page 417), this QosPolicy is used to attach a buffer of bytes to
Connext DDS's discovery meta-data.

PARTITION QosPolicy (Section 6.4.5
on page 323)

Adds string identifiers that are used for matching DataReaders and DataWriters for the same
Topic.

PRESENTATION QosPolicy (Section
6.4.6 on page 330)

Controls how Connext DDS presents data received by an application to the DataReaders of the
data.

6.2.4.1 Configuring QoS Settings when the Publisher is Created
As described in Creating Publishers (Section 6.2.2 on page 249), there are different ways to create a Publisher, depending on how you want to specify its QoS (with or without a QoS Profile).
l

l

In Creating a Publisher with Default QosPolicies (Section Figure 6.2 on page 250) we saw an
example of how to explicitly create a Publisher with default QosPolicies. It used the special constant, DDS_PUBLISHER_QOS_DEFAULT, which indicates that the default QoS values for a
Publisher should be used. Default Publisher QosPolicies are configured in the DomainParticipant;
you can change them with the DomainParticipant’s set_default_publisher_qos() or set_default_
publisher_qos_with_profile() operation (see Getting and Setting Default QoS for Child Entities
(Section 8.3.6.5 on page 568)).
To create a Publisher with non-default QoS settings, without using a QoS profile, see Figure 6.3
Creating a Publisher with Non-Default QosPolicies (not from a profile) on the next page. It uses the
DomainParticipant’s get_default_publisher_qos() method to initialize a DDS_PublisherQos structure. Then the policies are modified from their default values before the QoS structure is passed to
create_publisher().

252

6.2.4.1 Configuring QoS Settings when the Publisher is Created

l

l

You can also create a Publisher and specify its QoS settings via a QoS Profile. To do so, call create_publisher_with_profile(), as seen in Figure 6.4 Creating a Publisher with a QoS Profile on the
next page.
If you want to use a QoS profile, but then make some changes to the QoS before creating the Publisher, call the DomainParticipantFactory’s get_publisher_qos_from_profile(), modify the QoS
and use the modified QoS structure when calling create_publisher(), as seen in Figure 6.5 Getting
QoS Values from a Profile, Changing QoS Values, Creating a Publisher with Modified QoS Values
on the facing page.

For more information, see Creating Publishers (Section 6.2.2 on page 249) and Configuring QoS with
XML (Section Chapter 17 on page 791).
Figure 6.3 Creating a Publisher with Non-Default QosPolicies (not from a profile)
DDS_PublisherQos publisher_qos;1
// get defaults
if (participant->get_default_publisher_qos(publisher_qos) != DDS_RETCODE_OK){
// handle error
}
// make QoS changes here
// for example, this changes the ENTITY_FACTORY QoS
publisher_qos.entity_factory.autoenable_created_entities = DDS_BOOLEAN_FALSE;
// create the publisher
DDSPublisher* publisher = participant->create_publisher(publisher_qos,
NULL, DDS_STATUS_MASK_NONE);
if (publisher == NULL) {
// handle error
}

Figure 6.4 Creating a Publisher with a QoS Profile
// create the publisher with QoS profile
DDSPublisher* publisher = participant->create_publisher_with_profile(
“MyPublisherLibary”, “MyPublisherProfile”,
NULL, DDS_STATUS_MASK_NONE);
if (publisher == NULL) {
// handle error
}

1For the C API, you need to use DDS_PublisherQos_INITIALIZER or DDS_PublisherQos_initialize().

See Special QosPolicy Handling Considerations for C (Section 4.2.2 on page 168)
253

6.2.4.2 Comparing QoS Values

Figure 6.5 Getting QoS Values from a Profile, Changing QoS Values, Creating a Publisher
with Modified QoS Values
DDS_PublisherQos publisher_qos;1
// Get publisher QoS from profile
retcode = factory->get_publisher_qos_from_profile(publisher_qos,
“PublisherLibrary”, “PublisherProfile”);
if (retcode != DDS_RETCODE_OK) {
// handle error
}
// Makes QoS changes here
// New entity_factory autoenable_created_entities will be true
publisher_qos.entity_factory.autoenable_created_entities = DDS_BOOLEAN_TRUE;
// create the publisher with modified QoS
DDSPublisher* publisher = participant->create_publisher(
“Example Foo”, type_name, publisher_qos,
NULL, DDS_STATUS_MASK_NONE);
if (publisher == NULL) {
// handle error
}

6.2.4.2 Comparing QoS Values
The equals() operation compares two Publisher’s DDS_PublisherQoS structures for equality. It takes two
parameters for the two Publisher’s QoS structures to be compared, then returns TRUE is they are equal
(all values are the same) or FALSE if they are not equal.
6.2.4.3 Changing QoS Settings After the Publisher Has Been Created
There are 2 ways to change an existing Publisher’s QoS after it is has been created—again depending on
whether or not you are using a QoS Profile.
l

l

To change an existing Publisher’s QoS programmatically (that is, without using a QoS profile): get_
qos() and set_qos(). See the example code in Figure 6.6 Changing the Qos of an Existing Publisher
on the next page. It retrieves the current values by calling the Publisher’s get_qos() operation. Then
it modify the value and call set_qos() to apply the new value. Note, however, that some QosPolicies
cannot be changed after the Publisher has been enabled—this restriction is noted in the descriptions
of the individual QosPolicies.
You can also change a Publisher’s (and all other Entities’) QoS by using a QoS Profile and calling
set_qos_with_profile(). For an example, see Figure 6.7 Changing the QoS of an Existing Publisher

1For the C API, you need to use DDS_PublisherQos_INITIALIZER or DDS_PublisherQos_initialize().

See Special QosPolicy Handling Considerations for C (Section 4.2.2 on page 168)
254

6.2.4.4 Getting and Setting the Publisher’s Default QoS Profile and Library

with a QoS Profile on the next page. For more information, see Configuring QoS with XML (Section Chapter 17 on page 791).
Figure 6.6 Changing the Qos of an Existing Publisher
DDS_PublisherQos publisher_qos;1
// Get current QoS. publisher points to an existing DDSPublisher.
if (publisher->get_qos(publisher_qos) != DDS_RETCODE_OK) {
// handle error
}
// make changes
// New entity_factory autoenable_created_entities will be true
publisher_qos.entity_factory.autoenable_created_entities =DDS_BOOLEAN_TRUE;
// Set the new QoS
if (publisher->set_qos(publisher_qos) != DDS_RETCODE_OK ) {
// handle error
}

Figure 6.7 Changing the QoS of an Existing Publisher with a QoS Profile
retcode = publisher->set_qos_with_profile(
“PublisherProfileLibrary”,”PublisherProfile”);
if (retcode != DDS_RETCODE_OK) {
// handle error
}

6.2.4.4 Getting and Setting the Publisher’s Default QoS Profile and Library
You can retrieve the default QoS profile used to create Publishers with the get_default_profile() operation.
You can also get the default library for Publishers, as well as the library that contains the Publisher’s
default profile (these are not necessarily the same library); these operations are called get_default_library
() and get_default_library_profile(), respectively. These operations are for informational purposes only
(that is, you do not need to use them as a precursor to setting a library or profile.) For more information,
see Configuring QoS with XML (Section Chapter 17 on page 791).
virtual const char * get_default_library ()
const char * get_default_profile ()
const char * get_default_profile_library ()

There are also operations for setting the Publisher’s default library and profile:

1For the C API, you need to use DDS_PublisherQos_INITIALIZER or DDS_PublisherQos_initialize().

See Special QosPolicy Handling Considerations for C (Section 4.2.2 on page 168)
255

6.2.4.5 Getting and Setting Default QoS for DataWriters

DDS_ReturnCode_t set_default_library (const char *
DDS_ReturnCode_t set_default_profile (const char *
const char * profile_name)

library_name)
library_name,

These operations only affect which library/profile will be used as the default the next time a default Publisher library/profile is needed during a call to one of this Publisher’s operations.
When calling a Publisher operation that requires a profile_name parameter, you can use NULL to refer to
the default profile. (This same information applies to setting a default library.) If the default library/profile
is not set, the Publisher inherits the default from the DomainParticipant.
set_default_profile() does not set the default QoS for DataWriters created by the Publisher; for this functionality, use the Publisher’s set_default_datawriter_qos_with_profile(), see Getting and Setting Default
QoS for DataWriters (Section 6.2.4.5 below) (you may pass in NULL aftercalling the Publisher’s set_
default_profile()).
set_default_profile() does not set the default QoS for newly created Publishers; for this functionality, use
the DomainParticipant’s set_default_publisher_qos_with_profile() operation, see Getting and Setting
Default QoS for Child Entities (Section 8.3.6.5 on page 568).
6.2.4.5 Getting and Setting Default QoS for DataWriters
These operations set the default QoS that will be used for new DataWriters if create_datawriter() is
called with DDS_DATAWRITER_QOS_DEFAULT as the qos parameter:
DDS_ReturnCode_t set_default_datawriter_qos (const DDS_DataWriterQos &qos)
DDS_ReturnCode_t set_default_datawriter_qos_with_profile (
const char *library_name,
const char *profile_name)

The above operations may potentially allocate memory, depending on the sequences contained in some
QoS policies.
To get the default QoS that will be used for creating DataWriters if create_datawriter() is called with
DDS_PARTICIPANT_QOS_DEFAULT as the qos parameter:
DDS_ReturnCode_t get_default_datawriter_qos (DDS_DataWriterQos & qos)

This operation gets the QoS settings that were specified on the last successful call to set_default_
datawriter_qos() or set_default_datawriter_qos_with_profile(), or if the call was never made, the
default values listed in DDS_DataWriterQos.
Note: It is not safe to set the default DataWriter QoS values while another thread may be simultaneously
calling get_default_datawriter_qos(), set_default_datawriter_qos(), or create_datawriter() with
DDS_DATAWRITER_QOS_DEFAULT as the qos parameter. It is also not safe to get the default
DataWriter QoS values while another thread may be simultaneously calling set_default_datawriter_qos
().

256

6.2.4.6 Other Publisher QoS-Related Operations

6.2.4.6 Other Publisher QoS-Related Operations
l

Copying a Topic’s QoS into a DataWriter’s QoS
This method is provided as a convenience for setting the values in a DataWriterQos structure before
using that structure to create a DataWriter. As explained in Setting Topic QosPolicies (Section 5.1.3
on page 204), most of the policies in a TopicQos structure do not apply directly to the Topic itself,
but to the associated DataWriters and DataReaders of that Topic. The TopicQos serves as a single
container where the values of QosPolicies that must be set compatibly across matching DataWriters
and DataReaders can be stored.
Thus instead of setting the values of the individual QosPolicies that make up a DataWriterQos structure every time you need to create a DataWriter for a Topic, you can use the Publisher’s copy_
from_topic_qos() operation to “import” the Topic’s QosPolicies into a DataWriterQos structure.
This operation copies the relevant policies in the TopicQos to the corresponding policies in the
DataWriterQos.

l

This copy operation will often be used in combination with the Publisher’s get_default_
datawriter_qos() and the Topic’s get_qos() operations. The Topic’s QoS values are merged on top
of the Publisher’s default DataWriter QosPolicies with the result used to create a new DataWriter,
or to set the QoS of an existing one (see Setting DataWriter QosPolicies (Section 6.3.15 on page
300)).
Copying a Publisher’s QoS

l

C API users should use the DDS_PublisherQos_copy() operation rather than using structure assignment when copying between two QoS structures. The copy() operation will perform a deep copy so
that policies that allocate heap memory such as sequences are copied correctly. In C++, C++/CLI,
C# and Java, a copy constructor is provided to take care of sequences automatically.
Clearing QoS-Related Memory
Some QosPolicies contain sequences that allocate memory dynamically as they grow or shrink. The
C API’s DDS_PublisherQos_finalize() operation frees the memory used by sequences but otherwise
leaves the QoS unchanged. C API users should call finalize() on all DDS_PublisherQos objects
before they are freed, or for QoS structures allocated on the stack, before they go out of scope. In
C++, C++/CLI, C# and Java, the memory used by sequences is freed in the destructor.

6.2.5 Setting Up PublisherListeners
Like all Entities, Publishers may optionally have Listeners. Listeners are user-defined objects that implement a DDS-defined interface (i.e. a pre-defined set of callback functions). Listeners provide the means for
Connext DDS to notify applications of any changes in Statuses (events) that may be relevant to it. By writing the callback functions in the Listener and installing the Listener into the Publisher, applications can be
notified to handle the events of interest. For more general information on Listeners and Statuses, see Listeners (Section 4.4 on page 177).

257

6.2.5 Setting Up PublisherListeners

Note: Some operations cannot be used within a listener callback, see Restricted Operations in Listener
Callbacks (Section 4.5.1 on page 185).
As illustrated in Publication Module (Section Figure 6.1 on page 244), the PublisherListener interface
extends the DataWriterListener interface. In other words, the PublisherListener interface contains all the
functions in the DataWriterListener interface. There are no Publisher-specific statuses, and thus there are
no Publisher-specific functions.
Instead, the methods of a PublisherListener will be called back for changes in the Statuses of any of the
DataWriters that the Publisher has created. This is only true if the DataWriter itself does not have a
DataWriterListener installed, see Setting Up DataWriterListeners (Section 6.3.4 on page 269). If a
DataWriterListener has been installed and has been enabled to handle a Status change for the DataWriter,
then Connext DDS will call the method of the DataWriterListener instead.
If you want a Publisher to handle status events for its DataWriters, you can set up a PublisherListener during the Publisher’s creation or use the set_listener() method after the Publisher is created. The last parameter is a bit-mask with which you should set which Status events that the PublisherListener will handle.
For example,
DDS_StatusMask mask = DDS_OFFERED_DEADLINE_MISSED_STATUS |
DDS_OFFERED_INCOMPATIBLE_QOS_STATUS;
publisher = participant->create_publisher(
DDS_PUBLISHER_QOS_DEFAULT, listener, mask);

or
DDS_StatusMask mask = DDS_OFFERED_DEADLINE_MISSED_STATUS |
DDS_OFFERED_INCOMPATIBLE_QOS_STATUS;
publisher->set_listener(listener, mask);

As previously mentioned, the callbacks in the PublisherListener act as ‘default’ callbacks for all the
DataWriters contained within. When Connext DDS wants to notify a DataWriter of a relevant Status
change (for example, PUBLICATION_MATCHED), it first checks to see if the DataWriter has the corresponding DataWriterListener callback enabled (such as the on_publication_matched() operation). If
so, Connext DDS dispatches the event to the DataWriterListener callback. Otherwise, Connext DDS dispatches the event to the corresponding PublisherListener callback.
A particular callback in a DataWriter is not enabled if either:
l

l

The application installed a NULL DataWriterListener (meaning there are no callbacks for the
DataWriter at all).
The application has disabled the callback for a DataWriterListener. This is done by turning off the
associated status bit in the mask parameter passed to the set_listener() or create_datawriter() call
when installing the DataWriterListener on the DataWriter. For more information on DataWriterListeners, see Setting Up DataWriterListeners (Section 6.3.4 on page 269).

258

6.2.6 Finding a Publisher’s Related DDS Entities

Similarly, the callbacks in the DomainParticipantListener act as ‘default’ callbacks for all the Publishers
that belong to it. For more information on DomainParticipantListeners, see Setting Up DomainParticipantListeners (Section 8.3.5 on page 560).
For example, Example Code to Create a Publisher with a Simple Listener (Section Figure 6.8 below)
shows how to create a Publisher with a Listener that simply prints the events it receives.
Figure 6.8 Example Code to Create a Publisher with a Simple Listener
class MyPublisherListener : public DDSPublisherListener {
public:
virtual void on_offered_deadline_missed(
DDSDataWriter* writer,
const DDS_OfferedDeadlineMissedStatus& status);
virtual void on_liveliness_lost(
DDSDataWriter* writer,
const DDS_LivelinessLostStatus& status);
virtual void on_offered_incompatible_qos(
DDSDataWriter* writer,
const DDS_OfferedIncompatibleQosStatus& status);
virtual void on_publication_matched(
DDSDataWriter* writer,
const DDS_PublicationMatchedStatus& status);
virtual void on_reliable_writer_cache_changed(
DDSDataWriter* writer,
const DDS_ReliableWriterCacheChangedStatus& status);
virtual void on_reliable_reader_activity_changed (
DDSDataWriter* writer,
const DDS_ReliableReaderActivityChangedStatus& status);
};
void MyPublisherListener::on_offered_deadline_missed(
DDSDataWriter* writer,
const DDS_OfferedDeadlineMissedStatus& status)
{
printf(“on_offered_deadline_missed\n”);
}
// ...Implement all remaining listeners in a similar manner...
DDSPublisherListener *myPubListener = new MyPublisherListener();
DDSPublisher* publisher =
participant->create_publisher(DDS_PUBLISHER_QOS_DEFAULT,
myPubListener, DDS_STATUS_MASK_ALL);

6.2.6 Finding a Publisher’s Related DDS Entities
These Publisher operations are useful for obtaining a handle to related Entities:
l
l

get_participant(): Gets the DomainParticipant with which a Publisher was created.
lookup_datawriter(): Finds a DataWriter created by the Publisher with a Topic of a particular
name. Note that in the event that multiple DataWriters were created by the same Publisher with the

259

6.2.7 Waiting for Acknowledgments in a Publisher

same Topic, any one of them may be returned by this method. (In the Modern C++ API this method
is a freestanding function, dds::pub::find())
l

DDS_Publisher_as_Entity(): This method is provided for C applications and is necessary when
invoking the parent class Entity methods on Publishers. For example, to call the Entity method get_
status_changes() on a Publisher, my_pub, do the following:
DDS_Entity_get_status_changes(DDS_Publisher_as_Entity(my_pub))

DDS_Publisher_as_Entity() is not provided in the C++, C++/CLI, C# and Java APIs because the objectoriented features of those languages make it unnecessary.

6.2.7 Waiting for Acknowledgments in a Publisher
The Publisher’s wait_for_acknowledgments() operation blocks the calling thread until either all data written by the Publisher’s reliable DataWriters is acknowledged or the duration specified by the max_wait
parameter elapses, whichever happens first.
Note that if a thread is blocked in the call to wait_for_acknowledgments() on a Publisher and a different
thread writes new DDS samples on any of the Publisher’s reliable DataWriters, the new DDS samples
must be acknowledged before unblocking the thread that is waiting on wait_for_acknowledgments().
DDS_ReturnCode_t wait_for_acknowledgments (const DDS_Duration_t &

max_wait)

This operation returns DDS_RETCODE_OK if all the DDS samples were acknowledged, or DDS_
RETCODE_TIMEOUT if the max_wait duration expired first.
There is a similar operation available for individual DataWriters, see Waiting for Acknowledgments in a
DataWriter (Section 6.3.11 on page 288).
The reliability protocol used by Connext DDS is discussed in Reliable Communications (Section Chapter
10 on page 629).

6.2.8 Statuses for Publishers
There are no statuses specific to the Publisher itself. The following statuses can be monitored by the PublisherListener for the Publisher’s DataWriters.
l

OFFERED_DEADLINE_MISSED Status (Section 6.3.6.5 on page 277)

l

LIVELINESS_LOST Status (Section 6.3.6.4 on page 276)

l

OFFERED_INCOMPATIBLE_QOS Status (Section 6.3.6.6 on page 277)

l

PUBLICATION_MATCHED Status (Section 6.3.6.7 on page 278)

l

RELIABLE_WRITER_CACHE_CHANGED Status (DDS Extension) (Section 6.3.6.8 on page
279)

260

6.2.9 Suspending and Resuming Publications

l

RELIABLE_READER_ACTIVITY_CHANGED Status (DDS Extension) (Section 6.3.6.9 on
page 281)

6.2.9 Suspending and Resuming Publications
The operations suspend_publications() and resume_publications() provide a hint to Connext DDS that
multiple data-objects within the Publisher are about to be written. Connext DDS does not currently use this
hint.

6.3 DataWriters
To create a DataWriter, you need a DomainParticipant and a Topic.
You need a DataWriter for each Topic that you want to publish. Once you have a DataWriter, you can
use it to perform the operations listed in Table 6.3 DataWriter Operations. The most important operation is
write(), described in Writing Data (Section 6.3.8 on page 283). For more details on all operations, see the
API Reference HTML documentation.
DataWriters are created by using operations on a DomainParticipant or a Publisher, as described in Creating DataWriters (Section 6.3.1 on page 266). If you use the DomainParticipant’s operations, the
DataWriter will belong to an implicit Publisher that is automatically created by the middleware. If you use
a Publisher’s operations, the DataWriter will belong to that Publisher. So either way, the DataWriter
belongs to a Publisher.
Note: Some operations cannot be used within a listener callback, see Restricted Operations in Listener
Callbacks (Section 4.5.1 on page 185).

261

6.3 DataWriters

Table 6.3 DataWriter Operations
Working with ...

Operation

Description

Reference

assert_liveliness

Manually asserts the liveliness of the DataWriter.

Asserting Liveliness
(Section 6.3.17 on page
311)

enable

Enables the DataWriter.

Enabling DDS Entities
(Section 4.1.2 on page 154)

equals

Compares two DataWriter’s QoS structures for equality.

Comparing QoS Values
(Section 6.3.15.2 on page
305)

get_qos

Gets the QoS.

Setting DataWriter
QosPolicies (Section 6.3.15
on page 300)

lookup_instance

Gets a handle, given an instance. (Useful for keyed data
types only.)

Looking Up an Instance
Handle (Section 6.3.14.3 on
page 299)

set_qos

Modifies the QoS.

Setting DataWriter
QosPolicies (Section 6.3.15
on page 300)

set_qos_with_
profile

Modifies the QoS based on a QoS profile.

Setting DataWriter
QosPolicies (Section 6.3.15
on page 300)

get_listener

Gets the currently installed Listener.

set_listener

Replaces the Listener.

DataWriters

Setting Up
DataWriterListeners
(Section 6.3.4 on page 269)

262

6.3 DataWriters

Table 6.3 DataWriter Operations
Working with ...

Operation

Reference

dispose

States that the instance no longer exists. (Useful for keyed
data types only.)

dispose_w_
timestamp

Same as dispose, but allows the application to override the
automatic source_timestamp. (Useful for keyed data types
only.)

flush

Makes the batch available to be sent on the network.

Flushing Batches of DDS
Data Samples (Section 6.3.9
on page 287)

get_key_value

Maps an instance_handle to the corresponding key.

Getting the Key Value for
an Instance (Section
6.3.14.4 on page 299)

narrow

A type-safe way to cast a pointer. This takes a
DDSDataWriter pointer and ‘narrows’ it to a
‘FooDataWriter’ where ‘Foo’ is the related data type.

Using a Type-Specific
DataWriter (FooDataWriter)
(Section 6.3.7 on page 281)

register_instance

States the intent of the DataWriter to write values of the
data-instance that matches a specified key. Improves the
performance of subsequent writes to the instance. (Useful
for keyed data types only.)

register_instance_
w_
timestamp

Like register_instance, but allows the application to
override the automatic source_timestamp. (Useful for
keyed data types only.)

unregister_
instance

Reverses register_instance. Relinquishes the ownership of
the instance. (Useful for keyed data types only.)

unregister_
instance_w_
timestamp

Like unregister_instance, but allows the application to
override the automatic source_timestamp. (Useful for
keyed data types only.)

write

Writes a new value for a data-instance.

write_w_
timestamp

Same as write, but allows the application to override the
automatic source_timestamp.

FooDataWriter
(See Using a TypeSpecific DataWriter
(FooDataWriter) (Section
6.3.7 on page 281))

Description

Disposing of Data (Section
6.3.14.2 on page 299)

Registering and
Unregistering Instances
(Section 6.3.14.1 on page
297)

Writing Data (Section 6.3.8
on page 283)

263

6.3 DataWriters

Table 6.3 DataWriter Operations
Working with ...

FooDataWriter
(See Using a TypeSpecific DataWriter
(FooDataWriter) (Section
6.3.7 on page 281))

Matched Subscriptions

Status

Operation

Description

Reference

write_w_params

Same as write, but allows the application to specify
parameters such as source timestamp and instance handle.

Writing Data (Section 6.3.8
on page 283)

dispose_w_
params

Same as dispose, but allows the application to specify
parameters such as source timestamp and instance handle..

Disposing of Data (Section
6.3.14.2 on page 299)

register_w_params

Same as register, but allows the application to specify
parameters such as source timestamp, instance handle.

unregister_w_
params

Same as unregister, but allows the application to specify
parameters such as source timestamp, and instance handle.

get_matched_
subscriptions

Gets a list of subscriptions that have a matching Topic and
compatible QoS. These are the subscriptions currently
associated with the DataWriter.

get_matched_
subscription_data

Gets information on a subscription with a matching Topic
and compatible QoS.

get_matched_
subscription_
locators

Gets a list of locators for subscriptions that have a
matching Topic and compatible QoS. These are the
subscriptions currently associated with the DataWriter.

Registering and
Unregistering Instances
(Section 6.3.14.1 on page
297)

Finding Matching
Subscriptions (Section
6.3.16.1 on page 309)

get_matched_
subscription_
participant_data

Gets information about the DomainParticipant of a
matching subscription.

Finding the Matching
Subscription’s
ParticipantBuiltinTopicData
(Section 6.3.16.2 on page
311)

get_status_
changes

Gets a list of statuses that have changed since the last time
the application read the status or the listeners were called.

Getting Status and Status
Changes (Section 4.1.4 on
page 157)

264

6.3 DataWriters

Table 6.3 DataWriter Operations
Working with ...

Operation

Description

get_liveliness_
lost_status

Gets LIVELINESS_LOST status.

get_offered_
deadline_
missed_status

Gets OFFERED_DEADLINE_MISSED status.

Reference

get_offered_
incompatible_qos_ Gets OFFERED_INCOMPATIBLE_QOS status.
status
get_publication_
match_
status

Status
cont'd

Gets PUBLICATION_MATCHED_QOS status.

get_reliable_
writer_
cache_changed_
status

Gets RELIABLE_WRITER_CACHE_CHANGED
status

get_reliable_
reader_
activity_changed_
status

Gets RELIABLE_READER_ACTIVITY_CHANGED
status

get_datawriter_
cache_
status

Gets DATA_WRITER_CACHE_status

get_datawriter_
protocol_status

Gets DATA_WRITER_PROTOCOL status

get_matched_
subscription_
datawriter_
protocol_status

Gets DATA_WRITER_PROTOCOL status for this
DataWriter, per matched subscription identified by the
subscription_handle.

get_matched_
subscription_
datawriter_
protocol_status_
by_locator

Statuses for DataWriters
(Section 6.3.6 on page 271)

Statuses for DataWriters
(Section 6.3.6 on page 271)
Gets DATA_WRITER_PROTOCOL status for this
DataWriter, per matched subscription as identified by a
locator.

265

6.3.1 Creating DataWriters

Table 6.3 DataWriter Operations
Working with ...

Other

Operation

Description

get_publisher

Gets the Publisher to which the DataWriter belongs.

get_topic

Get the Topic associated with the DataWriter.

Blocks the calling thread until either all data written by the
wait_for_
DataWriter is acknowledged by all matched Reliable
acknowledgements DataReaders, or until the a specified timeout duration,
max_wait, elapses.

Reference
Finding Related DDS
Entities (Section 6.3.16.3
on page 311)
Waiting for
Acknowledgments in a
DataWriter (Section 6.3.11
on page 288)

6.3.1 Creating DataWriters
Before you can create a DataWriter, you need a DomainParticipant, a Topic, and optionally, a Publisher.
DataWriters are created by calling create_datawriter() or create_datawriter_with_profile()—these
operations exist for DomainParticipants and Publishers. If you use the DomainParticipant to create a
DataWriter, it will belong to the implicit Publisher described in Creating Publishers Explicitly vs. Implicitly (Section 6.2.1 on page 248). If you use a Publisher’s operations to create a DataWriter, it will belong
to that Publisher.
A QoS profile is way to use QoS settings from an XML file or string. With this approach, you can change
QoS settings without recompiling the application. For details, see Configuring QoS with XML (Section
Chapter 17 on page 791).
Note: In the Modern C++ API DataWriters provide constructors whose first argument is a Publisher. The
only required arguments are the publisher and the topic.
DDSDataWriter* create_datawriter (
DDSTopic *topic,
const DDS_DataWriterQos &qos,
DDSDataWriterListener *listener,
DDS_StatusMask mask)
DDSDataWriter * create_datawriter_with_profile(
DDSTopic * topic,
const char * library_name,
const char * profile_name,
DDSDataWriterListener * listener,
DDS_StatusMask mask)

Where:
topic

The Topic that the DataWriter will publish. This must have been previously created by the same
DomainParticipant.

266

6.3.1 Creating DataWriters

qos

If you want the default QoS settings (described in the API Reference HTML documentation),
use the constant DDS_DATAWRITER_QOS_DEFAULT for this parameter (see Figure 6.9
Creating a DataWriter with Default QosPolicies and a Listener on the facing page ). If you
want to customize any of the QosPolicies, supply a QoS structure (see Setting DataWriter
QosPolicies (Section 6.3.15 on page 300)).

Note: If you use DDS_DATAWRITER_QOS_DEFAULT for the qos parameter, it is not safe
to create the DataWriter while another thread may be simultaneously calling the Publisher’sset_
default_datawriter_qos() operation.
listener

Listeners are callback routines. Connext DDS uses them to notify your application of specific
events (status changes) that may occur with respect to the DataWriter. The listener parameter
may be set to NULL; in this case, the PublisherListener (or if that is NULL, the
DomainParticipantListener) will be used instead. For more information, see Setting Up
DataWriterListeners (Section 6.3.4 on page 269)

mask

This bit-mask indicates which status changes will cause the Listener to be invoked. The bits set
in the mask must have corresponding callbacks implemented in the Listener. If you use NULL
for the Listener, use DDS_STATUS_MASK_NONE for this parameter. If the Listener
implements all callbacks, use DDS_STATUS_MASK_ALL. For information on statuses, see
Listeners (Section 4.4 on page 177).

library_name A QoS Library is a named set of QoS profiles. See URL Groups (Section 17.8 on page 814).
profile_name A QoS profile groups a set of related QoS, usually one per entity. See URL Groups (Section
17.8 on page 814)

For more examples on how to create a DataWriter, see Configuring QoS Settings when the DataWriter is
Created (Section 6.3.15.1 on page 303)
After you create a DataWriter, you can use it to write data. See Writing Data (Section 6.3.8 on page 283).
Note: When a DataWriter is created, only those transports already registered are available to the
DataWriter. The built-in transports are implicitly registered when (a) the DomainParticipant is enabled,
(b) the first DataWriter is created, or (c) you look up a built-in data reader, whichever happens first.

267

6.3.2 Getting All DataWriters

Figure 6.9 Creating a DataWriter with Default QosPolicies and a Listener
// MyWriterListener is user defined, extends DDSDataWriterListener
DDSDataWriterListener* writer_listener = new MyWriterListener();
DDSDataWriter* writer = publisher->create_datawriter(
topic,
DDS_DATAWRITER_QOS_DEFAULT,
writer_listener,
DDS_STATUS_MASK_ALL);
if (writer == NULL) {
// ... error
};
// narrow it for your specific data type
FooDataWriter* foo_writer = FooDataWriter::narrow(writer);

6.3.2 Getting All DataWriters
To retrieve all the DataWriters created by the Publisher, use the Publisher’s get_all_datawriters() operation:
DDS_ReturnCode_t get_all_datawriters(DDS_Publisher* self,
struct DDS_DataWriterSeq* writers);

In the Modern C++ API, use the freestanding function rti::pub::find_datawriters().

6.3.3 Deleting DataWriters
(Note: in the Modern C++ API, Entities are automatically destroyed, see Creating and Deleting DDS Entities (Section 4.1.1 on page 153))
To delete a single DataWriter, use the Publisher’s delete_datawriter() operation:
DDS_ReturnCode_t delete_datawriter (
DDSDataWriter *a_datawriter)

Note: A DataWriter cannot be deleted within its own writer listener callback, see Restricted Operations in
Listener Callbacks (Section 4.5.1 on page 185)
To delete all of a Publisher's DataWriters, use the Publisher's delete_contained_entities() operation (see
Deleting Contained DataWriters (Section 6.2.3.1 on page 251)).
6.3.3.1 Special Instructions for deleting DataWriters if you are using the ‘Timestamp’ APIs
and BY_SOURCE_TIMESTAMP Destination Order:
This section only applies when the DataWriter’s DestinationOrderQosPolicy’s kind is BY_SOURCE_
TIMESTAMP.

268

6.3.4 Setting Up DataWriterListeners

Calls to delete_datawriter() may fail if your application has previously used the “with timestamp” APIs
(write_w_timestamp(), register_instance_w_timestamp(), unregister_instance_w_timestamp(), or dispose_w_timestamp()) with a timestamp that is larger than the time at which delete_datawriter() is called.
To prevent delete_datawriter() from failing in this situation, either:
l

Change the WriterDataLifeCycle QoS Policy so that Connext DDS will not auto-dispose unregistered instances:
writer_qos.writer_data_lifecycle.
autodispose_unregistered_instances =
DDS_BOOLEAN_FALSE;

or
l

Explicitly call unregister_instance_w_timestamp() for all instances modified with the *_w_
timestamp() APIs before calling delete_datawriter().

6.3.4 Setting Up DataWriterListeners
DataWriters may optionally have Listeners. Listeners are essentially callback routines and provide the
means for Connext DDS to notify your application of the occurrence of events (status changes) relevant to
the DataWriter. For more general information on Listeners, see Listeners (Section 4.4 on page 177).
Note: Some operations cannot be used within a listener callback, see Restricted Operations in Listener
Callbacks (Section 4.5.1 on page 185).
If you do not implement a DataWriterListener, the associated PublisherListener is used instead. If that Publisher also does not have a Listener, then the DomainParticipant’s Listener is used if one exists (see Setting Up PublisherListeners (Section 6.2.5 on page 257) and Setting Up DomainParticipantListeners
(Section 8.3.5 on page 560)).
Listeners are typically set up when the DataWriter is created (see Publishers (Section 6.2 on page 243)).
You can also set one up after creation by using the set_listener() operation. Connext DDS will invoke a
DataWriter’s Listener to report the status changes listed in Table 6.4 DataWriterListener Callbacks (if the
Listener is set up to handle the particular status, see Setting Up DataWriterListeners (Section 6.3.4
above)).
Table 6.4 DataWriterListener Callbacks
This DataWriterListener
callback...
on_instance_replaced()

... is triggered by ...
A replacement of an existing instance by a new instance; see Configuring DataWriter Instance
Replacement (Section 6.5.20.2 on page 407)

269

6.3.5 Checking DataWriter Status

Table 6.4 DataWriterListener Callbacks
This DataWriterListener
callback...

... is triggered by ...

on_liveliness_lost

A change to LIVELINESS_LOST Status (Section 6.3.6.4 on page 276)

on_offered_deadline_missed

A change to OFFERED_DEADLINE_MISSED Status (Section 6.3.6.5 on page 277)

on_offered_incompatible_qos

A change to OFFERED_INCOMPATIBLE_QOS Status (Section 6.3.6.6 on page 277)

on_publication_matched

A change to PUBLICATION_MATCHED Status (Section 6.3.6.7 on page 278)

on_reliable_writer_cache_
changed

A change to RELIABLE_WRITER_CACHE_CHANGED Status (DDS Extension) (Section 6.3.6.8
on page 279)

on_reliable_reader_activity_
changed

A change to RELIABLE_READER_ACTIVITY_CHANGED Status (DDS Extension) (Section
6.3.6.9 on page 281)

6.3.5 Checking DataWriter Status
You can access an individual communication status for a DataWriter with the operations shown in Table
6.5 DataWriter Status Operations.
Table 6.5 DataWriter Status Operations
Use this operation...
get_datawriter_cache_status

...to retrieve this status:
DATA_WRITER_CACHE_STATUS (Section 6.3.6.2 on page 272)

get_datawriter_protocol_status
get_matched_subscription_datawriter_protocol_
status

DATA_WRITER_PROTOCOL_STATUS (Section 6.3.6.3 on page 273)

get_matched_subscription_datawriter_protocol_
status_by_locator
get_liveliness_lost_status

LIVELINESS_LOST Status (Section 6.3.6.4 on page 276)

get_offered_deadline_missed_status

OFFERED_DEADLINE_MISSED Status (Section 6.3.6.5 on page 277)

get_offered_incompatible_qos_status

OFFERED_INCOMPATIBLE_QOS Status (Section 6.3.6.6 on page 277)

get_publication_match_status

PUBLICATION_MATCHED Status (Section 6.3.6.7 on page 278)

get_reliable_writer_cache_changed_status

RELIABLE_WRITER_CACHE_CHANGED Status (DDS Extension) (Section
6.3.6.8 on page 279)

270

6.3.6 Statuses for DataWriters

Table 6.5 DataWriter Status Operations
Use this operation...

...to retrieve this status:

get_reliable_reader_activity_changed_status

RELIABLE_READER_ACTIVITY_CHANGED Status (DDS Extension) (Section
6.3.6.9 on page 281)

get_status_changes

A list of what changed in all of the above.

These methods are useful in the event that no Listener callback is set to receive notifications of status
changes. If a Listener is used, the callback will contain the new status information, in which case calling
these methods is unlikely to be necessary.
The get_status_changes() operation provides a list of statuses that have changed since the last time the
status changes were ‘reset.’ A status change is reset each time the application calls the corresponding get_
*_status(), as well as each time Connext DDS returns from calling the Listener callback associated with
that status.
For more on status, see Setting Up DataWriterListeners (Section 6.3.4 on page 269), Statuses for
DataWriters (Section 6.3.6 below), and Listeners (Section 4.4 on page 177).

6.3.6 Statuses for DataWriters
There are several types of statuses available for a DataWriter. You can use the get_*_status() operations
(Setting DataWriter QosPolicies (Section 6.3.15 on page 300)) to access them, or use a DataWriterListener (Setting Up DataWriterListeners (Section 6.3.4 on page 269)) to listen for changes in their values. Each status has an associated data structure and is described in more detail in the following sections.
l

APPLICATION_ACKNOWLEDGMENT_STATUS (Section 6.3.6.1 on the facing page)

l

DATA_WRITER_CACHE_STATUS (Section 6.3.6.2 on the facing page)

l

DATA_WRITER_PROTOCOL_STATUS (Section 6.3.6.3 on page 273)

l

LIVELINESS_LOST Status (Section 6.3.6.4 on page 276)

l

OFFERED_DEADLINE_MISSED Status (Section 6.3.6.5 on page 277)

l

OFFERED_INCOMPATIBLE_QOS Status (Section 6.3.6.6 on page 277)

l

PUBLICATION_MATCHED Status (Section 6.3.6.7 on page 278)

l

l

RELIABLE_WRITER_CACHE_CHANGED Status (DDS Extension) (Section 6.3.6.8 on page
279)
RELIABLE_READER_ACTIVITY_CHANGED Status (DDS Extension) (Section 6.3.6.9 on
page 281)

271

6.3.6.1 APPLICATION_ACKNOWLEDGMENT_STATUS

6.3.6.1 APPLICATION_ACKNOWLEDGMENT_STATUS
This status indicates that a DataWriter has received an application-level acknowledgment for a DDS
sample, and triggers a DataWriter callback:
void DDSDataWriterListener::on_application_acknowledgment(
DDSDataWriter * writer,
const DDS_AcknowledgmentInfo & info)

on_application_acknowledgment() is called when a DDS sample is application-level acknowledged. It
provides identities of the DDS sample and the acknowledging DataReader, as well as user-specified
response data sent from the DataReader by the acknowledgment message—see Table 6.6 DDS_AcknowledgmentInfo.
Table 6.6 DDS_AcknowledgmentInfo
Type

Field Name

Description

DDS_InstanceHandle_t

subscription_handle

Subscription handle of the acknowledging DataReader.

struct DDS_SampleIdentity_t

sample_identity

Identity of the DDS sample being acknowledged.

DDS_Boolean

valid_response_data

Flag indicating validity of the user response data in the acknowledgment.

struct DDS_AckResponseData_t

response_data

User data payload of application-level acknowledgment message.

This status is only applicable when the DataWriter’s Reliability QosPolicy’s acknowledgment_kind is
DDS_APPLICATION_AUTO_ACKNOWLEDGMENT_MODE or DDS_APPLICATION_
EXPLICIT_ACKNOWLEDGMENT_MODE.
6.3.6.2 DATA_WRITER_CACHE_STATUS
This status keeps track of the number of DDS samples in the DataWriter’s queue.
This status does not have an associated Listener. You can access this status by calling the DataWriter’s
get_datawriter_cache_status() operation, which will return the status structure described in Table 6.7
DDS_DataWriterCacheStatus.
Table 6.7 DDS_DataWriterCacheStatus
Type

Field Name

Description

DDS_
Long

sample_count_
peak

Highest number of DDS samples in the DataWriter’s queue over the lifetime of the DataWriter.

DDS_
Long

sample_count

Current number of DDS samples in the DataWriter’s queue (including DDS unregister and dispose
samples)

272

6.3.6.3 DATA_WRITER_PROTOCOL_STATUS

6.3.6.3 DATA_WRITER_PROTOCOL_STATUS
This status includes internal protocol related metrics (such as the number of DDS samples pushed, pulled,
filtered) and the status of wire-protocol traffic.
l

l

l

Pulled DDS samples are DDS samples sent for repairs (that is, DDS samples that had to be resent),
for late joiners, and all DDS samples sent by the local DataWriter when push_on_write (in
DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347)) is
DDS_BOOLEAN_FALSE.
Pushed DDS samples are DDS samples sent on write() when push_on_write is DDS_
BOOLEAN_TRUE.
Filtered DDS samples are DDS samples that are not sent due to DataWriter filtering (time-based filtering and ContentFilteredTopics).

This status does not have an associated Listener. You can access this status by calling the following operations on the DataWriter (all of which return the status structure described in Table 6.8 DDS_DataWriterProtocolStatus):
l

l

l

get_datawriter_protocol_status() returns the sum of the protocol status for all the matched subscriptions for the DataWriter.
get_matched_subscription_datawriter_protocol_status() returns the protocol status of a particular matched subscription, identified by a subscription_handle.
get_matched_subscription_datawriter_protocol_status_by_locator() returns the protocol status
of a particular matched subscription, identified by a locator. (See Locator Format (Section 14.2.1.1
on page 714).)

Note: Status for a remote entity is only kept while the entity is alive. Once a remote entity is no longer
alive, its status is deleted. If you try to get the matched subscription status for a remote entity that is no
longer alive, the ‘get status’ call will return an error.

273

6.3.6.3 DATA_WRITER_PROTOCOL_STATUS

Table 6.8 DDS_DataWriterProtocolStatus
Type

Field Name

Description

pushed_sample_
count

The number of user DDS samples pushed on write from a local DataWriter to a matching remote
DataReader.

pushed_sample_
count_change

The incremental change in the number of user DDS samples pushed on write from a local
DataWriter to a matching remote DataReader since the last time the status was read.

pushed_sample_
bytes

The number of bytes of user DDS samples pushed on write from a local DataWriter to a matching
remote DataReader.

pushed_sample_
bytes_change

The incremental change in the number of bytes of user DDS samples pushed on write from a local
DataWriter to a matching remote DataReader since the last time the status was read.

sent_heartbeat_
count

The number of Heartbeats sent between a local DataWriter and matching remote DataReaders.

sent_heartbeat_
count_change

The incremental change in the number of Heartbeats sent between a local DataWriter and matching
remote DataReaders since the last time the status was read.

sent_heartbeat_
bytes

The number of bytes of Heartbeats sent between a local DataWriter and matching remote
DataReader.

sent_heartbeat_
bytes_change

The incremental change in the number of bytes of Heartbeats sent between a local DataWriter and
matching remote DataReaders since the last time the status was read.

pulled_sample_
count

The number of user DDS samples pulled from local DataWriter by matching DataReaders.

pulled_sample_
count_change

The incremental change in the number of user DDS samples pulled from local DataWriter by
matching DataReaders since the last time the status was read.

pulled_sample_
bytes

The number of bytes of user DDS samples pulled from local DataWriter by matching
DataReaders.

pulled_sample_
bytes_change

The incremental change in the number of bytes of user DDS samples pulled from local DataWriter
by matching DataReaders since the last time the status was read.

received_ack_
count

The number of ACKs from a remote DataReader received by a local DataWriter.

received_ack_
count_change

The incremental change in the number of ACKs from a remote DataReader received by a local
DataWriter since the last time the status was read.

received_ack_
bytes

The number of bytes of ACKs from a remote DataReader received by a local DataWriter.

received_ack_
bytes_change

The incremental change in the number of bytes of ACKs from a remote DataReader received by a
local DataWriter since the last time the status was read.

DDS_LongLong

DDS_LongLong

DDS_LongLong

DDS_LongLong

274

6.3.6.3 DATA_WRITER_PROTOCOL_STATUS

Table 6.8 DDS_DataWriterProtocolStatus
Type

Field Name

Description

received_nack_
count

The number of NACKs from a remote DataReader received by a local DataWriter.

received_nack_
count_change

The incremental change in the number of NACKs from a remote DataReader received by a local
DataWriter since the last time the status was read.

received_nack_
bytes

The number of bytes of NACKs from a remote DataReader received by a local DataWriter.

received_nack_
bytes_change

The incremental change in the number of bytes of NACKs from a remote DataReader received by
a local DataWriter since the last time the status was read.

sent_gap_count

The number of GAPs sent from local DataWriter to matching remote DataReaders.

sent_gap_count_
change

The incremental change in the number of GAPs sent from local DataWriter to matching remote
DataReaders since the last time the status was read.

sent_gap_bytes

The number of bytes of GAPs sent from local DataWriter to matching remote DataReaders.

sent_gap_bytes_
change

The incremental change in the number of bytes of GAPs sent from local DataWriter to matching
remote DataReaders since the last time the status was read.

rejected_sample_
count

The number of times a DDS sample is rejected for unanticipated reasons in the send path.

rejected_sample_
count_change

The incremental change in the number of times a DDS sample is rejected due to exceptions in the
send path since the last time the status was read.

send_window_
size

Current maximum number of outstanding DDS samples allowed in the DataWriter's queue.

DDS_LongLong

DDS_LongLong

DDS_LongLong

DDS_Long

275

6.3.6.4 LIVELINESS_LOST Status

Table 6.8 DDS_DataWriterProtocolStatus
Type

Field Name

Description

first_available_
sample_
Sequence number of the first available DDS sample in the DataWriter's reliability queue.
sequence_number
last_available_
sample_
Sequence number of the last available DDS sample in the DataWriter's reliability queue.
sequence_number
first_
unacknowledged_
Sequence number of the first unacknowledged DDS sample in the DataWriter's reliability queue.
sample_
sequence_number

DDS_
SequenceNumber_
first_available_
t
sample_virtual_
Virtual sequence number of the first available DDS sample in the DataWriter's reliability queue.
sequence_number
last_available_
sample_virtual_
Virtual sequence number of the last available DDS sample in the DataWriter's reliability queue.
sequence_number

first_
unacknowledged_
Virtual sequence number of the first unacknowledged DDS sample in the DataWriter's reliability
sample_
queue.
virtual_sequence_
number

DDS_
SequenceNumber_
t

first_
unacknowledged_
Instance Handle of the matching remote DataReader for which the DataWriter has kept the first
sample_
available DDS sample in the reliability queue.
subscription_
handle
first_unelapsed_
keep_duration_
Sequence number of the first DDS sample kept in the DataWriter's queue whose keep_duration
sample_
(applied when disable_positive_acks is set) has not yet elapsed.
sequence_number

6.3.6.4 LIVELINESS_LOST Status
A change to this status indicates that the DataWriter failed to signal its liveliness within the time specified
by the LIVELINESS QosPolicy (Section 6.5.13 on page 382).
It is different than the RELIABLE_READER_ACTIVITY_CHANGED Status (DDS Extension) (Section 6.3.6.9 on page 281) status that provides information about the liveliness of a DataWriter’s matched
DataReaders; this status reflects the DataWriter’s own liveliness.
The structure for this status appears in Table 6.9 DDS_LivelinessLostStatus.
276

6.3.6.5 OFFERED_DEADLINE_MISSED Status

Table 6.9 DDS_LivelinessLostStatus
Type

Field Name

DDS_Long total_count

Description
Cumulative number of times the DataWriter failed to explicitly signal its liveliness within the liveliness period.

DDS_Long total_count_change The change in total_count since the last time the Listener was called or the status was read.

The DataWriterListener’s on_liveliness_lost() callback is invoked when this status changes. You can also
retrieve the value by calling the DataWriter’s get_liveliness_lost_status() operation.
6.3.6.5 OFFERED_DEADLINE_MISSED Status
A change to this status indicates that the DataWriter failed to write data within the time period set in its
DEADLINE QosPolicy (Section 6.5.5 on page 363).
The structure for this status appears in Table 6.10 DDS_OfferedDeadlineMissedStatus.
Table 6.10 DDS_OfferedDeadlineMissedStatus
Type

Field Name

Description

DDS_Long

total_count

Cumulative number of times the DataWriter failed to write within its offered deadline.

DDS_Long

total_count_change

The change in total_count since the last time the Listener was called or the status was read.

DDS_InstanceHandle_ last_instance_
t
handle

Handle to the last data-instance in the DataWriter for which an offered deadline was
missed.

The DataWriterListener’s on_offered_deadline_missed() operation is invoked when this status changes.
You can also retrieve the value by calling the DataWriter’s get_deadline_missed_status() operation.
6.3.6.6 OFFERED_INCOMPATIBLE_QOS Status
A change to this status indicates that the DataWriter discovered a DataReader for the same Topic, but that
DataReader had requested QoS settings incompatible with this DataWriter’s offered QoS.
The structure for this status appears in Table 6.11 DDS_OfferedIncompatibleQoSStatus.

277

6.3.6.7 PUBLICATION_MATCHED Status

Table 6.11 DDS_OfferedIncompatibleQoSStatus
Type

Field
Name

Description

DDS_Long

total_
count

Cumulative number of times the DataWriter discovered a DataReader for the same Topic with a requested
QoS that is incompatible with that offered by the DataWriter.

DDS_Long

total_
count_
change

The change in total_count since the last time the Listener was called or the status was read.

DDS_QosPolicyId_ last_
The ID of the QosPolicy that was found to be incompatible the last time an incompatibility was detected.
t
policy_id (Note: if there are multiple incompatible policies, only one of them is reported here.)
DDS_
policies
QosPolicyCountSeq

A list containing—for each policy—the total number of times that the DataWriter discovered a
DataReader for the same Topic with a requested QoS that is incompatible with that offered by the
DataWriter.

The DataWriterListener’s on_offered_incompatible_qos() callback is invoked when this status changes.
You can also retrieve the value by calling the DataWriter’s get_offered_incompatible_qos_status() operation.
6.3.6.7 PUBLICATION_MATCHED Status
A change to this status indicates that the DataWriter discovered a matching DataReader.
A ‘match’ occurs only if the DataReader and DataWriter have the same Topic, same data type (implied by
having the same Topic), and compatible QosPolicies. In addition, if user code has directed Connext DDS
to ignore certain DataReaders, then those DataReaders will never be matched. See Ignoring Publications
and Subscriptions (Section 16.4.2 on page 786) for more on setting up a DomainParticipant to ignore specific DataReaders.
The structure for this status appears in Table 6.12 DDS_PublicationMatchedStatus.

278

6.3.6.8 RELIABLE_WRITER_CACHE_CHANGED Status (DDS Extension)

Table 6.12 DDS_PublicationMatchedStatus
Type

DDS_Long

Field Name

Description

total_count

Cumulative number of times the DataWriter discovered a "match" with a DataReader.

total_count_change

The change in total_count since the last time the Listener was called or the status was
read.

current_count

The number of DataReaders currently matched to the DataWriter.

current_count_peak

The highest value that current_count has reached until now.

current_count_change

The change in current_count since the last time the listener was called or the status was
read.

DDS_InstanceHandle_ last_subscription_
t
handle

Handle to the last DataReader that matched the DataWriter causing the status to
change.

The DataWriterListener’s on_publication_matched() callback is invoked when this status changes. You
can also retrieve the value by calling the DataWriter’s get_publication_match_status() operation.
6.3.6.8 RELIABLE_WRITER_CACHE_CHANGED Status (DDS Extension)
A change to this status indicates that the number of unacknowledged DDS samples1 in a reliable
DataWriter's cache has reached one of these trigger points:
l
l

l

The cache is empty (contains no unacknowledged DDS samples)
The cache is full (the number of unacknowledged DDS samples has reached the value specified in
DDS_ResourceLimitsQosPolicy::max_samples)
The number of unacknowledged DDS samples has reached a high or low watermark. See the high_
watermark and low_watermark fields in Table 6.37 DDS_RtpsReliableWriterProtocol_t of the
DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).

For more about the reliable protocol used by Connext DDS and specifically, what it means for a DDS
sample to be ‘unacknowledged,’ see Reliable Communications (Section Chapter 10 on page 629).
The structure for this status appears in Table 6.13 DDS_ReliableWriterCacheChangedStatus.The supporting structure, DDS_ReliableWriterCacheEventCount, is described in Table 6.14 DDS_ReliableWriterCacheEventCount.

1If batching is enabled, this still refers to a number of DDS samples, not batches.

279

6.3.6.8 RELIABLE_WRITER_CACHE_CHANGED Status (DDS Extension)

Table 6.13 DDS_ReliableWriterCacheChangedStatus
Type

DDS_
ReliableWriterCacheEventCount

Field Name

Description

empty_reliable_
writer_
cache

How many times the reliable DataWriter's cache of unacknowledged DDS samples
has become empty.

full_reliable_
writer_
cache

How many times the reliable DataWriter's cache of unacknowledged DDS samples
has become full.

low_watermark_
reliable_writer_
cache

How many times the reliable DataWriter's cache of unacknowledged DDS samples
has fallen to the low watermark.

high_watermark_
How many times the reliable DataWriter's cache of unacknowledged DDS samples
reliable_writer_
has risen to the high watermark.
cache
unacknowledged_
The current number of unacknowledged DDS samples in the DataWriter's cache.
sample_count
DDS_Long

unacknowledged_
sample_count_
The highest value that unacknowledged_sample_count has reached until now.
peak

Table 6.14 DDS_ReliableWriterCacheEventCount
Type

Field Name

Description

DDS_Long

total_count

The total number of times the event has occurred.

DDS_Long

total_count_change

The number of times the event has occurred since the Listener was last invoked or the status read.

The DataWriterListener’s on_reliable_writer_cache_changed() callback is invoked when this status
changes. You can also retrieve the value by calling the DataWriter’s get_reliable_writer_cache_
changed_status() operation.
If a reliable DataWriter's send window is finite, with both RtpsReliableWriterProtocol_t.min_send_window_size and RtpsReliableWriterProtocol_t.max_send_window_size set to positive values, then full_
reliable_writer_cache_status counts the number of times the unacknowledged DDS sample count
reaches the send window size.

280

6.3.6.9 RELIABLE_READER_ACTIVITY_CHANGED Status (DDS Extension)

6.3.6.9 RELIABLE_READER_ACTIVITY_CHANGED Status (DDS Extension)
This status indicates that one or more reliable DataReaders has become active or inactive.
This status is the reciprocal status to the LIVELINESS_CHANGED Status (Section 7.3.7.4 on page 475)
on the DataReader. It is different than LIVELINESS_LOST Status (Section 6.3.6.4 on page 276) status
on the DataWriter, in that the latter informs the DataWriter about its own liveliness; this status informs the
DataWriter about the liveliness of its matched DataReaders.
A reliable DataReader is considered active by a reliable DataWriter with which it is matched if that
DataReader acknowledges the DDS samples that it has been sent in a timely fashion. For the definition of
"timely" in this context, see DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3
on page 347).
This status is only used for DataWriters whose RELIABILITY QosPolicy (Section 6.5.19 on page 400)
is set to RELIABLE. For best-effort DataWriters, all counts in this status will remain at zero.
The structure for this status appears in Table 6.15 DDS_ReliableReaderActivityChangedStatus.
Table 6.15 DDS_ReliableReaderActivityChangedStatus
Type

DDS_Long

Field Name

Description

active_count

The current number of reliable readers currently matched with this reliable DataWriter.

inactive_count

The number of reliable readers that have been dropped by this reliable DataWriter because they failed
to send acknowledgments in a timely fashion.

active_count_
change

The change in the number of active reliable DataReaders since the Listener was last invoked or the
status read.

inactive_count_ The change in the number of inactive reliable DataReaders since the Listener was last invoked or the
change
status read.
DDS_
last_instance_
InstanceHandle_
handle
t

The instance handle of the last reliable DataReader to be determined to be inactive.

The DataWriterListener’s on_reliable_reader_activity_changed() callback is invoked when this status
changes. You can also retrieve the value by calling the DataWriter’s get_reliable_reader_activity_
changed_status() operation.

6.3.7 Using a Type-Specific DataWriter (FooDataWriter)

(Note: this section does not apply to the Modern C++ API where a DataWriter's data type is part of its template definition: DataWriter)

281

6.3.7 Using a Type-Specific DataWriter (FooDataWriter)

Recall that a Topic is bound to a data type that specifies the format of the data associated with the Topic.
Data types are either defined dynamically or in code generated from definitions in IDL or XML; see Data
Types and DDS Data Samples (Section Chapter 3 on page 23). For each of your application's generated
data types, such as 'Foo', there will be a FooDataWriter class (or a set of functions in C). This class allows
the application to use a type-safe interface to interact with DDS samples of type 'Foo'. You will use the
FooDataWriter's write() operation used to send data. For dynamically defined data-types, you will use the
DynamicDataWriter class.
In fact, you will use the FooDataWriter any time you need to perform type-specific operations, such as
registering or writing instances. Table 6.3 DataWriter Operations indicates which operations must be
called using FooDataWriter. For operations that are not type-specific, you can call the operation using
either a FooDataWriter or a DDSDataWriter object1.
You may notice that the Publisher’s create_datawriter() operation returns a pointer to an object of type
DDSDataWriter; this is because the create_datawriter() method is used to create DataWriters of any
data type. However, when executed, the function actually returns a specialization (an object of a derived
class) of the DataWriter that is specific for the data type of the associated Topic. For a Topic of type ‘Foo’,
the object actually returned by create_datawriter() is a FooDataWriter.
To safely cast a generic DDSDataWriter pointer to a FooDataWriter pointer, you should use the static
narrow() method of the FooDataWriter class. The narrow() method will return NULL if the generic
DDSDataWriter pointer is not pointing at an object that is really a FooDataWriter.
For instance, if you create a Topic bound to the type ‘Alarm’, all DataWriters created for that Topic will
be of type ‘AlarmDataWriter.’ To access the type-specific methods of AlarmDataWriter, you must cast
the generic DDSDataWriter pointer returned by create_datawriter(). For example:
DDSDataWriter* writer = publisher->create_datawriter(topic,writer_qos, NULL, NULL);
AlarmDataWriter *alarm_writer = AlarmDataWriter::narrow(writer);
if (alarm_writer == NULL) {
// ... error
};

In the C API, there is also a way to do the opposite of narrow(). FooDataWriter_as_datawriter() casts
a FooDataWriter as a DDSDataWriter, and FooDataReader_as_datareader() casts a FooDataReader as
a DDSDataReader.

1In the C API, the non type-specific operations must be called using a DDS_DataWriter pointer.

282

6.3.8 Writing Data

6.3.8 Writing Data
The write() operation informs Connext DDS that there is a new value for a data-instance to be published
for the corresponding Topic. By default, calling write() will send the data immediately over the network
(assuming that there are matched DataReaders). However, you can configure and execute operations on
the DataWriter’s Publisher to buffer the data so that it is sent in a batch with data from other DataWriters
or even to prevent the data from being sent. Those sending “modes” are configured using the
PRESENTATION QosPolicy (Section 6.4.6 on page 330) as well as the Publisher’s suspend/resume_
publications() operations. The actual transport-level communications may be done by a separate, lowerpriority thread when the Publisher is configured to send the data for its DataWriters. For more information
on threads, see Connext DDS Threading Model (Section Chapter 19 on page 837).
When you call write(), Connext DDS automatically attaches a stamp of the current time that is sent with
the DDS data sample to the DataReader(s). The timestamp appears in the source_timestamp field of the
DDS_SampleInfo structure that is provided along with your data using DataReaders (see The
SampleInfo Structure (Section 7.4.6 on page 504)).
DDS_ReturnCode_t write (const Foo &instance_data,
const DDS_InstanceHandle_t &handle)

You can use an alternate DataWriter operation called write_w_timestamp(). This performs the same
action as write(), but allows the application to explicitly set the source_timestamp. This is useful when
you want the user application to set the value of the timestamp instead of the default clock used by Connext DDS.
DDS_ReturnCode_t write_w_timestamp (
const Foo &instance_data,
const DDS_InstanceHandle_t &handle,
const DDS_Time_t &source_timestamp)

Note that, in general, the application should not mix these two ways of specifying timestamps. That is, for
each DataWriter, the application should either always use the automatic timestamping mechanism (by calling the normal operations) or always specify a timestamp (by calling the “w_timestamp” variants of the
operations). Mixing the two methods may result in not receiving sent data.
You can also use an alternate DataWriter operation, write_w_params(), which performs the same action
as write(), but allows the application to explicitly set the fields contained in the DDS_WriteParams structure, see Table 6.16 DDS_WriteParams_t.

283

6.3.8 Writing Data

Table 6.16 DDS_WriteParams_t
Type

Field
Name

Description
Allows retrieving the actual value of those fields that were automatic.

DDS_Boolean

replace_
auto

When this field is set to true, the fields that were configured with an automatic value (for example, DDS_
AUTO_SAMPLE_IDENTITY in identity) receive their actual value after write_w_params is called.
Identity of the DDS sample being written. The identity consists of a pair (Virtual Writer GUID, Virtual
Sequence Number).
When the value DDS_AUTO_SAMPLE_IDENTITY is used, the write_w_params() operation will
determine the DDS sample identity as follows:
l

DDS_
SampleIdentity_ identity
t

l

The Virtual Writer GUID (writer_guid) is the virtual GUID associated with the
DataWriter writing the DDS sample. This virtual GUID is configured using the member virtual_guid in DATA_WRITER_PROTOCOL_STATUS (Section 6.3.6.3 on
page 273).
The Virtual Sequence Number (sequence_number) is increased by one with respect to
the previous value.

The virtual sequence numbers for a given virtual GUID must be strictly monotonically increasing. If you try
to write a DDS sample with a sequence number smaller or equal to the last sequence number, the write
operation will fail.
A DataReader can inspect the identity of a received DDS sample by accessing the fields original_
publication_virtual_guid and original_publication_virtual_sequence_number in The SampleInfo
Structure (Section 7.4.6 on page 504).
The identity of another DDS sample related to this one.
The value of this field identifies another DDS sample that is logically related to the one that is written.

DDS_
related_
SampleIdentity_ sample_
t
identity

For example, the DataWriter created by a Replier (sets Introduction to the Request-Reply Communication
Pattern (Section Chapter 22 on page 874)) uses this field to associate the identity of the DDS request sample
to reponse sample.
To specify that there is no related DDS sample identity use the value DDS_UNKNOWN_SAMPLE_
IDENTITY,
A DataReader can inspect the related DDS sample identity of a received DDS sample by accessing the fields
related_original_publication_virtual_guid and related_original_publication_virtual_sequence_
number in The SampleInfo Structure (Section 7.4.6 on page 504).
Source timestamp that will be associated to the DDS sample that is written.

DDS_Time

source_
If source_timestamp is set to DDS_TIMER_INVALID, the middleware will assign the value.
timestamp
A DataReader can inspect the source_timestamp value of a received DDS sample by accessing the field
source_timestamp The SampleInfo Structure (Section 7.4.6 on page 504).

284

6.3.8 Writing Data

Table 6.16 DDS_WriteParams_t
Type

Field
Name

DDS_
InstanceHandle_ handle
t

Description
The instance handle.
This value can be either the handle returned by a previous call to register_instance() or the special value
DDS_HANDLE_NIL.
Positive integer designating the relative priority of the DDS sample, used to determine the transmission order
of pending transmissions.

DDS_Long

priority

To use publication priorities, the DataWriter’s PUBLISH_MODE QosPolicy (DDS Extension)
(Section 6.5.18 on page 397) must be set for asynchronous publishing and the DataWriter must use a
FlowController with a highest-priority first scheduling_policy.
For Multi-channel DataWriters, the publication priority of a DDS sample may be used as a filter criteria for
determining channel membership.
For more information, see Prioritized DDS Samples (Section 6.6.4 on page 428).
Flags for the DDS sample, represented as a 32-bit integer, of which only the 16 least-significant bits are
used.
RTI reserves least-significant bits [0-7] for middleware-specific usage. The application can use leastsignificant bits [8-15].
The first bit, REDELIVERED_SAMPLE, is reserved to mark a DDS sample as redelivered when using RTI
Queuing Service.

DDS_Long

flag

The second bit, INTERMEDIATE_REPLY_SEQUENCE_SAMPLE, is used to indicate that a response
DDS sample is not the last response DDS sample for a given request. This bit is usually set by Connext
DDS Repliers sending multiple responses for a request.
The third bit, REPLICATE_SAMPLE, indicates if a sample must be broad- cast by one Queuing Service
replica to other replicas.
The fourth bit, LAST_SHARED_READER_QUEUE_SAMPLE, indicates that a sample is the last sample
in a SharedReaderQueue for a QueueConsumer DataReader.
An application can inspect the flags associated with a received DDS sample by checking the field flag field in
The SampleInfo Structure (Section 7.4.6 on page 504).
Default 0 (no flags are set)

struct DDS_
GUID_t

source_
guid

Identifies the application logical data source associated with the sample being written.

struct DDS_
GUID_t

related_
source_
guid

Identifies the application logical data source that is related to the sample being written.

struct DDS_
GUID_t

related_
reader_
guid

Identifies a DataReader that is logically related to the sample that is being written.

285

6.3.8.1 Blocking During a write()

Note: Prioritized DDS samples are not supported when using the Java, Ada, or .NET APIs. Therefore the
priority field in DDS_WriteParams_t does not exist when using these APIs.
When using the C API, a newly created variable of type DDS_WriteParams_t should be initialized by setting it to DDS_WRITEPARAMS_DEFAULT.
The write() operation also asserts liveliness on the DataWriter, the associated Publisher, and the associated DomainParticipant. It has the same effect with regards to liveliness as an explicit call to assert_liveliness(), see Asserting Liveliness (Section 6.3.17 on page 311) and the LIVELINESS QosPolicy (Section
6.5.13 on page 382). Maintaining liveliness is important for DataReaders to know that the DataWriter
still exists and for the proper behavior of the OWNERSHIP QosPolicy (Section 6.5.15 on page 389).
See also: Clock Selection (Section 8.6 on page 619).
6.3.8.1 Blocking During a write()
The write() operation may block if the RELIABILITY QosPolicy (Section 6.5.19 on page 400) kind is
set to Reliable and the modification would cause data to be lost or cause one of the limits specified in the
RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) to be exceeded. Specifically, write() may
block in the following situations (note that the list may not be exhaustive), even if its HISTORY
QosPolicy (Section 6.5.10 on page 376) is KEEP_LAST:
l

l

l

If max_samples1 < max_instances, the DataWriter may block regardless of the depth field in the
HISTORY QosPolicy (Section 6.5.10 on page 376).
If max_samples < (max_instances * depth), in the situation where the max_samples resource
limit is exhausted, Connext DDS may discard DDS samples of some other instance, as long as at
least one DDS sample remains for such an instance. If it is still not possible to make space available
to store the modification, the writer is allowed to block.
If min_send_window_size < max_samples, it is possible for the send_window_size limit to be
reached before Connext DDS is allowed to discard DDS samples, in which case the DataWriter
will block.

This operation may also block when using BEST_EFFORT Reliability (RELIABILITY QosPolicy (Section 6.5.19 on page 400)) and ASYNCHRONOUS Publish Mode (PUBLISH_MODE QosPolicy (DDS
Extension) (Section 6.5.18 on page 397)) QoS settings. In this case, the DataWriter will queue DDS
samples until they are sent by the asynchronous publishing thread. The number of DDS samples that can
be stored is determined by the HISTORY QosPolicy (Section 6.5.10 on page 376). If the asynchronous
thread does not send DDS samples fast enough (such as when using a slow FlowController (FlowControllers (DDS Extension) (Section 6.6 on page 422))), the queue may fill up. In that case, subsequent write
calls will block.

1max_samples in is DDS_ResourceLimitsQosPolicy

286

6.3.9 Flushing Batches of DDS Data Samples

If this operation does block for any of the above reasons, the RELIABILITY max_blocking_time configures the maximum time the write operation may block (waiting for space to become available). If max_
blocking_time elapses before the DataWriter can store the modification without exceeding the limits, the
operation will fail and return RETCODE_TIMEOUT.

6.3.9 Flushing Batches of DDS Data Samples
The flush() operation makes a batch of DDS data samples available to be sent on the network.
DDS_ReturnCode_t flush ()

If the DataWriter’s PUBLISH_MODE QosPolicy (DDS Extension) (Section 6.5.18 on page 397) kind is
not ASYNCHRONOUS, the batch will be sent on the network immediately in the context of the calling
thread.
If the DataWriter’s PublishModeQosPolicy kind is ASYNCHRONOUS, the batch will be sent in the context of the asynchronous publishing thread.
The flush() operation may block based on the conditions described in Blocking During a write() (Section
6.3.8.1 on the previous page).
If this operation does block, the max_blocking_time in the RELIABILITY QosPolicy (Section 6.5.19 on
page 400) configures the maximum time the write operation may block (waiting for space to become available). If max_blocking_time elapses before the DataWriter is able to store the modification without
exceeding the limits, the operation will fail and return TIMEOUT.
For more information on batching, see the BATCH QosPolicy (DDS Extension) (Section 6.5.2 on page
341).

6.3.10 Writing Coherent Sets of DDS Data Samples
A publishing application can request that a set of DDS data-sample changes be propagated in such a way
that they are interpreted at the receivers' side as a cohesive set of modifications. In this case, the receiver
will only be able to access the data after all the modifications in the set are available at the subscribing end.
This is useful in cases where the values are inter-related. For example, suppose you have two datainstances representing the ‘altitude’ and ‘velocity vector’ of the same aircraft. If both are changed, it may
be important to ensure that reader see both together (otherwise, it may erroneously interpret that the aircraft
is on a collision course).
To use this mechanism in C, Traditional C++, Java and .NET:
1. Call the Publisher’s begin_coherent_changes() operation to indicate the start a coherent set.
2. For each DDS sample in the coherent set: call the FooDataWriter’s write() operation.
3. Call the Publisher’s end_coherent_changes() operation to terminate the set.

287

6.3.11 Waiting for Acknowledgments in a DataWriter

In the Modern C++ API:
1. Instantiate a dds::pub::CoherentSet passing a publisher to the constructor
2. For each DDS sample in the coherent set call dds::pub::DataWriter::write().
3. Let the dds::pub::CoherentSet destructor terminate the set or explicitly call dds::pub::CoherentSet::end()
Calls to begin_coherent_changes() and end_coherent_changes() can be nested.
See also: the coherent_access field in the PRESENTATION QosPolicy (Section 6.4.6 on page 330).

6.3.11 Waiting for Acknowledgments in a DataWriter
The DataWriter’s wait_for_acknowledgments() operation blocks the calling thread until either all data
written by the reliable DataWriter is acknowledged by (a) all reliable DataReaders that are matched and
alive and (b) by all required subscriptions (see Required Subscriptions (Section 6.3.13 on page 294)), or
until the duration specified by the max_wait parameter elapses, whichever happens first.
Note that if a thread is blocked in the call to wait_for_acknowledgments() on a DataWriter and a different thread writes new DDS samples on the same DataWriter, the new DDS samples must be acknowledged before unblocking the thread waiting on wait_for_acknowledgments().
DDS_ReturnCode_t wait_for_acknowledgments (
const DDS_Duration_t & max_wait)

This operation returns DDS_RETCODE_OK if all the DDS samples were acknowledged, or DDS_
RETCODE_TIMEOUT if the max_wait duration expired first.
If the DataWriter does not have its RELIABILITY QosPolicy (Section 6.5.19 on page 400) kind set to
RELIABLE, the operation will immediately return DDS_RETCODE_OK.
There is a similar operation available at the Publisher level, see Waiting for Acknowledgments in a Publisher (Section 6.2.7 on page 260).
The reliability protocol used by Connext DDS is discussed in Reliable Communications (Section Chapter
10 on page 629). The application acknowledgment mechanism is discussed in Application Acknowledgment (Section 6.3.12 below) and Guaranteed Delivery of Data (Section Chapter 13 on page 695).

6.3.12 Application Acknowledgment
The RELIABILITY QosPolicy (Section 6.5.19 on page 400) determines whether or not data published
by a DataWriter will be reliably delivered by Connext DDS to matching DataReaders. The reliability protocol used by Connext DDS is discussed in Reliable Communications (Section Chapter 10 on page 629).
With protocol-level reliability alone, the producing application knows that the information is received by
the protocol layer on the consuming side. However, the producing application cannot be certain that the

288

6.3.12.1 Application Acknowledgment Kinds

consuming application read that information or was able to successfully understand and process it. The
information could arrive in the consumer’s protocol stack and be placed in the DataReader cache but the
consuming application could either crash before it reads it from the cache, not read its cache, or read the
cache using queries or conditions that prevent that particular DDS data sample from being accessed. Furthermore, the consuming application could access the DDS sample, but not be able to interpret its meaning
or process it in the intended way.
The mechanism to let a DataWriter know to keep the DDS sample around, not just until it has been
acknowledged by the reliability protocol, but until the application has been able to process the DDS
sample is aptly called Application Acknowledgment. A reliable DataWriter will keep the DDS samples
until the application acknowledges the DDS samples. When the subscriber application is restarted, the middleware will know that the application did not acknowledge successfully processing the DDS samples and
will resend them.
6.3.12.1 Application Acknowledgment Kinds
Connext DDS supports three kinds of application acknowledgment, which is configured in the
RELIABILITY QosPolicy (Section 6.5.19 on page 400)):
1. DDS_PROTOCOL_ACKNOWLEDGMENT_MODE (Default): In essence, this mode is identical
to using no application-level acknowledgment. DDS samples are acknowledged according to the
Real-Time Publish-Subscribe (RTPS) reliability protocol. RTPS AckNack messages will acknowledge that the middleware received the DDS sample.
2. DDS_APPLICATION_AUTO_ACKNOWLEDGMENT_MODE: DDS samples are automatically acknowledged by the middleware after the subscribing application accesses them, either
through calling take() or read() on the DDS sample. The DDS samples are acknowledged after
return_loan() is called.
3. DDS_APPLICATION_EXPLICIT_ACKNOWLEDGMENT_MODE: DDS samples are acknowledged after the subscribing application explicitly calls acknowledge on the DDS sample. This can
be done by either calling the DataReader’s acknowledge_sample() or acknowledge_all() operations. When using acknowledge_sample(), the application will provide the DDS_SampleInfo to
identify the DDS sample being acknowledge. When using acknowledge_all, all the DDS samples
that have been read or taken by the reader will be acknowledged.
Note: Even in DDS_APPLICATION_EXPLICIT_ACKNOWLEDGMENT_MODE, some DDS
samples may be automatically acknowledged. This is the case when DDS samples are filtered out
by the reader using time-based filter, or using content filters. Additionally, when the reader is explicitly configured to use KEEP_LAST history kind, DDS samples may be replaced in the reader
queue due to resource constraints. In that case, the DDS sample will be automatically acknowledged
by the middleware if it has not been read by the application before it was replaced. To truly guarantee successful processing of DDS samples, it is recommended to use KEEP_ALL history kind.

289

6.3.12.2 Explicitly Acknowledging a Single DDS Sample (C++)

6.3.12.2 Explicitly Acknowledging a Single DDS Sample (C++)
void MyReaderListener::on_data_available(DDSDataReader *reader)
{
Foo sample;
DDS_SampleInfo info;
FooDataReader* fooReader = FooDataReader::narrow(reader);
DDS_ReturnCode_t retcode = fooReader->take_next_sample(
sample, info);
if (retcode == DDS_RETCODE_OK) {
if (info.valid_data) {
// Process sample
...
retcode = reader->acknowledge_sample(info);
if (retcode != DDS_RETCODE_OK) {
// Error
}
}
} else {
// Not OK or NO DATA
}
}

6.3.12.3 Explicitly Acknowledging All DDS samples (C++)
void MyReaderListener::on_data_available(DDSDataReader *reader)
{
...
// Loop while samples available
for(;;) {
retcode = string_reader->take_next_sample(
sample, info);
if (retcode == DDS_RETCODE_NO_DATA) {
// No more samples
break;
}
// Process sample
...
}
retcode = reader->acknowledge_all();
if (retcode != DDS_RETCODE_OK) {
// Error
}
}

6.3.12.4 Notification of Delivery with Application Acknowledgment
A DataWriter can get notification of delivery with Application Acknowledgment using two different mechanisms:

290

6.3.12.5 Application-Level Acknowledgment Protocol

l

DataWriter's wait_for_acknowledgments() operation
A DataWriter can use the wait_for_acknowledgments() operation to be notified when all the DDS
samples in the DataWriter’s queue have been acknowledged. See Waiting for Acknowledgments in
a DataWriter (Section 6.3.11 on page 288).
retCode = fooWriter->write(sample, DDS_HANDLE_NIL);
if (retCode != DDS_RETCODE_OK) {
// Error
}
retcode = writer->wait_for_acknowledgments(timeout);
if (retCode != DDS_RETCODE_OK) {
if (retCode == DDS_RETCODE_TIMEOUT) {
// Timeout: Sample not acknowledged yet
} else {
// Error
}
}

l

Using wait_for_acknowledgments() does not provide a way to get delivery notifications on a per
DataReader and DDS sample basis. If your application requires acknowledgment of message
receipt, use the the second mechanism described below.
DataWriter's listener callback on_application_acknowledgment()
An application can install a DataWriter listener callback on_application_acknowledgment() to
receive a notification when a DDS sample is acknowledged by a DataReader. As part of this notification, you can access:
l The subscription handle of the acknowledging DataReader.
l

The Identity of the DDS sample being acknowledged.

l

The response data associated with the DDS sample being acknowledged.

For more information, see APPLICATION_ACKNOWLEDGMENT_STATUS (Section 6.3.6.1
on page 272).
6.3.12.5 Application-Level Acknowledgment Protocol
When the subscribing application confirms it has successfully processed a DDS sample, an AppAck
RTPS message is sent to the publishing application. This message will be resent until the publishing application confirms receipt of the AppAck message by sending an AppAckConf RTPS message. See Figures
Figure 6.10 AppAck RTPS Messages Sent when Application Acknowledges a DDS Sample on the
facing page through Figure 6.12 AppAck RTPS Messages Sent as a Sequence of Intervals, Combined to
Optimize for Bandwidth on page 293.

291

6.3.12.5 Application-Level Acknowledgment Protocol

Figure 6.10 AppAck RTPS Messages Sent when Application Acknowledges a DDS Sample

Figure 6.11 AppAck RTPS Messages Resent Until Acknowledged Through AppAckConf
RTPS Message

292

6.3.12.6 Periodic and Non-Periodic AppAck Messages

Figure 6.12 AppAck RTPS Messages Sent as a Sequence of Intervals, Combined to Optimize
for Bandwidth

6.3.12.6 Periodic and Non-Periodic AppAck Messages
You can configure whether AppAck RTPS messages are sent immediately or periodically through the
DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511). The
samples_per_app_ack (Section on page 515) (in Table 7.20 DDS_RtpsReliableReaderProtocol_t) determines the minimum number of DDS samples acknowledged by one application-level Acknowledgment message. The middleware will not send an AppAck message until it has at least this many DDS samples
pending acknowledgment. By default, samples_per_app_ack is 1 and the AppAck RTPS message is
sent immediately. Independently, the app_ack_period (Section on page 514) (in Table 7.20 DDS_RtpsReliableReaderProtocol_t) determines the rate at which a DataReader will send AppAck messages.
6.3.12.7 Application Acknowledgment and Persistence Service
Application Acknowledgment is fully supported by RTI Persistence Service. The combination of Application Acknowledgment and Persistence Service is actually a common configuration. In addition to keeping
DDS samples available until fully acknowledged, Persistence Service, when used in peer-to-peer mode,
can take advantage of AppAck messages to avoid sending duplicate messages to the subscribing application. Because AppAck messages are sent to all matching writers, when the subscriber acknowledges the
original publisher, Persistence Service will also be notified of this event and will not send out duplicate
messages. This is illustrated in Figure 6.13 Application Acknowledgment and Persistence Service on the
facing page.

293

6.3.12.8 Application Acknowledgment and Routing Service

Figure 6.13 Application Acknowledgment and Persistence Service

6.3.12.8 Application Acknowledgment and Routing Service
Application Acknowledgment is supported by RTI Routing Service: That is, Routing Service will acknowledge the DDS sample it has processed. Routing Service is an active participant in the Connext DDS system and not transparent to the publisher or subscriber. As such, Routing Service will acknowledge to the
publisher, and the subscriber will acknowledge to Routing Service. However, the publisher will not get a
notification from the subscriber directly.

6.3.13 Required Subscriptions
The DURABILITY QosPolicy (Section 6.5.7 on page 368) specifies whether acknowledged DDS
samples need to be kept in the DataWriter’s queue and made available to late-joining applications. When a
late joining application is discovered, available DDS samples will be sent to the late joiner. With the Durability QoS alone, there is no way to specify or characterize the intended consumers of the information and
you do not have control over which DDS samples will be kept for late-joining applications. If while waiting for late-joining applications, the middleware needs to free up DDS samples, it will reclaim DDS
samples if they have been previously acknowledged by active/matching readers.
There are scenarios where you know a priori that a particular set of applications will join the system: e.g., a
logging service or a known processing application. The Required Subscription feature is designed to keep
data until these known late joining applications acknowledge the data.

294

6.3.13.1 Named, Required and Durable Subscriptions

Another use case is when DataReaders become temporarily inactive due to not responding to heartbeats,
or when the subscriber temporarily became disconnected and purged from the discovery database. In both
cases, the DataWriter will no longer keep the DDS sample for this DataReader. The Required Subscription feature will keep the data until these known DataReaders have acknowledged the data.
To use Required Subscriptions, the DataReaders and DataWriters must have their RELIABILITY
QosPolicy (Section 6.5.19 on page 400) kind set to RELIABLE.
6.3.13.1 Named, Required and Durable Subscriptions
Before describing the Required Subscriptions, it is important to understand a few concepts:
l

l

l

Named Subscription: Through the ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9
on page 374), each DataReader can be given a specific name. This name can be used by tools to
identify a specific DataReader. Additionally, the DataReader can be given a role_name. For
example: LOG_APP_1 DataReader belongs to the logger applications (role_name =
“LOGGER”).
Required Subscription is a named subscription to which a DataWriter is configured to deliver data
to. This is true even if the DataReaders serving those subscriptions are not available yet. The
DataWriter must store the DDS sample until it has been acknowledged by all active reliable
DataReaders and acknowledged by all required subscriptions. The DataWriter is not waiting for a
specific DataReader, rather it is waiting for DataReaders belonging to the required subscription by
setting their role_name to the subscription name.
Durable Subscription is a required subscription where DDS samples are stored and forwarded by
an external service. In this case, the required subscription is served by RTI Persistence Service. See
Configuring Durable Subscriptions in Persistence Service (Section 27.9 on page 955).

6.3.13.2 Durability QoS and Required Subscriptions
The DURABILITY QosPolicy (Section 6.5.7 on page 368) and the Required Subscriptions feature complement each other.
The DurabilityQosPolicy determines whether or not Connext DDS will store and deliver previously
acknowledged DDS samples to new DataReaders that join the network later. You can specify to either
not make the DDS samples available (DDS_VOLATILE_DURABILITY_QOS kind), or to make them
available and declare you are storing the DDS samples in memory (DDS_TRANSIENT_LOCAL_
DURABILITY_QOS or DDS_TRANSIENT_DURABILITY_QOS kind) or in permanent storage
(DDS_PERSISTENT_DURABILITY_QOS).
Required subscriptions help answer the question of when a DDS sample is considered acknowledged
before the DurabilityQosPolicy determines whether to keep it. When required subscriptions are used, a
DDS sample is considered acknowledged by a DataWriter when both the active DataReaders and a
quorum of required subscriptions have acknowledged the DDS sample. (Acknowledging a DDS sample

295

6.3.13.3 Required Subscriptions Configuration

can be done either at the protocol or application level—see Application Acknowledgment (Section 6.3.12
on page 288)).
6.3.13.3 Required Subscriptions Configuration
Each DataReader can be configured to be part of a named subscription, by giving it a role_name using
the ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9 on page 374). A DataWriter can then
be configured using the AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on page 337)
(required_matched_endpoint_groups) with a list of required named subscriptions identified by the role_
name. Additionally, the DataWriter can be configured with a quorum or minimum number of DataReaders from a given named subscription that must receive a DDS sample.
When configured with a list of required subscriptions, a DataWriter will store a DDS sample until the
DDS sample is acknowledged by all active reliable DataReaders, as well as all required subscriptions.
When a quorum is specified, a minimum number of DataReaders of the required subscription must
acknowledge a DDS sample in order for the DDS sample to be considered acknowledged. Specifying a
quorum provides a level of redundancy in the system as multiple applications or services acknowledge
they have received the DDS sample. Each individual DataReader is identified using its own virtual GUID
(see DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511)).

6.3.14 Managing Data Instances (Working with Keyed Data Types)
This section applies only to data types that use keys, see DDS Samples, Instances, and Keys (Section 2.3.1
on page 14). Using the following operations for non-keyed types has no effect.
Topics come in two flavors: those whose associated data type has specified some fields as defining the
‘key,’ and those whose associated data type has not. An example of a data-type that specifies key fields is
shown in Data Type with a Key (Section Figure 6.14 below).
Figure 6.14 Data Type with a Key
typedef struct Flight {
long
flightId; //@key
string departureAirport;
string arrivalAirport;
Time_t departureTime;
Time_t estimatedArrivalTime;
Location_t currentPosition;
};

If the data type has some fields that act as a ‘key,’ the Topic essentially defines a collection of datainstances whose values can be independently maintained. In Figure 6.14 Data Type with a Key above, the
flightId is the ‘key’. Different flights will have different values for the key. Each flight is an instance of the
Topic. Each write() will update the information about a single flight. DataReaders can be informed when
new flights appear or old ones disappear.

296

6.3.14.1 Registering and Unregistering Instances

Since the key fields are contained within the data structure, Connext DDS could examine the key fields
each time it needs to determine which data-instance is being modified. However, for performance and
semantic reasons, it is better for your application to declare all the data-instances it intends to modify—
prior to actually writing any DDS samples. This is known as registration, described below in Registering
and Unregistering Instances (Section 6.3.14.1 below).
The register_instance() operation provides a handle to the instance (of type DDS_InstanceHandle_t)
that can be used later to refer to the instance.
6.3.14.1 Registering and Unregistering Instances
If your data type has a key, you may improve performance by registering an instance (data associated with
a particular value of the key) before you write data for the instance. You can do this for any number of
instances up the maximum number of instances configured in the DataWriter’s RESOURCE_LIMITS
QosPolicy (Section 6.5.20 on page 405). Instance registration is completely optional.
Registration tells Connext DDS that you are about to modify (write or dispose of) a specific instance. This
allows Connext DDS to pre-configure itself to process that particular instance, which can improve performance.
If you write without registering, you can pass the NIL instance handle as part of the write() call.
If you register the instance first, Connext DDS can look up the instance beforehand and return a handle to
that instance. Then when you pass this handle to the write() operation, Connext DDS no longer needs to
analyze the data to check what instance it is for. Instead, it can directly update the instance pointed to by
the instance handle.
In summary, by registering an instance, all subsequent write() calls to that instance become more efficient.
If you only plan to write once to a particular instance, registration does not ‘buy’ you much in performance, but in general, it is good practice.
To register an instance, use the DataWriter’s register_instance() operation. For best performance, it
should be invoked prior to calling any operation that modifies the instance, such as write(), write_w_
timestamp(), dispose(), or dispose_w_timestamp().
When you are done using that instance, you can unregister it. To unregister an instance, use the
DataWriter’s unregister_instance() operation. Unregistering tells Connext DDS that the DataWriter does
not intend to modify that data-instance anymore, allowing Connext DDS to recover any resources it allocated for the instance. It does not delete the instance; that is done with the dispose_instance() operation, see
Disposing of Data (Section 6.3.14.2 on page 299). autodispose_unregistered_instances (Section on page
419) in the WRITER_DATA_LIFECYCLE QoS Policy (Section 6.5.27 on page 419) controls whether
instances are automatically disposed when they are unregistered.
unregister_instance() should only be used on instances that have been previously registered. The use of
these operations is illustrated in Figure 6.15 Registering an Instance on the facing page.

297

6.3.14.1 Registering and Unregistering Instances

Figure 6.15 Registering an Instance
Flight myFlight;
// writer is a previously-created FlightDataWriter
myFlight.flightId = 265;
DDS_InstanceHandle_t fl265Handle =
writer->register_instance(myFlight);
...
// Each time we update the flight, we can pass the handle
myFlight.departureAirport
= “SJC”;
myFlight.arrivalAirport
= “LAX”;
myFlight.departureTime
= {120000, 0};
myFlight.estimatedArrivalTime = {130200, 0};
myFlight.currentPosition
= { {37, 20}, {121, 53} };
if (writer->write(myFlight, fl265Handle) != DDS_RETCODE_OK) {
// ... handle error
}
// After updating the flight, it can be unregistered
if (writer->unregister_instance(myFlight, fl265Handle) !=
DDS_RETCODE_OK) {
// ... handle error
}

Once an instance has been unregistered, and assuming that no other DataWriters are writing values for the
instance, the matched DataReaders will eventually get an indication that the instance no longer has any
DataWriters. This is communicated to the DataReaders by means of the DDS_SampleInfo that accompanies each DDS data-sample (see The SampleInfo Structure (Section 7.4.6 on page 504)). Once there
are no DataWriters for the instance, the DataReader will see the value of DDS_InstanceStateKind for
that instance to be NOT_ALIVE_NO_WRITERS.
The unregister_instance() operation may affect the ownership of the data instance (see the
OWNERSHIP QosPolicy (Section 6.5.15 on page 389)). If the DataWriter was the exclusive owner of
the instance, then calling unregister_instance() relinquishes that ownership, and another DataWriter can
become the exclusive owner of the instance.
The unregister_instance() operation indicates only that a particular DataWriter no longer has anything to
say about the instance.
Note that this is different than the dispose() operation discussed in the next section, which informs
DataReaders that the data-instance is no longer “alive.” The state of an instance is stored in the DDS_
SampleInfo structure that accompanies each DDS sample of data that is received by a DataReader. User
code can access the instance state to see if an instance is “alive”—meaning there is at least one DataWriter
that is publishing DDS samples for the instance, see Instance States (Section 7.4.6.4 on page 507).
See also:

298

6.3.14.2 Disposing of Data

l

Unregistering vs. Disposing: (Section on page 420).

l

Use Cases for Unregistering without Disposing: (Section on page 420).

6.3.14.2 Disposing of Data
The dispose() operation informs DataReaders that, as far as the DataWriter knows, the data-instance no
longer exists and can be considered “not alive.” When the dispose() operation is called, the instance state
stored in the DDS_SampleInfo structure, accessed through DataReaders, will change to NOT_ALIVE_
DISPOSED for that particular instance.
See Unregistering vs. Disposing: (Section on page 420).
By default, instances are automatically disposed when they are unregistered. This behavior is controlled
by the autodispose_unregistered_instances (Section on page 419) field in the WRITER_DATA_
LIFECYCLE QoS Policy (Section 6.5.27 on page 419).
For example, in a flight tracking system, when a flight lands, a DataWriter may dispose the data-instance
corresponding to the flight. In that case, all DataReaders who are monitoring the flight will see the
instance state change to NOT_ALIVE_DISPOSED, indicating that the flight has landed.
If a particular instance is never disposed, its instance state will eventually change from ALIVE to NOT_
ALIVE_NO_WRITERS once all the DataWriters that were writing that instance unregister the instance
or lose their liveliness. For more information on DataWriter liveliness, see the LIVELINESS QosPolicy
(Section 6.5.13 on page 382).
See also:
l

Propagating Serialized Keys with Disposed-Instance Notifications (Section 6.5.3.5 on page 356).

l

Use Cases for Unregistering without Disposing: (Section on page 420).

6.3.14.3 Looking Up an Instance Handle
Some operations, such as write(), require an instance_handle parameter. If you need to get such as
handle, you can call the FooDataWriter’s lookup_instance() operation, which takes an instance as a parameter and returns a handle to that instance. This is useful for keyed data types.
DDS_InstanceHandle_t lookup_instance (const Foo & key_holder)

The instance must have already been registered (see Registering and Unregistering Instances (Section
6.3.14.1 on page 297)). If the instance is not registered, this operation returns DDS_HANDLE_NIL.
6.3.14.4 Getting the Key Value for an Instance
Once you have an instance handle (using register_instance() or lookup_instance()), you can use the
DataWriter’s get_key_value() operation to retrieve the value of the key of the corresponding instance.

299

6.3.15 Setting DataWriter QosPolicies

The key fields of the data structure passed into get_key_value() will be filled out with the original values
used to generate the instance handle. The key fields are defined when the data type is defined, see DDS
Samples, Instances, and Keys (Section 2.3.1 on page 14) for more information.
Following our example in Figure 6.15 Registering an Instance on page 298, register_instance() returns a
DDS_InstanceHandle_t (fl265Handle) that can be used in the call to the FlightDataWriter’s get_key_
value() operation. The value of the key is returned in a structure of type Flight with the flightId field filled
in with the integer 265.
See also: Propagating Serialized Keys with Disposed-Instance Notifications (Section 6.5.3.5 on page
356).

6.3.15 Setting DataWriter QosPolicies
The DataWriter’s QosPolicies control its resources and behavior.
The DDS_DataWriterQos structure has the following format:
DDS_DataWriterQos struct {
DDS_DurabilityQosPolicy
DDS_DurabilityServiceQosPolicy
DDS_DeadlineQosPolicy
DDS_LatencyBudgetQosPolicy
DDS_LivelinessQosPolicy
DDS_ReliabilityQosPolicy
DDS_DestinationOrderQosPolicy
DDS_HistoryQosPolicy
DDS_ResourceLimitsQosPolicy
DDS_TransportPriorityQosPolicy
DDS_LifespanQosPolicy
DDS_UserDataQosPolicy
DDS_OwnershipQosPolicy
DDS_OwnershipStrengthQosPolicy
DDS_WriterDataLifecycleQosPolicy
// extensions to the DDS standard:
DDS_DataWriterResourceLimitsQosPolicy
DDS_DataWriterProtocolQosPolicy
DDS_TransportSelectionQosPolicy
DDS_TransportUnicastQosPolicy
DDS_PublishModeQosPolicy
DDS_PropertyQosPolicy
DDS_ServiceQosPolicy
DDS_BatchQosPolicy
DDS_MultiChannelQosPolicy
DDS_AvailabilityQosPolicy
DDS_EntityNameQosPolicy
DDS_TypeSupportQosPolicy
} DDS_DataWriterQos;

durability;
durability_service;
deadline;
latency_budget;
liveliness;
reliability;
destination_order;
history;
resource_limits;
transport_priority;
lifespan;
user_data;
ownership;
ownership_strength;
writer_data_lifecycle;
writer_resource_limits;
protocol;
transport_selection;
unicast;
publish_mode;
property;
service;
batch;
multi_channel;
availability;
publication_name;
type_support;

Note: set_qos() cannot always be used within a listener callback, see Restricted Operations in Listener
Callbacks (Section 4.5.1 on page 185).

300

6.3.15 Setting DataWriter QosPolicies

Table 6.17 DataWriter QosPolicies summarizes the meaning of each policy. (They appear alphabetically
in the table.) For information on why you would want to change a particular QosPolicy, see the referenced
section. For defaults and valid ranges, please refer to the API Reference HTML documentation.
Table 6.17 DataWriter QosPolicies
QosPolicy

Description
This QoS policy is used in the context of two features:
Availability QoS Policy and Collaborative DataWriters (Section 6.5.1.1 on page 338)
AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on page 337)

Availability

For Collaborative DataWriters, Availability specifies the group of DataWriters expected to collaboratively
provide data and the timeouts that control when to allow data to be available that may skip DDS samples.
For Required Subscriptions, Availability configures a set of Required Subscriptions on a DataWriter.
See AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on page 337)

Batch

Specifies and configures the mechanism that allows Connext DDS to collect multiple DDS user data samples
to be sent in a single network packet, to take advantage of the efficiency of sending larger packets and thus
increase effective throughput. See BATCH QosPolicy (DDS Extension) (Section 6.5.2 on page 341).

DataWriterProtocol

This QosPolicy configures the Connext DDS on-the-network protocol, RTPS. See DATA_WRITER_
PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).

DataWriterResourceLimits

Controls how many threads can concurrently block on a write() call of this DataWriter. See DATA_
WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 6.5.4 on page 359).
For a DataReader, it specifies the maximum expected elapsed time between arriving DDS data samples.

Deadline

For a DataWriter, it specifies a commitment to publish DDS samples with no greater elapsed time between
them.
See DEADLINE QosPolicy (Section 6.5.5 on page 363).

DestinationOrder

Controls how Connext DDS will deal with data sent by multiple DataWriters for the same topic. Can be set
to "by reception timestamp" or to "by source timestamp". See DESTINATION_ORDER QosPolicy (Section
6.5.6 on page 365).

Durability

Specifies whether or not Connext DDS will store and deliver data that were previously published to new
DataReaders. See DURABILITY QosPolicy (Section 6.5.7 on page 368).

DurabilityService

Various settings to configure the external Persistence Service1 used by Connext DDS for DataWriters with
a Durability QoS setting of Persistent Durability. See DURABILITY SERVICE QosPolicy (Section 6.5.8 on
page 372).

1Persistence Service is provided with the Connext DDS Professional, Evaluation, and Basic package

types.
301

6.3.15 Setting DataWriter QosPolicies

Table 6.17 DataWriter QosPolicies
QosPolicy

Description

EntityName

Assigns a name to a DataWriter. See ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9 on page
374).

History

Specifies how much data must to stored by Connext DDS for the DataWriter or DataReader. This
QosPolicy affects the RELIABILITY QosPolicy (Section 6.5.19 on page 400) as well as the DURABILITY
QosPolicy (Section 6.5.7 on page 368). See HISTORY QosPolicy (Section 6.5.10 on page 376).

LatencyBudget

Suggestion to Connext DDS on how much time is allowed to deliver data. See LATENCYBUDGET QoS
Policy (Section 6.5.11 on page 380).

Lifespan

Specifies how long Connext DDS should consider data sent by an user application to be valid. See
LIFESPAN QoS Policy (Section 6.5.12 on page 381).

Liveliness

Specifies and configures the mechanism that allows DataReaders to detect when DataWriters become
disconnected or "dead." See LIVELINESS QosPolicy (Section 6.5.13 on page 382).

MultiChannel

Configures a DataWriter’s ability to send data on different multicast groups (addresses) based on the value of
the data. See MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386).

Ownership

Along with OwnershipStrength, specifies if DataReaders for a topic can receive data from multiple
DataWriters at the same time. See OWNERSHIP QosPolicy (Section 6.5.15 on page 389).

OwnershipStrength

Used to arbitrate among multiple DataWriters of the same instance of a Topic when Ownership QosPolicy is
EXCLUSIVE. See OWNERSHIP_STRENGTH QosPolicy (Section 6.5.16 on page 393).

Partition

Adds string identifiers that are used for matching DataReaders and DataWriters for the same Topic. See
PARTITION QosPolicy (Section 6.4.5 on page 323).

Property

Stores name/value (string) pairs that can be used to configure certain parameters of Connext DDS that are not
exposed through formal QoS policies. It can also be used to store and propagate application-specific
name/value pairs, which can be retrieved by user code during discovery. See PROPERTY QosPolicy (DDS
Extension) (Section 6.5.17 on page 394).

PublishMode

Specifies how Connext DDS sends application data on the network. By default, data is sent in the user thread
that calls the DataWriter’s write() operation. However, this QosPolicy can be used to tell Connext DDS to
use its own thread to send the data. See PUBLISH_MODE QosPolicy (DDS Extension) (Section 6.5.18 on
page 397).

Reliability

Specifies whether or not Connext DDS will deliver data reliably. See RELIABILITY QosPolicy (Section
6.5.19 on page 400).

ResourceLimits

Controls the amount of physical memory allocated for Entities, if dynamic allocations are allowed, and how
they occur. Also controls memory usage among different instance values for keyed topics. See RESOURCE_
LIMITS QosPolicy (Section 6.5.20 on page 405).

Service

Intended for use by RTI infrastructure services. User applications should not modify its value. See SERVICE
QosPolicy (DDS Extension) (Section 6.5.21 on page 408).

302

6.3.15.1 Configuring QoS Settings when the DataWriter is Created

Table 6.17 DataWriter QosPolicies
QosPolicy

Description

TransportPriority

Set by a DataWriter to tell Connext DDS that the data being sent is a different "priority" than other data. See
TRANSPORT_PRIORITY QosPolicy (Section 6.5.22 on page 409).

TransportSelection

Allows you to select which physical transports a DataWriter or DataReader may use to send or receive its
data. See TRANSPORT_SELECTION QosPolicy (DDS Extension) (Section 6.5.23 on page 411).

TransportUnicast

Specifies a subset of transports and port number that can be used by an Entity to receive data. See
TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412).

TypeSupport

Used to attach application-specific value(s) to a DataWriter or DataReader. These values are passed to the
serialization or deserialization routine of the associated data type. Also controls whether padding bytes are set
to 0 during serialization. See TYPESUPPORT QosPolicy (DDS Extension) (Section 6.5.25 on page 416).

UserData

Along with Topic Data QosPolicy and Group Data QosPolicy, used to attach a buffer of bytes to Connext
DDS's discovery meta-data. See USER_DATA QosPolicy (Section 6.5.26 on page 417).

WriterDataLifeCycle

Controls how a DataWriter handles the lifecycle of the instances (keys) that the DataWriter is registered to
manage. See WRITER_DATA_LIFECYCLE QoS Policy (Section 6.5.27 on page 419).

Many of the DataWriter QosPolicies also apply to DataReaders (see DataReaders (Section 7.3 on page
459)). For a DataWriter to communicate with a DataReader, their QosPolicies must be compatible. Generally, for the QosPolicies that apply both to the DataWriter and the DataReader, the setting in the
DataWriter is considered an “offer” and the setting in the DataReader is a “request.” Compatibility means
that what is offered by the DataWriter equals or surpasses what is requested by the DataReader. Each
policy’s description includes compatibility restrictions. For more information on compatibility, see QoS
Requested vs. Offered Compatibility—the RxO Property (Section 4.2.1 on page 167).
Some of the policies may be changed after the DataWriter has been created. This allows the application to
modify the behavior of the DataWriter while it is in use. To modify the QoS of an already-created
DataWriter, use the get_qos() and set_qos() operations on the DataWriter. This is a general pattern for all
Entities, described in Changing the QoS for an Existing Entity (Section 4.1.7.3 on page 161).
6.3.15.1 Configuring QoS Settings when the DataWriter is Created
As described in Creating DataWriters (Section 6.3.1 on page 266), there are different ways to create a
DataWriter, depending on how you want to specify its QoS (with or without a QoS Profile).
l

In Creating a DataWriter with Default QosPolicies and a Listener (Section Figure 6.9 on page 268),
there is an example of how to create a DataWriter with default QosPolicies by using the special constant, DDS_DATAWRITER_QOS_DEFAULT, which indicates that the default QoS values for
a DataWriter should be used. The default DataWriter QoS values are configured in the Publisher or
DomainParticipant; you can change them with set_default_datawriter_qos() or set_default_

303

6.3.15.1 Configuring QoS Settings when the DataWriter is Created

datawriter_qos_with_profile(). Then any DataWriters created with the Publisher will use the new
default values. As described in Getting, Setting, and Comparing QosPolicies (Section 4.1.7 on page
158), this is a general pattern that applies to the construction of all Entities.
l

l

l

To create a DataWriter with non-default QoS without using a QoS Profile, see the example code in
Figure 6.16 Creating a DataWriter with Modified QosPolicies (not from a profile) below. It uses the
Publisher’s get_default_writer_qos() method to initialize a DDS_DataWriterQos structure. Then
the policies are modified from their default values before the structure is used in the create_
datawriter() method.
You can also create a DataWriter and specify its QoS settings via a QoS Profile. To do so, you will
call create_datawriter_with_profile(), as seen in Figure 6.17 Creating a DataWriter with a QoS
Profile on the next page.
If you want to use a QoS profile, but then make some changes to the QoS before creating the
DataWriter, call get_datawriter_qos_from_profile() and create_datawriter() as seen in Figure
6.18 Getting QoS Values from a Profile, Changing QoS Values, Creating a DataWriter with Modified QoS Values on the next page.

For more information, see Creating DataWriters (Section 6.3.1 on page 266) and Configuring QoS with
XML (Section Chapter 17 on page 791).
Figure 6.16 Creating a DataWriter with Modified QosPolicies (not from a profile)
DDS_DataWriterQos writer_qos;1
// initialize writer_qos with default values
publisher->get_default_datawriter_qos(writer_qos);
// make QoS changes
writer_qos.history.depth = 5;
// Create the writer with modified qos
DDSDataWriter * writer = publisher->create_datawriter(
topic, writer_qos, NULL, DDS_STATUS_MASK_NONE);
if (writer == NULL) {
// ... error
}
// narrow it for your specific data type
FooDataWriter* foo_writer = FooDataWriter::narrow(writer);

1Note: In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
304

6.3.15.2 Comparing QoS Values

Figure 6.17 Creating a DataWriter with a QoS Profile

// Create the datawriter
DDSDataWriter * writer =
publisher->create_datawriter_with_profile(
topic, “MyWriterLibrary”, “MyWriterProfile”,
NULL, DDS_STATUS_MASK_NONE);
if (writer == NULL) {
// ... error
};
// narrow it for your specific data type
FooDataWriter* foo_writer = FooDataWriter::narrow(writer);

Figure 6.18 Getting QoS Values from a Profile, Changing QoS Values, Creating a DataWriter
with Modified QoS Values
DDS_DataWriterQos writer_qos;1
// Get writer QoS from profile
retcode = factory->get_datawriter_qos_from_profile(
writer_qos, “WriterProfileLibrary”, “WriterProfile”);
if (retcode != DDS_RETCODE_OK) {
// handle error
}
// Makes QoS changes
writer_qos.history.depth = 5;
DDSDataWriter * writer = publisher->create_datawriter(
topic, writer_qos, NULL, DDS_STATUS_MASK_NONE);
if (participant == NULL) {
// handle error
}

6.3.15.2 Comparing QoS Values
The equals() operation compares two DataWriter’s DDS_DataWriterQoS structures for equality. It takes
two parameters for the two DataWriter’s QoS structures to be compared, then returns TRUE is they are
equal (all values are the same) or FALSE if they are not equal.
6.3.15.3 Changing QoS Settings After the DataWriter Has Been Created
There are two ways to change an existing DataWriter’s QoS after it is has been created—again depending
on whether or not you are using a QoS Profile.
1Note: In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
305

6.3.15.4 Using a Topic’s QoS to Initialize a DataWriter’s QoS

l

l

To change QoS programmatically (that is, without using a QoS Profile), use get_qos() and set_qos
(). See the example code in Figure 6.19 Changing the QoS of an Existing DataWriter (without a
QoS Profile) below. It retrieves the current values by calling the DataWriter’s get_qos() operation.
Then it modifies the value and calls set_qos() to apply the new value. Note, however, that some
QosPolicies cannot be changed after the DataWriter has been enabled—this restriction is noted in
the descriptions of the individual QosPolicies.
You can also change a DataWriter’s (and all other Entities’) QoS by using a QoS Profile and calling
set_qos_with_profile(). For an example, see Figure 6.20 Changing the QoS of an Existing
DataWriter with a QoS Profile below. For more information, see Configuring QoS with XML (Section Chapter 17 on page 791).

Figure 6.19 Changing the QoS of an Existing DataWriter (without a QoS Profile)
DDS_DataWriterQos writer_qos;1
// Get current QoS.
if (datawriter->get_qos(writer_qos) != DDS_RETCODE_OK) {
// handle error
}
// Makes QoS changes here
writer_qos.history.depth = 5;
// Set the new QoS
if (datawriter->set_qos(writer_qos) != DDS_RETCODE_OK ) {
// handle error
}

Figure 6.20 Changing the QoS of an Existing DataWriter with a QoS Profile

retcode = writer->set_qos_with_profile(
“WriterProfileLibrary”,”WriterProfile”);
if (retcode != DDS_RETCODE_OK) {
// handle error
}

6.3.15.4 Using a Topic’s QoS to Initialize a DataWriter’s QoS
Several DataWriter QosPolicies can also be found in the QosPolicies for Topics (see Setting Topic
QosPolicies (Section 5.1.3 on page 204)). The QosPolicies set in the Topic do not directly affect the
DataWriters (or DataReaders) that use that Topic. In many ways, some QosPolicies are a Topic-level
concept, even though the DDS standard allows you to set different values for those policies for different

1Note: In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
306

6.3.15.4 Using a Topic’s QoS to Initialize a DataWriter’s QoS

DataWriters and DataReaders of the same Topic. Thus, the policies in the DDS_TopicQos structure exist
as a way to help centralize and annotate the intended or suggested values of those QosPolicies. Connext
DDS does not check to see if the actual policies set for a DataWriter is aligned with those set in the Topic
to which it is bound.
There are many ways to use the QosPolicies’ values set in the Topic when setting the QosPolicies’ values
in a DataWriter. The most straightforward way is to get the values of policies directly from the Topic and
use them in the policies for the DataWriter, as shown in Figure 6.21 Copying Selected QoS from a Topic
when Creating a DataWriter below.
Figure 6.21 Copying Selected QoS from a Topic when Creating a DataWriter
DDS_DataWriterQos writer_qos;1
DDS_TopicQos topic_qos;
// topic and publisher already created
// get current QoS for the topic, default QoS for the writer
if (topic->get_qos(topic_qos) != DDS_RETCODE_OK) {
// handle error
}
if (publisher->get_default_datawriter_qos(writer_qos)
!= DDS_RETCODE_OK) {
// handle error
}
// Copy specific policies from topic QoS to writer QoS
writer_qos.deadline = topic_qos.deadline;
writer_qos.reliability = topic_qos.reliability;
// Create the DataWriter with the modified QoS
DDSDataWriter* writer = publisher->create_datawriter(topic,
writer_qos,NULL, DDS_STATUS_MASK_NONE);

1Note: In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
307

6.3.15.4 Using a Topic’s QoS to Initialize a DataWriter’s QoS

You can use the Publisher’s copy_from_topic_qos() operation to copy all of the common policies from
the Topic QoS to a DataWriter QoS. This is illustrated in Figure 6.22 Copying all QoS from a Topic when
Creating a DataWriter below.
Figure 6.22 Copying all QoS from a Topic when Creating a DataWriter
DDS_DataWriterQos writer_qos;1
DDS_TopicQos topic_qos;
// topic, publisher, writer_listener already created
if (topic->get_qos(topic_qos) != DDS_RETCODE_OK) {
// handle error
}
if (publisher->get_default_datawriter_qos(writer_qos)
!= DDS_RETCODE_OK)
{
// handle error
}
// copy relevant QoS from topic into writer’s qos
publisher->copy_from_topic_qos(writer_qos, topic_qos);
// Optionally, modify policies as desired
writer_qos.deadline.duration.sec = 1;
writer_qos.deadline.duration.nanosec = 0;
// Create the DataWriter with the modified QoS
DDSDataWriter* writer = publisher->create_datawriter(topic,
writer_qos, writer_listener, DDS_STATUS_MASK_ALL);

In another design pattern, you may want to start with the default QoS values for a DataWriter and override
them with the QoS values of the Topic. Figure 6.23 Combining Default Topic and DataWriter QoS
(Option 1) on the next page gives an example of how to do this.
Because this is a common pattern, Connext DDS provides a special macro, DDS_DATAWRITER_
QOS_USE_TOPIC_QOS, that can be used to indicate that the DataWriter should be created with the set
of QoS values that results from modifying the default DataWriter QosPolicies with the QoS values specified by the Topic. Figure 6.24 Combining Default Topic and DataWriter QoS (Option 2) on the next
page shows how the macro is used.
The code fragments shown in Figure 6.23 Combining Default Topic and DataWriter QoS (Option 1) on
the next page and Figure 6.24 Combining Default Topic and DataWriter QoS (Option 2) on the next page
result in identical QoS settings for the created DataWriter.

1Note: In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
308

6.3.16 Navigating Relationships Among DDS Entities

Figure 6.23 Combining Default Topic and DataWriter QoS (Option 1)
DDS_DataWriterQos writer_qos;1
DDS_TopicQos topic_qos;
// topic, publisher, writer_listener already created
if (topic->get_qos(topic_qos) != DDS_RETCODE_OK) {
// handle error
}
if (publisher->get_default_datawriter_qos(writer_qos)
!= DDS_RETCODE_OK) {
// handle error
}
if (publisher->copy_from_topic_qos(writer_qos, topic_qos)
!= DDS_RETCODE_OK) {
// handle error
}
// Create the DataWriter with the combined QoS
DDSDataWriter* writer =
publisher->create_datawriter(topic, writer_qos,
writer_listener,DDS_STATUS_MASK_ALL);

Figure 6.24 Combining Default Topic and DataWriter QoS (Option 2)

// topic, publisher, writer_listener already created
DDSDataWriter* writer = publisher->create_datawriter (topic,
DDS_DATAWRITER_QOS_USE_TOPIC_QOS,
writer_listener, DDS_STATUS_MASK_ALL);

For more information on the general use and manipulation of QosPolicies, see Getting, Setting, and Comparing QosPolicies (Section 4.1.7 on page 158).

6.3.16 Navigating Relationships Among DDS Entities
6.3.16.1 Finding Matching Subscriptions
The following DataWriter operations can be used to get information on the DataReaders that are currently
associated with the DataWriter (that is, the DataReaders to which Connext DDS will send the data written
by the DataWriter).

1Note: In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
309

6.3.16.1 Finding Matching Subscriptions

l

get_matched_subscriptions()

l

get_matched_subscription_data()

l

get_matched_subscription_locators()

get_matched_subscriptions() will return a sequence of handles to matched DataReaders. You can use
these handles in the get_matched_subscription_data() method to get information about the DataReader
such as the values of its QosPolicies.
get_matched_subscription_locators() retrieves a list of locators for subscriptions currently "associated"
with the DataWriter. Matched subscription locators include locators for all those subscriptions in the same
DDS domain that have a matching Topic, compatible QoS, and a common partition that the DomainParticipant has not indicated should be "ignored." These are the locators that Connext DDS uses to communicate with matching DataReaders. (See Locator Format (Section 14.2.1.1 on page 714).)
Note: In the Modern C++ API these operations are freestanding functions in the dds::pub or rti::pub
namespaces.
You can also get the DATA_WRITER_PROTOCOL_STATUS for matching subscriptions with these
operations (see DATA_WRITER_PROTOCOL_STATUS (Section 6.3.6.3 on page 273)):
l

get_matched_subscription_datawriter_protocol_status()

l

get_matched_subscription_datawriter_protocol_status_by_locator()

Notes:
l

l

l

Status/data for a matched subscription is only kept while the matched subscription is alive. Once a
matched subscription is no longer alive, its status is deleted. If you try to get the status/data for a
matched subscription that is no longer alive, the 'get status' or ' get data' call will return an error.
DataReaders that have been ignored using the DomainParticipant’s ignore_subscription() operation are not considered to be matched even if the DataReader has the same Topic and compatible
QosPolicies. Thus, they will not be included in the list of DataReaders returned by get_matched_
subscriptions() or get_matched_subscription_locators(). See Ignoring Publications and Subscriptions (Section 16.4.2 on page 786) for more on ignore_subscription().
The get_matched_subscription_data() operation does not retrieve the following information from
built-in-topic data structures: type_code, property, and content_filter_property. This information
is available through the on_data_available() callback (if a DataReaderListener is installed on the
SubscriptionBuiltinTopicDataDataReader). (bug 11914)

See also: Finding the Matching Subscription’s ParticipantBuiltinTopicData (Section 6.3.16.2 on the next
page)

310

6.3.16.2 Finding the Matching Subscription’s ParticipantBuiltinTopicData

6.3.16.2 Finding the Matching Subscription’s ParticipantBuiltinTopicData
get_matched_subscription_participant_data() allows you to get the DDS_ParticipantBuiltinTopicData
(see Table 16.1 Participant Built-in Topic’s Data Type (DDS_ParticipantBuiltinTopicData)) of a matched
subscription using a subscription handle.
This operation retrieves the information on a discovered DomainParticipant associated with the subscription that is currently matching with the DataWriter.The subscription handle passed into this operation
must correspond to a subscription currently associated with the DataWriter. Otherwise, the operation will
fail with RETCODE_BAD_PARAMETER. The operation may also fail with RETCODE_
PRECONDITION_NOT_MET if the subscription corresponds to the same DomainParticipant to which
the DataWriter belongs.
Use get_matched_subscriptions() (see Finding Matching Subscriptions (Section 6.3.16.1 on page 309))
to find the subscriptions that are currently matched with the DataWriter.
6.3.16.3 Finding Related DDS Entities
These operations are useful for obtaining a handle to various related Entities:
l

get_publisher()

l

get_topic()

get_publisher() returns the Publisher that created the DataWriter. get_topic() returns the Topic with
which the DataWriter is associated.

6.3.17 Asserting Liveliness
The assert_liveliness() operation can be used to manually assert the liveliness of the DataWriter without
writing data. This operation is only useful if the kind of LIVELINESS QosPolicy (Section 6.5.13 on
page 382) is MANUAL_BY_PARTICIPANT or MANUAL_BY_TOPIC.
How DataReaders determine if DataWriters are alive is configured using the LIVELINESS QosPolicy
(Section 6.5.13 on page 382). The lease_duration parameter of the LIVELINESS QosPolicy is a contract by the DataWriter to all of its matched DataReaders that it will send a packet within the time value of
the lease_duration to state that it is still alive.
There are three ways to assert liveliness. One is to have Connext DDS itself send liveliness packets periodically when the kind of LIVELINESS QosPolicy is set to AUTOMATIC. The other two ways to
assert liveliness, used when liveliness is set to MANUAL, are to call write() to send data or to call the
assert_liveliness() operation without sending data.

311

6.3.18 Turbo Mode and Automatic Throttling for DataWriter Performance—Experimental Features

6.3.18 Turbo Mode and Automatic Throttling for DataWriter Performance—
Experimental Features
This section describes two experimental features. The DataWriter has many QoS settings that can affect
the latency and throughput of outgoing data. There are QoS settings to control send window size (see
Understanding the Send Queue and Setting its Size (Section 10.3.2.1 on page 639)) and settings that
allow to aggregate multiple DDS samples together to reduce CPU and bandwidth utilization (see BATCH
QosPolicy (DDS Extension) (Section 6.5.2 on page 341) and FlowControllers (DDS Extension) (Section
6.6 on page 422)). The choice of settings that provide the best performance depends on several factors,
such as the frequency of writing data, the size of the data, or the condition of the network. If these factors
do not change over time, you can choose values for those QoS settings that best suit your system. If these
factors do change over time in your system, you can use the following properties to let Connext DDS automatically adjust the QoS settings as system conditions change:
l

l

l

dds.domain_participant.auto_throttle.enable: Configures the DomainParticipant to gather
internal measurements (during DomainParticipant creation) that are required for the Auto Throttle
feature. This allows DataWriters belonging to this DomainParticipant to use the Auto Throttle feature. Default: false.
dds.data_writer.auto_throttle.enable: Enables automatic throttling in the DataWriter so it can
automatically adjust the writing rate and the send window size; this minimizes the need for repair
DDS samples and improves latency. Default: false. For additional information on automatic throttling, see Turbo Mode: Automatically Adjusting the Number of Bytes in a Batch—Experimental
Feature (Section 6.5.2.4 on page 344).
Note: This property takes effect only in DataWriters that belong to a DomainParticipant that has set
the property dds.domain_participant.auto_throttle.enable (described above) to true.
dds.data_writer.enable_turbo_mode: Enables Turbo Mode and adjusts the batch max_data_bytes
(Section on page 342) (see BATCH QosPolicy (DDS Extension) (Section 6.5.2 on page 341))
based on how frequently the DataWriter writes data. Default: false. For additional information, see
Turbo Mode: Automatically Adjusting the Number of Bytes in a Batch—Experimental Feature (Section 6.5.2.4 on page 344).

The Built-in QoS profile BuiltinQosLibExp::Generic.AutoTuning enables both Turbo Mode and Auto
Throttling.

6.4 Publisher/Subscriber QosPolicies
This section provides detailed information on the QosPolicies associated with a Publisher. Note that Subscribers have the exact same set of policies. Table 6.2 Publisher QosPolicies provides a quick reference.
They are presented here in alphabetical order.

312

6.4.1 ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension)

l

ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension) (Section 6.4.1 below)

l

ENTITYFACTORY QosPolicy (Section 6.4.2 on page 315)

l

EXCLUSIVE_AREA QosPolicy (DDS Extension) (Section 6.4.3 on page 318)

l

GROUP_DATA QosPolicy (Section 6.4.4 on page 320)

l

PARTITION QosPolicy (Section 6.4.5 on page 323)

l

PRESENTATION QosPolicy (Section 6.4.6 on page 330)

6.4.1 ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension)
This QosPolicy is used to enable or disable asynchronous publishing and asynchronous batch flushing for
the Publisher.
This QosPolicy can be used to reduce amount of time spent in the user thread to send data. You can use it
to send large data reliably. Large in this context means that the data cannot be sent as a single packet by a
transport. For example, to send data larger than 63K reliably using UDP/IP, you must configure Connext
DDS to send the data using asynchronous Publishers.
If so configured, the Publisher will spawn two threads, one for asynchronous publishing and one for asynchronous batch flushing. The asynchronous publisher thread will be shared by all DataWriters (belonging
to this Publisher) that have their PUBLISH_MODE QosPolicy (DDS Extension) (Section 6.5.18 on page
397) kind set to ASYNCHRONOUS. The asynchronous publishing thread will then handle the data transmission chores for those DataWriters. This thread will only be spawned when the first of these
DataWriters is enabled.
The asynchronous batch flushing thread will be shared by all DataWriters (belonging to this Publisher)
that have batching enabled and max_flush_delay different than DURATION_INFINITE in BATCH
QosPolicy (DDS Extension) (Section 6.5.2 on page 341). This thread will only be spawned when the first
of these DataWriters is enabled.
This QosPolicy allows you to adjust the asynchronous publishing and asynchronous batch flushing threads
independently.
Batching and asynchronous publication are independent of one another. Flushing a batch on an asynchronous DataWriter makes it available for sending to the DataWriter's FlowControllers (DDS Extension)
(Section 6.6 on page 422). From the point of view of the FlowController, a batch is treated like one large
DDS sample.
Connext DDS will sometimes coalesce multiple DDS samples into a single network datagram. For
example, DDS samples buffered by a FlowController or sent in response to a negative acknowledgement
(NACK) may be coalesced. This behavior is distinct from DDS sample batching. DDS data samples sent
by different asynchronous DataWriters belonging to the same Publisher to the same destination will not be
coalesced into a single network packet. Instead, two separate network packets will be sent. Only DDS
samples written by the same DataWriter and intended for the same destination will be coalesced.

313

6.4.1.1 Properties

This QosPolicy includes the members in Table 6.18 DDS_AsynchronousPublisherQosPolicy.
Table 6.18 DDS_AsynchronousPublisherQosPolicy
Type

DDS_Boolean

disable_
asynchronous_
write

DDS_
ThreadSettings_ thread
t

DDS_
Boolean

Description

Field Name

disable_
asynchronous_
batch

DDS_
asynchronous_
ThreadSettings_
batch_thread
t

Disables asynchronous publishing. To write asynchronously, this field must be FALSE (the
default).

Settings for the publishing thread. These settings are OS-dependent (see the RTI Connext DDS
Core Libraries Platform Notes).

Disables asynchronous batch flushing. To flush asynchronously, this field must be FALSE (the
default).

Settings for the asynchronous batch flushing thread. These settings are OS-dependent (see the RTI
Connext DDS Core Libraries Platform Notes).

6.4.1.1 Properties
This QosPolicy cannot be modified after the Publisher is created.
Since it is only for Publishers, there are no compatibility restrictions for how it is set on the publishing and
subscribing sides.
6.4.1.2 Related QosPolicies
l

l

l

If disable_asynchronous_write is TRUE (not the default), then any DataWriters created from this
Publisher must have their PUBLISH_MODE QosPolicy (DDS Extension) (Section 6.5.18 on page
397) kind set to SYNCHRONOUS. (Otherwise create_datawriter() will return INCONSISTENT_
QOS.)
If disable_asynchronous_batch is TRUE (not the default), then any DataWriters created from this
Publisher must have max_flush_delay in BATCH QosPolicy (DDS Extension) (Section 6.5.2 on
page 341) set to DURATION_INFINITE. (Otherwise create_datawriter() will return
INCONSISTENT_QOS.)
DataWriters configured to use the MULTI_CHANNEL QosPolicy (DDS Extension) (Section
6.5.14 on page 386) do not support asynchronous publishing; an error is returned if a multi-channel
DataWriter is configured for asynchronous publishing.

6.4.1.3 Applicable DDS Entities
l

Publishers (Section 6.2 on page 243)

314

6.4.1.4 System Resource Considerations

6.4.1.4 System Resource Considerations
Two threads can potentially be created:
l

l

For asynchronous publishing, system resource usage depends on the activity of the asynchronous
thread controlled by the FlowController (see FlowControllers (DDS Extension) (Section 6.6 on
page 422)).
For asynchronous batch flushing, system resource usage depends on the activity of the asynchronous thread controlled by max_flush_delay in BATCH QosPolicy (DDS Extension) (Section
6.5.2 on page 341).

6.4.2 ENTITYFACTORY QosPolicy
This QosPolicy controls whether or not child Entities are created in the enabled state.
This QosPolicy applies to the DomainParticipantFactory, DomainParticipants, Publishers, and Subscribers, which act as ‘factories’ for the creation of subordinate Entities. A DomainParticipantFactory is
used to create DomainParticipants. A DomainParticipant is used to create both Publishers and Subscribers. A Publisher is used to create DataWriters, similarly a Subscriber is used to create DataReaders.
Entities can be created either in an ‘enabled’ or ‘disabled’ state. An enabled entity can actively participate
in communication. A disabled entity cannot be discovered or take part in communication until it is explicitly enabled. For example, Connext DDS will not send data if the write() operation is called on a disabled
DataWriter, nor will Connext DDS deliver data to a disabled DataReader. You can only enable a disabled
entity. Once an entity is enabled, you cannot disable it, see Enabling DDS Entities (Section 4.1.2 on page
154) about the enable() method.
The ENTITYFACTORY contains only one member, as illustrated in Table 6.19 DDS_EntityFactoryQosPolicy.
Table 6.19 DDS_EntityFactoryQosPolicy
Type

Field Name

Description
DDS_BOOLEAN_TRUE: enable Entities when they are created

DDS_Boolean

autoenable_created_entities
DDS_BOOLEAN_FALSE: do not enable Entities when they are created

The ENTITYFACTORY QosPolicy controls whether the Entities created from the factory are automatically enabled upon creation or are left disabled. For example, if a Publisher is configured to autoenable created Entities, then all DataWriters created from that Publisher will be automatically enabled.
Note: if an entity is disabled, then all of the child Entities it creates are also created in a disabled state,
regardless of the setting of this QosPolicy. However, enabling a disabled entity will enable all of its children if this QosPolicy is set to autoenable child Entities.

315

6.4.2.1 Example

Note: an entity can only be enabled; it cannot be disabled after its been enabled.
See Example (Section 6.4.2.1 below) for an example of how to set this policy.
There are various reasons why you may want to create Entities in the disabled state:
l

l

l

To get around a “chicken and egg”-type issue. Where you need to have an entity in order to modify
it, but you don’t want the entity to be used by Connext DDS until it has been modified.
For example, if you create a DomainParticipant in the enabled state, it will immediately start sending packets to other nodes trying to discover if other Connext DDS applications exist. However,
you may want to configure the built-in topic reader listener before discovery occurs. To do this, you
need to create a DomainParticipant in the disabled state because once enabled, discovery will
occur. If you set up the built-in topic reader listener after the DomainParticipant is enabled, you
may miss some discovery traffic.
You may want to create Entities without having them automatically start to work. This especially
pertains to DataReaders. If you create a DataReader in an enabled state and you are using
DataReaderListeners, Connext DDS will immediately search for matching DataWriters and callback the listener as soon as data is published. This may not be what you want to happen if your
application is still in the middle of initialization when data arrives.
So typically, you would create all Entities in a disabled state, and then when all parts of the application have been initialized, one would enable all Entities at the same time using the enable() operation on the DomainParticipant, see Enabling DDS Entities (Section 4.1.2 on page 154).
An entity’s existence is not advertised to other participants in the network until the entity is enabled.
Instead of sending an individual declaration packet to other applications announcing the existence of
the entity, Connext DDS can be more efficient in bundling multiple declarations into a single packet
when you enable all Entities at the same time.

See Enabling DDS Entities (Section 4.1.2 on page 154) for more information about enabled/disabled Entities.
6.4.2.1 Example
The code in Figure 6.25 Configuring a Publisher so that New DataWriters are Disabled on the next page
illustrates how to use the ENTITYFACTORY QoS.

316

6.4.2.2 Properties

Figure 6.25 Configuring a Publisher so that New DataWriters are Disabled
DDS_PublisherQos publisher_qos;1
// topic, publisher, writer_listener already created
if (publisher->get_qos(publisher_qos) != DDS_RETCODE_OK) {
// handle error
}
publisher_qos.entity_factory.autoenable_created_entities
= DDS_BOOLEAN_FALSE;
if (publisher->set_qos(publisher_qos) != DDS_RETCODE_OK) {
// handle error
}
// Subsequently created DataWriters are created disabled and
// must be explicitly enabled by the user-code
DDSDataWriter* writer = publisher->create_datawriter(topic,
DDS_DATAWRITER_QOS_DEFAULT, writer_listener, DDS_STATUS_MASK_ALL);
// now do other initialization
// Now explicitly enable the DataWriter, this will allow other
// applications to discover the DataWriter and for this application
// to send data when the DataWriter’s write() method is called
writer->enable();

6.4.2.2 Properties
This QosPolicy can be modified at any time.
It can be set differently on the publishing and subscribing sides.
6.4.2.3 Related QosPolicies
This QosPolicy does not interact with any other policies.
6.4.2.4 Applicable DDS Entities
l

DomainParticipantFactory (Section 8.2 on page 539)

l

DomainParticipants (Section 8.3 on page 547)

l

Publishers (Section 6.2 on page 243)

l

Subscribers (Section 7.2 on page 440)

6.4.2.5 System Resource Considerations
This QosPolicy does not significantly impact the use of system resources.
1Note in C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
317

6.4.3 EXCLUSIVE_AREA QosPolicy (DDS Extension)

6.4.3 EXCLUSIVE_AREA QosPolicy (DDS Extension)
This QosPolicy controls the creation and use of Exclusive Areas. An exclusive area (EA) is a mutex with
built-in deadlock protection when multiple EAs are in use. It is used to provide mutual exclusion among
different threads of execution. Multiple EAs allow greater concurrency among the internal and user threads
when executing Connext DDS code.
EAs allow Connext DDS to be multi-threaded while preventing threads from a classical deadlock scenario
for multi-threaded applications. EAs prevent a DomainParticipant's internal threads from deadlocking
with each other when executing internal code as well as when executing the code of user-registered
listener callbacks.
Within an EA, all calls to the code protected by the EA are single threaded. Each DomainParticipant, Publisher and Subscriber represents a separate EA. All DataWriters of the same Publisher and all DataReaders of the same Subscriber share the EA of its parent. This means that the DataWriters of the same
Publisher and the DataReaders of the same Subscriber are inherently single threaded.
Within an EA, there are limitations on how code protected by a different EA can be accessed. For
example, when data is being processed by user code received in the DataReaderListener of a Subscriber
EA, the user code may call the write() function of a DataWriter that is protected by the EA of its
Publisher. So you can send data in the function called to process received data. However, you cannot create Entities or call functions that are protected by the EA of the DomainParticipant. See Exclusive Areas
(EAs) (Section 4.5 on page 182) for the complete documentation on Exclusive Areas.
With this QoS, you can force a Publisher or Subscriber to share the same EA as its DomainParticipant.
Using this capability, the restriction of not being to create Entities in a DataReaderListener's on_data_
available() callback is lifted. However, the trade-off is that the application has reduced concurrency
through the Entities that share an EA.
Note that the restrictions on calling methods in a different EA only exists for user code that is called in
registered Listeners by internal DomainParticipant threads. User code may call all Connext DDS functions for any Entities from their own threads at any time.
The EXCLUSIVE_AREA includes a single member, as listed in Table 6.20 DDS_ExclusiveAreaQosPolicy. For the default value, please see the API Reference HTML documentation.
Table 6.20 DDS_ExclusiveAreaQosPolicy
Type

Field Name

Description
DDS_BOOLEAN_FALSE:
subordinates will not use the same EA

DDS_Boolean

use_shared_exclusive_area
DDS_BOOLEAN_TRUE:
subordinates will use the same EA

318

6.4.3.1 Example

The implications and restrictions of using a private or shared EA are discussed in Exclusive Areas (EAs)
(Section 4.5 on page 182). The basic trade-off is concurrency versus restrictions on which methods can be
called in user, listener, callback functions. To summarize:
Behavior when the Publisher or Subscriber’s use_shared_exclusive_area is set to FALSE:
l

l

The creation of the Publisher/Subscriber will create an EA that will be used only by the Publisher/Subscriber and the DataWriters/DataReaders that belong to them.
Consequences: This setting maximizes concurrency at the expense of creating a mutex for the Publisher or Subscriber. In addition, using a separate EA may restrict certain Connext DDS operations
(see Operations Allowed within Listener Callbacks (Section 4.4.5 on page 182)) from being called
from the callbacks of Listeners attached to those Entities and the Entities that they create. This limitation results from a built-in deadlock protection mechanism.

Behavior when the Publisher or Subscriber’s use_shared_exclusive_area is set to TRUE:
l

l

The creation of the Publisher/Subscriber does not create a new EA. Instead, the Publisher/Subscriber, along with the DataWriters/DataReaders that they create, will use a common EA
shared with the DomainParticipant.
Consequences: By sharing the same EA among multiple Entities, you may decrease the amount of
concurrency in the application, which can adversely impact performance. However, this setting does
use less resources and allows you to call almost any operation on any Entity within a listener callback (see Exclusive Areas (EAs) (Section 4.5 on page 182) for full details).

6.4.3.1 Example
The code in Figure 6.26 Creating a Publisher with a Shared Exclusive Area on the facing page illustrates
how to change the EXCLUSIVE_AREA policy.

319

6.4.3.2 Properties

Figure 6.26 Creating a Publisher with a Shared Exclusive Area
DDS_PublisherQos publisher_qos;1
// domain, publisher_listener have been previously created
if (participant->get_default_publisher_qos(publisher_qos) !=
DDS_RETCODE_OK) {
// handle error
}
publisher_qos.exclusive_area.use_shared_exclusive_area = DDS_BOOLEAN_TRUE;
DDSPublisher* publisher = participant->create_publisher(publisher_qos,
publisher_listener, DDS_STATUS_MASK_ALL);

6.4.3.2 Properties
This QosPolicy cannot be modified after the Entity has been created.
It can be set differently on the publishing and subscribing sides.
6.4.3.3 Related QosPolicies
This QosPolicy does not interact with any other policies.
6.4.3.4 Applicable DDS Entities
l

Publishers (Section 6.2 on page 243)

l

Subscribers (Section 7.2 on page 440)

6.4.3.5 System Resource Considerations
This QosPolicy affects the use of operating-system mutexes. When use_shared_exclusive_area is
FALSE, the creation of a Publisher or Subscriber will create an operating-system mutex.

6.4.4 GROUP_DATA QosPolicy
This QosPolicy provides an area where your application can store additional information related to the Publisher and Subscriber. This information is passed between applications during discovery (see Discovery
(Section Chapter 14 on page 709)) using built-in-topics (see Built-In Topics (Section Chapter 16 on page
772)). How this information is used will be up to user code. Connext DDS does not do anything with the
information stored as GROUP_DATA except to pass it to other applications.

1Note in C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
320

6.4.4.1 Example

Use cases are often application-to-application identification, authentication, authorization, and encryption
purposes. For example, applications can use this QosPolicy to send security certificates to each other for
RSA-type security.
The value of the GROUP_DATA QosPolicy is sent to remote applications when they are first discovered,
as well as when the Publisher or Subscriber’s set_qos() method is called after changing the value of the
GROUP_DATA. User code can set listeners on the built-in DataReaders of the built-in Topics used by
Connext DDS to propagate discovery information. Methods in the built-in topic listeners will be called
whenever new DomainParticipants, DataReaders, and DataWriters are found. Within the user callback,
you will have access to the GROUP_DATA that was set for the associated Publisher or Subscriber.
Currently, GROUP_DATA of the associated Publisher or Subscriber is only propagated with the information that declares a DataWriter or DataReader. Thus, you will need to access the value of GROUP_
DATA through DDS_PublicationBuiltinTopicData or DDS_SubscriptionBuiltinTopicData (see Built-In
Topics (Section Chapter 16 on page 772)).
The structure for the GROUP_DATA QosPolicy includes just one field, as seen in Table 6.21 DDS_
GroupDataQosPolicy. The field is a sequence of octets that translates to a contiguous buffer of bytes
whose contents and length is set by the user. The maximum size for the data are set in the DOMAIN_
PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593).
Table 6.21 DDS_GroupDataQosPolicy
Type
DDS_OctetSeq

Field Name
value

Description
Empty by default

This policy is similar to the USER_DATA QosPolicy (Section 6.5.26 on page 417) and TOPIC_DATA
QosPolicy (Section 5.2.1 on page 209) that apply to other types of Entities.
6.4.4.1 Example
One possible use of GROUP_DATA is to pass some credential or certificate that your subscriber application can use to accept or reject communication with the DataWriters that belong to the Publisher (or vice
versa, where the publisher application can validate the permission of DataReaders of a Subscriber to
receive its data). The value of the GROUP_DATA of the Publisher is propagated in the ‘group_data’ field
of the DDS_PublicationBuiltinTopicData that is sent with the declaration of each DataWriter. Similarly,
the value of the GROUP_DATA of the Subscriber is propagated in the ‘group_data’ field of the DDS_
SubscriptionBuiltinTopicData that is sent with the declaration of each DataReader.
When Connext DDS discovers a DataWriter/DataReader, the application can be notified of the discovery
of the new entity and retrieve information about the DataWriter/DataReader QoS by reading the
DCPSPublication or DCPSSubscription built-in topics (see Built-In Topics (Section Chapter 16 on page
772)). Your application can then examine the GROUP_DATA field in the built-in Topic and decide

321

6.4.4.2 Properties

whether or not the DataWriter/DataReader should be allowed to communicate with local DataReaders/DataWriters. If communication is not allowed, the application can use the DomainParticipant’s
ignore_publication() or ignore_subscription() operation to reject the newly discovered remote entity as
one with which the application allows Connext DDS to communicate. See Figure 16.2, “Ignoring Publications,” on page 16-12 for an example of how to do this.
The code in Figure 6.27 Creating a Publisher with GROUP_DATA below illustrates how to change the
GROUP_DATA policy.
Figure 6.27 Creating a Publisher with GROUP_DATA
DDS_PublisherQos publisher_qos;1
int i = 0;
// Bytes that will be used for the group data. In this case, 8 bytes
// of some information that is meaningful to the user application
char myGroupData[GROUP_DATA_SIZE] =
{ 0x34, 0xaa, 0xfe, 0x31, 0x7a, 0xf2, 0x34, 0xaa};
// assume domainparticipant and publisher_listener already created
if (participant->get_default_publisher_qos(publisher_qos) !=
DDS_RETCODE_OK) {
// handle error
}
// Must set the size of the sequence first
publisher_qos.group_data.value.maximum(GROUP_DATA_SIZE);
publisher_qos.group_data.value.length(GROUP_DATA_SIZE);
for (i = 0; i < GROUP_DATA_SIZE; i++) {
publisher_qos.group_data.value[i] = myGroupData[i]
}
DDSPublisher* publisher = participant->create_publisher( publisher_qos,
publisher_listener, DDS_STATUS_MASK_ALL);

6.4.4.2 Properties
This QosPolicy can be modified at any time.
It can be set differently on the publishing and subscribing sides.
6.4.4.3 Related QosPolicies
l

TOPIC_DATA QosPolicy (Section 5.2.1 on page 209)

l

USER_DATA QosPolicy (Section 6.5.26 on page 417)

1Note in C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
322

6.4.4.4 Applicable DDS Entities

l

DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
on page 593)

6.4.4.4 Applicable DDS Entities
l

Publishers (Section 6.2 on page 243)

l

Subscribers (Section 7.2 on page 440)

6.4.4.5 System Resource Considerations
The maximum size of the GROUP_DATA is set in the publisher_group_data_max_length and subscriber_group_data_max_length fields of the DOMAIN_PARTICIPANT_RESOURCE_LIMITS
QosPolicy (DDS Extension) (Section 8.5.4 on page 593). Because Connext DDS will allocate memory
based on this value, you should only increase this value if you need to. If your system does not use
GROUP_DATA, then you can set this value to zero to save memory. Setting the value of the GROUP_
DATA QosPolicy to hold data longer than the value set in the [publisher/subscriber]_group_data_
max_length fields will result in failure and an INCONSISTENT_QOS_POLICY return code.
However, should you decide to change the maximum size of GROUP_DATA, you must make certain
that all applications in the DDS domain have changed the value of [publisher/subscriber]_group_data_
max_length to be the same. If two applications have different limits on the size of GROUP DATA, and
one application sets the GROUP_DATA QosPolicy to hold data that is greater than the maximum size set
by another application, then the matching DataWriters and DataReaders of the Publisher and Subscriber
between the two applications will not connect. This is also true for the TOPIC_DATA (TOPIC_DATA
QosPolicy (Section 5.2.1 on page 209)) and USER_DATA (USER_DATA QosPolicy (Section 6.5.26
on page 417)) QosPolicies.

6.4.5 PARTITION QosPolicy
The PARTITION QoS provides another way to control which DataWriters will match—and thus communicate with—which DataReaders. It can be used to prevent DataWriters and DataReaders that would
have otherwise matched with the same Topic and compatible QosPolicies from talking to each other.
Much in the same way that only applications within the same DDS domain will communicate with each
other, only DataWriters and DataReaders that belong to the same partition can talk to each other.
The PARTITION QoS applies to Publishers and Subscribers, therefore the DataWriters and DataReaders
belong to the partitions as set on the Publishers and Subscribers that created them. The mechanism implementing the PARTITION QoS is relatively lightweight, and membership in a partition can be dynamically
changed. Unlike the creation and destruction of DomainParticipants, there is no spawning and killing of
threads or allocation and deallocation of memory when Publishers and Subscribers add or remove themselves from partitions.
The PARTITION QoS consists of a set of partition names that identify the partitions of which the Entity is
a member. These names are simply strings, and DataWriters and DataReaders are considered to be in the

323

6.4.5 PARTITION QosPolicy

same partition if they have at least one partition name in common in the PARTITION QoS set on their
Publishers or Subscribers. By default, Publishers and Subscribers belong to a single partition whose name
is the empty string, ““.
Conceptually each partition name can be thought of as defining a “visibility plane” within the DDS
domain. DataWriters will make their data available on all the visibility planes that correspond to its Publisher’s partition names, and the DataReaders will see the data that is placed on any of the visibility planes
that correspond to its Subscriber’s partition names.
Figure 6.28 Controlling Visibility of Data with the PARTITION QoS below illustrates the concept of
PARTITION QoS. In this figure, all DataWriters and DataReaders belong to the same DDS domain and
refer to the same Topic. DataWriter1 is configured to belong to three partitions: partition_A, partition_B,
and partition_C. DataWriter2 belongs to partition_C and partition_D.
Figure 6.28 Controlling Visibility of Data with the PARTITION QoS

Similarly, DataReader1 is configured to belong to partition_A and partition_B, and DataReader2 belongs
only to partition_C. Given this topology, the data written by DataWriter1 is visible in partitions A, B, and
C. The oval tagged with the number “1” represents one DDS data sample written by DataWriter1.
Similarly, the data written by DataWriter2 is visible in partitions C and D. The oval tagged with the number “2” represents one DDS data sample written by DataWriter2.
The result is that the data written by DataWriter1 will be received by both DataReader1 and
DataReader2, but the data written by DataWriter2 will only be visible by DataReader2.
Publishers and Subscribers always belong to a partition. By default, Publishers and Subscribers belong to
a single partition whose name is the empty string, ““. If you set the PARTITION QoS to be an empty set,
Connext DDS will assign the Publisher or Subscriber to the default partition, ““. Thus, for the example
above, without using the PARTITION QoS, DataReaders 1 and 2 would have received all DDS data
DDS samples written by DataWriters 1 and 2.

324

6.4.5.1 Rules for PARTITION Matching

6.4.5.1 Rules for PARTITION Matching
On the Publisher side, the PARTITION QosPolicy associates a set of strings (partition names) with the
Publisher. On the Subscriber side, the application also uses the PARTITION QoS to associate partition
names with the Subscriber.
Taking into account the PARTITION QoS, a DataWriter will communicate with a DataReader if and
only if the following conditions apply:
1. The DataWriter and DataReader belong to the same DDS domain. That is, their respective
DomainParticipants are bound to the same DDS domain ID (see Creating a DomainParticipant (Section 8.3.1 on page 556)).
2. The DataWriter and DataReader have matching Topics. That is, each is associated with a Topic
with the same topic_name and data type.
3. The QoS offered by the DataWriter is compatible with the QoS requested by the DataReader.
4. The application has not used the ignore_participant(), ignore_datareader(), or ignore_
datawriter() APIs to prevent the association (see Restricting Communication—Ignoring Entities
(Section 16.4 on page 784)).
5. The Publisher to which the DataWriter belongs and the Subscriber to which the DataReader
belongs must have at least one matching partition name.
The last condition reflects the visibility of the data introduced by the PARTITION QoS. Matching partition names is done by string comparison, thus partition names are case sensitive.
Note: Failure to match partitions is not considered an incompatible QoS and does not trigger any listeners
or change any status conditions.
6.4.5.2 Pattern Matching for PARTITION Names
You may also add strings that are regular expressions1 to the PARTITION QosPolicy. A regular expression does not define a set of partitions to which the Publisher or Subscriber belongs, as much as it is used
in the partition matching process to see if a remote entity has a partition name that would be matched with
the regular expression. That is, the regular expressions in the PARTITION QoS of a Publisher are never
matched against those found in the PARTITION QoS of a Subscriber. Regular expressions are always
matched against “concrete” partition names. Thus, a concrete partition name may not contain any reserved
characters that are used to define regular expressions, for example ‘*’, ‘.’, ‘+’, etc.
For more on regular expressions, see SQL Extension: Regular Expression Matching (Section 5.4.6.5 on
page 228).

1As defined by the POSIX fnmatch API (1003.2-1992 section B.6).
325

6.4.5.3 Example

If a PARTITION QoS only contains regular expressions, then the Publisher or Subscriber will be
assigned automatically to the default partition with the empty string name (““). Thus, do not be fooled into
thinking that a PARTITION QoS that only contains the string “*” matches another PARTITION QoS that
only contains the string “*”. Yes, the Publisher will match the Subscriber, but it is because they both
belong to the default ““ partition.
DataWriters and DataReaders are considered to have a partition in common if the sets of partitions that
their associated Publishers and Subscribers have defined have:
At least one concrete partition name in common
A regular expression in one Entity that matches a concrete partition name in another Entity
The programmatic representation of the PARTITION QoS is shown in Table 6.22 DDS_PartitionQosPolicy. The QosPolicy contains the single string sequence, name. Each element in the sequence
can be a concrete name or a regular expression. The Entity will be assigned to the default ““ partition if the
sequence is empty.
Table 6.22 DDS_PartitionQosPolicy
Type

Field Name

Description
Empty by default.

DDS_StringSeq

name
There can be up to 64 names, with a maximum of 256 characters summed across all names.

You can have one long partition string of 256 chars, or multiple shorter strings that add up to 256 or less
characters. For example, you can have one string of 4 chars and one string of 252 chars.
6.4.5.3 Example
Since the set of partitions for a Publisher or Subscriber can be dynamically changed, the Partition
QosPolicy is useful to control which DataWriters can send data to which DataReaders and vice versa—
even if all of the DataWriters and DataReaders are for the same topic. This facility is useful for creating
temporary separation groups among Entities that would otherwise be connected to and exchange data each
other.
Note when using Partitions and Durability: If a Publisher changes partitions after startup, it is possible for a
reliable, late-joining DataReader to receive data that was written for both the original and the new partition. For example, suppose a DataWriter with TRANSIENT_LOCAL Durability initially writes DDS
samples with Partition A, but later changes to Partition B. In this case, a reliable, late-joining DataReader
configured for Partition B will receive whatever DDS samples have been saved for the DataWriter. These
may include DDS samples which were written when the DataWriter was using Partition A.
The code in Figure 6.29 Setting Partition Names on a Publisher on the next page illustrates how to change
the PARTITION policy.

326

6.4.5.3 Example

Figure 6.29 Setting Partition Names on a Publisher
DDS_PublisherQos publisher_qos;1
// domain, publisher_listener have been previously created
if (participant->get_default_publisher_qos(publisher_qos) !=
DDS_RETCODE_OK) {
// handle error
}
// Set the partition QoS
publisher_qos.partition.name.maximum(3);
publisher_qos.partition.name.length(3);
publisher_qos.partition.name[0] = DDS_String_dup(“partition_A”);
publisher_qos.partition.name[1] = DDS_String_dup(“partition_B”);
publisher_qos.partition.name[2] = DDS_String_dup(“partition_C”);
DDSPublisher* publisher = participant->create_publisher(
publisher_qos, publisher_listener, DDS_STATUS_MASK_ALL);

The ability to dynamically control which DataWriters are matched to which DataReaders (of the same
Topic) offered by the PARTITION QoS can be used in many different ways. Using partitions, connectivity can be controlled based on location-based partitioning, access-control groups, purpose, or a combination of these and other application-defined criteria. We will examine some of these options via
concrete examples.
Example of location-based partitions. Assume you have a set of Topics in a traffic management system
such as “TrafficAlert,” “AccidentReport,” and “CongestionStatus.” You may want to control the visibility
of these Topics based on the actual location to which the information applies. You can do this by placing
the Publisher in a partition that represents the area to which the information applies. This can be done
using a string that includes the city, state, and country, such as “USA/California/Santa Clara.” A Subscriber can then choose whether it wants to see the alerts in a single city, the accidents in a set of states, or
the congestion status across the US. Some concrete examples are shown in Table 6.23 Example of Using
Location-Based Partitions.
Table 6.23 Example of Using Location-Based Partitions
Publisher Partitions
Specify a single partition name
using the pattern:

Subscriber Partitions
Specify multiple partition names, one
per region of interest

Result
Limits the visibility of the data to Subscribers that express
interest in the geographical region.

“//”

1Note in C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
327

6.4.5.3 Example

Table 6.23 Example of Using Location-Based Partitions
Publisher Partitions
“USA/California/Santa Clara”

Subscriber Partitions

Result

(Subscriber participant is irrelevant
here.)

Send only information for Santa Clara, California.

“USA/California/Santa Clara”

Receive only information for Santa Clara, California.

“USA/California/Santa Clara”
Receive information for Santa Clara or Sunnyvale, California.
“USA/California/Sunnyvale”
(Publisher partition is irrelevant
here.)

“USA/California/*”
Receive information for California or Nevada.
“USA/Nevada/*”
“USA/California/*”
“USA/Nevada/Reno”

Receive information for California and two cities in Nevada.

“USA/Nevada/Las Vegas”

Example of access-control group partitions. Suppose you have an application where access to the information must be restricted based on reader membership to access-control groups. You can map this groupcontrolled visibility to partitions by naming all the groups (e.g. executives, payroll, financial, general-staff,
consultants, external-people) and assigning the Publisher to the set of partitions that represents which
groups should have access to the information. The Subscribers specify the groups to which they belong,
and the partition-matching behavior will ensure that the information is only distributed to Subscribers
belonging to the appropriate groups. Some concrete examples are shown in Table 6.24 Example of
Access-Control Group Partitions.
Table 6.24 Example of Access-Control Group Partitions
Publisher Partitions

Subscriber Partitions

Result

Specify several partition names, one Specify multiple partition names, one per
per group that is allowed access:
group to which the Subscriber belongs.

Limits the visibility of the data to Subscribers that
belong to the access-groups specified by the Publisher.

“payroll”

Makes information available only to Subscribers that
have access to either financial or payroll information.

(Subscriber participant is irrelevant here.)
“financial”
(Publisher participant is irrelevant
here.)

“executives”
“financial”

Gain access to information that is intended for
executives or people with access to the finances.

A slight variation of this pattern could be used to confine the information based on security levels.
Example of purpose-based partitions: Assume an application containing subsystems that can be used for
multiple purposes, such as training, simulation, and real use. In some occasions it is convenient to be able

328

6.4.5.4 Properties

to dynamically switch the subsystem from operating in the “simulation world” to the “training world” or to
the “real world.” For supervision purposes, it may be convenient to observe multiple worlds, so that you
can compare the each one’s results. This can be accomplished by setting a partition name in the Publisher
that represents the “world” to which it belongs and a set of partition names in the Subscriber that model the
worlds that it can observe.
6.4.5.4 Properties
This QosPolicy can be modified at any time.
Strictly speaking, this QosPolicy does not have request-offered semantics, although it is matched between
DataWriters and DataReaders, and communication is established only if there is a match between partition
names.
6.4.5.5 Related QosPolicies
l

DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
on page 593).

6.4.5.6 Applicable DDS Entities
l

Publishers (Section 6.2 on page 243)

l

Subscribers (Section 7.2 on page 440)

6.4.5.7 System Resource Considerations
Partition names are propagated along with the declarations of the DataReaders and the DataWriters and
can be examined by user code through built-in topics (see Built-In Topics (Section Chapter 16 on page
772)). Thus the sum-total length of the partition names will impact the bandwidth needed to transmit those
declarations, as well as the memory used to store them.
The maximum number of partitions and the maximum number of characters that can be used for the sumtotal length of all partition names are configured using the max_partitions and max_partition_cumulative_
characters fields of the DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension)
(Section 8.5.4 on page 593). Setting more partitions or using longer names than allowed by those limits
will result in failure and an INCONSISTENT_QOS_POLICY return code.
However, should you decide to change the maximum number of partitions or maximum cumulative length
of partition names, then you must make certain that all applications in the DDS domain have changed the
values of max_partitions and max_partition_cumulative_characters to be the same. If two applications
have different values for those settings, and one application sets the PARTITION QosPolicy to hold more
partitions or longer names than set by another application, then the matching DataWriters and DataReaders of the Publisher and Subscriber between the two applications will not connect. This similar to the
restrictions for the GROUP_DATA (GROUP_DATA QosPolicy (Section 6.4.4 on page 320)), USER_

329

6.4.6 PRESENTATION QosPolicy

DATA (USER_DATA QosPolicy (Section 6.5.26 on page 417)), and TOPIC_DATA (TOPIC_DATA
QosPolicy (Section 5.2.1 on page 209)) QosPolicies.

6.4.6 PRESENTATION QosPolicy
Usually DataReaders will receive data in the order that it was sent by a DataWriter. In addition, data is
presented to the DataReader as soon as the application receives the next value expected.
Sometimes, you may want a set of data for the same DataWriter to be presented to the receiving
DataReader only after ALL the elements of the set have been received, but not before. You may also
want the data to be presented in a different order than it was received. Specifically, for keyed data, you
may want Connext DDS to present the data in keyed or instance order.
The Presentation QosPolicy allows you to specify different scopes of presentation: within a DataWriter,
across instances of a DataWriter, and even across different DataWriters of a publisher. It also controls
whether or not a set of changes within the scope must be delivered at the same time or delivered as soon as
each element is received.
There are three components to this QoS, the boolean flag coherent_access, the boolean flag ordered_
access, and an enumerated setting for the access_scope. The structure used is shown in Table 6.25 DDS_
PresentationQosPolicy.
Table 6.25 DDS_PresentationQosPolicy
Type

Field
Name

Description
Controls the granularity used when coherent_access and/or ordered_access are TRUE.
If both coherent_access and ordered_access are FALSE, access_scope’s setting has no effect.
l

DDS_INSTANCE_PRESENTATION_QOS:
Queue is ordered/sorted per instance
l

DDS_Presentation_
access_
QosPolicyAccessScopeKind scope

DDS_TOPIC_PRESENTATION_QOS:
Queue is ordered/sorted per topic (across all instances)
l

DDS_GROUP_PRESENTATION_QOS:
Queue is ordered/sorted per topic across all instances belonging to DataWriter (or
DataReaders) within the same Publisher (or Subscriber). Not supported for coherent_
access = TRUE.
l

DDS_HIGHEST_OFFERED_PRESENTATION_QOS: Only applies to Subscribers.
With this setting, the Subscriber will use the access scope specified by each remote
Publisher.

330

6.4.6.1 Coherent Access

Table 6.25 DDS_PresentationQosPolicy
Type

Field
Name

Description
Controls whether Connext DDS will preserve the groupings of changes made by the
publishing application by means of begin_coherent_changes() and end_coherent_changes().
l

DDS_Boolean

DDS_BOOLEAN_FALSE:Coherency is not preserved. The value of access_scope is
ignored.

coherent_
access
l

DDS_BOOLEAN_TRUE:Changes made to instances within each DataWriter will be
available to the DataReader as a coherent set, based on the value of access_scope. Not
supported for access_scope = GROUP.
Controls whether Connext DDS will preserve the order of changes.
l

DDS_Boolean

DDS_BOOLEAN_FALSE:The order of DDS samples is only preserved for each
instance, not across instances. The value of access_scope is ignored.

ordered_
access
l

DDS_BOOLEAN_TRUE:The order of DDS samples from a DataWriter is preserved,
based on the value set in access_scope.

6.4.6.1 Coherent Access
A 'coherent set' is a set of DDS data-sample modifications that must be propagated in such a way that they
are interpreted at the receiver's side as a consistent set; that is, the receiver will only be able to access the
data after all the modifications in the set are available at the subscribing end.
Coherency enables a publishing application to change the value of several data-instances and have those
changes be seen atomically (as a cohesive set) by the readers.
Setting coherent_access to TRUE only behaves as described in the DDS specification when the
DataWriter and DataReader are configured for reliable delivery. Non-reliable DataReaders will never
receive DDS samples that belong to a coherent set.
To send a coherent set of DDS data samples, the publishing application uses the Publisher’s begin_coherent_changes() and end_coherent_changes() operations (see Writing Coherent Sets of DDS Data
Samples (Section 6.3.10 on page 287)).
l

l

If coherent_access is TRUE, then the access_scope controls the maximum extent of the coherent
changes, as follows:
If access_scope is INSTANCE, the use of begin_coherent_changes() and end_coherent_changes
() has no effect on how the subscriber can access the data. This is because, with the scope limited to

331

6.4.6.2 Ordered Access

each instance, changes to separate instances are considered independent and thus cannot be grouped
by a coherent change.
l

If access_scope is TOPIC, then coherent changes (indicated by their enclosure within calls to
begin_coherent_changes() and end_coherent_changes()) will be made available as such to each
remote DataReader independently. That is, changes made to instances within the each individual
DataWriter will be available as a coherent set with respect to other changes to instances in that same
DataWriter, but will not be grouped with changes made to instances belonging to a different
DataWriter.

If access_scope is GROUP, coherent changes made to instances through a DataWriter attached to a common Publisher are made available as a unit to remote subscribers. Coherent access with GROUP access
scope is currently not supported.
6.4.6.2 Ordered Access
If ordered_access is TRUE, then access_scope controls the scope of the order in which DDS samples are
presented to the subscribing application, as follows:
l

l

l

l

If access_scope is INSTANCE, the relative order of DDS samples sent by a DataWriter is only preserved on an per-instance basis. If two DDS samples refer to the same instance (identified by Topic
and a particular value for the key) then the order in which they are stored in the DataReader’s queue
is consistent with the order in which the changes occurred. However, if the two DDS samples
belong to different instances, the order in which they are presented may or may not match the order
in which the changes occurred.
If access_scope is TOPIC, the relative order of DDS samples sent by a DataWriter is preserved for
all DDS samples of all instances. The coherent grouping and/or order in which DDS samples appear
in the DataReader’s queue is consistent with the grouping/order in which the changes occurred—
even if the DDS samples affect different instances.
If access_scope is GROUP, the scope spans all instances belonging to DataWriters within the same
Publisher—even if they are instances of different topics. Changes made to instances via DataWriters
attached to the same Publisher are made available to Subscribers on the same order they occurred.
If access_scope is HIGHEST_OFFERED, the Subscriber will use the access scope specified by
each remote Publisher.

The data stored in the DataReader is accessed by the DataReader’s read()/take() APIs. The application
does not have to access the DDS data samples in the same order as they are stored in the queue. How the
application actually gets the data from the DataReader is ultimately under the control of the user code, see
Using DataReaders to Access Data (Read & Take) (Section 7.4 on page 491).

332

6.4.6.3 Example

6.4.6.3 Example
Coherency is useful in cases where the values are inter-related (for example, if there are two data-instances
representing the altitude and velocity vector of the same aircraft and both are changed, it may be useful to
communicate those values in a way the reader can see both together; otherwise, it may e.g., erroneously
interpret that the aircraft is on a collision course).
Ordered access is useful when you need to ensure that DDS samples appear on the DataReader’s queue in
the order sent by one or multiple DataWriters within the same Publisher.
To illustrate the effect of the PRESENTATION QosPolicy with TOPIC and INSTANCE access scope,
assume the following sequence of DDS samples was written by the DataWriter: {A1, B1, C1, A2, B2,
C2}. In this example, A, B, and C represent different instances (i.e., different keys). Assume all of these
DDS samples have been propagated to the DataReader’s history queue before your application invokes
the read() operation. The DDS data-sample sequence returned depends on how the PRESENTATION
QoS is set, as shown in Table 6.26 Effect of ordered_access for access_scope INSTANCE and TOPIC.
Table 6.26 Effect of ordered_access for access_scope INSTANCE and TOPIC
Sequence retrieved via “read()”.
PRESENTATION QoS

Order sent was {A1, B1, C1, A2, B2, C2}
Order received was {A1, A2, B1, B2, C1, C2}

ordered_access = FALSE

access_scope = 

{A1, A2, B1, B2, C1, C2}

ordered_access = TRUE

access_scope = INSTANCE

{A1, A2, B1, B2, C1, C2}

ordered_access = TRUE

access_scope = TOPIC

{A1, B1, C1, A2, B2, C2}

To illustrate the effect of a PRESENTATION QosPolicy with GROUP access_scope, assume the following sequence of DDS samples was written by two DataWriters, W1 and W2, within the same Publisher: {(W1,A1), (W2,B1), (W1,C1), (W2,A2), (W1,B2), (W2,C2)}. As in the previous example, A, B,
and C represent different instances (i.e., different keys). With access_scope set to INSTANCE or TOPIC,
the middleware cannot guarantee that the application will receive the DDS samples in the same order they
were published by W1 and W2. With access_scope set to GROUP, the middleware is able to provide the
DDS samples in order to the application as long as the read()/take() operations are invoked within a
begin_access()/end_access() block (see Beginning and Ending Group-Ordered Access (Section 7.2.5 on
page 453)).

333

6.4.6.4 Properties

Table 6.27 Effect of ordered_access for access_scope GROUP
Sequence retrieved via “read()”.
PRESENTATION QoS
Order sent was {(W1,A1), (W2,B1), (W1,C1), (W2,A2), (W1,B2), (W2,C2)}
ordered_access = FALSE
or
access_scope = TOPIC or
INSTANCE

The order across DataWriters will not be preserved. DDS samples may be delivered in multiple
orders. For example:
{(W1,A1), (W1,C1), (W1,B2), (W2,B1), (W2,A2), (W2,C2)}
{(W1,A1), (W2,B1), (W1,B2), (W1,C1), (W2,A2), (W2,C2)}

ordered_access = TRUE

DDS samples are delivered in the same order they were published:

access_scope = GROUP

{(W1,A1), (W2,B1), (W1,C1), (W2,A2), (W1,B2), (W2,C2)}

6.4.6.4 Properties
This QosPolicy cannot be modified after the Publisher or Subscriber is enabled.
This QoS must be set compatibly between the DataWriter’s Publisher and the DataReader’s Subscriber.
The compatible combinations are shown in Table 6.28 Valid Combinations of ordered_access and access_
scope, with Subscriber’s ordered_access = False and Table 6.29 Valid Combinations of ordered_access
and access_scope, with Subscriber’s ordered_access = True for ordered_access and Table 6.30 Valid
Combinations of Presentation Coherent Access and Access Scope for coherent_access.
Table 6.28 Valid Combinations of ordered_access and access_scope, with Subscriber’s
ordered_access = False
Subscriber Requests:
{ordered_access/access_scope}
False/Instance

False/Topic

False/Group

False/Highest

False/Instance

4

incompatible

incompatible

4

False/Topic

4

4

incompatible

4

False/Group

4

4

4

4

True/Instance

4

incompatible

incompatible

4

True/Topic

4

4

incompatible

4

True/Group

4

4

4

4

Publisher offers:

334

6.4.6.5 Related QosPolicies

Table 6.29 Valid Combinations of ordered_access and access_scope, with Subscriber’s
ordered_access = True
Subscriber Requests:
{ordered_access/access_scope}
True/Instance

True/Topic

True/Group

True/Highest

False/Instance

incompatible

incompatible

incompatible

incompatible

False/Topic

incompatible

incompatible

incompatible

incompatible

False/Group

incompatible

incompatible

incompatible

incompatible

True/Instance

4

incompatible

incompatible

4

True/Topic

4

4

incompatible

4

True/Group

4

4

4

4

Publisher offers:

Table 6.30 Valid Combinations of Presentation Coherent Access and Access Scope
Subscriber requests:
{coherent_access/access_scope}
False/Instance

False/Topic

True/Instance

True/Topic

False/Instance

4

incompatible

incompatible

incompatible

False/Topic

4

4

incompatible

incompatible

True/Instance

4

incompatible

4

incompatible

True/Topic

4

4

4

4

Publisher offers:

6.4.6.5 Related QosPolicies
l

l

The DESTINATION_ORDER QosPolicy (Section 6.5.6 on page 365) is closely related and also
affects the ordering of DDS data samples on a per-instance basis when there are multiple
DataWriters.
The DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511)
may be used to configure the DDS sample ordering process in the Subscribers configured with
GROUP or HIGHEST_OFFERED access_scope.

335

6.4.6.6 Applicable DDS Entities

6.4.6.6 Applicable DDS Entities
l

Publishers (Section 6.2 on page 243)

l

Subscribers (Section 7.2 on page 440)

6.4.6.7 System Resource Considerations
The use of this policy does not significantly impact the usage of resources.

6.5 DataWriter QosPolicies
This section provides detailed information about the QosPolicies associated with a DataWriter. Table 6.17
DataWriter QosPolicies provides a quick reference. They are presented here in alphabetical order.
l

AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on the next page)

l

BATCH QosPolicy (DDS Extension) (Section 6.5.2 on page 341)

l

DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347)

l

DATA_WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 6.5.4 on page
359)

l

DEADLINE QosPolicy (Section 6.5.5 on page 363)

l

DESTINATION_ORDER QosPolicy (Section 6.5.6 on page 365)

l

DURABILITY QosPolicy (Section 6.5.7 on page 368)

l

DURABILITY SERVICE QosPolicy (Section 6.5.8 on page 372)

l

ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9 on page 374)

l

HISTORY QosPolicy (Section 6.5.10 on page 376)

l

LATENCYBUDGET QoS Policy (Section 6.5.11 on page 380)

l

LIFESPAN QoS Policy (Section 6.5.12 on page 381)

l

LIVELINESS QosPolicy (Section 6.5.13 on page 382)

l

MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386)

l

OWNERSHIP QosPolicy (Section 6.5.15 on page 389)

l

OWNERSHIP_STRENGTH QosPolicy (Section 6.5.16 on page 393)

l

PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394)

l

PUBLISH_MODE QosPolicy (DDS Extension) (Section 6.5.18 on page 397)

l

RELIABILITY QosPolicy (Section 6.5.19 on page 400)

l

RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)

336

6.5.1 AVAILABILITY QosPolicy (DDS Extension)

l

SERVICE QosPolicy (DDS Extension) (Section 6.5.21 on page 408)

l

TRANSPORT_PRIORITY QosPolicy (Section 6.5.22 on page 409)

l

TRANSPORT_SELECTION QosPolicy (DDS Extension) (Section 6.5.23 on page 411)

l

TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412)

l

TYPESUPPORT QosPolicy (DDS Extension) (Section 6.5.25 on page 416)

l

USER_DATA QosPolicy (Section 6.5.26 on page 417)

l

WRITER_DATA_LIFECYCLE QoS Policy (Section 6.5.27 on page 419)

6.5.1 AVAILABILITY QosPolicy (DDS Extension)
This QoS policy configures the availability of data and it is used in the context of two features:
l

l

Collaborative DataWriters (Availability QoS Policy and Collaborative DataWriters (Section 6.5.1.1
on the facing page))
Required Subscriptions (Availability QoS Policy and Required Subscriptions (Section 6.5.1.2 on
page 339))

It contains the members listed in Table 6.31 DDS_AvailabilityQosPolicy.
Table 6.31 DDS_AvailabilityQosPolicy
Type

Field Name

Description
Enables support for required subscriptions in a DataWriter.

DDS_Boolean

enable_required_
subscriptions

max_data_
struct
availability_
DDS_Duration_t
waiting_time

For Collaborative DataWriters: Not applicable.
For Required Subscriptions: See Table 6.34 Configuring Required Subscriptions with DDS_
AvailabilityQosPolicy.
Defines how much time to wait before delivering a DDS sample to the application without having
received some of the previous DDS samples.
For Collaborative DataWriters: See Table 6.33 Configuring Collaborative DataWriters with
DDS_AvailabilityQosPolicy.
For Required Subscriptions: Not applicable.

max_endpoint_
struct
availability_
DDS_Duration_t
waiting_time

Defines how much time to wait to discover DataWriters providing DDS samples for the same
data source.
For Collaborative DataWriters: See Table 6.33 Configuring Collaborative DataWriters with
DDS_AvailabilityQosPolicy.
For Required Subscriptions: Not applicable.

337

6.5.1.1 Availability QoS Policy and Collaborative DataWriters

Table 6.31 DDS_AvailabilityQosPolicy
Type

Field Name

Description
A sequence of endpoint groups, described in Table 6.32 struct DDS_EndpointGroup_t.

struct
DDS_EndpointGroupSeq

required_matched_
endpoint_groups

For Collaborative DataWriters: See Table 6.33 Configuring Collaborative DataWriters with
DDS_AvailabilityQosPolicy.
For Required Subscriptions: See Table 6.34 Configuring Required Subscriptions with DDS_
AvailabilityQosPolicy

Table 6.32 struct DDS_EndpointGroup_t
Type

char *

Field
Name
role_
name

Description
Defines the role name of the endpoint group.
If used in the AvailabilityQosPolicy on a DataWriter, it specifies the name that identifies a Required Subscription.
Defines the minimum number of members that satisfies the endpoint group.

quorum_ If used in the AvailabilityQosPolicy on a DataWriter, it specifies the number of DataReaders with a specific role name
count
that must acknowledge a DDS sample before the DDS sample is considered to be acknowledged by the Required
Subscription.

int

6.5.1.1 Availability QoS Policy and Collaborative DataWriters
The Collaborative DataWriters feature allows you to have multiple DataWriters publishing DDS samples
from a common logical data source. The DataReaders will combine the DDS samples coming from the
DataWriters in order to reconstruct the correct order at the source. The Availability QosPolicy allows you
to configure the DDS sample combination (synchronization) process in the DataReader.
Each DDS sample published in a DDS domain for a given logical data source is uniquely identified by a
pair (virtual GUID, virtual sequence number). DDS samples from the same data source (same virtual
GUID) can be published by different DataWriters.
A DataReader will deliver a DDS sample (VGUIDn, VSNm) to the application if one of the following
conditions is satisfied:
l
l

(GUIDn, SNm-1) has already been delivered to the application.
All the known DataWriters publishing VGUIDn have announced that they do not have (VGUIDn,
VSNm-1).

338

6.5.1.2 Availability QoS Policy and Required Subscriptions

l

None of the known DataWriters publishing VGUIDn have announced potential availability of
(VGUIDn, VSNm-1) and both timeouts in this QoS policy have expired.

A DataWriter announces potential availability of DDS samples by using virtual heartbeats. The frequency
at which virtual heartbeats are sent is controlled by the protocol parameters virtual_heartbeat_period (Section on page 350) and samples_per_virtual_heartbeat (Section on page 350) (see Table 6.37 DDS_
RtpsReliableWriterProtocol_t).
Table 6.33 Configuring Collaborative DataWriters with DDS_AvailabilityQosPolicy describes the fields
of this policy when used for a Collaborative DataWriter.
For further information, see Collaborative DataWriters (Section Chapter 11 on page 670).
Table 6.33 Configuring Collaborative DataWriters with DDS_AvailabilityQosPolicy
Field
Name

Description for Collaborative DataWriters
Defines how much time to wait before delivering a DDS sample to the application without having received some of the
previous DDS samples.

max_data_ A DDS sample identified by (VGUIDn, VSNm) will be delivered to the application if this timeout expires for the DDS
availability_ sample and the following two conditions are satisfied:
waiting_
None of the known DataWriters publishing VGUIDn have announced potential availability of (VGUIDn, VSNm-1).
time
The DataWriters for all the endpoint groups specified in required_matched_endpoint_groups (Section on the previous page)
have been discovered or max_endpoint_availability_waiting_time (Section on the facing page) has expired.
Defines how much time to wait to discover DataWriters providing DDS samples for the same data source.
max_
endpoint_
availability_
waiting_
time

The set of endpoint groups that are required to provide DDS samples for a data source can be configured using required_
matched_endpoint_groups (Section on the previous page).
A non-consecutive DDS sample identified by (GUIDn, SNm) cannot be delivered to the application unless the DataWriters
for all the endpoint groups in required_matched_endpoint_groups (Section on the previous page) are discovered or this
timeout expires.
Specifies the set of endpoint groups that are expected to provide DDS samples for the same data source.

required_
matched_
endpoint_
groups

The quorum count in a group represents the number of DataWriters that must be discovered for that group before the
DataReader is allowed to provide non consecutive DDS samples to the application.
A DataWriter becomes a member of an endpoint group by configuring the role_name in the DataWriter’s ENTITY_
NAME QosPolicy (DDS Extension) (Section 6.5.9 on page 374).
The DataWriters created by RTI Persistence Service have a predefined role_name of ‘PERSISTENCE_SERVICE’. For
other DataWriters, the role_name is not set by default.

6.5.1.2 Availability QoS Policy and Required Subscriptions
In the context of Required Subscriptions, the Availability QosPolicy can be used to configure a set of
required subscriptions on a DataWriter.

339

6.5.1.3 Properties

Required Subscriptions are preconfigured, named subscriptions that may leave and subsequently rejoin the
network from time to time, at the same or different physical locations. Any time a required subscription is
disconnected, any DDS samples that would have been delivered to it are stored for delivery if and when
the subscription rejoins the network.
Table 6.34 Configuring Required Subscriptions with DDS_AvailabilityQosPolicy describes the fields of
this policy when used for a Required Subscription.
For further information, see Required Subscriptions (Section 6.3.13 on page 294).
Table 6.34 Configuring Required Subscriptions with DDS_AvailabilityQosPolicy
Field Name
enable_required_
subscriptions

Description for Required Subscriptions
Enables support for Required Subscriptions in a DataWriter.

max_data_availability_
waiting_time

Not applicable to Required Subscriptions.
max_endpoint_
availability_
waiting_time

A sequence of endpoint groups that specify the Required Subscriptions on a DataWriter.
Each Required Subscription is specified by a name and a quorum count.
required_
matched_
endpoint_groups

The quorum count represents the number of DataReaders that have to acknowledge the DDS sample before it can be
considered fully acknowledged for that Required Subscription.
A DataReader is associated with a Required Subscription by configuring the role_name in the DataReader’s
ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9 on page 374).

6.5.1.3 Properties
For DataWriters, all the members in this QosPolicy can be changed after the DataWriter is created except
for the member enable_required_subscriptions.
For DataReaders, this QosPolicy cannot be changed after the DataReader is created.
There are no compatibility restrictions for how it is set on the publishing and subscribing sides.
6.5.1.4 Related QosPolicies
l

ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9 on page 374)

l

DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4

340

6.5.1.5 Applicable DDS Entities

on page 593)
l

DURABILITY QosPolicy (Section 6.5.7 on page 368)

6.5.1.5 Applicable DDS Entities
l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.1.6 System Resource Considerations
The resource limits for the endpoint groups in required_matched_endpoint_groups are determined by
two values in the DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593):
l

max_endpoint_groups

l

max_endpoint_group_cumulative_characters

The maximum number of virtual writers (identified by a virtual GUID) that can be managed by a
DataReader is determined by the max_remote_virtual_writers in DATA_READER_RESOURCE_
LIMITS QosPolicy (DDS Extension) (Section 7.6.2 on page 517). When the Subscriber’s access_scope
is GROUP, max_remote_virtual_writers determines the maximum number of DataWriter groups supported by the Subscriber. Since the Subscriber may contain more than one DataReader, only the setting of
the first applies.

6.5.2 BATCH QosPolicy (DDS Extension)
This QosPolicy can be used to decrease the amount of communication overhead associated with the transmission and (in the case of reliable communication) acknowledgement of small DDS samples, in order to
increase throughput.
It specifies and configures the mechanism that allows Connext DDS to collect multiple user data DDS
samples to be sent in a single network packet, to take advantage of the efficiency of sending larger packets
and thus increase effective throughput.
This QosPolicy can be used to increase effective throughput dramatically for small data DDS samples.
Throughput for small DDS samples (size < 2048 bytes) is typically limited by CPU capacity and not by
network bandwidth. Batching many smaller DDS samples to be sent in a single large packet will increase
network utilization and thus throughput in terms of DDS samples per second.
It contains the members listed in Table 6.35 DDS_BatchQosPolicy.

341

6.5.2 BATCH QosPolicy (DDS Extension)

Table 6.35 DDS_BatchQosPolicy
Field
Name

Type

Description

DDS_
Boolean

enable

DDS_Long

max_data_
Before or when this limit is reached, the batch is automatically flushed.
bytes

Enables/disables batching.
Sets the maximum cumulative length of all serialized DDS samples in a batch.

The size does not include the meta-data associated with the batch DDS samples.
max_
samples

DDS_Long

struct DDS_
Duration_t

max_
flush_
delay

Sets the maximum number of DDS samples in a batch.
When this limit is reached, the batch is automatically flushed.
Sets the maximum flush delay.
When this duration is reached, the batch is automatically flushed.
The delay is measured from the time the first DDS sample in the batch is written by the application.
Sets the batch source timestamp resolution.
The value of this field determines how the source timestamp is associated with the DDS samples in a batch.
A DDS sample written with timestamp 't' inherits the source timestamp 't2' associated with the previous DDS
sample, unless ('t' - 't2') is greater than source_timestamp_resolution.

struct DDS_
Duration_t

source_
timestamp_ If source_timestamp_resolution is DURATION_INFINITE, every DDS sample in the batch will share the
resolution source timestamp associated with the first DDS sample.
If source_timestamp_resolution is zero, every DDS sample in the batch will contain its own source
timestamp corresponding to the moment when the DDS sample was written.
The performance of the batching process is better when source_timestamp_resolution is set to
DURATION_INFINITE.
Determines whether or not the write operation is thread-safe.

DDS_
Boolean

thread_
safe_write

If TRUE, multiple threads can call write on the DataWriter concurrently.
A setting of FALSE can be used to increase batching throughput for batches with many small DDS samples.

If batching is enabled (not the default), DDS samples are not immediately sent when they are written.
Instead, they get collected into a "batch." A batch always contains whole number of DDS samples—a
DDS sample will never be fragmented into multiple batches.
A batch is sent on the network ("flushed") when one of the following things happens:
l

User-configurable flushing conditions
l A batch size limit (max_data_bytes) is reached.
l

A number of DDS samples are in the batch (max_samples).

342

6.5.2.1 Synchronous and Asynchronous Flushing

l

l

l

A time-limit (max_flush_delay) is reached, as measured from the time the first DDS sample
in the batch is written by the application.
The application explicitly calls a DataWriter's flush() operation.

Non-user configurable flushing conditions:
l A coherent set starts or ends.
l

The number of DDS samples in the batch is equal to max_samples in RESOURCE_LIMITS
for unkeyed topics or max_samples_per_instance in RESOURCE_LIMITS for keyed topics.

Additional batching configuration takes place in the Publisher’s ASYNCHRONOUS_PUBLISHER
QosPolicy (DDS Extension) (Section 6.4.1 on page 313).
The flush() operation is described in Flushing Batches of DDS Data Samples (Section 6.3.9 on page
287).
6.5.2.1 Synchronous and Asynchronous Flushing
Usually, a batch is flushed synchronously:
l

l

l

l

l

When a batch reaches its application-defined size limit (max_data_bytes or max_samples) because
the application called write(), the batch is flushed immediately in the context of the writing thread.
When an application manually flushes a batch, the batch is flushed immediately in the context of the
calling thread.
When the first DDS sample in a coherent set is written, the batch in progress (without including the
DDS sample in the coherent set) is immediately flushed in the context of the writing thread.
When a coherent set ends, the batch in progress is immediately flushed in the context of the calling
thread.
When the number of DDS samples in a batch is equal to max_samples in RESOURCE_LIMITS
for unkeyed topics or max_samples_per_instance in RESOURCE_LIMITS for keyed topics, the
batch is flushed immediately in the context of the writing thread.

However, some behavior is asynchronous:
l

To flush batches based on a time limit (max_flush_delay), enable asynchronous batch flushing in
the ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension) (Section 6.4.1 on page 313)
of the DataWriter's Publisher. This will cause the Publisher to create an additional thread that will
be used to flush batches of that Publisher's DataWriters. This behavior is analogous to the way asynchronous publishing works.

343

6.5.2.2 Batching vs. Coalescing

l

You may also use batching alongside asynchronous publication with FlowControllers (DDS Extension) (Section 6.6 on page 422). These features are independent of one another. Flushing a batch on
an asynchronous DataWriter makes it available for sending to the DataWriter's FlowController.
From the point of view of the FlowController, a batch is treated like one large DDS sample.

6.5.2.2 Batching vs. Coalescing
Even when batching is disabled, Connext DDS will sometimes coalesce multiple DDS samples into a
single network datagram. For example, DDS samples buffered by a FlowController or sent in response to
a negative acknowledgement (NACK) may be coalesced. This behavior is distinct from DDS sample
batching.
DDS samples that are sent individually (not part of a batch) are always treated as separate DDS samples
by Connext DDS. Each DDS sample is accompanied by a complete RTPS header on the network
(although DDS samples may share UDP and IP headers) and (in the case of reliable communication) a
unique physical sequence number that must be positively or negatively acknowledged.
In contrast, batched DDS samples share an RTPS header and an entire batch is acknowledged —positively or negatively—as a unit, potentially reducing the amount of meta-traffic on the network and the
amount of processing per individual DDS sample.
Batching can also improve latency relative to simply coalescing. Consider two use cases:
1. A DataWriter is configured to write asynchronously with a FlowController. Even if the FlowController's rules would allow it to publish a new DDS sample immediately, the send will always happen in the context of the asynchronous publishing thread. This context switch can add latency to the
send path.
2. A DataWriter is configured to write synchronously but with batching turned on. When the batch is
full, it will be sent on the wire immediately, eliminating a thread context switch from the send path.
6.5.2.3 Batching and ContentFilteredTopics
When batching is enabled, content filtering is always done on the reader side.
6.5.2.4 Turbo Mode: Automatically Adjusting the Number of Bytes in a Batch—Experimental
Feature
Turbo Mode is an experimental feature that uses an intelligent algorithm that automatically adjusts the number of bytes in a batch at run time according to current system conditions, such as write speed (or write frequency) and DDS sample size. This intelligence is what gives it the ability to increase throughput at high
message rates and avoid negatively impacting message latency at low message rates.
To enable Turbo mode, set the DataWriter's property dds.data_writer.enable_turbo_mode to true.
Turbo mode is not enabled by default.

344

6.5.2.5 Performance Considerations

Note: If you explicitly enable batching by setting enable to TRUE in BatchQosPolicy, the value of the
turbo mode property is ignored and turbo mode is not used.
6.5.2.5 Performance Considerations
The purpose of batching is to increase throughput when writing small DDS samples at a high rate. In such
cases, throughput can be increased several-fold, approaching much more closely the physical limitations of
the underlying network transport.
However, collecting DDS samples into a batch implies that they are not sent on the network immediately
when the application writes them; this can potentially increase latency. However, if the application sends
data faster than the network can support, an increased proportion of the network's available bandwidth will
be spent on acknowledgements and DDS sample resends. In this case, reducing that overhead by turning
on batching could decrease latency while increasing throughput.
As a general rule, to improve batching throughput:
l

l

Set thread_safe_write to FALSE when the batch contains a big number of small DDS samples. If
you do not use a thread-safe write configuration, asynchronous batch flushing must be disabled.
Set source_timestamp_resolution to DURATION_INFINITE. Note that you set this value, every
DDS sample in the batch will share the same source timestamp.

Batching affects how often piggyback heartbeats are sent; see heartbeats_per_max_samples in Table
6.37 DDS_RtpsReliableWriterProtocol_t.
6.5.2.6 Maximum Transport Datagram Size
Batches cannot be fragmented. As a result, the maximum batch size (max_data_bytes) must be set no larger than the maximum transport datagram size. For example, a UDP datagram is limited to 64 KB, so any
batches sent over UDP must be less than or equal to that size.
6.5.2.7 Properties
This QosPolicy cannot be modified after the DataWriter is enabled.
Since it is only for DataWriters, there are no compatibility restrictions for how it is set on the publishing
and subscribing sides.
All batching configuration occurs on the publishing side. A subscribing application does not configure anything specific to receive batched DDS samples, and in many cases, it will be oblivious to whether the DDS
samples it processes were received individually or as part of a batch.
Consistency rules:

345

6.5.2.8 Related QosPolicies

l

l

l

max_samples must be consistent with max_data_bytes: they cannot both be set to LENGTH_
UNLIMITED.
If max_flush_delay is not DURATION_INFINITE, disable_asynchronous_batch in the
ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension) (Section 6.4.1 on page 313)
must be FALSE.
If thread_safe_write is FALSE, source_timestamp_resolution must be DURATION_INFINITE.

6.5.2.8 Related QosPolicies
To flush batches based on a time limit, enable batching in the ASYNCHRONOUS_PUBLISHER
QosPolicy (DDS Extension) (Section 6.4.1 on page 313) of the DataWriter's Publisher.
Be careful when configuring a DataWriter's LIFESPAN QoS Policy (Section 6.5.12 on page 381) with a
duration shorter than the batch flush period (max_flush_delay). If the batch does not fill up before the
flush period elapses, the short duration will cause the DDS samples to be lost without being sent.
Do not configure the DataReader’s or DataWriter’s HISTORY QosPolicy (Section 6.5.10 on page 376)
to be shallower than the DataWriter's maximum batch size (max_samples). When the HISTORY
QosPolicy is shallower on the DataWriter, some DDS samples may not be sent. When the HISTORY
QosPolicy is shallower on the DataReader, DDS samples may be dropped before being provided to the
application.
The initial and maximum numbers of batches that a DataWriter will manage is set in the DATA_
WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 6.5.4 on page 359).
The maximum number of DDS samples that a DataWriter can store is determined by the value max_
samples in the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) and max_batches in the
DATA_WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 6.5.4 on page 359). The
limit that is reached first is applied.
The amount of resources required for batching depends on the configuration of the RESOURCE_LIMITS
QosPolicy (Section 6.5.20 on page 405) and the DATA_WRITER_RESOURCE_LIMITS QosPolicy
(DDS Extension) (Section 6.5.4 on page 359). See System Resource Considerations (Section 6.5.2.10
below).
6.5.2.9 Applicable DDS Entities
l

DataWriters (Section 6.3 on page 261)

6.5.2.10 System Resource Considerations
l

Batching requires additional resources to store the meta-data associated with the DDS samples in the
batch.

346

6.5.3 DATA_WRITER_PROTOCOL QosPolicy (DDS Extension)

l

l

For unkeyed topics, the meta-data will be at least 8 bytes, with a maximum of 20 bytes.

l

For keyed topics, the meta-data will be at least 8 bytes, with a maximum of 52 bytes.

Other resource considerations are described in Related QosPolicies (Section 6.5.2.8 on the previous
page).

6.5.3 DATA_WRITER_PROTOCOL QosPolicy (DDS Extension)
Connext DDS uses a standard protocol for packet (user and meta data) exchange between applications.
The DataWriterProtocol QosPolicy gives you control over configurable portions of the protocol, including
the configuration of the reliable data delivery mechanism of the protocol on a per DataWriter basis.
These configuration parameters control timing and timeouts, and give you the ability to trade off between
speed of data loss detection and repair, versus network and CPU bandwidth used to maintain reliability.
It is important to tune the reliability protocol on a per DataWriter basis to meet the requirements of the enduser application so that data can be sent between DataWriters and DataReaders in an efficient and optimal
manner in the presence of data loss. You can also use this QosPolicy to control how Connext DDS
responds to "slow" reliable DataReaders or ones that disconnect or are otherwise lost.
This policy includes the members presented in Table 6.36 DDS_DataWriterProtocolQosPolicy and Table
6.37 DDS_RtpsReliableWriterProtocol_t. For defaults and valid ranges, please refer to the API Reference
HTML documentation.
For details on the reliability protocol used by Connext DDS, see Reliable Communications (Section
Chapter 10 on page 629). See the RELIABILITY QosPolicy (Section 6.5.19 on page 400) for more
information on per-DataReader/DataWriter reliability configuration. The HISTORY QosPolicy (Section
6.5.10 on page 376) and RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) also play
important roles in the DDS reliability protocol.

347

6.5.3 DATA_WRITER_PROTOCOL QosPolicy (DDS Extension)

Table 6.36 DDS_DataWriterProtocolQosPolicy
Type

Field
Name

Description
The virtual GUID (Global Unique Identifier) is used to uniquely identify the same DataWriter across
multiple incarnations. In other words, this value allows Connext DDS to remember information about a
DataWriter that may be deleted and then recreated.

Connext DDS uses the virtual GUID to associate a durable writer history to a DataWriter.
DDS_GUID_t

virtual_
guid

Persistence Service1 uses the virtual GUID to send DDS samples on behalf of the original DataWriter.
A DataReader persists its state based on the virtual GUIDs of matching remote DataWriters.
For more information, see Durability and Persistence Based on Virtual GUIDs (Section 12.2 on page 680).
By default, Connext DDS will assign a virtual GUID automatically. If you want to restore the state of the
durable writer history after a restart, you can retrieve the value of the writer's virtual GUID using the
DataWriter’s get_qos() operation, and set the virtual GUID of the restarted DataWriter to the same value.
Determines the DataWriter’s RTPS object ID, according to the DDS-RTPS Interoperability Wire Protocol.
Only the last 3 bytes are used; the most significant byte is ignored.

DDS_
UnsignedLong

rtps_
object_id

DDS_Boolean

push_on_
write

DDS_Boolean

disable_
positive_
acks

The rtps_host_id, rtps_app_id, and rtps_instance_id in the WIRE_PROTOCOL QosPolicy (DDS
Extension) (Section 8.5.9 on page 610), together with the 3 least significant bytes in rtps_object_id, and
another byte assigned by Connext DDS to identify the entity type, forms the BuiltinTopicKey in
PublicationBuiltinTopicData.
Controls when a DDS sample is sent after write() is called on a DataWriter. If TRUE, the DDS sample is
sent immediately; if FALSE, the DDS sample is put in a queue until an ACK/NACK is received from a
reliable DataReader.
Determines whether matching DataReaders send positive acknowledgements (ACKs) to the DataWriter.
When TRUE, the DataWriter will keep DDS samples in its queue for ACK-disabled readers for a minimum
keep duration (see Disabling Positive Acknowledgements (Section 6.5.3.3 on page 354)).
When strict reliability is not required, setting this to TRUE reduces overhead network traffic.

aPersistence Service is included with the Connext DDS Professional, Evaluation, and Basic package types. It saves

DDS data samples so they can be delivered to subscribing applications that join the system at a later time (see
Introduction to RTI Persistence Service (Section Chapter 26 on page 933)).

348

6.5.3 DATA_WRITER_PROTOCOL QosPolicy (DDS Extension)

Table 6.36 DDS_DataWriterProtocolQosPolicy
Type

Field
Name

Description
Controls whether or not the key-hash is propagated on the wire with DDS samples.
This field only applies to keyed writers.

Connext DDS associates a key-hash (an internal 16-byte representation) with each key.
When FALSE, the key-hash is sent on the wire with every data instance.

DDS_Boolean

disable_
inline_
keyhash

When TRUE, the key-hash is not sent on the wire (so the readers must compute the value using the received
data).
If the reader is CPU bound, sending the key-hash on the wire may increase performance, because the reader
does not have to get the key-hash from the data.
If the writer is CPU bound, sending the key-hash on the wire may decrease performance, because it requires
more bandwidth (16 more bytes per DDS sample).

Setting disable_inline_keyhash to TRUE is not compatible with using RTI Database Integration Service or RTI
Recording Service.

Controls whether or not the serialized key is propagated on the wire with dispose notifications.
This field only applies to keyed writers.
DDS_Boolean

serialize_
key_
with_
dispose

RTI recommends setting this field to TRUE if there are DataReaders with propagate_dispose_of_
unregistered_instances (in the DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section
7.6.1 on page 511)) also set to TRUE.
Important: When this field TRUE, batching will not be compatible with RTI Data Distribution Service 4.3e,
4.4b, or 4.4c—the DataReaders will receive incorrect data and/or encounter deserialization errors.

DDS_Boolean

propagate_
app_
ack_with_
no_
response

DDS_
rtps_
RtpsReliable
reliable_
WriterProtocol_
writer
t

Controls whether or not a DataWriter receives on_application_acknowledgment() notifications with an
empty or invalid response.
When FALSE, on_application_acknowledgment() will not be invoked if the DDS sample being
acknowledged has an empty or invalid response.

This structure includes the fields in Table 6.37 DDS_RtpsReliableWriterProtocol_t.

349

6.5.3 DATA_WRITER_PROTOCOL QosPolicy (DDS Extension)

Table 6.37 DDS_RtpsReliableWriterProtocol_t
Type
DDS_
Long

Field Name

Description

low_watermark

Queue levels that control when to switch between the regular and fast heartbeat rates (heartbeat_period
(Section below) and fast_heartbeat_period (Section below)). See High and Low Watermarks (Section 6.5.3.1
high_watermark on page 352).
heartbeat_period

fast_heartbeat_
DDS_
period
Duration_
t
late_joiner_
heartbeat_
period

Rates at which to send heartbeats to DataReaders with unacknowledged DDS samples. See Normal, Fast,
and Late-Joiner Heartbeat Periods (Section 6.5.3.2 on page 353) and How Often Heartbeats are Resent
(heartbeat_period) (Section 10.3.4.1 on page 645).

DDS_
The rate at which a reliable DataWriter will send virtual heartbeats. Virtual heartbeat informs the reliable
virtual_
Duration_
DataReader about the range of DDS samples currently present for each virtual GUID in the reliable writer's
heartbeat_period
t
queue. See Virtual Heartbeats (Section 6.5.3.6 on page 357).
DDS_
Long

samples_per_
virtual_
heartbeat

The number of DDS samples that a reliable DataWriter must publish before sending a virtual heartbeat.
See Virtual Heartbeats (Section 6.5.3.6 on page 357).
Maximum number of periodic heartbeats sent without receiving an ACK/NACK packet before marking a
DataReader ‘inactive.’

DDS_
Long

max_heartbeat_
retries

When a DataReader has not acknowledged all the DDS samples the reliable DataWriter has sent to it, and
max_heartbeat_retries number of periodic heartbeats have been sent without receiving any
ACK/NACK packets in return, the DataReader will be marked as inactive (not alive) and be ignored until it
resumes sending ACK/NACKs.
Note that piggyback heartbeats do not count towards this value.
See Controlling How Many Times Heartbeats are Resent (max_heartbeat_retries) (Section 10.3.4.4 on page
650).

DDS_
Boolean

Allows the DataWriter to treat DataReaders that send successive non-progressing NACK packets as
inactivate_
inactive.
nonprogressing_
See Treating Non-Progressing Readers as Inactive Readers (inactivate_nonprogressing_readers) (Section
readers
10.3.4.5 on page 650).
A piggyback heartbeat is sent every [current send-window size/heartbeats_per_max_samples] number of
DDS samples written.

DDS_
Long

heartbeats_per_
max_samples

If set to zero, no piggyback heartbeat will be sent.
If the current send-window size is LENGTH_UNLIMITED, 100 million is assumed as the value in the
calculation.
See Configuring the Send Window Size (Section 6.5.3.4 on page 355)

350

6.5.3 DATA_WRITER_PROTOCOL QosPolicy (DDS Extension)

Table 6.37 DDS_RtpsReliableWriterProtocol_t
Type

Field Name

Description
Minimum delay to respond to an ACK/NACK.

DDS_
min_nack_
Duration_
response_delay
t

When a reliable DataWriter receives an ACK/NACK from a DataReader, the DataWriter can choose to
delay a while before it sends repair DDS samples or a heartbeat. This set the value of the minimum delay.
See Coping with Redundant Requests for Missing DDS Samples (max_nack_response_delay) (Section
10.3.4.6 on page 651).
Maximum delay to respond to a ACK/NACK.

DDS_
max_nack_
Duration_
response_delay
t

This sets the value of maximum delay between receiving an ACK/NACK and sending repair DDS samples
or a heartbeat.
A longer wait can help prevent storms of repair packets if many DataReaders send NACKs at the same
time. However, it delays the repair, and hence increases the latency of the communication.
See Coping with Redundant Requests for Missing DDS Samples (max_nack_response_delay) (Section
10.3.4.6 on page 651).
How long consecutive NACKs are suppressed.

DDS_
nack_
Duration_ suppression_
t
duration

When a reliable DataWriter receives consecutive NACKs within a short duration, this may trigger the
DataWriter to send redundant repair messages. This value sets the duration during which consecutive
NACKs are ignored, thus preventing redundant repairs from being sent.
Maximum bytes in a repair package.

DDS_
Long

max_bytes_per_ When a reliable DataWriter resends DDS samples, the total package size is limited to this value. Note: The
nack_
reliable DataWriter will always send at least one sample.
response
See Controlling Packet Size for Resent DDS Samples (max_bytes_per_nack_response) (Section 10.3.4.3
on page 649).
disable_
positive_acks_
min_sample_
keep_
duration

DDS_
Duration_
t
disable_
positive_acks_
max_sample_
keep_
duration

DDS_
Boolean

disable_
positive_acks_
enable_
adaptive_
sample_keep_
duration

Minimum duration that a DDS sample will be kept in the DataWriter’s queue for ACK-disabled
DataReaders.
See Disabling Positive Acknowledgements (Section 6.5.3.3 on page 354) and Disabling Positive
Acknowledgements (disable_positive_acks_min_sample_keep_duration) (Section 10.3.4.7 on page 652).

Maximum duration that a DDS sample will be kept in the DataWriter’s queue for ACK-disabled readers.

Enables automatic dynamic adjustment of the ‘keep duration’ in response to network congestion.

351

6.5.3.1 High and Low Watermarks

Table 6.37 DDS_RtpsReliableWriterProtocol_t
Type

DDS_
Long

DDS_
Long

DDS_
Long

Field Name

Description

disable_
positive_acks_
increase_
sample_
keep_duration_
factor

When the ‘keep duration’ is dynamically controlled, the lengthening of the ‘keep duration’ is controlled by
this factor, which is expressed as a percentage.

disable_
positive_acks_
decrease_
sample_
keep_duration_
factor

When the ‘keep duration’ is dynamically controlled, the shortening of the ‘keep duration’ is controlled by
this factor, which is expressed as a percentage.

min_send_
window_size

When the adaptive algorithm determines that the keep duration should be increased, this factor is multiplied
with the current keep duration to get the new longer keep duration. For example, if the current keep duration
is 20 milliseconds, using the default factor of 150% would result in a new keep duration of 30 milliseconds.

When the adaptive algorithm determines that the keep duration should be decreased, this factor is multiplied
with the current keep duration to get the new shorter keep duration. For example, if the current keep duration
is 20 milliseconds, using the default factor of 95% would result in a new keep duration of 19 milliseconds.

Minimum and maximum size for the window of outstanding DDS samples.

max_send_
window_size

See Configuring the Send Window Size (Section 6.5.3.4 on page 355).

send_window_
decrease_
factor

Scales the current send-window size down by this percentage to decrease the effective send-rate in response
to received negative acknowledgement.
See Configuring the Send Window Size (Section 6.5.3.4 on page 355).
Controls whether or not periodic heartbeat messages are sent over multicast.

DDS_
Boolean

enable_
multicast_
periodic_
heartbeat

When enabled, if a reader has a multicast destination, the writer will send its periodic HEARTBEAT
messages to that destination.
Otherwise, if not enabled or the reader does not have a multicast destination, the writer will send its periodic
HEARTBEATs over unicast.

DDS_
Long

multicast_
resend_
threshold

Sets the minimum number of requesting readers needed to trigger a multicast resend.

DDS_
Long

send_window_
increase_
factor

Scales the current send-window size up by this percentage to increase the effective send-rate when a duration
has passed without any received negative acknowledgements.

send_window_
update_
period

Period in which DataWriter checks for received negative acknowledgements and conditionally increases the
send-window size when none are received.

DDS_
Duration

See Resending Over Multicast (Section 6.5.3.7 on page 357).

See Configuring the Send Window Size (Section 6.5.3.4 on page 355)

See Configuring the Send Window Size (Section 6.5.3.4 on page 355)

6.5.3.1 High and Low Watermarks
When the number of unacknowledged DDS samples in the current send-window of a reliable DataWriter
meets or exceeds high_watermark (Section on page 350), the RELIABLE_WRITER_CACHE_
352

6.5.3.2 Normal, Fast, and Late-Joiner Heartbeat Periods

CHANGED Status (DDS Extension) (Section 6.3.6.8 on page 279) will be changed appropriately, a
listener callback will be triggered, and the DataWriter will start heartbeating its matched DataReaders at
fast_heartbeat_period (Section on page 350)
When the number of DDS samples meets or falls below low_watermark (Section on page 350), the
RELIABLE_WRITER_CACHE_CHANGED Status (DDS Extension) (Section 6.3.6.8 on page 279)
will be changed appropriately, a listener callback will be triggered, and the heartbeat rate will return to the
"normal" rate (heartbeat_period (Section on page 350)).
Having both high and low watermarks (instead of one) helps prevent rapid flickering between the rates,
which could happen if the number of DDS samples hovers near the cut-off point.
Increasing the high and low watermarks will make the DataWriters less aggressive about seeking acknowledgments for sent data, decreasing the size of traffic spikes but slowing performance.
Decreasing the watermarks will make the DataWriters more aggressive, increasing both network utilization and performance.
If batching is used, high_watermark (Section on page 350) and low_watermark (Section on page 350)
refer to batches, not DDS samples.
When min_send_window_size (Section on the previous page) and max_send_window_size (Section on
the previous page) are not equal, the low and high watermarks are scaled down linearly to stay within the
current send-window size. The value provided by configuration corresponds to the high and low watermarks for the max_send_window_size (Section on the previous page).
6.5.3.2 Normal, Fast, and Late-Joiner Heartbeat Periods
The normal heartbeat_period (Section on page 350) is used until the number of DDS samples in the reliable DataWriter’s queue meets or exceeds high_watermark (Section on page 350); then fast_heartbeat_
period (Section on page 350) is used. Once the number of DDS samples meets or drops below low_watermark (Section on page 350), the normal rate (heartbeat_period (Section on page 350)) is used again.
l

fast_heartbeat_period (Section on page 350) must be <= heartbeat_period (Section on page 350)

Increasing fast_heartbeat_period (Section on page 350) increases the speed of discovery, but results in a larger surge of traffic when the DataWriter is waiting for acknowledgments.
Decreasing heartbeat_period (Section on page 350) decreases the steady state traffic on the wire, but may
increase latency by decreasing the speed of repairs for lost packets when the writer does not have very
many outstanding unacknowledged DDS samples.
Having two periodic heartbeat rates, and switching between them based on watermarks:

353

6.5.3.3 Disabling Positive Acknowledgements

l

l

l

Ensures that all DataReaders receive all their data as quickly as possible (the sooner they receive a
heartbeat, the sooner they can send a NACK, and the sooner the DataWriter can send repair DDS
samples);
Helps prevent the DataWriter from overflowing its resource limits (as its queue starts the fill, the
DataWriter sends heartbeats faster, prompting the DataReaders to acknowledge sooner, allowing
the DataWriter to purge these acknowledged DDS samples from its queue);
Tunes the amount of network traffic. (Heartbeats and NACKs use up network bandwidth like any
other traffic; decreasing the heartbeat rates, or increasing the threshold before the fast rate starts, can
smooth network traffic—at the expense of discovery performance).

The late_joiner_heartbeat_period (Section on page 350) is used when a reliable DataReader joins after a
reliable DataWriter (with non-volatile Durability) has begun publishing DDS samples. Once the late-joining DataReader has received all cached DDS samples, it will be serviced at the same rate as other reliable
DataReaders.
l

late_joiner_heartbeat_period (Section on page 350) must be <= heartbeat_period (Section on page
350)

6.5.3.3 Disabling Positive Acknowledgements
When strict reliable communication is not required, you can configure Connext DDS so that it does not
send positive acknowledgements (ACKs). In this case, reliability is maintained solely based on negative
acknowledgements (NACKs). The removal of ACK traffic may improve middleware performance. For
example, when sending DDS samples over multicast, ACK-storms that previously may have hindered
DataWriters and consumed overhead network bandwidth are now precluded.
By default, DataWriters and DataReaders are configured with positive ACKS enabled. To disable ACKs,
either:
l

l

Configure the DataWriter to disable positive ACKs for all matching DataReaders (by setting disable_positive_acks to TRUE in the DATA_WRITER_PROTOCOL QosPolicy (DDS Extension)
(Section 6.5.3 on page 347)).
Disable ACKs for individual DataReaders (by setting disable_positive_acks to TRUE in the
DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511)).

If ACKs are disabled, instead of the DataWriter holding a DDS sample in its send queue until all of its
DataReaders have ACKed it, the DataWriter will hold a DDS sample for a configurable duration. This
“keep-duration" starts when a DDS sample is written. When this time elapses, the DDS sample is logically
considered as acknowledged by its ACK-disabled readers.
The length of the "keep-duration" can be static or dynamic, depending on how rtps_reliable_writer.disable_positive_acks_enable_adaptive_sample_keep_duration is set.

354

6.5.3.4 Configuring the Send Window Size

l

l

When the length is static, the "keep-duration" is set to the minimum (rtps_reliable_writer.disable_
positive_acks_min_sample_keep_duration).
When the length is dynamic, the "keep-duration" is dynamically adjusted between the minimum and
maximum durations (rtps_reliable_writer.disable_positive_acks_min_sample_keep_duration
and rtps_reliable_writer.disable_positive_acks_max_sample_keep_duration).

Dynamic adjustment maximizes throughput and reliability in response to current network conditions: when
the network is congested, durations are increased to decrease the effective send rate and relieve the congestion; when the network is not congested, durations are decreased to increase the send rate and maximize throughput.
You should configure the minimum "keep-duration" to allow at least enough time for a possible NACK to
be received and processed. When a DataWriter has both matching ACK-disabled and ACK-enabled
DataReaders, it holds a DDS sample in its queue until all ACK-enabled DataReaders have ACKed it and
the "keep-duration" has elapsed.
See also: Disabling Positive Acknowledgements (disable_positive_acks_min_sample_keep_duration) (Section 10.3.4.7 on page 652).
6.5.3.4 Configuring the Send Window Size
When a reliable DataWriter writes a DDS sample, it keeps the DDS sample in its queue until it has
received acknowledgements from all of its subscribing DataReaders. The number of these outstanding
DDS samples is referred to as the DataWriter's "send window." Once the number of outstanding DDS
samples has reached the send window size, subsequent writes will block until an outstanding DDS sample
is acknowledged.
Configuration of the send window sets a minimum and maximum size, which may be unlimited. The min
and max send windows can be the same. When set differently, the send window will dynamically change
in response to detected network congestion, as signaled by received negative acknowledgements. When
NACKs are received, the DataWriter responds to the slowed reader by decreasing the send window by the
send_window_decrease_factor to throttle down its effective send rate. The send window will not be
decreased to less than the min_send_window_size. After a period (send_window_update_period) during which no NACKs are received, indicating that the reader is catching up, the DataWriter will increase
the send window size to increase the effective send rate by the percentage specified by send_window_
increase_factor. The send window will increase to no greater than the max_send_window_size.
When both min_send_window_size and max_send_window_size are unlimited, either the resource limits max_samples in RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) (for non-batching) or
max_batches in DATA_WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 6.5.4
on page 359) (for batching) serves as the effective max_send_window_size.

355

6.5.3.5 Propagating Serialized Keys with Disposed-Instance Notifications

When either max_samples (for non-batching) or max_batches (for batching) is less than max_send_window_size, it serves as the effective max_send_window_size. If it is also less than min_send_window_
size, then effectively both min and max send-window sizes are equal to max_samples or max_batches.
6.5.3.5 Propagating Serialized Keys with Disposed-Instance Notifications
This section describes the interaction between these two fields:
l

l

serialize_key_with_dispose in DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347)
propagate_dispose_of_unregistered_instances in DATA_READER_PROTOCOL QosPolicy
(DDS Extension) (Section 7.6.1 on page 511)

RTI recommends setting serialize_key_with_dispose to TRUE if there are DataReaders with propagate_dispose_of_unregistered_instances also set to TRUE. However, it is permissible to set one to TRUE
and the other to FALSE. The following examples will help you understand how these fields work.
See also: Disposing of Data (Section 6.3.14.2 on page 299).
Example 1
1. DataWriter’s serialize_key_with_dispose = FALSE
2. DataReader’s propagate_dispose_of_unregistered_instances = TRUE
3. DataWriter calls dispose() before writing any DDS samples
4. DataReader calls take() and receives a disposed-instance notification (without a key)
5. DataReader calls get_key_value(), which returns an error because there is no key associated with
the disposed-instance notification
Example 2
1. DataWriter’s serialize_key_with_dispose = TRUE
2. DataReader’s propagate_dispose_of_unregistered_instances = FALSE
3. DataWriter calls dispose() before writing any DDS samples
4. DataReader calls take(), which does not return any DDS samples because none were written, and it
does not receive any disposed-instance notifications because propagate_dispose_of_unregistered_
instances = FALSE

356

6.5.3.6 Virtual Heartbeats

Example 3
1. DataWriter’s serialize_key_with_dispose = TRUE
2. DataReader’s propagate_dispose_of_unregistered_instances = TRUE
3. DataWriter calls dispose() before writing any DDS samples
4. DataReader calls take() and receives the disposed-instance notification
5. DataReader calls get_key_value() and receives the key for the disposed-instance notification
Example 4
1. DataWriter’s serialize_key_with_dispose = TRUE
2. DataReader’s propagate_dispose_of_unregistered_instances = TRUE
3. DataWriter calls write(), which writes a DDS sample with a key
4. DataWriter calls dispose(), which writes a disposed-instance notification with a key
5. DataReader calls take() and receives a DDS sample and a disposed-instance notification; both have
keys
6. DataReader calls get_key_value() with no errors
6.5.3.6 Virtual Heartbeats
Virtual heartbeats announce the availability of DDS samples with the Collaborative DataWriters feature
described in DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511),
where multiple DataWriters publish DDS samples from a common logical data-source (identified by a virtual GUID).
When PRESENTATION QosPolicy (Section 6.4.6 on page 330) access_scope is set to TOPIC or
INSTANCE on the Publisher, the virtual heartbeat contains information about the DDS samples contained
in the DataWriter queue.
When presentation access_scope is set to GROUP on the Publisher, the virtual heartbeat contains information about the DDS samples in the queues of all DataWriters that belong to the Publisher.
6.5.3.7 Resending Over Multicast
Given DataReaders with multicast destinations, when a DataReader sends a NACK to request for DDS
samples to be resent, the DataWriter can either resend them over unicast or multicast. Though resending
over multicast would save bandwidth and processing for the DataWriter, the potential problem is that there
could be DataReaders of the multicast group that did not request for any resends, yet they would have to
process, and drop, the resent DDS samples.

357

6.5.3.8 Example

Thus, to make each multicast resend more efficient, the multicast_resend_threshold is set as the minimum number of DataReaders of the same multicast group that the DataWriter must receive NACKs from
within a single response-delay duration. This allows the DataWriter to coalesce near-simultaneous unicast
resends into a multicast resend, and it allows a "vote" from DataReaders of a multicast group to exceed a
threshold before resending over multicast.
The multicast_resend_threshold must be set to a positive value. Note that a threshold of 1 means that all
resends will be sent over multicast. Also, note that a DataWriter with a zero NACK response-delay (i.e.,
both min_nack_response_delay and min_nackresponse_delay are zero) will resend over multicast only
if the threshold is 1.
6.5.3.8 Example
For information on how to use the fields in Table 6.37 DDS_RtpsReliableWriterProtocol_t, see Controlling Heartbeats and Retries with DataWriterProtocol QosPolicy (Section 10.3.4 on page 645).
The following describes a use case for when to change push_on_write to DDS_BOOLEAN_FALSE.
Suppose you have a system in which the data packets being sent is very small. However, you want the
data to be sent reliably, and the latency between the time that data is sent to the time that data is received is
not an issue. However, the total network bandwidth between the DataWriter and DataReader applications
is limited.
If the DataWriter sends a burst of data a a high rate, it is possible that it will overwhelm the limited bandwidth of the network. If you allocate enough space for the DataWriter to store the data burst being sent
(see RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)), then you can use the push_on_
write parameter of the DATA_WRITER_PROTOCOL QosPolicy to delay sending the data until the reliable DataReader asks for it.
By setting push_on_write to DDS_BOOLEAN_FALSE, when write() is called on the DataWriter, no
data is actually sent. Instead data is stored in the DataWriter’s send queue. Periodically, Connext DDS will
be sending heartbeats informing the DataReader about the data that is available. So every heartbeat period,
the DataReader will realize that the DataWriter has new data, and it will send an ACK/NACK, asking for
them.
When DataWriter receives the ACK/NACK packet, it will put together a package of data, up to the size
set by the parameter max_bytes_per_nack_response, to be sent to the DataReader. This method not
only self-throttles the send rate, but also uses network bandwidth more efficiently by eliminating redundant
packet headers when combining several small packets into one larger one. Please note that the DataWriter
will always send at least one sample.
6.5.3.9 Properties
This QosPolicy cannot be modified after the DataWriter is created.

358

6.5.3.10 Related QosPolicies

Since it is only for DataWriters, there are no compatibility restrictions for how it is set on the publishing
and subscribing sides.
When setting the fields in this policy, the following rules apply. If any of these are false, Connext DDS
returns DDS_RETCODE_INCONSISTENT_POLICY:
l

min_nack_response_delay <= max_nack_response_delay

l

fast_heartbeat_period <= heartbeat_period

l

late_joiner_heartbeat_period <= heartbeat_period

l

low_watermark < high_watermark

l

l

If batching is disabled:
l heartbeats_per_max_samples <= writer_qos.resource_limits.max_samples
If batching is enabled:
l heartbeats_per_max_samples <= writer_qos.resource_limits.max_batches

6.5.3.10 Related QosPolicies
l

DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511)

l

HISTORY QosPolicy (Section 6.5.10 on page 376)

l

RELIABILITY QosPolicy (Section 6.5.19 on page 400)

6.5.3.11 Applicable DDS Entities
l

DataWriters (Section 6.3 on page 261)

6.5.3.12 System Resource Considerations
A high max_bytes_per_nack_response may increase the instantaneous network bandwidth required to
send a single burst of traffic for resending dropped packets.

6.5.4 DATA_WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension)
This QosPolicy defines various settings that configure how DataWriters allocate and use physical memory
for internal resources.
It includes the members in Table 6.38 DDS_DataWriterResourceLimitsQosPolicy. For defaults and valid
ranges, please refer to the API Reference HTML documentation.

359

6.5.4 DATA_WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 6.38 DDS_DataWriterResourceLimitsQosPolicy
Type

Field
Name

Description

DDS_Long

initial_
concurrent_
blocking_
threads

Initial number of threads that are allowed to concurrently block on the write() call on the same
DataWriter.

DDS_Long

max_
concurrent_
blocking_
threads

Maximum number of threads that are allowed to concurrently block on write() call on the same
DataWriter.

DDS_Long

max_
remote_
reader_
filters

Maximum number of remote DataReaders for which this DataWriter will perform content-based
filtering.

DDS_Long

initial_
batches

Initial number of batches that a DataWriter will manage if batching is enabled.
Maximum number of batches that a DataWriter will manage if batching is enabled.

DDS_Long

max_batches When batching is enabled, the maximum number of DDS samples that a DataWriter can store is
limited by this value and max_samples in RESOURCE_LIMITS QosPolicy (Section 6.5.20 on
page 405).

DDS_DataWriter
instance_
ResourceLimits
replacement
InstanceReplacementKind

Sets the kinds of instances allowed to be replaced when a DataWriter reaches instance resource
limits. (See Configuring DataWriter Instance Replacement (Section 6.5.20.2 on page 407)

DDS_Boolean

replace_
empty_
instances

DDS_Boolean

Whether to register automatically instances written with non-NIL handle that are not yet
autoregister_
registered, which will otherwise return an error. This can be especially useful if the instance has
instances
been replaced.

DDS_Long

initial_
virtual_
writers

Whether to replace empty instances during instance replacement. (See Configuring DataWriter
Instance Replacement (Section 6.5.20.2 on page 407)

Initial number of virtual writers supported by a DataWriter.

360

6.5.4 DATA_WRITER_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 6.38 DDS_DataWriterResourceLimitsQosPolicy
Type

Field
Name

Description
Maximum number of virtual writers supported by a DataWriter.

DDS_Long

Sets the maximum number of unique virtual writers supported by a DataWriter, where virtual
max_virtual_
writers are added when DDS samples are written with the virtual writer GUID.
writers
This field is especially relevant in the configuration of Persistence Service1 DataWriters, since
they publish information on behalf of multiple virtual writers.

DDS_Long

max_
remote_
readers

DDS_Long

max_app_
The maximum number of application-level acknowledging remote readers supported by a
ack_remote_
DataWriter.
readers

The maximum number of remote readers supported by a DataWriter.

DataWriters must allocate internal structures to handle the simultaneous blocking of threads trying to call
write() on the same DataWriter, for the storage used to batch small DDS samples, and for content-based
filters specified by DataReaders.
Most of these internal structures start at an initial size and by default, will grow as needed by dynamically
allocating additional memory. You may set fixed, maximum sizes for these internal structures if you want
to bound the amount of memory that a DataWriter can use. By setting the initial size to the maximum size,
you will prevent Connext DDS from dynamically allocating any memory after the creation of the
DataWriter.
When setting the fields in this policy, the following rule applies. If this is false, Connext DDS returns
DDS_RETCODE_INCONSISTENT_POLICY:
l

max_concurrent_blocking_threads >= initial_concurrent_blocking_threads

The initial_concurrent_blocking_threads is used to allocate necessary initial system resources. If necessary, it will be increased automatically up to the max_concurrent_blocking_threads limit.
Every user thread calling write() on a DataWriter may use a semaphore that will block the thread when
the DataWriter’s send queue is full. Because user code may set a timeout, each thread must use a different
semaphore. See the max_blocking_time parameter of the RELIABILITY QosPolicy (Section 6.5.19 on

aPersistence Service is included with the Connext DDS Professional, Evaluation, and Basic package types. It saves DDS

data samples so they can be delivered to subscribing applications that join the system at a later time (see Introduction
to RTI Persistence Service (Section Chapter 26 on page 933)).

361

6.5.4.1 Example

page 400). This QoS is offered so that the user application can control the dynamic allocation of system
resources by Connext DDS.
If you do not mind if Connext DDS dynamically allocates semaphores when needed, then you can set the
max_concurrent_blocking_threads parameter to some large value like MAX_INT. However, if you
know exactly how many threads will be calling write() on the same DataWriter, and you do not want Connext DDS to allocate any system resources or memory after initialization, then you should set:
max_concurrent_blocking_threads = initial_concurrent_blocking_threads = NUM

(where NUM is the number of threads that could possibly block concurrently).
Each DataWriter can perform content-based data filtering for up to max_remote_reader_filters number
of DataReaders.
Values for max_remote_reader_filters may be.
l

l

l

0: The DataWriter will not perform filtering for any DataReader, which means the DataReader will
have to filter the data itself.
1 to (231-2): The DataWriter will filter for up to the specified number of DataReaders. In addition,
the Datawriter will store the result of the filtering per DDS sample per DataReader.
DDS_LENGTH_UNLIMITED: The DataWriter will filter for up to (231)-2 DataReaders.
However, in this case, the DataWriter will not store the filtering result per DDS sample per
DataReader. Thus, if a DDS sample is resent (such as due to a loss of reliable communication), the
DDS sample will be filtered again.

For more information, see ContentFilteredTopics (Section 5.4 on page 212).
6.5.4.1 Example
If there are multiple threads that can write on the same DataWriter, and the write() operation may block
(based on reliability_qos.max_blocking_time and HISTORY settings), you may want to set initial_concurrent_blocking_threads to the most likely number of threads that will block on the same DataWriter at
the same time, and set max_concurrent_blocking_threads to the maximum number of threads that could
potentially block in the worst case.
6.5.4.2 Properties
This QosPolicy cannot be modified after the DataWriter is created.
Since it is only for DataWriters, there are no compatibility restrictions for how it is set on the publishing
and subscribing sides.

362

6.5.4.3 Related QosPolicies

6.5.4.3 Related QosPolicies
l

BATCH QosPolicy (DDS Extension) (Section 6.5.2 on page 341)

l

RELIABILITY QosPolicy (Section 6.5.19 on page 400)

l

HISTORY QosPolicy (Section 6.5.10 on page 376)

6.5.4.4 Applicable DDS Entities
l

DataWriters (Section 6.3 on page 261)

6.5.4.5 System Resource Considerations
Increasing the values in this QosPolicy will cause more memory usage and more system resource usage.

6.5.5 DEADLINE QosPolicy
On a DataWriter, this QosPolicy states the maximum period in which the application expects to call write
() on the DataWriter, thus publishing a new DDS sample. The application may call write() faster than the
rate set by this QosPolicy.
On a DataReader, this QosPolicy states the maximum period in which the application expects to receive
new values for the Topic. The application may receive data faster than the rate set by this QosPolicy.
The DEADLINE QosPolicy has a single member, shown in Table 6.39 DDS_DeadlineQosPolicy. For
the default and valid range, please refer to the API Reference HTML documentation.
Table 6.39 DDS_DeadlineQosPolicy
Type

Field Name

Description
For DataWriters: maximum time between writing a new value of an instance.

DDS_Duration_t

period
For DataReaders: maximum time between receiving new values for an instance.

You can use this QosPolicy during system integration to ensure that applications have been coded to meet
design specifications. You can also use it during run time to detect when systems are performing outside of
design specifications. Receiving applications can take appropriate actions to prevent total system failure
when data is not received in time. For topics on which data is not expected to be periodic, the deadline
period should be set to an infinite value.
For keyed topics, the DEADLINE QoS applies on a per-instance basis. An application must call write()
for each known instance of the Topic within the period specified by the DEADLINE on the DataWriter
or receive a new value for each known instance within the period specified by the DEADLINE on the
DataReader. For a DataWriter, the deadline period begins when the instance is first written or registered.
For a DataReader, the deadline period begins when the first DDS sample is received.

363

6.5.5.1 Example

Connext DDS will modify the OFFERED_DEADLINE_MISSED_STATUS and call the associated
method in the DataWriterListener (see OFFERED_DEADLINE_MISSED Status (Section 6.3.6.5 on
page 277)) if the application fails to write() a value for an instance within the period set by the
DEADLINE QosPolicy of the DataWriter.
Similarly, Connext DDS will modify the REQUESTED_DEADLINE_MISSED_STATUS and call the
associated method in the DataReaderListener (see REQUESTED_DEADLINE_MISSED Status (Section
7.3.7.5 on page 476)) if the application fails to receive a value for an instance within the period set by the
DEADLINE QosPolicy of the DataReader.
For DataReaders, the DEADLINE QosPolicy and the TIME_BASED_FILTER QosPolicy (Section
7.6.4 on page 526) may interact such that even though the DataWriter writes DDS samples fast enough to
fulfill its commitment to its own DEADLINE QosPolicy, the DataReader may see violations of its
DEADLINE QosPolicy. This happens because Connext DDS will drop any packets received within the
minimum_separation set by the TIME_BASED_FILTER—packets that could satisfy the DataReader’s
deadline.
To avoid triggering the DataReader’s deadline even though the matched DataWriter is meeting its own
deadline, set your QoS parameters to meet the following relationship:
reader deadline period >= reader minimum_separation + writer deadline period

Although you can set the DEADLINE QosPolicy on Topics, its value can only be used to initialize the
DEADLINE QosPolicies of either a DataWriter or DataReader. It does not directly affect the operation of
Connext DDS, see Setting Topic QosPolicies (Section 5.1.3 on page 204).
6.5.5.1 Example
Suppose you have a time-critical piece of data that should be updated at least once every second. You can
set the DEADLINE period to 1 second on both the DataWriter and DataReader. If there is no update
within that time, the DataWriter will get an on_offered_deadline_missed Listener callback, and the
DataReader will get on_requested_deadline_missed, so that both sides can handle the error situation
properly.
Note that in practice, there will be latency and jitter in the time between when data is send and when data
is received. Thus even if the DataWriter is sending data at exactly 1 second intervals, the DataReader may
not receive the data at exactly 1 second intervals. More likely, it will DataReader will receive the data at 1
second plus a small variable quantity of time. Thus you should accommodate this practical reality in choosing the DEADLINE period as well as the actual update period of the DataWriter or your application may
receive false indications of failure.
The DEADLINE QosPolicy also interacts with the OWNERSHIP QosPolicy when OWNERSHIP is set
to EXCLUSIVE. If a DataReader fails to receive data from the highest strength DataWriter within its
requested DEADLINE, then the DataReaders can fail-over to lower strength DataWriters, see the
OWNERSHIP QosPolicy (Section 6.5.15 on page 389).

364

6.5.5.2 Properties

6.5.5.2 Properties
This QosPolicy can be changed at any time.
The deadlines on the two sides must be compatible.
DataWriter’s DEADLINE period <= the DataReader’s DEADLINE period.
That is, the DataReader cannot expect to receive DDS samples more often than the DataWriter commits
to sending them.
If the DataReader and DataWriter have compatible deadlines, Connext DDS monitors this “contract” and
informs the application of any violations. If the deadlines are incompatible, both sides are informed and
communication does not occur. The ON_OFFERED_INCOMPATIBLE_QOS and the ON_
REQUESTED_INCOMPATIBLE_QOS statuses will be modified and the corresponding Listeners
called for the DataWriter and DataReader respectively.
6.5.5.3 Related QosPolicies
l

LIVELINESS QosPolicy (Section 6.5.13 on page 382)

l

OWNERSHIP QosPolicy (Section 6.5.15 on page 389)

l

TIME_BASED_FILTER QosPolicy (Section 7.6.4 on page 526)

6.5.5.4 Applicable DDS Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.5.5 System Resource Considerations
A Connext DDS-internal thread will wake up at least by the DEADLINE period to check to see if the
deadline was missed. It may wake up faster if the last DDS sample that was published or sent was close to
the last time that the deadline was checked. Therefore a short period will use more CPU to wake and
execute the thread checking the deadline.

6.5.6 DESTINATION_ORDER QosPolicy
When multiple DataWriters send data for the same topic, the order in which data from different
DataWriters are received by the applications of different DataReaders may be different. Thus different
DataReaders may not receive the same "last" value when DataWriters stop sending data.
This policy controls how each subscriber resolves the final value of a data instance that is written by multiple DataWriters (which may be associated with different Publishers) running on different nodes.

365

6.5.6 DESTINATION_ORDER QosPolicy

This QosPolicy can be used to create systems that have the property of "eventual consistency." Thus intermediate states across multiple applications may be inconsistent, but when DataWriters stop sending
changes to the same topic, all applications will end up having the same state.
Each DDS sample includes two timestamps: a source timestamp and a destination timestamp. The source
timestamp is recorded by the DataWriter application when the data was written. The destination timestamp
is recorded by the DataReader application when the data was received.
This QoS includes the member in Table 6.40 DDS_DestinationOrderQosPolicy.
Table 6.40 DDS_DestinationOrderQosPolicy
Type

Field Name

Description
Can be either:

DDS_DestinationOrderQosPolicyKind

kind

DDS_BY_RECEPTION_TIMESTAMP_DESTINATIONORDER_QOS
DDS_BY_SOURCE_TIMESTAMP_DESTINATIONORDER_QOS
Allowed tolerance between source timestamps of consecutive DDS samples.

DDS_Duration_t

source_timestamp_
tolerance

Only applies when kind (above) is DDS_BY_SOURCE_TIMESTAMP_
DESTINATIONORDER_QOS.

Each DataReader can set this QoS to:
l

DDS_BY_RECEPTION_TIMESTAMP_DESTINATIONORDER_QOS

l

Assuming the OWNERSHIP_STRENGTH allows it, the latest received value for the instance
should be the one whose value is kept. Data will be delivered by a DataReader in the order in
which it was received (which may lead to inconsistent final values).
DDS_BY_SOURCE_TIMESTAMP_DESTINATIONORDER_QOS
Assuming the OWNERSHIP_STRENGTH allows it, within each instance, the source_timestamp
shall be used to determine the most recent information. This is the only setting that, in the case of
concurrent same-strength DataWriters updating the same instance, ensures all subscribers will end
up with the same final value for the instance.
Data will be delivered by a DataReader in the order in which it was sent. If data arrives on the network with a source timestamp earlier than the source timestamp of the last data delivered, the new
data will be dropped. This ordering therefore works best when system clocks are relatively synchronized among writing machines.
Not all data sent by multiple DataWriters may be delivered to a DataReader and not all DataReaders will see the same data sent by DataWriters. However, all DataReaders will see the same "final"
data when DataWriters "stop" sending data.

366

6.5.6.1 Properties

l

For a DataWriter with kind
DDS_BY_SOURCE_TIMESTAMP_DESTINATIONORDER_QOS:
When writing a DDS sample, its timestamp must not be less than the timestamp of the previously written DDS sample. However, if it is less than the timestamp of the previously written DDS sample but the difference is less than this tolerance, the DDS sample will use the
previously written DDS sample's timestamp as its timestamp. Otherwise, if the difference is
greater than this tolerance, the write will fail.
See also: Special Instructions for deleting DataWriters if you are using the ‘Timestamp’ APIs
and BY_SOURCE_TIMESTAMP Destination Order: (Section 6.3.3.1 on page 268).

l

A DataReader with kind
DDS_BY_SOURCE_TIMESTAMP_DESTINATIONORDER_QOS will accept a DDS
sample only if only if the source timestamp is no farther in the future from the reception
timestamp than this tolerance. Otherwise, the DDS sample is rejected.

Although you can set the DESTINATION_ORDER QosPolicy on Topics, its value can only be used to
initialize the DESTINATION_ORDER QosPolicies of either a DataWriter or DataReader. It does not directly affect the operation of Connext DDS, see Setting Topic QosPolicies (Section 5.1.3 on page 204).
6.5.6.1 Properties
This QosPolicy cannot be modified after the Entity is enabled.
This QoS must be set compatibly between the DataWriter and the DataReader. The compatible combinations are shown in Table 6.41 Valid Reader/Writer Combinations of DestinationOrder.
Table 6.41 Valid Reader/Writer Combinations of DestinationOrder
DataReader requests:
Destination Order
BY_SOURCE

BY_RECEPTION

BY_SOURCE

4

4

BY_RECEPTION

incompatible

4

DataWriter offers:

If this QosPolicy is set incompatibly, the ON_OFFERED_INCOMPATIBLE_QOS and ON_
REQUESTED_INCOMPATIBLE_QOS statuses will be modified and the corresponding Listeners
called for the DataWriter and DataReader respectively.

367

6.5.6.2 Related QosPolicies

6.5.6.2 Related QosPolicies
l

OWNERSHIP QosPolicy (Section 6.5.15 on page 389)

l

HISTORY QosPolicy (Section 6.5.10 on page 376)

6.5.6.3 Applicable DDS Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.6.4 System Resource Considerations
The use of this policy does not significantly impact the use of resources.

6.5.7 DURABILITY QosPolicy
Because the publish-subscribe paradigm is connectionless, applications can create publications and subscriptions in any way they choose. As soon as a matching pair of DataWriters and DataReaders exist,
then data published by the DataWriter will be delivered to the DataReader. However, a DataWriter may
publish data before a DataReader has been created. For example, before you subscribe to a magazine,
there have been past issues that were published.
The DURABILITY QosPolicy controls whether or not, and how, published DDS samples are stored by
the DataWriter application for DataReaders that are found after the DDS samples were initially written.
DataReaders use this QoS to request DDS samples that were published before they were created. The analogy is for a new subscriber to a magazine to ask for issues that were published in the past. These are
known as ‘historical’ DDS data samples. (Reliable DataReaders may wait for these historical DDS
samples, see Checking DataReader Status and StatusConditions (Section 7.3.5 on page 468).)
This QosPolicy can be used to help ensure that DataReaders get all data that was sent by DataWriters,
regardless of when it was sent. This QosPolicy can increase system tolerance to failure conditions.
Exactly how many DDS samples are stored by the DataWriter or requested by the DataReader is controlled using the HISTORY QosPolicy (Section 6.5.10 on page 376).
For more information, please see Mechanisms for Achieving Information Durability and Persistence (Section Chapter 12 on page 675).
The possible settings for this QoS are:
l

DDS_VOLATILE_DURABILITY_QOS
Connext DDSis not required to send and will not deliver any DDS data samples to DataReaders

368

6.5.7 DURABILITY QosPolicy

that are discovered after the DDS samples were initially published.
l

DDS_TRANSIENT_LOCAL_DURABILITY_QOS
Connext DDSwill store and send previously published DDS samples for delivery to newly discovered DataReaders as long as the DataWriter still exists. For this setting to be effective, you must
also set the RELIABILITY QosPolicy (Section 6.5.19 on page 400) kind to Reliable (not Best
Effort). Which particular DDS samples are kept depends on other QoS settings such as HISTORY
QosPolicy (Section 6.5.10 on page 376) and RESOURCE_LIMITS QosPolicy (Section 6.5.20 on
page 405).

l

DDS_TRANSIENT_DURABILITY_QOS
Connext DDSwill store previously published DDS samples in memory using Persistence Service, which
will send the stored data to newly discovered DataReaders. Which particular DDS samples are kept
and sent by Persistence Service depends on the HISTORY QosPolicy (Section 6.5.10 on page
376) and RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) of the Persistence Service DataWriters. These QosPolicies can be configured in the Persistence Service configuration file
or through the DURABILITY SERVICE QosPolicy (Section 6.5.8 on page 372) of the
DataWriters configured with DDS_TRANSIENT_DURABILITY_QOS.

l

DDS_PERSISTENT_DURABILITY_QOS
Connext DDSwill store previously published DDS samples in permanent storage, like a disk, using
Persistence Service, which will send the stored data to newly discovered DataReaders. Which particular
DDS samples are kept and sent by Persistence Service depends on the HISTORY QosPolicy (Section 6.5.10 on page 376) and RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) in
the Persistence Service DataWriters. These QosPolicies can be configured in the Persistence Service configuration file or through the DURABILITY SERVICE QosPolicy (Section 6.5.8 on page
372) of the DataWriters configured with DDS_PERSISTENT_DURABILITY_QOS.

This QosPolicy includes the members in Table 6.42 DDS_DurabilityQosPolicy. For default settings,
please refer to the API Reference HTML documentation.

369

6.5.7.1 Example

Table 6.42 DDS_DurabilityQosPolicy
Type

Field Name

Description
DDS_VOLATILE_DURABILITY_QOS:
Do not save or deliver old DDS samples.

DDS_
kind
DurabilityQosPolicyKind

DDS_TRANSIENT_LOCAL_DURABILITY_QOS:
Save and deliver old DDS samples if the DataWriter still exists.
DDS_TRANSIENT_DURABILITY_QOS:
Save and deliver old DDS samples using a memory-based service.
DDS_PERSISTENCE_DURABILITY_QOS:
Save and deliver old DDS samples using disk-based service.
Whether or not a TRANSIENT or PERSISTENT DataReader should receive DDS samples
directly from a TRANSIENT or PERSISTENT DataWriter.
When TRUE, a TRANSIENT or PERSISTENT DataReader will receive DDS samples
directly from the original DataWriter. The DataReader may also receive DDS samples from
Persistence Service1 but the duplicates will be filtered by the middleware.

DDS_Boolean

direct_
communication When FALSE, a TRANSIENT or PERSISTENT DataReader will receive DDS samples only
from the DataWriter created by Persistence Service. This ‘relay communication’ pattern
provides a way to guarantee eventual consistency.
See RTI Persistence Service (Section 12.5.1 on page 692).
This field only applies to DataReaders.

With this QoS policy alone, there is no way to specify or characterize the intended consumers of the
information. With TRANSIENT_LOCAL, TRANSIENT, or PERSISTENT durability a DataWriter can
be configured to keep DDS samples around for late-joiners. However, there is no way to know when the
information has been consumed by all the intended recipients.
Information durability can be combined with required subscriptions in order to guarantee that DDS
samples are delivered to a set of required subscriptions. For additional details on required subscriptions see
Required Subscriptions (Section 6.3.13 on page 294) and AVAILABILITY QosPolicy (DDS Extension)
(Section 6.5.1 on page 337).
6.5.7.1 Example
Suppose you have a DataWriter that sends data sporadically and its DURABILITY kind is set to
VOLATILE. If a new DataReader joins the system, it won’t see any data until the next time that write()
is called on the DataWriter. If you want the DataReader to receive any data that is valid, old or new, both

1Persistence Service is included with the Connext DDS Professional, Evaluation, and Basic package types. It saves

DDS data samples so they can be delivered to subscribing applications that join the system at a later time (see
Introduction to RTI Persistence Service (Section Chapter 26 on page 933)).

370

6.5.7.2 Properties

sides should set their DURABILITY kind to TRANSIENT_LOCAL. This will ensure that the
DataReader gets some of the previous DDS samples immediately after it is enabled.
6.5.7.2 Properties
This QosPolicy cannot be modified after the Entity has been created.
The DataWriter and DataReader must use compatible settings for this QosPolicy. To be compatible, the
DataWriter and DataReader must use one of the valid combinations shown in Table 6.43 Valid Combinations of Durability ‘kind’.
If this QosPolicy is found to be incompatible, the ON_OFFERED_INCOMPATIBLE_QOS and ON_
REQUESTED_INCOMPATIBLE_QOS statuses will be modified and the corresponding Listeners
called for the DataWriter and DataReader respectively.
Table 6.43 Valid Combinations of Durability ‘kind’
DataReader requests:
VOLATILE

DataWriter offers:

TRANSIENT_LOCAL

TRANSIENT

PERSISTENT

VOLATILE

4

incompatible

incompatible

incompatible

TRANSIENT_
LOCAL

4

4

incompatible

incompatible

TRANSIENT

4

4

4

incompatible

PERSISTENT

4

4

4

4

6.5.7.3 Related QosPolicies
l

HISTORY QosPolicy (Section 6.5.10 on page 376)

l

RELIABILITY QosPolicy (Section 6.5.19 on page 400)

l

DURABILITY SERVICE QosPolicy (Section 6.5.8 on the facing page)

l

AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on page 337)

6.5.7.4 Applicable Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

371

6.5.7.5 System Resource Considerations

6.5.7.5 System Resource Considerations
Using this policy with a setting other than VOLATILE will cause Connext DDS to use CPU and network bandwidth to send old DDS samples to matching, newly discovered DataReaders. The actual
amount of resources depends on the total size of data that needs to be sent.
The maximum number of DDS samples that will be kept on the DataWriter’s queue for late-joiners and/or
required subscriptions is determined by max_samples in RESOURCE_LIMITS Qos Policy.
System Resource Considerations With Required Subscriptions”
By default, when TRANSIENT_LOCAL durability is used in combination with required subscriptions, a
DataWriter configured with KEEP_ALL in the HISTORY QosPolicy (Section 6.5.10 on page 376) will
keep the DDS samples in its cache until they are acknowledged by all the required subscriptions. After the
DDS samples are acknowledged by the required subscriptions they will be marked as reclaimable, but they
will not be purged from the DataWriter’s queue until the DataWriter needs these resources for new DDS
samples. This may lead to a non efficient resource utilization, specially when max_samples is high or
even UNLIMITED.
The DataWriter’s behavior can be changed to purge DDS samples after they have been acknowledged by
all the active/matching DataReaders and all the required subscriptions configured on the DataWriter. To
do so, set the dds.data_writer.history.purge_samples_after_acknowledgment property to 1 (see
PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394)).

6.5.8 DURABILITY SERVICE QosPolicy
This QosPolicy is only used if the DURABILITY QosPolicy (Section 6.5.7 on page 368) is
PERSISTENT or TRANSIENT and you are using Persistence Service, which is included with the Connext DDS Professional, Evaluation, and Basic package types. It is used to store and possibly forward the
data sent by the DataWriter to DataReaders that are created after the data was initially sent.
This QosPolicy configures certain parameters of Persistence Service when it operates on the behalf of the
DataWriter, such as how much data to store. Specifically, this QosPolicy configures the HISTORY and
RESOURCE_LIMITS used by the fictitious DataReader and DataWriter used by Persistence Service.
Note however, that by default, Persistence Service will ignore the values in the DURABILITY SERVICE
QosPolicy (Section 6.5.8 above) and must be configured to use those values.
For more information, please see:
l

Mechanisms for Achieving Information Durability and Persistence (Section Chapter 12 on page
675)

l

Introduction to RTI Persistence Service (Section Chapter 26 on page 933)

l

Configuring Persistence Service (Section Chapter 27 on page 934)

372

6.5.8 DURABILITY SERVICE QosPolicy

This QosPolicy includes the members in Table 6.44 DDS_DurabilityServiceQosPolicy. For default values, please refer to the API Reference HTML documentation.
Table 6.44 DDS_DurabilityServiceQosPolicy
Type

Field
Name

Description
How long to keep all information regarding an instance.

DDS_Duration_t

service_
cleanup_
delay

Can be:
Zero (default): Purge disposed instances from Persistence Service
immediately. However, this will only happen if use_durability_service = 1.
INFINITE: Do not purge disposed instances.

DDS_
history_kind
HistoryQosPolicyKind

Settings to use for the HISTORY QosPolicy (Section 6.5.10 on

page 376) when recouping durable data.
DDS_Long

history_depth
max_samples

DDS_Long

max_
instances

Settings to use for the RESOURCE_LIMITS QosPolicy (Section

6.5.20 on page 405) when feeding data to a late joiner.
max_
samples_per_
instance

The service_cleanup_delay in this QosPolicy controls when Persistence Service may remove all information
regarding a data-instances. Information on a data-instance is maintained until all of the following conditions are met:
1. The instance has been explicitly disposed
(instance_state = NOT_ALIVE_DISPOSED).
2.

All samples for the disposed instance have been acknowledged, including the dispose sample itself.

3. A time interval longer that DurabilityService QosPolicy’s service_cleanup_delay has elapsed since
the time that Connext DDS detected that the previous two conditions were met. (Note: Only values
of zero or INFINITE are currently supported for service_cleanup_delay.)
The service_cleanup_delay field is useful in the situation where your application disposes an instance and
it crashes before it has a chance to complete additional tasks related to the disposition. Upon restart, your
application may ask for initial data to regain its state and the delay introduced by service_cleanup_delay
will allow your restarted application to receive the information about the disposed instance and complete
any interrupted tasks.

373

6.5.8.1 Properties

Although you can set the DURABILITY_SERVICE QosPolicy on a Topic, this is only useful as a means
to initialize the DURABILITY_SERVICE QosPolicy of a DataWriter. A Topic’s DURABILITY_
SERVICE setting does not directly affect the operation of Connext DDS, see Setting Topic QosPolicies
(Section 5.1.3 on page 204).
6.5.8.1 Properties
This QosPolicy cannot be modified after the Entity has been enabled.
It does not apply to DataReaders, so there is no requirement for setting it compatibly on the sending and
receiving sides.
6.5.8.2 Related QosPolicies
l

DURABILITY QosPolicy (Section 6.5.7 on page 368)

l

HISTORY QosPolicy (Section 6.5.10 on page 376)

l

RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)

6.5.8.3 Applicable Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

6.5.8.4 System Resource Considerations
Since this QosPolicy configures the HISTORY and RESOURCE_LIMITS used by the fictitious
DataReader and DataWriter used by Persistence Service, it does have some impact on resource usage.

6.5.9 ENTITY_NAME QosPolicy (DDS Extension)
The ENTITY_NAME QosPolicy assigns a name and role name to a DomainParticipant, Publisher, Subscriber, DataReader, or DataWriter.
How the name is used is strictly application-dependent.
It is useful to attach names that are meaningful to the user. These names (except for Publishers and Subscribers) are propagated during discovery so that applications can use these names to identify, in a usercontext, the entities that it discovers. Also, Connext DDS tools will print the names of discovered entities
(except for Publishers and Subscribers).
The role_name identifies the role of the entity. It is used by the Collaborative DataWriter feature (see
Availability QoS Policy and Collaborative DataWriters (Section 6.5.1.1 on page 338)). With Durable Subscriptions, role_name is used to specify to which Durable Subscription the DataReader belongs. (see
Availability QoS Policy and Required Subscriptions (Section 6.5.1.2 on page 339).

374

6.5.9.1 Properties

This QosPolicy contains the members listed in Table 6.45 DDS_EntityNameQoSPolicy.
Table 6.45 DDS_EntityNameQoSPolicy
Type

Field
Name

Description
A null-terminated string up to 255 characters in length.

char * name
To set this in XML, see Entity Names (Section 17.4.8 on page 809).
A null-terminated string up to 255 characters in length.
To set this in XML, see Entity Names (Section 17.4.8 on page 809).
char *

role_
name

For Collaborative DataWriters, this name is used to specify to which endpoint group the DataWriter belongs. See.
Availability QoS Policy and Collaborative DataWriters (Section 6.5.1.1 on page 338).
For Required and Durable Subscriptions this name is used to specify to which Subscription the DataReader belongs.
See Required Subscriptions (Section 6.3.13 on page 294).

These names will appear in the built-in topic for the entity (see the tables in Built-in DataReaders (Section
16.2 on page 773)).
Prior to get_qos(), if the name and/or role_name field in this QosPolicy is not null, Connext DDS
assumes the memory to be valid and big enough and may write to it. If that is not desired, set name and/or
role_name to NULL before calling get_qos() and Connext DDS will allocate adequate memory for name.
When you call the destructor of entity’s QoS structure (DomainParticipantQos, DataReaderQos, or
DataWriterQos) (in C++, C++/CLI, and C#) or Qos_finalize() (in C), Connext DDS will attempt
to free the memory used for name and role_name if it is not NULL. If this behavior is not desired, set
name and/or role_name to NULL before you call the destructor of entity’s QoS structure or DomainParticipantQos_finalize().
6.5.9.1 Properties
This QosPolicy cannot be modified after the entity is enabled.
6.5.9.2 Related QosPolicies
l

None

6.5.9.3 Applicable Entities
l

DomainParticipants (Section 8.3 on page 547)

l

Publishers (Section 6.2 on page 243)

l

Subscribers (Section 7.2 on page 440)

375

6.5.9.4 System Resource Considerations

l

DataReaders (Section 7.3 on page 459)

l

DataWriters (Section 6.3 on page 261)

6.5.9.4 System Resource Considerations
If the value of name in this QosPolicy is not NULL, some memory will be consumed in storing the information in the database, but should not significantly impact the use of resource.

6.5.10 HISTORY QosPolicy
This QosPolicy configures the number of DDS samples that Connext DDS will store locally for
DataWriters and DataReaders. For keyed Topics, this QosPolicy applies on a per instance basis, so that
Connext DDS will attempt to store the configured value of DDS samples for every instance (see DDS
Samples, Instances, and Keys (Section 2.3.1 on page 14) for a discussion of keys and instances).
It includes the members seen in Table 6.46 DDS_HistoryQosPolicy. For defaults and valid ranges, please
refer to the API Reference HTML documentation.
Table 6.46 DDS_HistoryQosPolicy
Type

Field
Name

DDS_
HistoryQosPolicyKind

kind

DDS_Long

depth

Description
DDS_KEEP_LAST_HISTORY_QOS: keep the last depth number of DDS samples per instance.
DDS_KEEP_ALL_HISTORY_QOS: keep all DDS samples.1
If kind = DDS_KEEP_LAST_HISTORY_QOS, this is how many DDS samples to keep per instance.2
if kind = DDS_KEEP_ALL_HISTORY_QOS, this value is ignored.

1Connext DDS will store up to the value of the max_samples_per_instance parameter of the RESOURCE_LIMITS

QosPolicy (Section 6.5.20 on page 405).
2depth must be <= max_samples_per_instance parameter of the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on

page 405)

376

6.5.10 HISTORY QosPolicy

Table 6.46 DDS_HistoryQosPolicy
Field
Name

Type

Description
Specifies how a DataWriter should handle previously written DDS samples for a new DataReader.
When a new DataReader matches a DataWriter, the DataWriter can be configured to perform content-based
filtering on previously written DDS samples stored in the DataWriter queue for the new DataReader.
May be:
l

DDS_
RefilterQosPolicyKind

DDS_NONE_REFILTER_QOS Do not filter existing DDS samples for a new DataReader. The
DataReader will do the filtering.

refilter
l

DDS_ALL_REFILTER_QOS Filter all existing DDS samples for a newly matched DataReader.
l

DDS_ON_DEMAND_REFILTER_QOS Filter existing DDS samples only when they are requested
by the DataReader.
(An extension to the DDS standard.)

The kind determines whether or not to save a configured number of DDS samples or all DDS samples. It
can be set to either of the following:
l

DDS_KEEP_LAST_HISTORY_QOSConnext DDS attempts to keep the latest values of the
data-instance and discard the oldest ones when the limit as set by the depth parameter is reached;
new data will overwrite the oldest data in the queue. Thus the queue acts like a circular buffer of
length depth.
l For a DataWriter: Connext DDS attempts to keep the most recent depth DDS samples of
each instance (identified by a unique key) managed by the DataWriter.
l

l

For a DataReader: Connext DDS attempts to keep the most recent depth DDS samples
received for each instance (identified by a unique key) until the application takes them via the
DataReader's take() operation. See Accessing DDS Data Samples with Read or Take (Section 7.4.3 on page 493) for a discussion of the difference between read() and take().

DDS_KEEP_ALL_HISTORY_QOSConnext DDS attempts to keep all of the DDS samples of a
Topic.
l For a DataWriter: Connext DDS attempts to keep all DDS samples published by the
DataWriter.
l

For a DataReader: Connext DDS attempts to keep all DDS samples received by the
DataReader for a Topic (both keyed and non-keyed) until the application takes them via the
DataReader's take() operation. See Accessing DDS Data Samples with Read or Take (Sec-

377

6.5.10 HISTORY QosPolicy

tion 7.4.3 on page 493) for a discussion of the difference between read() and take().
l

The value of the depth parameter is ignored.

The above descriptions say “attempts to keep” because the actual number of DDS samples kept is subject
to the limitations imposed by the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405). All of
the DDS samples of all instances of a Topic share a single physical queue that is allocated for a DataWriter
or DataReader. The size of this queue is configured by the RESOURCE_LIMITS QosPolicy. If there are
many difference instances for a Topic, it is possible that the physical queue may run out of space before the
number of DDS samples reaches the depth for all instances.
In the KEEP_ALL case, Connext DDS can only keep as many DDS samples for a Topic (independent of
instances) as the size of the allocated queue. Connext DDS may or may not allocate more memory when
the queue is filled, depending on the settings in the RESOURCE_LIMITS QoSPolicy of the DataWriter
or DataReader.
This QosPolicy interacts with the RELIABILITY QosPolicy (Section 6.5.19 on page 400) by controlling
whether or not Connext DDS guarantees that ALL of the data sent is received or if only the last N data values
sent are guaranteed to be received (a reduced level of reliability using the KEEP_LAST setting).
However, the physical sizes of the send and receive queues are not controlled by the History QosPolicy.
The memory allocation for the queues is controlled by the RESOURCE_LIMITS QosPolicy (Section
6.5.20 on page 405). Also, the amount of data that is sent to new DataReaders who have configured their
DURABILITY QosPolicy (Section 6.5.7 on page 368) to receive previously published data is controlled
by the History QosPolicy.
What happens when the physical queue is filled depends both on the setting for the HISTORY QosPolicy
as well as the RELIABILITY QosPolicy.
l

DDS_KEEP_LAST_HISTORY_QOS
l If RELIABILITY is BEST_EFFORT: When the number of DDS samples for an instance in
the queue reaches the value of depth, a new DDS sample for the instance will replace the oldest DDS sample for the instance in the queue.
l

l

If RELIABILITY is RELIABLE: When the number of DDS samples for an instance in the
queue reaches the value of depth, a new DDS sample for the instance will replace the oldest
DDS sample for the instance in the queue—even if the DDS sample being overwritten has
not been fully acknowledged as being received by all reliable DataReaders. This implies that
the discarded DDS sample may be lost by some reliable DataReaders. Thus, when using the
KEEP_LAST setting, strict reliability is not guaranteed. See Reliable Communications (Section Chapter 10 on page 629) for a complete discussion on Connext DDS’s reliable protocol.

DDS_KEEP_ALL_HISTORY_QOS
l If RELIABILITY is BEST_EFFORT: If the number of DDS samples for an instance in the
queue reaches the value of the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page

378

6.5.10.1 Example

405)’s max_samples_per_instance field, a new DDS sample for the instance will replace the
oldest DDS sample for the instance in the queue (regardless of instance).
l

If RELIABILITY is RELIABLE: When the number of DDS samples for an instance in the
queue reaches the value of the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page
405)’s max_samples_per_instance field, then:
l

l

For a DataWriter—a new DDS sample for the instance will replace the oldest DDS
sample for the instance in the sending queue—only if the DDS sample being overwritten has been fully acknowledged as being received by all reliable DataReaders. If
the oldest DDS sample for the instance has not been fully acknowledged, the write()
operation trying to enter a new DDS sample for the instance into the sending queue will
block (for the max_blocking_time specified in the RELIABLE QosPolicy).
For a DataReader—a new DDS sample received by the DataReader will be discarded.
Because the DataReader will not acknowledge the discarded DDS sample, the
DataWriter is forced to resend the DDS sample. Hopefully, the next time the DDS
sample is received, there is space for the instance in the DataReader’s queue to store
(and accept, thus acknowledge) the DDS sample. A DDS sample will remain in the
DataReader’s queue for one of two reasons. The more common reason is that the user
application has not removed the DDS sample using the DataReader’s take() method.
Another reason is that the DDS sample has been received out of order and is not available to be taken or read by the user application until all older DDS samples have been
received.

Although you can set the HISTORY QosPolicy on Topics, its value can only be used to initialize the
HISTORY QosPolicies of either a DataWriter or DataReader. It does not directly affect the operation of
Connext DDS, see Setting Topic QosPolicies (Section 5.1.3 on page 204).
6.5.10.1 Example
To achieve strict reliability, you must (1) set the DataWriter’s and DataReader’s HISTORY QosPolicy to
KEEP_ALL, and (2) set the DataWriter’s and DataReader’s RELIABILITY QosPolicy to
RELIABLE.
See Reliable Communications (Section Chapter 10 on page 629) for a complete discussion on Connext
DDS’s reliable protocol.
See Controlling Queue Depth with the History QosPolicy (Section 10.3.3 on page 644).
6.5.10.2 Properties
This QosPolicy cannot be modified after the Entity has been enabled.
There is no requirement that the publishing and subscribing sides use compatible values.

379

6.5.10.3 Related QosPolicies

6.5.10.3 Related QosPolicies
l

BATCH QosPolicy (DDS Extension) (Section 6.5.2 on page 341)Do not configure the
DataReader’s depth to be shallower than the DataWriter's maximum batch size (batch_max_
data_size). Because batches are acknowledged as a group, a DataReader that cannot process an
entire batch will lose the remaining DDS samples in it.

l

RELIABILITY QosPolicy (Section 6.5.19 on page 400)

l

RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)

6.5.10.4 Applicable Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.10.5 System Resource Considerations
While this QosPolicy does not directly affect the system resources used by Connext DDS, the
RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) that must be used in conjunction with the
HISTORY QosPolicy (Section 6.5.10 on page 376) will affect the amount of memory that Connext DDS
will allocate for a DataWriter or DataReader.

6.5.11 LATENCYBUDGET QoS Policy
This QosPolicy can be used by a DDS implementation to change how it processes and sends data that has
low latency requirements. The DDS specification does not mandate whether or how this parameter is used.
Connext DDS uses it to prioritize the sending of asynchronously published data; see
ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension) (Section 6.4.1 on page 313).
This QosPolicy also applies to Topics. The Topic’s setting for the policy is ignored unless you explicitly
make the DataWriter use it.
It contains the single member listed in Table 6.47 DDS_LatencyBudgetQosPolicy.
Table 6.47 DDS_LatencyBudgetQosPolicy
Type
DDS_
Duration_t

Field
Name
duration

Description
Provides a hint as to the maximum acceptable delay from the time the data is written to the time it is received by
the subscribing applications.

380

6.5.11.1 Applicable Entities

6.5.11.1 Applicable Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.12 LIFESPAN QoS Policy
The purpose of this QoS is to avoid delivering stale data to the application by specifying how long the data
written by a DataWriter is considered valid.
Each data sample written by a DataWriter has an associated expiration time beyond which the data should
not be delivered to any application. Once the sample expires, the data will be removed from the
DataWriter and DataReader caches.
The expiration time of each sample from the DataWriter's cache is computed by adding the duration specified by this QoS policy to the time when the sample is added to the DataWriter's cache. This timestamp
is not necessarily equal to the sample's source timestamp that can be provided by the user using the
DataWriter's write_w_timestamp() or write_w_params() APIs.
The expiration time of each sample from the DataReader's cache is computed by adding the duration to
the reception timestamp.
The Lifespan QosPolicy can be used to control how much data is stored by Connext DDS. Even if it is
configured to store "all" of the data sent or received for a topic (see the HISTORY QosPolicy (Section
6.5.10 on page 376)), the total amount of data it stores may be limited by the Lifespan QosPolicy.
You may also use the Lifespan QosPolicy to ensure that applications do not receive or act on data, commands or messages that are too old and have "expired.”
It includes the single member listed in Table 6.48 DDS_LifespanQosPolicy. For the default and valid
range, please refer to the API Reference HTML documentation.
Table 6.48 DDS_LifespanQosPolicy
Type
DDS_Duration_t

Field Name
duration

Description
Maximum duration for the data's validity.

Although you can set the LIFESPAN QosPolicy on Topics, its value can only be used to initialize the
LIFESPAN QosPolicies of DataWriters. The Topic’s setting for this QosPolicy does not directly affect the
operation of Connext DDS, see Setting Topic QosPolicies (Section 5.1.3 on page 204).

381

6.5.12.1 Properties

6.5.12.1 Properties
This QoS policy can be modified after the entity is enabled.
It does not apply to DataReaders, so there is no requirement that the publishing and subscribing sides use
compatible values.
6.5.12.2 Related QoS Policies
l

l

BATCH QosPolicy (DDS Extension) (Section 6.5.2 on page 341)Be careful when configuring a
DataWriter with a Lifespan duration shorter than the batch flush period (batch_flush_delay). If
the batch does not fill up before the flush period elapses, the short duration will cause the DDS
samples to be lost without being sent.
DURABILITY QosPolicy (Section 6.5.7 on page 368)

6.5.12.3 Applicable Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

6.5.12.4 System Resource Considerations
The use of this policy does not significantly impact the use of resources.

6.5.13 LIVELINESS QosPolicy
The LIVELINESS QosPolicy specifies how Connext DDS determines whether a DataWriter is “alive.” A
DataWriter’s liveliness is used in combination with the OWNERSHIP QosPolicy (Section 6.5.15 on page
389) to maintain ownership of an instance (note that the DEADLINE QosPolicy (Section 6.5.5 on page
363) is also used to change ownership when a DataWriter is still alive). That is, for a DataWriter to own
an instance, the DataWriter must still be alive as well as honoring its DEADLINE contract.
It includes the members in Table 6.49 DDS_LivelinessQosPolicy. For defaults and valid ranges, please
refer to the API Reference HTML documentation.

382

6.5.13 LIVELINESS QosPolicy

Table 6.49 DDS_LivelinessQosPolicy
Type

Field
Name

Description
DDS_AUTOMATIC_LIVELINESS_QOS:
Connext DDS will automatically assert liveliness for the DataWriter at least as often as the
lease_duration.

DDS_
kind
LivelinessQosPolicyKind

DDS_MANUAL_BY_PARTICIPANT_LIVELINESS_QOS:
The DataWriter is assumed to be alive if any Entity within the same DomainParticipant has
asserted its liveliness.
DDS_MANUAL_BY_TOPIC_LIVELINESS_QOS:
Your application must explicitly assert the liveliness of the DataWriter within the lease_
duration.
The timeout by which liveliness must be asserted for the DataWriter or the DataWriter will be
considered inactive or not alive.

DDS_Duration_t

lease_
duration

Additionally, for DataReaders, the lease_duration also specifies the maximum period at which
Connext DDS will check to see if the matching DataWriter is still alive.

A DataReader will consider a DataWriter not alive if the DataWriter does
not assert its liveliness within the DataWriter's lease_duration, not the
DataReader's lease_duration.
The number of assertions a DataWriter will send during a lease_duration period.

DDS_Long

assertions_
per_lease_
duration

This field only applies to DataWriters using DDS_AUTOMATIC_LIVELINESS_QOS kind
and it is not considered during QoS compatibility checks.
The default value is 3. A higher value will make the liveliness mechanism more robust against
packet losses, but it will also increase the network traffic.

Setting a DataWriter’s kind of LIVELINESS specifies the mechanism that will be used to assert liveliness
for the DataWriter. The DataWriter’s lease_duration then specifies the maximum period at which packets
that indicate that the DataWriter is still alive are sent to matching DataReaders.
The various mechanisms are:
l

DDS_AUTOMATIC_LIVELINESS_QOS:
The DomainParticipant is responsible for automatically sending packets to indicate that the
DataWriter is alive; this will be done at the rate determined by the assertions_per_lease_duration
and lease_duration values. This setting is appropriate when the primary failure mode is that the publishing application itself dies. It does not cover the case in which the application is still alive but in
an erroneous state–allowing the DomainParticipant to continue to assert liveliness for the
DataWriter but preventing threads from calling write() on the DataWriter.

383

6.5.13 LIVELINESS QosPolicy

As long as the internal threads spawned by Connext DDS for a DomainParticipant are running,
then the liveliness of the DataWriter will be asserted regardless of the state of the rest of the application.
This setting is certainly the most convenient, if the least accurate, method of asserting liveliness for a
DataWriter.
l

DDS_MANUAL_BY_PARTICIPANT_LIVELINESS_QOS:
Connext DDS will assume that as long as the user application has asserted the liveliness of at least
one DataWriter belonging to the same DomainParticipant or the liveliness of the DomainParticipant itself, then this DataWriter is also alive.
This setting allows the user code to control the assertion of liveliness for an entire group of
DataWriters with a single operation on any of the DataWriters or their DomainParticipant. Its a
good balance between control and convenience.

l

DDS_MANUAL_BY_TOPIC_LIVELINESS_QOS:
The DataWriter is considered alive only if the user application has explicitly called operations that
assert the liveliness for that particular DataWriter.
This setting forces the user application to assert the liveliness for a DataWriter which gives the user
application great control over when other applications can consider the DataWriter to be inactive,
but at the cost of convenience.

With the MANUAL_BY_[TOPIC,PARTICIPANT] settings, user application code can assert the liveliness
of DataWriters either explicitly by calling the assert_liveliness() operation on the DataWriter (as well as
the DomainParticipant for the MANUAL_BY_PARTICIPANT setting) or implicitly by calling write() on
the DataWriter. If the application does not use either of the methods mentioned at least once every lease_
duration, then the subscribing application may assume that the DataWriter is no longer alive. Sending
data MANUAL_BY_TOPIC will cause an assert message to be sent between the DataWriter and its
matched DataReaders.
Publishing applications will monitor their DataWriters to make sure that they are honoring their
LIVELINESS QosPolicy by asserting their liveliness at least at the period set by the lease_duration. If
Connext DDS finds that a DataWriter has failed to have its liveliness asserted by its lease_duration, an
internal thread will modify the DataWriter’s LIVELINESS_LOST_STATUS and trigger its on_liveliness_
lost() DataWriterListener callback if a listener exists, see Listeners (Section 4.4 on page 177).
Setting the DataReader’s kind of LIVELINESS requests a specific mechanism for the publishing application to maintain the liveliness of DataWriters. The subscribing application may want to know that the publishing application is explicitly asserting the liveliness of the matching DataWriter rather than inferring its
liveliness through the liveliness of its DomainParticipant or its sibling DataWriters.

384

6.5.13.1 Example

The DataReader’s lease_duration specifies the maximum period at which matching DataWriters must
have their liveliness asserted. In addition, in the subscribing application Connext DDS uses an internal
thread that wakes up at the period set by the DataReader’s lease_duration to see if the DataWriter’s
lease_duration has been violated.
When a matching DataWriter is determined to be dead (inactive), Connext DDS will modify the
LIVELINESS_CHANGED_STATUS of each matching DataReader and trigger that DataReader’s on_
liveliness_changed() DataReaderListener callback (if a listener exists).
Although you can set the LIVELINESS QosPolicy on Topics, its value can only be used to initialize the
LIVELINESS QosPolicies of either a DataWriter or DataReader. It does not directly affect the operation
of Connext DDS, see Setting Topic QosPolicies (Section 5.1.3 on page 204).
For more information on Liveliness, see Maintaining DataWriter Liveliness for kinds AUTOMATIC and
MANUAL_BY_PARTICIPANT (Section 14.3.1.2 on page 724).
6.5.13.1 Example
You can use LIVELINESS QosPolicy during system integration to ensure that applications have been
coded to meet design specifications. You can also use it during run time to detect when systems are performing outside of design specifications. Receiving applications can take appropriate actions in response to
disconnected DataWriters.
The LIVELINESS QosPolicy can be used to manage fail-over when the OWNERSHIP QosPolicy (Section 6.5.15 on page 389) is set to EXCLUSIVE. This implies that the DataReader will only receive data
from the highest strength DataWriter that is alive (active). When that DataWriter’s liveliness expires, then
Connext DDS will start delivering data from the next highest strength DataWriter that is still alive.
6.5.13.2 Properties
This QosPolicy cannot be modified after the Entity has been enabled.
The DataWriter and DataReader must use compatible settings for this QosPolicy. To be compatible, both
of the following conditions must be true:
The DataWriter and DataReader must use one of the valid combinations shown in Table 6.50 Valid Combinations of Liveliness ‘kind’.
DataWriter’s lease_duration <= DataReader’s lease_duration.
If this QosPolicy is found to be incompatible, the ON_OFFERED_INCOMPATIBLE_QOS and ON_
REQUESTED_INCOMPATIBLE_QOS statuses will be modified and the corresponding Listeners
called for the DataWriter and DataReader respectively.

385

6.5.13.3 Related QosPolicies

Table 6.50 Valid Combinations of Liveliness ‘kind’
DataReader requests:
MANUAL_
BY_
TOPIC

DataWriter
offers:

MANUAL_BY_
PARTICIPANT

AUTOMATIC

MANUAL_BY_TOPIC

4

4

4

MANUAL_BY_
PARTICIPANT

incompatible

4

4

AUTOMATIC

incompatible

incompatible

4

6.5.13.3 Related QosPolicies
l

DEADLINE QosPolicy (Section 6.5.5 on page 363)

l

OWNERSHIP QosPolicy (Section 6.5.15 on page 389)

l

OWNERSHIP_STRENGTH QosPolicy (Section 6.5.16 on page 393)

6.5.13.4 Applicable Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.13.5 System Resource Considerations
An internal thread in Connext DDS will wake up periodically to check the liveliness of all the
DataWriters. This happens both in the application that contains the DataWriters at the lease_duration set
on the DataWriters as well as the applications that contain the DataReaders at the lease_duration set on
the DataReaders. Therefore, as lease_duration becomes smaller, more CPU will be used to wake up
threads and perform checks. A short lease_duration (or a high assertions_per_lease_duration) set on
DataWriters may also use more network bandwidth because liveliness packets are being sent at a higher
rate—this is especially true when LIVELINESS kind is set to AUTOMATIC.

6.5.14 MULTI_CHANNEL QosPolicy (DDS Extension)
This QosPolicy is used to partition the data published by a DataWriter across multiple channels. A channel is defined by a filter expression and a sequence of multicast locators.
By using this QosPolicy, a DataWriter can be configured to send data to different multicast groups based
on the content of the data. Using syntax similar to those used in Content-Based Filters, you can associate

386

6.5.14 MULTI_CHANNEL QosPolicy (DDS Extension)

different multicast addresses with filter expressions that operate on the values of the fields within the data.
When your application’s code calls write(), data is sent to any multicast address for which the data passes
the filter.
See Multi-channel DataWriters (Section Chapter 18 on page 824) for complete documentation on multichannel DataWriters.
Note:Durable writer history is not supported for multi-channel DataWriters (see Multi-channel
DataWriters (Section Chapter 18 on page 824)); an error is reported if a multi-channel DataWriter tries to
configure Durable Writer History.
This QosPolicy includes the members presented in Table 6.51 DDS_MultiChannelQosPolicy, Table 6.52
DDS_ChannelSettings_t, and Table 6.53 DDS_TransportMulticastSettings_t. For defaults and valid
ranges, please refer to the API Reference HTML documentation.
Table 6.51 DDS_MultiChannelQosPolicy
Type

Field
Name

Description

DDS_
A sequence of channel settings used to configure the channels’ properties. If the length of the sequence is
channels
ChannelSettingsSeq
zero, the QosPolicy will be ignored. See Table 6.52 DDS_ChannelSettings_t.
Name of the filter class used to describe the filter expressions1. The following values are supported:
filter_
name

char *

DDS_SQLFILTER_NAME (see SQL Filter Expression Notation (Section 5.4.6 on page 222))
DDS_STRINGMATCHFILTER_NAME (see STRINGMATCH Filter Expression Notation (Section
5.4.7 on page 231))

Table 6.52 DDS_ChannelSettings_t
Type

Field
Name

Description
A sequence of multicast settings used to configure the multicast addresses associated with a
channel. The sequence cannot be empty.

DDS_
multicast_
The maximum number of multicast locators in a channel is limited to four. (A locator is
TransportMulticastSettingsSeq settings
defined by a transport alias, a multicast address and a port.) See Table 6.53 DDS_
TransportMulticastSettings_t.

1 In Java and C#, you can access the names of the built-in filters by using

DomainParticipant.SQLFILTER_NAME and DomainParticipant.STRINGMATCHFILTER_NAME.

387

6.5.14 MULTI_CHANNEL QosPolicy (DDS Extension)

Table 6.52 DDS_ChannelSettings_t
Field
Name

Type

Description
A logical expression used to determine the data that will be published in the channel.

filter_
This string cannot be NULL. An empty string always evaluates to TRUE.
expression
See SQL Filter Expression Notation (Section 5.4.6 on page 222) and STRINGMATCH
Filter Expression Notation (Section 5.4.7 on page 231) for expression syntax.

char *

A positive integer designating the relative priority of the channel, used to determine the
transmission order of pending transmissions. Larger numbers have higher priority.

DDS_Long

priority

To use publication priorities, the DataWriter’s PUBLISH_MODE QosPolicy (DDS
Extension) (Section 6.5.18 on page 397) must be set for asynchronous publishing and the
DataWriter must use a FlowController that is configured for highest-priority-first (HPF)
scheduling.
See Prioritized DDS Samples (Section 6.6.4 on page 428).
Note: Prioritized DDS samples are not supported when using the Java, Ada, or .NET APIs.
Therefore the priority field does not exist when using these APIs.

Table 6.53 DDS_TransportMulticastSettings_t
Type

Field
Name

Description

DDS_
StringSeq

transports

A sequence of transport aliases that specifies which transport should be used to publish multicast messages
for this channel.

char *

receive_
address

A multicast group address on which DataReaders subscribing to this channel will receive data.

DDS_Long

receive_port

The multicast port on which DataReaders subscribing to this channel will receive data.

The format of the filter_expression should correspond to one of the following filter classes:
l
l

DDS_SQLFILTER_NAME (see SQL Filter Expression Notation (Section 5.4.6 on page 222))
DDS_STRINGMATCHFILTER_NAME (see STRINGMATCH Filter Expression Notation (Section 5.4.7 on page 231)

A DataReader can use the ContentFilteredTopic API (see Using a ContentFilteredTopic (Section 5.4.5
on page 219)) to subscribe to a subset of the channels used by a DataWriter.

388

6.5.14.1 Example

6.5.14.1 Example
See Multi-channel DataWriters (Section Chapter 18 on page 824).
6.5.14.2 Properties
This QosPolicy cannot be modified after the DataWriter is created.
It does not apply to DataReaders, so there is no requirement that the publishing and subscribing sides use
compatible values.
6.5.14.3 Related Qos Policies
l

DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
on page 593)

6.5.14.4 Applicable Entities
l

DataWriters (Section 6.3 on page 261)

6.5.14.5 System Resource Considerations
The following fields in the DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593) configure the resources associated with the channels stored in the
MULTI_CHANNEL QosPolicy:
l

channel_seq_max_length

l

channel_filter_expression_max_length

For information about partitioning topic data across multiple channels, please refer to Multi-channel
DataWriters (Section Chapter 18 on page 824).

6.5.15 OWNERSHIP QosPolicy
The OWNERSHIP QosPolicy specifies whether a DataReader receive data for an instance of a Topic sent
by multiple DataWriters.
For non-keyed Topics, there is only one instance of the Topic.
This policy includes the single member shown in Table 6.54 DDS_OwnershipQosPolicy.

389

6.5.15 OWNERSHIP QosPolicy

Table 6.54 DDS_OwnershipQosPolicy
Type

Field Name

Description
DDS_SHARED_OWNERSHIP_QOS or

DDS_OwnershipQosPolicyKind

kind
DDS_EXCLUSIVE_OWNERSHIP_QOS

The kind of OWNERSHIP can be set to one of two values:
l

SHARED Ownership
When OWNERSHIP is SHARED, and multiple DataWriters for the Topic publishes the value of
the same instance, all the updates are delivered to subscribing DataReaders. So in effect, there is no
“owner;” no single DataWriter is responsible for updating the value of an instance. The subscribing
application will receive modifications from all DataWriters.

l

EXCLUSIVE Ownership
When OWNERSHIP is EXCLUSIVE, each instance can only be owned by one DataWriter at a
time. This means that a single DataWriter is identified as the exclusive owner whose updates are
allowed to modify the value of the instance for matching DataWriters. Other DataWriters may submit modifications for the instance, but only those made by the current owner are passed on to the
DataReaders. If a non-owner DataWriter modifies an instance, no error or notification is made; the
modification is simply ignored. The owner of the instance can change dynamically.
Note for non-keyed Topics, EXCLUSIVE ownership implies that DataReaders will pay attention
to only one DataWriter at a time because there is only a single instance. For keyed Topics,
DataReaders may actually receive data from multiple DataWriters when different DataWriters own
different instances of the Topic.

This QosPolicy is often used to help users build systems that have redundant elements to safeguard against
component or application failures. When systems have active and hot standby components, the Ownership
QosPolicy can be used to ensure that data from standby applications are only delivered in the case of the
failure of the primary.
The Ownership QosPolicy can also be used to create data channels or topics that are designed to be taken
over by external applications for testing or maintenance purposes.
Although you can set the OWNERSHIP QosPolicy on Topics, its value can only be used to initialize the
OWNERSHIP QosPolicies of either a DataWriter or DataReader. It does not directly affect the operation
of Connext DDS, see Setting Topic QosPolicies (Section 5.1.3 on page 204).

390

6.5.15.1 How Connext DDS Selects which DataWriter is the Exclusive Owner

6.5.15.1 How Connext DDS Selects which DataWriter is the Exclusive Owner
When OWNERSHIP is EXCLUSIVE, the owner of an instance at any given time is the DataWriter with
the highest OWNERSHIP_STRENGTH QosPolicy (Section 6.5.16 on page 393) that is “alive” as
defined by the LIVELINESS QosPolicy (Section 6.5.13 on page 382)) and has not violated the
DEADLINE QosPolicy (Section 6.5.5 on page 363) of the DataReader. OWNERSHIP_STRENGTH is
simply an integer set by the DataWriter.
If the Topic’s data type is keyed (see DDS Samples, Instances, and Keys (Section 2.3.1 on page 14)),
EXCLUSIVE ownership is determined on a per-instance basis. That is, the DataWriter owner of each
instance is considered separately. A DataReader can receive values written by a lower strength
DataWriter as long as those values are for instances that are not being written by a higher-strength
DataWriter.
If there are multiple DataWriters with the same OWNERSHIP_STRENGTH writing to the same instance,
Connext DDS resolves the tie by choosing the DataWriter with the smallest GUID (Globally Unique Identifier, see Simple Participant Discovery (Section 14.1.1 on page 710).). This means that different
DataReaders (in different applications) of the same Topic will all choose the same DataWriter as the
owner when there are multiple DataWriters with the same strength.
The owner of an instance can change when:
l
l

l
l

A DataWriter with a higher OWNERSHIP_STRENGTH publishes a value for the instance.
The OWNERSHIP_STRENGTH of the owning DataWriter is dynamically changed to be less than
the strength of an existing DataWriter of the instance.
The owning DataWriter stops asserting its LIVELINESS (the DataWriter dies).
The owning DataWriter violates the DEADLINE QosPolicy by not updating the value of the
instance within the period set by the DEADLINE.

Note however, the change of ownership is not synchronous across different DataReaders in different participants. That is, DataReaders in different applications may not determine that the ownership of an
instance has changed at exactly the same time.
6.5.15.2 Example
OWNERSHIP is really a property that is shared between DataReaders and DataWriters of a Topic.
However, in a system, some Topics will be exclusively owned and others will be shared. System requirements will determine which are which.
An example of a Topic that may be shared is one that is used by applications to publish alarm messages. If
the application detects an anomalous condition, it will use a DataWriter to write a Topic “Alarm.” Another
application that records alarms into a system log file will have a DataReader that subscribes to “Alarm.” In
this example, any number of applications can publish the “Alarm” message. There is no concept that only

391

6.5.15.3 Properties

one application at a time is allowed to publish the “Alarm” message, so in this case, the OWNERSHIP of
the DataWriters and DataReaders should be set to SHARED.
In a different part of the system, EXCLUSIVE OWNERSHIP may be used to implement redundancy in
support of fault tolerance. Say, the distributed system controls a traffic system. It monitors traffic and
changes the information posted on signs, the operation of metering lights, and the timing of traffic lights.
This system must be tolerant to failure of any part of the system including the application that actually
issues commands to change the lights at a particular intersection.
One way to implement fault tolerance is to create the system redundantly both in hardware and software.
So if a piece of the running system fails, a backup can take over. In systems where failover from the
primary to backup system must be seamless and transparent, the actual mechanics of failover must be fast,
and the redundant component must immediately pickup where the failed component left off. For the network connections of the component, Connext DDS can provided redundant DataWriter and
DataReaders.
In this case, you would not want the DataReaders to receive redundant messages from the redundant
DataWriters. Instead you will want the DataReaders to only receive messages from the primary application and only from a backup application when a failure occurs. To continue our example, if we have
redundant applications that all try to control the lights at an intersection, we would want the DataReaders
on the light to receive messages only from the primary application. To do so, we should configure the
DataWriters and DataReaders to have EXCLUSIVE OWNERSHIP and set the OWNERSHIP_
STRENGTH differently on different redundant applications to distinguish between primary and backup
systems.
6.5.15.3 Properties
This QosPolicy cannot be modified after the Entity is enabled.
It must be set to the same kind on both the publishing and subscribing sides. If a DataWriter and
DataReader of the same topic are found to have different kinds set for the OWNERSHIP QoS, the ON_
OFFERED_INCOMPATIBLE_QOS and ON_REQUESTED_INCOMPATIBLE_QOS statuses
will be modified and the corresponding Listeners called for the DataWriter and DataReader respectively.
6.5.15.4 Related QosPolicies
l

DEADLINE QosPolicy (Section 6.5.5 on page 363)

l

LIVELINESS QosPolicy (Section 6.5.13 on page 382)

l

OWNERSHIP_STRENGTH QosPolicy (Section 6.5.16 on the next page)

392

6.5.15.5 Applicable Entities

6.5.15.5 Applicable Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.15.6 System Resource Considerations
This QosPolicy does not significantly impact the use of system resources.

6.5.16 OWNERSHIP_STRENGTH QosPolicy
The OWNERSHIP_STRENGTH QosPolicy is used to rank DataWriters of the same instance of a Topic,
so that Connext DDS can decide which DataWriter will have ownership of the instance when the
OWNERSHIP QosPolicy (Section 6.5.15 on page 389) is set to EXCLUSIVE.
It includes the member in Table 6.55 DDS_OwnershipStrengthQosPolicy. For the default and valid range,
please refer to the API Reference HTML documentation.
Table 6.55 DDS_OwnershipStrengthQosPolicy
Type
DDS_Long

Field Name
value

Description
The strength value used to arbitrate among multiple DataWriters.

This QosPolicy only applies to DataWriters when EXCLUSIVE OWNERSHIP is used. The strength is
simply an integer value, and the DataWriter with the largest value is the owner. A deterministic method is
used to decide which DataWriter is the owner when there are multiple DataWriters that have equal
strengths. See How Connext DDS Selects which DataWriter is the Exclusive Owner (Section 6.5.15.1 on
page 391) for more details.
6.5.16.1 Example
Suppose there are two DataWriters sending DDS samples of the same Topic instance, one as the main
DataWriter, and the other as a backup. If you want to make sure the DataReader always receive from the
main one whenever possible, then set the main DataWriter to use a higher ownership_strength value
than the one used by the backup DataWriter.
6.5.16.2 Properties
This QosPolicy can be changed at any time.
It does not apply to DataReaders, so there is no requirement that the publishing and subscribing sides use
compatible values.

393

6.5.16.3 Related QosPolicies

6.5.16.3 Related QosPolicies
l

OWNERSHIP QosPolicy (Section 6.5.15 on page 389)

6.5.16.4 Applicable Entities
l

DataWriters (Section 6.3 on page 261)

6.5.16.5 System Resource Considerations
The use of this policy does not significantly impact the use of resources.

6.5.17 PROPERTY QosPolicy (DDS Extension)
The PROPERTY QosPolicy stores name/value (string) pairs that can be used to configure certain parameters of Connext DDS that are not exposed through formal QoS policies.
It can also be used to store and propagate application-specific name/value pairs that can be retrieved by
user code during discovery. This is similar to the USER_DATA QosPolicy, except this policy uses (name,
value) pairs, and you can select whether or not a particular pair should be propagated (included in the
built-in topic).
It includes the member in Table 6.56 DDS_PropertyQosPolicy.
Table 6.56 DDS_PropertyQosPolicy
Type
DDS_
PropertySeq

Field
Name
value

Description
A sequence of: (name, value) pairs and booleans that indicate whether the pair should be propagated (included in
the entity’s built-in topic upon discovery).

The Property QoS stores name/value pairs for an Entity. Both the name and value are strings. Certain configurable parameters for Entities that do not have a formal DDS QoS definition may be configured via this
QoS by using a pre-defined name and the desired setting in string form.
You can manipulate the sequence of properties (name, value pairs) with the standard methods available for
sequences. You can also use the helper class, DDSPropertyQosPolicyHelper, which provides another way
to work with a PropertyQosPolicy object.
The PropertyQosPolicy may be used to configure:
l

Durable writer history (see How To Configure Durable Writer History (Section 12.3.2 on page
683))

394

6.5.17 PROPERTY QosPolicy (DDS Extension)

l

l

Durable reader state (see How To Configure a DataReader for Durable Reader State (Section 12.4.4
on page 690))
Built-in and extension Transport Plugins (see Setting Builtin Transport Properties with the PropertyQosPolicy (Section 15.6 on page 748), Setting Up a Transport with the Property QoS (Section
25.2 on page 915), Configuring the TCP Transport (Section 35.1 on page 993)).

l

Automatic registration of built-in types (see Registering Built-in Types (Section 3.2.1 on page 30))

l

Clock Selection (Section 8.6 on page 619)

l

l

Turbo Mode and Automatic Throttling for DataWriter Performance—Experimental Features (Section 6.3.18 on page 312)
Location or content of your license from RTI (see License Management, in the Getting Started
Guide)

In addition, you can add your own name/value pairs to the Property QoS of an Entity. You may also use
this QosPolicy to direct Connext DDS to propagate these name/value pairs with the discovery information
for the Entity. Applications that discover the Entity can then access the user-specific name/value pairs in
the discovery information of the remote Entity. This allows you to add meta-information about an Entity
for application-specific use, for example, authentication/authorization certificates (which can also be done
using the User or Group Data QoS).
Reasons for using the PropertyQosPolicy include:
l

l

Some features can only be configured through the PropertyQosPolicy, not through other QoS or
API.s For example, Durable Reader State, Durable Writer History, Built-in Types, Monotonic
Clock.
Alternative way to configure built-in transports settings. For example, to use non-default values for
the built-in transports without using the PropertyQosPolicy, you would have to create a DomainParticipant disabled, change the built-in transport property settings, then enable the DomainParticipant.
Using the PropertyQosPolicy to configure built-in transport settings will save you the work of
enabling and disabling the DomainParticipant. Also, transport settings are not a QoS and therefore
cannot be configured through an XML file. By configuring built-in transport settings through the
PropertyQosPolicy instead, XML files can be used.
When using the Java or .NET APIs, transport configuration must take place through the
PropertyQosPolicy (not through the transport property structures).

395

6.5.17 PROPERTY QosPolicy (DDS Extension)

l
l

l

l

Alternative way to support multiple instances of built-in transports (without using Transport API).
Alternative way to dynamically load extension transports (such as RTI Secure WAN Transport1 or
RTI TCP Transport2) or user-created transport plugins in C/C++ language bindings. If the extension
or user-created transport plugin is installed using the transport API instead, the library that extra transport library/code will need to be linked into your application and may require recompilation.
Allows full pluggable transport configuration for non-C/C++ language bindings (Java, C++/CLI,
C#, etc.) The pluggable transport API is not available in those languages. Without using PropertyQosPolicy, you cannot use extension transports (such as RTI Secure WAN Transport) and you
cannot create your own custom transport.
Alternative way to provide a license for platforms that do not support a file system, or if a default
license location is not feasible and environment variables are not supported.

The PropertyQosPolicyHelper operations are described in Table 6.57 PropertyQoSPolicyHelper Operations. For more information, see the API Reference HTML documentation.
Table 6.57 PropertyQoSPolicyHelper Operations
Operation

Description

get_number_of_properties

Gets the number of properties in the input policy.

assert_property

Asserts the property identified by name in the input policy. (Either adds it, or replaces an existing one.)

add_property

Adds a new property to the input policy.

assert_pointer_property

Asserts the property identified by name in the input policy.
Used when the property to store is a pointer.

add_pointer_property

Adds a new property to the input policy.
Used when the property to store is a pointer.

lookup_property

Searches for a property in the input policy given its name.

remove_property

Removes a property from the input policy.

get_properties

Retrieves a list of properties whose names match the input prefix.

1RTI Secure WAN Transport is an optional component that is installed separately.
2RTI TCP Transport is included with your Connext DDS distribution but is not a built-in transport and therefore not

enabled by default.

396

6.5.17.1 Properties

6.5.17.1 Properties
This QosPolicy can be changed at any time.
There is no requirement that the publishing and subscribing sides use compatible values.
6.5.17.2 Related QosPolicies
l

DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
on page 593)

6.5.17.3 Applicable Entities
l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

l

DomainParticipants (Section 8.3 on page 547)

6.5.17.4 System Resource Considerations
The DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on
page 593) contains several fields for configuring the resources associated with the properties stored in this
QosPolicy.

6.5.18 PUBLISH_MODE QosPolicy (DDS Extension)
This QosPolicy determines the DataWriter’s publishing mode, either asynchronous or synchronous.
The publishing mode controls whether data is written synchronously—in the context of the user thread
when calling write(), or asynchronously—in the context of a separate thread internal to Connext DDS.
DataWriters do not perform sender-side filtering. Any filtering, such as time-based or
content-based filtering, takes place on the DataReader side.
Note: Asynchronous

Each Publisher spawns a single asynchronous publishing thread (set in its ASYNCHRONOUS_
PUBLISHER QosPolicy (DDS Extension) (Section 6.4.1 on page 313)) to serve all its asynchronous
DataWriters.
When data is written asynchronously, a FlowController (FlowControllers (DDS Extension) (Section 6.6
on page 422)), identified by flow_controller_name, can be used to shape the network traffic. The
FlowController's properties determine when the asynchronous publishing thread is allowed to send data
and how much.
The fastest way for Connext DDS to send data is for the user thread to execute the middleware code that
actually sends the data itself. However, there are times when user applications may need or want an
internal middleware thread to send the data instead. For instance, for sending large data reliably, an asyn-

397

6.5.18 PUBLISH_MODE QosPolicy (DDS Extension)

chronous thread must be used (see ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension) (Section 6.4.1 on page 313)).
This QosPolicy can select a FlowController to prioritize or shape the data flow sent by a DataWriter to
DataReaders. Shaping a data flow usually means limiting the maximum data rates with which the middleware will send data for a DataWriter. The FlowController will buffer data sent faster than the maximum
rate by the DataWriter, and then only send the excess data when the user send rate drops below the maximum rate.
If kind is set to DDS_ASYNCHRONOUS_PUBLISH_MODE_QOS, the flow controller referred to by
flow_controller_name must exist. Otherwise, the setting will be considered inconsistent.
This QosPolicy includes the members in Table 6.58 DDS_PublishModeQosPolicy. For the defaults,
please refer to the API Reference HTML documentation.
Table 6.58 DDS_PublishModeQosPolicy
Type

Field
Name

Description
Either:

DDS_
PublishMode
kind
QosPolicyKind

l

DDS_ASYNCHRONOUS_PUBLISH_MODE_QOS
l

DDS_SYNCHRONOUS_PUBLISH_MODE_QOS
Name of the associated flow controller.
There are three built-in FlowControllers:
l

DDS_DEFAULT_FLOW_CONTROLLER_NAME
char*

flow_
controller_
name

l

DDS_FIXED_RATE_FLOW_CONTROLLER_NAME
l

DDS_ON_DEMAND_FLOW_CONTROLLER_NAME
You may also create your own FlowControllers.
See FlowControllers (DDS Extension) (Section 6.6 on page 422).
A positive integer designating the relative priority of the DataWriter, used to determine the transmission
order of pending writes.

DDS_Long

priority

To use publication priorities, this QosPolicy’s kind must be DDS_ASYNCHRONOUS_PUBLISH_
MODE_QOS and the DataWriter must use a FlowController with a highest-priority first (HPF)
scheduling_policy.
See Prioritized DDS Samples (Section 6.6.4 on page 428).
Note: Prioritized DDS samples are not supported when using the Java, Ada, or .NET APIs. Therefore the
priority field does not exist when using these APIs.

398

6.5.18.1 Properties

The maximum number of DDS samples that will be coalesced depends on NDDS_Transport_Property_
t::gather_send_buffer_count_max (each DDS sample requires at least 2-4 gather-send buffers). Performance can be improved by increasing NDDS_Transport_Property_t::gather_send_buffer_count_
max. Note that the maximum value is operating system dependent.
Connext DDS queues DDS samples until they can be sent by the asynchronous publishing thread (as
determined by the corresponding FlowController).
The number of DDS samples that will be queued is determined by the HISTORY QosPolicy (Section
6.5.10 on page 376): when using KEEP_LAST, the most recent depth DDS samples are kept in the
queue.
Once unsent DDS samples are removed from the queue, they are no longer available to the asynchronous
publishing thread and will therefore never be sent.
Unless flow_controller_name points to one of the built-in FlowControllers, finalizing the DataWriterQos
will also free the string pointed to by flow_controller_name. Therefore, you should use DDS_String_
dup() before passing the string to flow_controller_name, or reset flow_controller_name to NULL
before the destructing /finalizing the QoS.
Advantages of Asynchronous Publishing:
Asynchronous publishing may increase latency, but offers the following advantages:
l

l

l

The write() call does not make any network calls and is therefore faster and more deterministic. This
becomes important when the user thread is executing time-critical code.
When data is written in bursts or when sending large data types as multiple fragments, a flow controller can throttle the send rate of the asynchronous publishing thread to avoid flooding the network.
Asynchronously written DDS samples for the same destination will be coalesced into a single network packet which reduces bandwidth consumption.

6.5.18.1 Properties
This QosPolicy cannot be modified after the Publisher is created.
Since it is only for DataWriters, there are no compatibility restrictions for how it is set on the publishing
and subscribing sides.
6.5.18.2 Related QosPolicies
l

ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension) (Section 6.4.1 on page 313)

l

HISTORY QosPolicy (Section 6.5.10 on page 376)

399

6.5.18.3 Applicable Entities

6.5.18.3 Applicable Entities
l

DataWriters (Section 6.3 on page 261)

6.5.18.4 System Resource Considerations
See Configuring Resource Limits for Asynchronous DataWriters (Section 6.5.20.1 on page 406).
System resource usage depends on the settings in the corresponding FlowController (see FlowControllers
(DDS Extension) (Section 6.6 on page 422)).

6.5.19 RELIABILITY QosPolicy
This RELIABILITY QosPolicy determines whether or not data published by a DataWriter will be reliably
delivered by Connext DDS to matching DataReaders. The reliability protocol used by Connext DDS is
discussed in Reliable Communications (Section Chapter 10 on page 629).
The reliability of a connection between a DataWriter and DataReader is entirely user configurable. It can
be done on a per DataWriter/DataReader connection. A connection may be configured to be "best effort"
which means that Connext DDS will not use any resources to monitor or guarantee that the data sent by a
DataWriter is received by a DataReader.
For some use cases, such as the periodic update of sensor values to a GUI displaying the value to a person,
"best effort" delivery is often good enough. It is certainly the fastest, most efficient, and least resourceintensive (CPU and network bandwidth) method of getting the newest/latest value for a topic from
DataWriters to DataReaders. But there is no guarantee that the data sent will be received. It may be lost
due to a variety of factors, including data loss by the physical transport such as wireless RF or even Ethernet. Packets received out of order are dropped and a SAMPLE_LOST Status (Section 7.3.7.7 on page
478) is generated.
However, there are data streams (topics) in which you want an absolute guarantee that all data sent by a
DataWriter is received reliably by DataReaders. This means that Connext DDS must check whether or
not data was received, and repair any data that was lost by resending a copy of the data as many times as it
takes for the DataReader to receive the data.
Connext DDS uses a reliability protocol configured and tuned by these QoS policies:
l

HISTORY QosPolicy (Section 6.5.10 on page 376),

l

DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347),

l

DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511),

l

RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)

400

6.5.19 RELIABILITY QosPolicy

The Reliability QoS policy is simply a switch to turn on the reliability protocol for a
DataWriter/DataReader connection. The level of reliability provided by Connext DDS is determined by
the configuration of the aforementioned QoS policies.
You can configure Connext DDS to deliver ALL data in the order they were sent (also known as absolute
or strict reliability). Or, as a trade-off for less memory, CPU, and network usage, you can choose a
reduced level of reliability where only the last N values are guaranteed to be delivered reliably to
DataReaders (where N is user-configurable). With the reduced level of reliability, there are no guarantees
that the data sent before the last N are received. Only the last N data packets are monitored and repaired if
necessary.
It includes the members in Table 6.59 DDS_ReliabilityQosPolicy. For defaults and valid ranges, please
refer to the API Reference HTML documentation.
Table 6.59 DDS_ReliabilityQosPolicy
Type

Field Name

Description
Can be either:
l

DDS_
ReliabilityQosPolicyKind

DDS_BEST_EFFORT_RELIABILITY_QOS:
DDS data samples are sent once and missed samples are acceptable.

kind
l

DDS_RELIABLE_RELIABILITY_QOS:
Connext DDS will make sure that data sent is received and missed DDS
samples are resent.

DDS_Duration_t

max_blocking_
time

How long a DataWriter can block on a write() when the send queue is full due to
unacknowledged messages. (Has no meaning for DataReaders.)
Kind of reliable acknowledgment.
Only applies when kind is RELIABLE.
Sets the kind of acknowledgments supported by a DataWriter and sent by DataReader.
Possible values:

DDS_ReliabilityQosPolicy- acknowledgment_
AcknowledgmentModeKind kind

l

DDS_PROTOCOL_ACKNOWLEDGMENT_MODE
l

DDS_APPLICATION_AUTO_ACKNOWLEDGMENT_MODE
l

DDS_APPLICATION_EXPLICIT_ACKNOWLEDGMENT_MODE
See Application Acknowledgment Kinds (Section 6.3.12.1 on page 289)

The kind of RELIABILITY can be either:

401

6.5.19 RELIABILITY QosPolicy

l

BEST_EFFORT
Connext DDS will send DDS data samples only once to DataReaders. No effort or resources are
spent to track whether or not sent DDS samples are received. Minimal resources are used. This is
the most deterministic method of sending data since there is no indeterministic delay that can be introduced by buffering or resending data. DDS data samples may be lost. This setting is good for periodic data.

l

RELIABLE
Connext DDS will send DDS samples reliably to DataReaders–buffering sent data until they have
been acknowledged as being received by DataReaders and resending any DDS samples that may
have been lost during transport. Additional resources configured by the HISTORY and
RESOURCE_LIMITS QosPolicies may be used. Extra packets will be sent on the network to
query (heartbeat) and acknowledge the receipt of DDS samples by the DataReader. This setting is a
good choice when guaranteed data delivery is required; for example, sending events or commands.

To send large data reliably, you will also need to set the PUBLISH_MODE QosPolicy (DDS
Extension) (Section 6.5.18 on page 397) kind to DDS_ASYNCHRONOUS_PUBLISH_
MODE_QOS. Large in this context means that the data cannot be sent as a single packet by a
transport (for example, data larger than 63K when using UDP/IP).
While a DataWriter sends data reliably, the HISTORY QosPolicy (Section 6.5.10 on page 376) and
RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) determine how many DDS samples can
be stored while waiting for acknowledgements from DataReaders. A DDS sample that is sent reliably is
entered in the DataWriter’s send queue awaiting acknowledgement from DataReaders. How many DDS
samples that the DataWriter is allowed to store in the send queue for a data-instance depends on the kind
of the HISTORY QoS as well as the max_samples_per_instance and max_samples parameter of the
RESOURCE_LIMITS QoS.
If the HISTORY kind is KEEP_LAST, then the DataWriter is allowed to have the HISTORY depth
number of DDS samples per instance of the Topic in the send queue. Should the number of unacknowledge DDS samples in the send queue for a data-instance reach the HISTORY depth, then the next
DDS sample written by the DataWriter for the instance will overwrite the oldest DDS sample for the
instance in the queue. This implies that an unacknowledged DDS sample may be overwritten and thus
lost. So even if the RELIABILITY kind is RELIABLE, if the HISTORY kind is KEEP_LAST, it is
possible that some data sent by the DataWriter will not be delivered to the DataReader. What is guaranteed is that if the DataWriter stops writing, the last N DDS samples that the DataWriter wrote will be
delivered reliably; where n is the value of the HISTORY depth.
However, if the HISTORY kind is KEEP_ALL, then when the send queue is filled with acknowledged
DDS samples (either due to the number of unacknowledged DDS samples for an instance reaching the
RESOURCE_LIMITS max_samples_per_instance value or the total number of unacknowledged DDS
samples have reached the size of the send queue as specified by RESOURCE_LIMITS max_samples),

402

6.5.19.1 Example

the next write() operation on the DataWriter will block until either a DDS sample in the queue has been
fully acknowledged by DataReaders and thus can be overwritten or a timeout of RELIABILITY max_
blocking_period has been reached.
If there is still no space in the queue when max_blocking_time is reached, the write() call will return a
failure with the error code DDS_RETCODE_TIMEOUT.
Thus for strict reliability—a guarantee that all DDS data samples sent by a DataWriter are received by
DataReaders—you must use a RELIABILITY kind of RELIABLE and a HISTORY kind of KEEP_
ALL for both the DataWriter and the DataReader.
Although you can set the RELIABILITY QosPolicy on Topics, its value can only be used to initialize the
RELIABILITY QosPolicies of either a DataWriter or DataReader. It does not directly affect the operation
of Connext DDS, see Setting Topic QosPolicies (Section 5.1.3 on page 204).
6.5.19.1 Example
This QosPolicy is used to achieve reliable communications, which is discussed in Reliable Communications (Section Chapter 10 on page 629) and Enabling Reliability (Section 10.3.1 on page 637).
6.5.19.2 Properties
This QosPolicy cannot be modified after the Entity has been enabled.
The DataWriter and DataReader must use compatible settings for this QosPolicy. To be compatible, the
DataWriter and DataReader must use one of the valid combinations for the Reliability kind (see Table
6.60 Valid Combinations of Reliability ‘kind’), and one of the valid combinations for the acknowledgment_kind (see Table 6.61 Valid Combinations of Reliability ‘acknowledgment_kind’):
Table 6.60 Valid Combinations of Reliability ‘kind’
DataReader requests:
BEST_EFFORT

RELIABLE

BEST_EFFORT

4

incompatible

RELIABLE

4

4

DataWriter offers:

403

6.5.19.3 Related QosPolicies

Table 6.61 Valid Combinations of Reliability ‘acknowledgment_kind’
DataReader requests:
APPLICATION_
AUTO

PROTOCOL

DataWriter offers:

APPLICATION_
EXPLICIT

PROTOCOL

4

incompatible

incompatible

APPLICATION_AUTO

4

4

4

APPLICATION_EXPLICIT

4

4

4

If this QosPolicy is found to be incompatible, statuses ON_OFFERED_INCOMPATIBLE_QOS and
ON_REQUESTED_INCOMPATIBLE_QOS will be modified and the corresponding Listeners called
for the DataWriter and DataReader, respectively.
There are no compatibility issues regarding the value of max_blocking_wait, since it does not apply to
DataReaders.
6.5.19.3 Related QosPolicies
l

HISTORY QosPolicy (Section 6.5.10 on page 376)

l

PUBLISH_MODE QosPolicy (DDS Extension) (Section 6.5.18 on page 397)

l

RESOURCE_LIMITS QosPolicy (Section 6.5.20 on the next page)

6.5.19.4 Applicable Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.19.5 System Resource Considerations
Setting the kind to RELIABLE will cause Connext DDS to use up more resources to monitor and maintain a reliable connection between a DataWriter and all of its reliable DataReaders. This includes the use
of extra CPU and network bandwidth to send and process heartbeat, ACK/NACK, and repair packets (see
Reliable Communications (Section Chapter 10 on page 629)).
Setting max_blocking_time to a non-zero number may block the sending thread when the
RELIABILITY kind is RELIABLE.

404

6.5.20 RESOURCE_LIMITS QosPolicy

6.5.20 RESOURCE_LIMITS QosPolicy
For the reliability protocol (and the DURABILITY QosPolicy (Section 6.5.7 on page 368)), this
QosPolicy determines the actual maximum queue size when the HISTORY QosPolicy (Section 6.5.10 on
page 376) is set to KEEP_ALL.
In general, this QosPolicy is used to limit the amount of system memory that Connext DDS can allocate. For
embedded real-time systems and safety-critical systems, pre-determination of maximum memory usage is
often required. In addition, dynamic memory allocation could introduce non-deterministic latencies in timecritical paths.
This QosPolicy can be set such that an entity does not dynamically allocate any more memory after its initialization phase.
It includes the members in Table 6.62 DDS_ResourceLimitsQosPolicy. For defaults and valid ranges,
please refer to the API Reference HTML documentation.
Table 6.62 DDS_ResourceLimitsQosPolicy
Type
DDS_
Long

Field
Name
max_
samples

Description
Maximum number of live DDS samples that Connext DDS can store for a DataWriter/DataReader. This is a
physical limit.
Maximum number of instances that can be managed by a DataWriter/DataReader.

DDS_
Long

max_
instances

For DataReaders, max_instances must be <= max_total_instances in the DATA_READER_RESOURCE_
LIMITS QosPolicy (DDS Extension) (Section 7.6.2 on page 517).
See also: Example (Section 6.5.20.3 on page 407).

DDS_
Long

Maximum number of DDS samples of any one instance that Connext DDS will store for a DataWriter/DataReader.
max_
samples_ For keyed types and DataReaders, this value only applies to DDS samples with an instance state of DDS_ALIVE_
per_
INSTANCE_STATE.
instance
If a keyed Topic is not used, then max_samples_per_instance must equal max_samples.

DDS_
Long

initial_
samples

Initial number of DDS samples that Connext DDS will store for a DataWriter/DataReader. (DDS extension)

DDS_
Long

initial_
instances

Initial number of instances that can be managed by a DataWriter/DataReader. (DDS extension)

DDS_
Long

instance_
hash_
Number of hash buckets, which are used by Connext DDS to facilitate instance lookup. (DDS extension).
buckets

405

6.5.20.1 Configuring Resource Limits for Asynchronous DataWriters

One of the most important fields is max_samples, which sets the size and causes memory to be allocated
for the send or receive queues. For information on how this policy affects reliability, see Tuning Queue
Sizes and Other Resource Limits (Section 10.3.2 on page 638).
When a DataWriter or DataReader is created, the initial_instances and initial_samples parameters
determine the amount of memory first allocated for the those Entities. As the application executes, if more
space is needed in the send/receive queues to store DDS samples or as more instances are created, then
Connext DDS will automatically allocate memory until the limits of max_instances and max_samples are
reached.
You may set initial_instances = max_instances and initial_samples = max_samples if you do not want
Connext DDS to dynamically allocate memory after initialization.
For keyed Topics, the max_samples_per_instance field in this policy represents maximum number of
DDS samples with the same key that are allowed to be stored by a DataWriter or DataReader. This is a
logical limit. The hard physical limit is determined by max_samples. However, because the theoretical
number of instances may be quite large (as set by max_instances), you may not want Connext DDS to
allocate the total memory needed to hold the maximum number of DDS samples per instance for all possible instances (max_samples_per_instance * max_instances) because during normal operations, the
application will never have to hold that much data for the Entity.
So it is possible that an Entity will hit the physical limit max_samples before it hits the max_samples_
per_instance limit for a particular instance. However, Connext DDS must be able to store max_samples_
per_instance for at least one instance. Therefore, max_samples_per_instance must be <= max_
samples.
If a keyed data type is not used, there is only a single instance of the Topic, so max_samples_per_
instance must equal max_samples.
Once a physical or logical limit is hit, then how Connext DDS deals with new DDS data samples being
sent or received for a DataWriter or DataReader is described in the HISTORY QosPolicy (Section 6.5.10
on page 376) setting of DDS_KEEP_ALL_HISTORY_QOS. It is closely tied to whether or not a reliable connection is being maintained.
Although you can set the RESOURCE_LIMITS QosPolicy on Topics, its value can only be used to initialize the RESOURCE_LIMITS QosPolicies of either a DataWriter or DataReader. It does not directly
affect the operation of Connext DDS, see Setting Topic QosPolicies (Section 5.1.3 on page 204).
6.5.20.1 Configuring Resource Limits for Asynchronous DataWriters
When using an asynchronous Publisher, if a call to write() is blocked due to a resource limit, the block
will last until the timeout period expires, which will prevent others from freeing the resource. To avoid this
situation, make sure that the DomainParticipant’s outstanding_asynchronous_sample_allocation in the
DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page
593) is always greater than the sum of all asynchronous DataWriters’ max_samples.

406

6.5.20.2 Configuring DataWriter Instance Replacement

6.5.20.2 Configuring DataWriter Instance Replacement
When the max_instances limit is reached, a DataWriter will try to make space for a new instance by replacing an existing instance according to the instance replacement kind set in instance_replacement. For the
sake of instance replacement, an instance is considered to be unregistered, disposed, or alive. The oldest
instance of the specified kind, if such an instance exists, would be replaced with the new instance. Also, all
DDS samples of a replaced instance must already have been acknowledged, such that removing the
instance would not deprive any existing reader from receiving them.
Since an unregistered instance is one that a DataWriter will not update any further, unregistered instances
are replaced before any other instance kinds. This applies for all instance_replacement kinds; for
example, the ALIVE_THEN_DISPOSED kind would first replace unregistered, then alive, and then disposed instances. The rest of the kinds specify one or two kinds (e.g DISPOSED and ALIVE_OR_
DISPOSED). For the single kind, if no unregistered instances are replaceable, and no instances of the specified kind are replaceable, then the instance replacement will fail. For the others specifying multiple kinds,
it either specifies to look for one kind first and then another kind (e.g. ALIVE_THEN_DISPOSED),
meaning if the first kind is found then that instance will be replaced, or it will replace either of the kinds
specified (e.g. ALIVE_OR_DISPOSED), whichever is older as determined by the time of instance registering, writing, or disposing.
If an acknowledged instance of the specified kind is found, the DataWriter will reclaim its resources for
the new instance. It will also invoke the DataWriterListener’s on_instance_replaced() callback (if
installed) and notify the user with the handle of the replaced instance, which can then be used to retrieve
the instance key from within the callback. If no replaceable instances are found, the new instance will fail
to be registered; the DataWriter may block, if the instance registration was done in the context of a write,
or it may return with an out-of-resources return code.
In addition, replace_empty_instances (in the DATA_WRITER_RESOURCE_LIMITS QosPolicy
(DDS Extension) (Section 6.5.4 on page 359)) configures whether instances with no DDS samples are eligible to be replaced. If this is set, then a DataWriter will first try to replace empty instances, even before
replacing unregistered instances.
6.5.20.3 Example
If you want to be able to store max_samples_per_instance for every instance, then you should set
max_samples >= max_instances * max_samples_per_instance

But if you want to save memory and you do not expect that the running application will ever reach the
case where it will see max_instances of instances, then you may use a smaller value for max_samples to
save memory.
In any case, there is a lower limit for max_samples:
max_samples >= max_samples_per_instance

407

6.5.20.4 Properties

If the HISTORY QosPolicy (Section 6.5.10 on page 376)’s kind is set to KEEP_LAST, then you
should set:
max_samples_per_instance = HISTORY.depth

6.5.20.4 Properties
This QosPolicy cannot be modified after the Entity is enabled.
There are no requirements that the publishing and subscribing sides use compatible values.
6.5.20.5 Related QosPolicies
l

HISTORY QosPolicy (Section 6.5.10 on page 376)

l

RELIABILITY QosPolicy (Section 6.5.19 on page 400)

l

For DataReaders, max_instances must be <= max_total_instances in the DATA_READER_
RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 7.6.2 on page 517)

6.5.20.6 Applicable Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.20.7 System Resource Considerations
Larger initial_* numbers will increase the initial system memory usage. Larger max_* numbers will
increase the worst-case system memory usage.
Increasing instance_hash_buckets speeds up instance-lookup time but also increases memory usage.

6.5.21 SERVICE QosPolicy (DDS Extension)
The SERVICE QosPolicy is intended for use by RTI infrastructure services. User applications should not
modify its value. It includes the member in Table 6.63 DDS_ServiceQosPolicy.

408

6.5.21.1 Properties

Table 6.63 DDS_ServiceQosPolicy
Type

Field Name

Description
Kind of service associated with the entity.
Possible values:
DDS_NO_SERVICE_QOS,
DDS_PERSISTENCE_SERVICE_QOS,

DDS_ServiceQosPolicyKind

kind

DDS_QUEUING_SERVICE_QOS,
DDS_ROUTING_SERVICE_QOS,
DDS_RECORDING_SERVICE_QOS,
DDS_REPLAY_SERVICE_QOS,
DDS_DATABASE_INTEGRATION_SERVICE_QOS

An application can determine the kind of service associated with a discovered DataWriter and
DataReader by looking at the service field in the PublicationBuiltinTopicData and SubscriptionBuiltinTopicData structures (see Chapter 16: Built-In Topics).
6.5.21.1 Properties
This QosPolicy cannot be modified after the Entity is enabled.
There are no requirements that the publishing and subscribing sides use compatible values.
6.5.21.2 Related QosPolicies
None
6.5.21.3 Applicable Entities
l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3)

6.5.21.4 System Resource Considerations
None.

6.5.22 TRANSPORT_PRIORITY QosPolicy
The TRANSPORT_PRIORITY QosPolicy is optional and only partially supported on certain OSs and
transports by RTI. However, its intention is to allow you to specify on a per-DataWriter or perDataReader basis that the data sent by a DataWriter or DataReader is of a different priority.

409

6.5.22.1 Example

DDS does not specify how a DDS implementation shall treat data of different priorities. It is often difficult
or impossible for DDS implementations to treat data of higher priority differently than data of lower priority, especially when data is being sent (delivered to a physical transport) directly by the thread that called
DataWriter’s write() operation. Also, many physical network transports themselves do not have an enduser controllable level of data packet priority.
In Connext DDS, for the UDPv4 built-in transport, the value set in the TRANSPORT_PRIORITY
QosPolicy is used in a setsockopt call to set the TOS (type of service) bits of the IPv4 header for datagrams sent by a DataWriter or DataReader. It is platform dependent on how and whether or not the setsockopt has an effect. On some platforms such as Windows and Linux, external permissions must be given
to the user application in order to set the TOS bits.
It is incorrect to assume that using the TRANSPORT_PRIORITY QosPolicy will have any effect at all on
the end-to-end delivery of data between a DataWriter and DataReader. All network elements such as
switches and routers must have the capability and be enabled to actually use the TOS bits to treat higherpriority packets differently. Thus the ability to use the TRANSPORT_PRIORITY QosPolicy must be
designed and configured at a system level; just turning it on in an application may have no effect at all.
It includes the member in Table 6.64 DDS_TransportPriorityQosPolicy. For the default and valid range,
please refer to the API Reference HTML documentation.
Table 6.64 DDS_TransportPriorityQosPolicy
Type
DDS_Long

Field Name
value

Description
Hint as to how to set the priority.

Connext DDS will propagate the value set on a per-DataWriter or per-DataReader basis to the transport
when the DataWriter publishes data. It is up to the implementation of the transport to do something with
the value, if anything.
You can set the TRANSPORT_PRIORITY QosPolicy on a Topic and use its value to initialize the
TRANSPORT_PRIORITY QosPolicies of DataWriters and DataReaders. The TRANSPORT_
PRIORITY QosPolicy of a Topic does not directly affect the operation of Connext DDS, see Setting
Topic QosPolicies (Section 5.1.3 on page 204).
6.5.22.1 Example
Should Connext DDS be configured with a transport that can use and will honor the concept of a prioritized message, then you would be able to create a DataWriter of a Topic whose DDS data samples,
when published, will be sent at a higher priority than other DataWriters that use the same transport.
6.5.22.2 Properties
This QosPolicy cannot be modified after the entity is created.

410

6.5.22.3 Related QosPolicies

6.5.22.3 Related QosPolicies
This QosPolicy does not interact with any other policies.
6.5.22.4 Applicable Entities
l

Topics (Section 5.1 on page 200)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.22.5 System Resource Considerations
The use of this policy does not significantly impact the use of resources. However, if a transport is implemented to use the value set by this policy, then there may be transport-specific issues regarding the
resources that the transport implementation itself uses.

6.5.23 TRANSPORT_SELECTION QosPolicy (DDS Extension)
The TRANSPORT_SELECTION QosPolicy allows you to select the transports that have been installed
with the DomainParticipant to be used by the DataWriter or DataReader.
An application may be simultaneously connected to many different physical transports, e.g., Ethernet, Infiniband, shared memory, VME backplane, and wireless. By default, the middleware will use up to 4 transports to deliver data from a DataWriter to a DataReader.
This QosPolicy can be used to both limit and control which of the application’s available transports may
be used by a DataWriter to send data or by a DataReader to receive data.
It includes the member in Table 6.65 DDS_TransportSelectionQosPolicy. For more information, please
refer to the API Reference HTML documentation.
Table 6.65 DDS_TransportSelectionQosPolicy
Type
DDS_StringSeq

Field Name
enabled_transports

Description
A sequence of aliases for the transports that may be used by the DataWriter or DataReader.

Connext DDS allows you to configure the transports that it uses to send and receive messages. A number
of built-in transports, such as UDPv4 and shared memory, are available as well as custom ones that you
may implement and install. Each transport will be installed in the DomainParticipant with one or more aliases.
To enable a DataWriter or DataReader to use a particular transport, add the alias to the enabled_transports sequence of this QosPolicy. An empty sequence is a special case, and indicates that all transports
installed in the DomainParticipant can be used by the DataWriter or DataReader.

411

6.5.23.1 Example

For more information on configuring and installing transports, please see the API Reference HTML documentation (from the Modules page, select RTI DDS API Reference, Pluggable Transports).
6.5.23.1 Example
Suppose a DomainParticipant has both UDPv4 and shared memory transports installed. If you want a particular DataWriter to publish its data only over shared memory, then you should use this QosPolicy to specify that restriction.
6.5.23.2 Properties
This QosPolicy cannot be modified after the Entity is created.
It can be set differently for the DataWriter and the DataReader.
6.5.23.3 Related QosPolicies
l

TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 below)

l

TRANSPORT_MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on page 529)

l

TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section 8.5.7 on page 606)

6.5.23.4 Applicable Entities
l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.23.5 System Resource Considerations
By restricting DataWriters from sending or DataReaders from receiving over certain transports, you may
decrease the load on those transports.

6.5.24 TRANSPORT_UNICAST QosPolicy (DDS Extension)
The TRANSPORT_UNICAST QosPolicy allows you to specify unicast network addresses to be used by
DomainParticipant, DataWriters and DataReaders for receiving messages.
Connext DDS may send data to a variety of Entities, not just DataReaders. DomainParticipants receive
messages to support the discovery process discussed in Discovery (Section Chapter 14 on page 709).
DataWriters may receive ACK/NACK messages to support the reliable protocol discussed in Reliable
Communications (Section Chapter 10 on page 629).
During discovery, each Entity announces to remote applications a list of (up to 4) unicast addresses to
which the remote application should send data (either user-data packets or reliable protocol meta-data such
as ACK/NACK and Heartbeats).

412

6.5.24 TRANSPORT_UNICAST QosPolicy (DDS Extension)

By default, the list of addresses is populated automatically with values obtained from the enabled transport
plugins allowed to be used by the Entity (see the TRANSPORT_BUILTIN QosPolicy (DDS Extension)
(Section 8.5.7 on page 606) and TRANSPORT_SELECTION QosPolicy (DDS Extension) (Section
6.5.23 on page 411)). Also, the associated ports are automatically determined (see Inbound Ports for User
Traffic (Section 14.5.2 on page 740)).
Use TRANSPORT_UNICAST QosPolicy to manually set the receive address list for an Entity. You may
optionally set a port to use a non-default receive port as well. Only the first 4 addresses will be used. Connext DDS will create a receive thread for every unique port number that it encounters (on a per transport
basis).
The QosPolicy structure includes the members in Table 6.66 DDS_TransportUnicastQosPolicy. For more
information and default values, please refer to the API Reference HTML documentation.
Table 6.66 DDS_TransportUnicastQosPolicy
Field
Name

Type
DDS_TransportUnicastSettingsSeq
(see Table 6.67 DDS_
TransportUnicastSettings_t)

value

Description

A sequence of up to 4 unicast settings that should be used by remote entities to address
messages to be sent to this Entity.

Table 6.67 DDS_TransportUnicastSettings_t
Type

Field
Name

Description

DDS_
A sequence of transport aliases that specifies which transports should be used to receive unicast messages for this
transports
StringSeq
Entity.
DDS_
Long

receive_
port

The port that should be used in the addressing of unicast messages destined for this Entity. A value of 0 will cause
Connext DDS to use a default port number based on domain and participant ids. See Ports Used for Discovery
(Section 14.5 on page 738).

A message sent to a unicast address will be received by a single node on the network (as opposed to a multicast address where a single message may be received by multiple nodes). This policy sets the unicast
addresses and ports that remote entities should use when sending messages to the Entity on which the
TRANSPORT_UNICAST QosPolicy is set.
Up to four “return” unicast addresses may be configured for an Entity. Instead of specifying addresses directly, you use the transports field of the DDS_TransportUnicastSetting_t to select the transports (using
their aliases) on which remote entities should send messages destined for this Entity. The addresses of the
selected transports will be the “return” addresses. See the API Reference HTML documentation about

413

6.5.24 TRANSPORT_UNICAST QosPolicy (DDS Extension)

configuring transports and aliases (from the Modules page, select RTI Connext DDS API Reference,
Pluggable Transports).
Note, a single transport may have more than one unicast address. For example, if a node has multiple network interface cards (NICs), then the UDPv4 transport will have an address for each NIC. When using the
TRANSPORT_UNICAST QosPolicy to set the return addresses, a single value for the DDS_TransportUnicastSettingsSeq may provide more than the four return addresses that Connext DDS currently
uses.
Whether or not you are able to configure the network interfaces that are allowed to be used by a transport
is up to the implementation of the transport. For the built-in UDPv4 transport, you may restrict an instance
of the transport to use a subset of the available network interfaces. See the API Reference HTML documentation for the built-in UDPv4 transport for more information.
For a DomainParticipant, this QoS policy sets the default list of addresses used by other applications to
send user data for local DataReaders.
For a reliable DataWriter, if set, the other applications will use the specified list of addresses to send reliable protocol packets (ACKS/NACKS) on the behalf of reliable DataReaders. Otherwise, if not set, the
other applications will use the addresses set by the DomainParticipant.
For a DataReader, if set, then other applications will use the specified list of addresses to send user data
(and reliable protocol packets for reliable DataReaders). Otherwise, if not set, the other applications will
use the addresses set by the DomainParticipant.
For a DataReader, if the port number specified by this QoS is the same as a port number specified by a
TRANSPORT_MULTICAST QoS, then the transport may choose to process data received both via multicast and unicast with a single thread. Whether or not a transport must use different threads to process data
received via multicast or unicast for the same port number depends on the implementation of the transport.
To use this QosPolicy, you also need to specify a port number. A port number of 0 will cause Connext
DDS to automatically use a default value. As explained in Ports Used for Discovery (Section 14.5 on page
738), the default port number for unicast addresses is based on the domain and participant IDs. Should you
choose to use a different port number, then for every unique port number used by Entities in your application, depending on the transport, Connext DDS may create a thread to process messages received for that
port on that transport. See Connext DDS Threading Model (Section Chapter 19 on page 837) for more
about threads.
Threads are created on a per-transport basis, so if this QosPolicy specifies multiple transports for a
receive_port, then a thread may be created for each transport for that unique port. Some transports may be
able to share a single thread for different ports, others can not. Different Entities can share the same port
number, and thus, the same thread will process all of the data for all of the Entities sharing the same port
number for a transport.

414

6.5.24.1 Example

Note: If a DataWriter is using the MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on
page 386), the unicast addresses specified in the TRANSPORT_UNICAST QosPolicy are ignored by that
DataWriter. The DataWriter will not publish DDS samples on those locators.
6.5.24.1 Example
You may use this QosPolicy to restrict an Entity from receiving data through a particular transport. For
example, on a multi-NIC (network interface card) system, you may install different transports for different
NICs. Then you can balance the network load between network cards by using different values for the
TRANSPORT_UNICAST QosPolicy for different DataReaders. Thus some DataReaders will receive
their data from one NIC and other DataReaders will receive their data from another.
6.5.24.2 Properties
This QosPolicy cannot be modified after the Entity is created.
It can be set differently for the DomainParticipant, the DataWriter and the DataReader.
6.5.24.3 Related QosPolicies
l

MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386)

l

TRANSPORT_SELECTION QosPolicy (DDS Extension) (Section 6.5.23 on page 411)

l

TRANSPORT_MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on page 529)

l

TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section 8.5.7 on page 606)

6.5.24.4 Applicable Entities
l

DomainParticipants (Section 8.3 on page 547)

l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

6.5.24.5 System Resource Considerations
Because this QosPolicy changes the transports on which messages are received for different Entities, the
bandwidth used on the different transports may be affected.
Depending on the implementation of a transport, Connext DDS may need to create threads to receive and
process data on a unique-port-number basis. Some transports can share the same thread to process data
received for different ports; others like UDPv4 must have different threads for different ports. In addition,
if the same port is used for both unicast and multicast, the transport implementation will determine whether
or not the same thread can be used to process both unicast and multicast data. For UDPv4, only one thread
is needed per port–independent of whether the data was received via unicast or multicast data. See Receive
Threads (Section 19.3 on page 839) for more information.
415

6.5.25 TYPESUPPORT QosPolicy (DDS Extension)

6.5.25 TYPESUPPORT QosPolicy (DDS Extension)
This policy can be used to modify the code generated by RTI Code Generator so that the [de]serialization
routines act differently depending on the information passed in via the object pointer. This policy also
determines if padding bytes are set to zero during serialization.
It includes the members in Table 6.68 DDS_TypeSupportQosPolicy.
Table 6.68 DDS_TypeSupportQosPolicy
Type

void *

Field
Name
plugin_
data

Description

Value to pass into the type plug-in's serialization/deserialization function. See Note below.
Determines whether or not the padding bytes will be set to zero during CDR serialization.
For a DomainParticipant: Configures how padding bytes are set when serializing data for the builtin topic
DataWriters and DataReaders.
For DataWriters and DataReaders: Configures how padding bytes are set when serializing data for that
entity.

May be:
cdr_
DDS_
padding_
CdrPaddingKind
l
kind
ZERO_CDR_PADDING (Padding bytes will be set to zero during CDR serialization)
l

NOT_SET_CDR_PADDING (Padding bytes will not be set to any value during CDR serialization)
l

AUTO_CDR_PADDING (For a DomainParticipant, the default behavior is NOT_SET_CDR_
PADDING. For a DataWriter or DataReader, the behavior is to inherit the value from the
DomainParticipant.)

Note: RTI generally recommends that you treat generated source files as compiler outputs
(analogous to object files) and that you do not modify them. RTI cannot support user changes to
generated source files. Furthermore, such changes would make upgrading to newer versions of
Connext DDS more difficult, as this generated code is considered to be a part of the middleware
implementation and consequently does change from version to version. The plugin_data field in
this QoS policy should be considered a back door, only to be used after careful design
consideration, testing, and consultation with your RTI representative.
6.5.25.1 Properties
This QoS policy may be modified after the DataWriter or DataReader is enabled.
It can be set differently for the DataWriter and DataReader.

416

6.5.25.2 Related QoS Policies

6.5.25.2 Related QoS Policies
None.
6.5.25.3 Applicable Entities
l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

l

DomainParticipants (Section 8.3 on page 547)

6.5.25.4 System Resource Considerations
None.

6.5.26 USER_DATA QosPolicy
This QosPolicy provides an area where your application can store additional information related to a
DomainParticipant, DataWriter, or DataReader. This information is passed between applications during
discovery (see Discovery (Section Chapter 14 on page 709)) using built-in-topics (see Built-In Topics (Section Chapter 16 on page 772)). How this information is used will be up to user code. Connext DDS does
not do anything with the information stored as USER_DATA except to pass it to other applications.
Use cases are usually for application-to-application identification, authentication, authorization, and encryption purposes. For example, applications can use Group or User Data to send security certificates to each
other for RSA-type security.
The value of the USER_DATA QosPolicy is sent to remote applications when they are first discovered, as
well as when the DomainParticipant, DataWriter or DataReader’s set_qos() methods are called after
changing the value of the USER_DATA. User code can set listeners on the built-in DataReaders of the
built-in Topics used by Connext DDS to propagate discovery information. Methods in the built-in topic
listeners will be called whenever new DomainParticipants, DataReaders, and DataWriters are found.
Within the user callback, you will have access to the USER_DATA that was set for the associated Entity.
Currently, USER_DATA of the associated Entity is only propagated with the information that declares a
DomainParticipant, DataWriter or DataReader. Thus, you will need to access the value of USER_
DATA through DDS_ParticipantBuiltinTopicData, DDS_PublicationBuiltinTopicData or DDS_SubscriptionBuiltinTopicData (see Built-In Topics (Section Chapter 16 on page 772)).
The structure for the USER_DATA QosPolicy includes just one field, as seen in Table 6.69 DDS_UserDataQosPolicy. The field is a sequence of octets that translates to a contiguous buffer of bytes whose contents and length is set by the user. The maximum size for the data are set in the DOMAIN_
PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593).

417

6.5.26.1 Example

Table 6.69 DDS_UserDataQosPolicy
Type
DDS_OctetSeq

Field Name
value

Description
Default: empty

This policy is similar to the GROUP_DATA QosPolicy (Section 6.4.4 on page 320) and TOPIC_DATA
QosPolicy (Section 5.2.1 on page 209) that apply to other types of Entities.
6.5.26.1 Example
One possible use of USER_DATA is to pass some credential or certificate that your subscriber application
can use to accept or reject communication with the DataWriters (or vice versa, where the publisher application can validate the permission of DataReaders to receive its data). Using the same method, an application (DomainParticipant) can accept or reject all connections from another application. The value of the
USER_DATA of the DomainParticipant is propagated in the ‘user_data’ field of the DDS_ParticipantBuiltinTopicData that is sent with the declaration of each DomainParticipant. Similarly, the
value of the USER_DATA of the DataWriter is propagated in the ‘user_data’ field of the DDS_PublicationBuiltinTopicData that is sent with the declaration of each DataWriter, and the value of the USER_
DATA of the DataReader is propagated in the ‘user_data’ field of the DDS_SubscriptionBuiltinTopicData that is sent with the declaration of each DataReader.
When Connext DDS discovers a DomainParticipant/DataWriter/DataReader, the application can be notified of the discovery of the new entity and retrieve information about the Entity’s QoS by reading the
DCPSParticipant, DCPSPublication or DCPSSubscription built-in topics (see Built-In Topics (Section
Chapter 16 on page 772)). The user application can then examine the USER_DATA field in the built-in
Topic and decide whether or not the remote Entity should be allowed to communicate with the local Entity.
If communication is not allowed, the application can use the DomainParticipant’s ignore_participant(),
ignore_publication() or ignore_subscription() operation to reject the newly discovered remote entity as
one with which the application allows Connext DDS to communicate. See Built-in DataReaders (Section
16.2 on page 773) for an example of how to do this.
6.5.26.2 Properties
This QosPolicy can be modified at any time. A change in the QosPolicy will cause Connext DDS to send
packets containing the new USER_DATA to all of the other applications in the DDS domain.
It can be set differently on the publishing and subscribing sides.
6.5.26.3 Related QosPolicies
l

TOPIC_DATA QosPolicy (Section 5.2.1 on page 209)

l

GROUP_DATA QosPolicy (Section 6.4.4 on page 320)

418

6.5.26.4 Applicable Entities

l

DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
on page 593)

6.5.26.4 Applicable Entities
l

DataWriters (Section 6.3 on page 261)

l

DataReaders (Section 7.3 on page 459)

l

DomainParticipants (Section 8.3 on page 547)

6.5.26.5 System Resource Considerations
The maximum size of the USER_DATA is set in the participant_user_data_max_length, writer_user_
data_max_length, and reader_user_data_max_length fields of the DOMAIN_PARTICIPANT_
RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593). Because Connext DDS
will allocated memory based on this value, you should only increase this value if you need to. If your system does not use USER_DATA, then you can set this value to 0 to save memory. Setting the value of the
USER_DATA QosPolicy to hold data longer than the value set in the [participant,writer,reader]_user_
data_max_length field will result in failure and an INCONSISTENT_QOS_POLICY return code.
However, should you decide to change the maximum size of USER_DATA, you must make certain that
all applications in the DDS domain have changed the value of [participant,writer,reader]_user_data_
max_length to be the same. If two applications have different limits on the size of USER_DATA, and
one application sets the USER_DATA QosPolicy to hold data that is greater than the maximum size set by
another application, then the DataWriters and DataReaders between the two applications will not connect.
The DomainParticipants may also reject connections from each other entirely. This is also true for the
GROUP_DATA (GROUP_DATA QosPolicy (Section 6.4.4 on page 320)) and TOPIC_DATA
(TOPIC_DATA QosPolicy (Section 5.2.1 on page 209)) QosPolicies.

6.5.27 WRITER_DATA_LIFECYCLE QoS Policy
This QoS policy controls how a DataWriter handles the lifecycle of the instances (keys) that the
DataWriter is registered to manage. This QoS policy includes the members in Table 6.70 DDS_WriterDataLifecycleQosPolicy.
Table 6.70 DDS_WriterDataLifecycleQosPolicy
Type

DDS_
Boolean

Field
Name

Description

autodispose_ RTI_TRUE (default): Instance is disposed when unregistered.
unregistered_
RTI_FALSE: Instance is not disposed when unregistered.
instances

419

6.5.27 WRITER_DATA_LIFECYCLE QoS Policy

Table 6.70 DDS_WriterDataLifecycleQosPolicy
Type

Field
Name

Description
Determines how long the DataWriter will maintain information regarding an instance that has been
unregistered.

struct
DDS_
Duration_
t

By default, the DataWriter resources associated with an instance (e.g., the space needed to remember the
autopurge_
Instance Key or KeyHash) are released lazily. This means the resources are only reclaimed when the space is
unregistered_
needed for another instance because max_instances (Section on page 405) (see RESOURCE_LIMITS
instance_
QosPolicy (Section 6.5.20 on page 405)) is exceeded. This behavior can be changed by setting autopurge_
delay
unregistered_instance_delay to a value other than INFINITE.
After this time elapses, the DataWriter will purge all internal information regarding the instance, including
historical DDS samples even if max_instances (Section on page 405) has not been reached.

You may use the DataWriter’s unregister() operation (Registering and Unregistering Instances (Section
6.3.14.1 on page 297)) to indicate that the DataWriter no longer wants to send data for a Topic. This QoS
controls whether or not Connext DDS automatically also calls dispose() (Disposing of Data (Section
6.3.14.2 on page 299)) on the behalf of the DataWriter for the data.
Unregistering vs. Disposing:
l

l

When an instance is unregistered, it means this particular DataWriter has no more information/data
on this instance.
When an instance is disposed, it means the instance is "dead"—there will no more information/data
from any DataWriter on this instance.

The behavior controlled by this QoS applies on a per instance (key) basis for keyed Topics, so when a
DataWriter unregisters an instance, Connext DDS also automatically disposes that instance. This is the
default behavior since autodispose_unregistered_instances defaults to TRUE.
Use Cases for Unregistering without Disposing:
There are situations in which you may want to set autodispose_unregistered_instances to FALSE, so
that unregistering will not automatically dispose the instance. For example:
l

In many cases where the ownership of a Topic is EXCLUSIVE (see the OWNERSHIP QosPolicy
(Section 6.5.15 on page 389)), DataWriters may want to relinquish ownership of a particular
instance of the Topic to allow other DataWriters to send updates for the value of that instance. In
this case, you may want a DataWriter to just unregister an instance—without disposing it (since
there are other writers). Unregistering an instance implies that the DataWriter no longer owns that
instance, but it is a stronger statement to say that instance no longer exists.

420

6.5.27.1 Properties

l

User applications may be coded to trigger on the disposal of instances, thus the ability to unregister
without disposing may be useful to properly maintain the semantic of disposal.

When you delete a DataWriter (Creating DataWriters (Section 6.3.1 on page 266)), all of the instances
managed by the DataWriter are automatically unregistered. Therefore, this QoS policy determines whether
or not all of the instances are disposed when the DataWriter is deleted when you call one of these operations:
l
l

l

Publisher’s delete_datawriter() (see Creating DataWriters (Section 6.3.1 on page 266))
Publisher’s delete_contained_entities() (see Deleting Contained DataWriters (Section 6.2.3.1 on
page 251))
DomainParticipant’s delete_contained_entities() (see Deleting Contained Entities (Section 8.3.3
on page 559))

When autodispose_unregistered_instances is TRUE, the middleware will clean up all the resources associated with an unregistered instance (most notably, the DDS sample history of non-volatile DataWriters)
when all the instance’s DDS samples have been acknowledged by all its live DataReaders, including the
DDS sample that indicates the unregistration. By default, autopurge_unregistered_instances_delay is
disabled (the delay is INFINITE). If the delay is set to zero, the DataWriter will clean up as soon as all the
DDS samples are acknowledged after the call to unregister(). A non-zero value for the delay can be useful in two ways:
l
l

To keep the historical DDS samples for late-joiners for a period of time.
In the context of discovery, if the applications temporarily lose the connection before the unregistration (which represents the remote entity destruction), to provide the DDS samples that indicate
the dispose and unregister actions once the connection is reestablished.

This delay can also be set for discovery data through these fields in the DISCOVERY_CONFIG
QosPolicy (DDS Extension) (Section 8.5.3 on page 585):
l

publication_writer_data_lifecycle.autopurge_unregistered_instances_delay

l

subscription_writer_data_lifecycle.autopurge_unregistered_instances_delay

6.5.27.1 Properties
It does not apply to DataReaders, so there is no requirement that the publishing and subscribing sides use
compatible values.
This QoS policy may be modified after the DataWriter is enabled.

421

6.5.27.2 Related QoS Policies

6.5.27.2 Related QoS Policies
l

None.

6.5.27.3 Applicable Entities
l

DataWriters (Section 6.3 on page 261)

6.5.27.4 System Resource Considerations
None.

6.6 FlowControllers (DDS Extension)
This section does not apply when using the separate add-on product, Ada Language Support,
which does not support FlowControllers.
A FlowController is the object responsible for shaping the network traffic by determining when attached
asynchronous DataWriters are allowed to write data.
You can use one of the built-in FlowControllers (and optionally modify their properties), create a custom
FlowController by using the DomainParticipant’s create_flowcontroller() operation (see Creating and
Deleting FlowControllers (Section 6.6.6 on page 433)), or create a custom FlowController by using the
DomainParticipant's PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394); see Creating and Configuring Custom FlowControllers with Property QoS (Section 6.6.5 on page 431).
To use a FlowController, you provide its name in the DataWriter’s PUBLISH_MODE QosPolicy (DDS
Extension) (Section 6.5.18 on page 397).
l

DDS_DEFAULT_FLOW_CONTROLLER_NAME
By default, flow control is disabled. That is, the built-in DDS_DEFAULT_FLOW_
CONTROLLER_NAME flow controller does not apply any flow control. Instead, it allows data to
be sent asynchronously as soon as it is written by the DataWriter.

l

DDS_FIXED_RATE_FLOW_CONTROLLER_NAME
The FIXED_RATE flow controller shapes the network traffic by allowing data to be sent only once
every second. Any accumulated DDS samples destined for the same destination are coalesced into
as few network packets as possible.

l

DDS_ON_DEMAND_FLOW_CONTROLLER_NAME
The ON_DEMAND flow controller allows data to be sent only when you call the FlowController’s

422

6.6 FlowControllers (DDS Extension)

trigger_flow() operation. With each trigger, all accumulated data since the previous trigger is sent
(across all Publishers or DataWriters). In other words, the network traffic shape is fully controlled
by the user. Any accumulated DDS samples destined for the same destination are coalesced into as
few network packets as possible.
This external trigger source is ideal for users who want to implement some form of closed-loop flow
control or who want to only put data on the wire every so many DDS samples (e.g., with the number of DDS samples based on NDDS_Transport_Property_t’s gather_send_buffer_count_max).
The default property settings for the built-in FlowControllers are described in the API Reference HTML
documentation.
DDS samples written by an asynchronous DataWriter are not sent in the context of the write() call.
Instead, Connext DDS puts the DDS samples in a queue for future processing. The FlowController associated with each asynchronous DataWriter determines when the DDS samples are actually sent.
Each FlowController maintains a separate FIFO queue for each unique destination (remote application).
DDS samples written by asynchronous DataWriters associated with the FlowController are placed in the
queues that correspond to the intended destinations of the DDS sample.
When tokens become available, a FlowController must decide which queue(s) to grant tokens first. This is
determined by the FlowController's scheduling_policy property (see Table 6.71 DDS_FlowControllerProperty_t). Once a queue has been granted tokens, it is serviced by the asynchronous publishing
thread. The queued up DDS samples will be coalesced and sent to the corresponding destination. The number of DDS samples sent depends on the data size and the number of tokens granted.
Table 6.71 DDS_FlowControllerProperty_t lists the properties for a FlowController.
Table 6.71 DDS_FlowControllerProperty_t
Type
DDS_
FlowControllerSchedulingPolicy

Field
Name

Description

scheduling_ Round robin, earliest deadline first, or highest priority first. See Flow Controller
policy
Scheduling Policies (Section 6.6.1 on the facing page).

DDS_
token_
FlowControllerTokenBucketProperty_
bucket
t

See Token Bucket Properties (Section 6.6.3 on page 426).

Table 6.72 FlowController Operations lists the operations available for a FlowController.

423

6.6.1 Flow Controller Scheduling Policies

Table 6.72 FlowController Operations
Operation

Description

get_property

Reference

Get and Set the FlowController properties.

Getting/Setting Properties for a Specific FlowController (Section
6.6.8 on page 435)

trigger_flow

Provides an external trigger to the FlowController.

Adding an External Trigger (Section 6.6.9 on page 435)

get_name

Returns the name of the FlowController.

get_
participant

Returns the DomainParticipant to which the
FlowController belongs.

set_property

Other FlowController Operations (Section 6.6.10 on page 435)

6.6.1 Flow Controller Scheduling Policies
l

Round Robin
(DDS_RR_FLOW_CONTROLLER_SCHED_POLICY) Perform flow control in a round-robin
(RR) fashion.
Whenever tokens become available, the FlowController distributes the tokens uniformly across all of
its (non-empty) destination queues. No destinations are prioritized. Instead, all destinations are
treated equally and are serviced in a round-robin fashion.

l

Earliest Deadline First
(DDS_EDF_FLOW_CONTROLLER_SCHED_POLICY) Perform flow control in an earliestdeadline-first (EDF) fashion.
A DDS sample's deadline is determined by the time it was written plus the latency budget of the
DataWriter at the time of the write call (as specified in the DDS_LatencyBudgetQosPolicy). The relative priority of a flow controller's destination queue is determined by the earliest deadline across all
DDS samples it contains.
When tokens become available, the FlowController distributes tokens to the destination queues in
order of their priority. In other words, the queue containing the DDS sample with the earliest deadline is serviced first. The number of tokens granted equals the number of tokens required to send the
first DDS sample in the queue. Note that the priority of a queue may change as DDS samples are
sent (i.e., removed from the queue). If a DDS sample must be sent to multiple destinations or two
DDS samples have an equal deadline value, the corresponding destination queues are serviced in a
round-robin fashion.

424

6.6.1 Flow Controller Scheduling Policies

With the default duration of 0 in the LatencyBudgetQosPolicy, using an EDF_FLOW_
CONTROLLER_SCHED_POLICY FlowController preserves the order in which you call write()
across the DataWriters associated with the FlowController.
Since the LatencyBudgetQosPolicy is mutable, a DDS sample written second may contain an earlier
deadline than the DDS sample written first if the DDS_LatencyBudgetQosPolicy’s duration is sufficiently decreased in between writing the two DDS samples. In that case, if the first DDS sample is
not yet written (still in queue waiting for its turn), it inherits the priority corresponding to the (earlier)
deadline from the second DDS sample.
In other words, the priority of a destination queue is always determined by the earliest deadline
among all DDS samples contained in the queue. This priority inheritance approach is required in
order to both honor the updated duration and to adhere to the DataWriter in-order data delivery
guarantee.
l

Highest Priority First
(DDS_HPF_FLOW_CONTROLLER_SCHED_POLICY) Perform flow control in an highest-priority-first (HPF) fashion.
Note: Prioritized DDS samples are not supported when using the Java, Ada, or .NET APIs. Therefore the Highest Priority First scheduling policy is not supported when using these APIs.
The next destination queue to service is determined by the publication priority of the DataWriter, the
channel of a multi-channel DataWriter, or individual DDS sample.
The relative priority of a flow controller's destination queue is determined by the highest publication
priority of all the DDS samples it contains.
When tokens become available, the FlowController distributes tokens to the destination queues in
order of their publication priority. The queue containing the DDS sample with the highest publication priority is serviced first. The number of tokens granted equals the number of tokens required
to send the first DDS sample in the queue. Note that a queue’s priority may change as DDS samples
are sent (i.e., as they are removed from the queue). If a DDS sample must be sent to multiple destinations or two DDS samples have the same publication priority, the corresponding destination
queues are serviced in a round-robin fashion.
This priority inheritance approach is required to both honor the designated publication priority and
adhere to the DataWriter’s in-order data delivery guarantee.
See also: Prioritized DDS Samples (Section 6.6.4 on page 428).

425

6.6.2 Managing Fast DataWriters When Using a FlowController

6.6.2 Managing Fast DataWriters When Using a FlowController
If a DataWriter is writing DDS samples faster than its attached FlowController can throttle, Connext DDS
may drop DDS samples on the writer’s side. This happens because the DDS samples may be removed
from the queue before the asynchronous publisher’s thread has a chance to send them. To work around
this problem, either:
l

Use reliable communication to block the write() call and thereby throttle your application.

l

Do not allow the queue to fill up in the first place.
The queue should be sized large enough to handle expected write bursts, so that no DDS samples
are dropped. Then in steady state, the FlowController will smooth out these bursts and the queue
will ideally have only one entry.

6.6.3 Token Bucket Properties
FlowControllers use a token-bucket approach for open-loop network flow control. The flow control characteristics are determined by the token bucket properties. The properties are listed in Table 6.73 DDS_
FlowControllerTokenBucketProperty_t ; see the API Reference HTML documentation for their defaults
and valid ranges.
Table 6.73 DDS_FlowControllerTokenBucketProperty_t
Type

Field Name

Description

DDS_Long

max_tokens

Maximum number of tokens than can accumulate in the token bucket. See max_tokens (Section 6.6.3.1
on the next page).

DDS_Long

tokens_added_
per_period

The number of tokens added to the token bucket per specified period. See tokens_added_per_period
(Section 6.6.3.2 on the next page).

DDS_Long

tokens_leaked_
per_period

The number of tokens removed from the token bucket per specified period. See tokens_leaked_per_
period (Section 6.6.3.3 on the next page).

DDS_
Duration_t

period

Period for adding tokens to and removing tokens from the bucket. See period (Section 6.6.3.4 on the
next page).

DDS_Long

bytes_per_token

Maximum number of bytes allowed to send for each token available. See bytes_per_token (Section
6.6.3.5 on page 428).

Asynchronously published DDS samples are queued up and transmitted based on the token bucket flow
control scheme. The token bucket contains tokens, each of which represents a number of bytes. DDS
samples can be sent only when there are sufficient tokens in the bucket. As DDS samples are sent, tokens
are consumed. The number of tokens consumed is proportional to the size of the data being sent. Tokens
are replenished on a periodic basis.

426

6.6.3.1 max_tokens

The rate at which tokens become available and other token bucket properties determine the network traffic
flow.
Note that if the same DDS sample must be sent to multiple destinations, separate tokens are required for
each destination. Only when multiple DDS samples are destined to the same destination will they be
coalesced and sent using the same token(s). In other words, each token can only contribute to a single network packet.
6.6.3.1 max_tokens
The maximum number of tokens in the bucket will never exceed this value. Any excess tokens are discarded. This property value, combined with bytes_per_token, determines the maximum allowable data
burst.
Use DDS_LENGTH_UNLIMITED to allow accumulation of an unlimited amount of tokens (and therefore potentially an unlimited burst size).
6.6.3.2 tokens_added_per_period
A FlowController transmits data only when tokens are available. Tokens are periodically replenished. This
field determines the number of tokens added to the token bucket with each periodic replenishment.
Available tokens are distributed to associated DataWriters based on the scheduling_policy. Use DDS_
LENGTH_UNLIMITED to add the maximum number of tokens allowed by max_tokens.
6.6.3.3 tokens_leaked_per_period
When tokens are replenished and there are sufficient tokens to send all DDS samples in the queue, this
property determines whether any or all of the leftover tokens remain in the bucket.
Use DDS_LENGTH_UNLIMITED to remove all excess tokens from the token bucket once all DDS
samples have been sent. In other words, no token accumulation is allowed. When new DDS samples are
written after tokens were purged, the earliest point in time at which they can be sent is at the next periodic
replenishment.
6.6.3.4 period
This field determines the period by which tokens are added or removed from the token bucket.
The special value DDS_DURATION_INFINITE can be used to create an on-demand FlowController,
for which tokens are no longer replenished periodically. Instead, tokens must be added explicitly by calling
the FlowController’s trigger_flow() operation. This external trigger adds tokens_added_per_period
tokens each time it is called (subject to the other property settings).

427

6.6.3.5 bytes_per_token

Once period is set to DDS_DURATION_INFINITE, it can no longer be reverted to a finite
period.
6.6.3.5 bytes_per_token
This field determines the number of bytes that can actually be transmitted based on the number of tokens.
Tokens are always consumed in whole by each DataWriter. That is, in cases where bytes_per_token is
greater than the DDS sample size, multiple DDS samples may be sent to the same destination using a
single token (regardless of the scheduling_policy).
Where fragmentation is required, the fragment size will be either (a) bytes_per_token or (b) the minimum
of the largest message sizes across all transports installed with the DataWriter, whichever is less.
Use DDS_LENGTH_UNLIMITED to indicate that an unlimited number of bytes can be transmitted per
token. In other words, a single token allows the recipient DataWriter to transmit all its queued DDS
samples to a single destination. A separate token is required to send to each additional destination.

6.6.4 Prioritized DDS Samples
Note: This feature is not supported when using the Ada API.
The Prioritized DDS Samples feature allows you to prioritize traffic that is in competition for transmission
resources. The granularity of this prioritization may be by DataWriter, by instance, or by individual DDS
sample.
Prioritized DDS Samples can improve latency in the following cases:
l

Low-Availability Links

l

With low-availability communication, unsent DDS samples may accumulate while the link is
unavailable. When the link is restored, a large number of DDS samples may be waiting for transmission. High priority DDS samples will be sent first.
Low-Bandwidth Links
With low-bandwidth communication, a temporary backlog may occur or the link may become congested with large DDS samples. High-priority DDS samples will be sent at the first available gap,
between the fragments of a large low-priority DDS sample.

l

Prioritized Topics
With limited bandwidth communication, some topics may be deemed to be of higher priority than
others on an ongoing basis, and DDS samples written to some topics should be given precedence
over others on transmission.

428

6.6.4.1 Designating Priorities

l

High Priority Events
Due to external rules or content analysis (e.g., perimeter violation or identification as a threat), the
priority of DDS samples is dynamically determined, and the priority assigned a given DDS sample
will reflect the urgency of its delivery.

To configure a DataWriter to use prioritized DDS samples:
l

l

Create a FlowController with the scheduling_policy property set to DDS_HPF_FLOW_
CONTROLLER_SCHED_POLICY.
Create a DataWriter with the PUBLISH_MODE QosPolicy (DDS Extension) (Section 6.5.18 on
page 397) kind set to ASYNCHRONOUS and flow_controller_name set to the name of the
FlowController.

A single FlowController may perform traffic shaping for multiple DataWriters and multiple DataWriter
channels. The FlowController’s configuration determines how often publication resources are scheduled,
how much data may be sent per period, and other transmission characteristics that determine the ultimate
performance of prioritized DDS samples.
When working with prioritized DDS samples, you should use these operations, which allow you to specify priority:
l
l

l

write_w_params() (see Writing Data (Section 6.3.8 on page 283))
unregister_instance_w_params() (see Registering and Unregistering Instances (Section 6.3.14.1
on page 297))
dispose_w_params() (see Disposing of Data (Section 6.3.14.2 on page 299))

If you use write(), unregister(), or dispose() instead of the _w_params() versions, the affected DDS
sample is assigned priority 0 (undefined priority). If you are using a multi-channel DataWriter with a priority filter, and you have no channel for priority 0, the DDS sample will be discarded.
6.6.4.1 Designating Priorities
For DataWriters and DataWriter channels, valid publication priority values are:
l

DDS_PUBLICATION_PRIORITY_UNDEFINED

l

DDS_PUBLICATION_PRIORITY_AUTOMATIC

l

Positive integers excluding zero

For individual DDS samples, valid publication priority values are 0 and positive integers.

429

6.6.4.2 Priority-Based Filtering

There are three ways to set the publication priority of a DataWriter or DataWriter channel:
1. For a DataWriter, publication priority is set in the priority field of its PUBLISH_MODE
QosPolicy (DDS Extension) (Section 6.5.18 on page 397). For a multi-channel DataWriter (see
MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386)), this value
will be the default publication priority for any member channel that has not been assigned a
specific value.
2. For a channel of a Multi-channel DataWriter, publication priority can be set in the DataWriter’s
MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386) in channels[].priority.
3. If a DataWriter or a channel of a Multi-channel DataWriter is configured for publication priority
inheritance (DDS_PUBLICATION_PRIORITY_AUTOMATIC), its publication priority is the
highest priority among all the DDS samples currently in the publication queue. When using publication priority inheritance, the publication priorities of individual DDS samples are set by calling
the write_w_params() operation, which takes a priority parameter.
The effective publication priority is determined from the interaction of the DataWriter, channel, and DDS
sample publication priorities, as shown in Table 6.74 Effective Publication Priority .
Table 6.74 Effective Publication Priority
Priority Setting Combinations
Writer Priority
Channel Priority

Undefined Don’t care

AUTOMATIC

Don’t care

Designated positive
integer > 0

Undefined AUTOMATIC

Undefined

Designated positive integer > 0

Undefined

Designated positive
integer > 0

Designated positive
integer > 0

Don’t care

Don’t care

DDS Sample
Priority1

DDS Sample
Priority2

Channel
Priority

Writer
Priority

DDS Sample Priority Don’t care

Effective Priority

Lowest
Priority

6.6.4.2 Priority-Based Filtering
The configuration methods explained above are sufficient to create multiple DataWriters, each with its
own assigned priority, all using the same FlowController configured for publication priority-based schedul-

1Highest sample priority among all DDS samples currently in the publication queue.
2Highest sample priority among all DDS samples currently in the publication queue.

430

6.6.5 Creating and Configuring Custom FlowControllers with Property QoS

ing. Such a configuration is sufficient to assign different priorities to individual topics, but it does not allow
different publication priorities to be assigned to published data within a Topic.
To assign different priorities to data within a DataWriter, you will need to use a Multi-channel DataWriter
and configure the channels with different priorities. Configuring the publication priorities of DataWriter
channels is explained above. To associate different priorities of data with different publication channels,
configure the channel[].filter_expression in the DataWriter’s MULTI_CHANNEL QosPolicy (DDS
Extension) (Section 6.5.14 on page 386). The filtering criteria that is available for evaluation by each channel is determined by the filter type, which is configured with the DataWriter’s filter_name (also in the
MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386)).
For example, using the built-in SQL-based content filter allows channel membership to be determined
based on the content of each DDS sample.
If you do not want to embed priority criteria within each DDS sample, you can use a built-in filter named
DDS_PRIFILTER_NAME that uses the publication priority that is provided when you call write_w_
params() (see Writing Data (Section 6.3.8 on page 283)). The filter’s expression syntax is:
@priority OP VAL

where OP can be < , <= , > , >= , = , or <> (standard relational operators), and VAL is a positive integer.
The filter supports multiple expressions, combined with the conjunctions AND and OR. You can use parentheses to disambiguate combinations of AND and OR in the same expression. For example:
@priority = 2 OR (@priority > 6 AND @priority < 10)

6.6.5 Creating and Configuring Custom FlowControllers with Property QoS
You can create and configure FlowControllers using the PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394). The properties must have a prefix of “dds.flow_controller.token_bucket”, followed by the name of the FlowController being created or configured. For example, if you want to
create/configure a FlowController named MyFC, all the properties for MyFC should have the prefix
“dds.flow_controller.token_bucket.MyFC“.
Table 6.75 FlowController Properties lists the properties that can be set for FlowControllers in the
DomainParticipant's PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394). A
FlowController with the name "dds.flow_controller.token_bucket." will be
implicitly created when at least one property using that prefix is specified. Then, to link a DataWriter to
your FlowController, use "dds.flow_controller.token_bucket." in the
DataWriter's publish_mode.flow_controller_name.

431

6.6.5.1 Example

Table 6.75 FlowController Properties
Property Name
prefix with ‘dds.flow_
controller.token_bucket.


Property Value Description

Specifies the scheduling policy to be used. (See Flow Controller Scheduling Policies
(Section 6.6.1 on page 424)) May be:
scheduling_policy

DDS_RR_FLOW_CONTROLLER_SCHED_POLICY
DDS_EDF_FLOW_CONTROLLER_SCHED_POLICY
DDS_HPF_FLOW_CONTROLLER_SCHED_POLICY
Maximum number of tokens than can accumulate in the token bucket.

token_bucket.max_tokens
Use -1 for unlimited.
Number of tokens added to the token bucket per specified period.
token_bucket.tokens_added_per_period
Use -1 for unlimited.
Number of tokens removed from the token bucket per specified period.
token_bucket.tokens_leaked_per_period
Use -1 for unlimited.
token_bucket.period.sec

Period for adding tokens to and removing tokens from the bucket in seconds.

token_bucket.period.nanosec

Period for adding tokens to and removing tokens from the bucket in nanoseconds.

token_bucket.bytes_per_token

Maximum number of bytes allowed to send for each token available.

6.6.5.1 Example
The following example shows how to set FlowController properties.
Note: Some lines in this example, such as dds.flow_controller.token_bucket.MyFlowController.scheduling_policy, are too long to fit on the page as one line; however in your
XML file, they each need to be on a single line.





dds.flow_controller.token_bucket.MyFlowController.scheduling_policy

DDS_RR_FLOW_CONTROLLER_SCHED_POLICY



dds.flow_controller.token_bucket.MyFlowController.token_bucket.period.sec

432

6.6.6 Creating and Deleting FlowControllers


100



dds.flow_controller.token_bucket.MyFlowController.
token_bucket.period.nanosec

0



dds.flow_controller.token_bucket.MyFlowController.token_bucket.tokens_added_per_period

2



dds.flow_controller.token_bucket.MyFlowController.token_bucket.tokens_leaked_per_period

2



dds.flow_controller.token_bucket.MyFlowController.token_bucket.bytes_per_token

1024







dds.flow_controller.token_bucket.MyFlowController

ASYNCHRONOUS_PUBLISH_MODE_QOS



6.6.6 Creating and Deleting FlowControllers
(Note: in the Modern C++ API FlowControllers have reference semantics, see Creating and Deleting Entities)
If you do not want to use one of the three built-in FlowControllers described in FlowControllers (DDS
Extension) (Section 6.6 on page 422), you can create your own with the DomainParticipant’s create_
flowcontroller() operation:
DDSFlowController* create_flowcontroller
(const char * name,
const DDS_FlowControllerProperty_t & property)

433

6.6.7 Getting/Setting Default FlowController Properties

To associate a FlowController with a DataWriter, you set the FlowController’s name in the PUBLISH_
MODE QosPolicy (DDS Extension) (Section 6.5.18 on page 397) (flow_controller_name).
A single FlowController may service multiple DataWriters, even if they belong to a different Publisher.
The FlowController’s property structure determines how the FlowController shapes the network traffic.
Name of the FlowController to create. A DataWriter is associated with a DDSFlowController
name
by name. Limited to 255 characters.
Properties to be used for creating the FlowController. The special value DDS_FLOW_
CONTROLLER_PROPERTY_DEFAULT can be used to indicate that the FlowController
property
should be created with the default DDS_FlowControllerProperty_t set in the DomainParticipant.
Note: If you use DDS_FLOW_CONTROLLER_PROPERTY_DEFAULT, it is not safe to create the
FlowController while another thread may be simultaneously calling set_default_flowcontroller_property
() or looking for that FlowController with lookup_flowcontroller().
To delete an existing FlowController, use the DomainParticipant’s delete_flowcontroller() operation:
DDS_ReturnCode_t delete_flowcontroller (DDSFlowController *

fc)

The FlowController must belong this the DomainParticipant and not have any attached DataWriters or
the delete call will return an error (PRECONDITION_NOT_MET).

6.6.7 Getting/Setting Default FlowController Properties
To get the default DDS_FlowControllerProperty_t values, use this operation on the DomainParticipant:
DDS_ReturnCode_t get_default_flowcontroller_property
(DDS_FlowControllerProperty_t & property)

The retrieved property will match the set of values specified on the last successful call to the DomainParticipant’s set_default_flowcontroller_property(), or if the call was never made, the default values listed
in DDS_FlowControllerProperty_t.
To change the default DDS_FlowControllerProperty_t values used when a new FlowController is created,
use this operation on the DomainParticipant:
DDS_ReturnCode_t set_default_flowcontroller_property
(const DDS_FlowControllerProperty_t & property)

The special value DDS_FLOW_CONTROLLER_PROPERTY_DEFAULT may be passed for the property to indicate that the default property should be reset to the default values the factory would use if set_
default_flowcontroller_property() had never been called.
Note: It is not safe to set the default FlowController properties while another thread may be simultaneously
calling get_default_flowcontroller_property(), set_default_flowcontroller_property(), or create_

434

6.6.8 Getting/Setting Properties for a Specific FlowController

flowcontroller() with DDS_FLOW_CONTROLLER_PROPERTY_DEFAULT as the qos parameter. It
is also not safe to get the default FlowController properties while another thread may be simultaneously
calling get_default_flowcontroller_property().

6.6.8 Getting/Setting Properties for a Specific FlowController
To get the properties of a FlowController, use the FlowController’s get_property() operation:
DDS_ReturnCode_t DDSFlowController::get_property
(struct DDS_FlowControllerProperty_t & property)

To change the properties of a FlowController, use the FlowController’s set_property() operation:
DDS_ReturnCode_t DDSFlowController::set_property
(const struct DDS_FlowControllerProperty_t &

property)

Once a FlowController has been instantiated, only its token_bucket property can be changed. The
scheduling_policy is immutable. A new token.period only takes effect at the next scheduled token distribution time (as determined by its previous value).
The special value DDS_FLOW_CONTROLLER_PROPERTY_DEFAULT can be used to match the
current default properties set in the DomainParticipant.

6.6.9 Adding an External Trigger
Typically, a FlowController uses an internal trigger to periodically replenish its tokens. The period by
which this trigger is called is determined by the period property setting.
The trigger_flow() function provides an additional, external trigger to the FlowController. This trigger
adds tokens_added_per_period tokens each time it is called (subject to the other property settings of the
FlowController).
DDS_ReturnCode_t trigger_flow ()

An on-demand FlowController can be created with a DDS_DURATION_INFINITE as period, in which
case the only trigger source is external (i.e. the FlowController is solely triggered by the user on demand).
trigger_flow() can be called on both a strict on-demand FlowController and a hybrid FlowController
(internally and externally triggered).

6.6.10 Other FlowController Operations
If you have the FlowController object and need its name, call the FlowController’s get_name() operation:
const char* DDSFlowController::get_name( )

Conversely, if you have the name of the FlowController and need the FlowController object, call the
DomainPartipant’s lookup_flowcontroller() operation:

435

6.6.10 Other FlowController Operations

DDSFlowController* lookup_flowcontroller (const char *

name)

To get a FlowController’s DomainParticipant, call the FlowController’s get_participant() operation:
DDSDomainParticipant* get_participant ( )

Note: It is not safe to lookup a FlowController description while another thread is creating that FlowController

436

Chapter 7 Receiving Data
This section discusses how to create, configure, and use Subscribers and DataReaders to receive
data. It describes how these objects interact, as well as the types of operations that are available for
them.
This section includes:
The goal of this section is to help you become familiar with the Entities you need for receiving
data. For up-to-date details such as formal parameters and return codes on any mentioned operations, please see the Connext DDS API Reference HTML documentation.

7.1 Preview: Steps to Receiving Data
There are three ways to receive data:
l

l

l

Your application can explicitly check for new data by calling a DataReader’s read() or take
() operation. This method is also known as polling for data.
Your application can be notified asynchronously whenever new DDS data samples arrive—
this is done with a Listener on either the Subscriber or the DataReader. Connext DDS will
invoke the Listener’s callback routine when there is new data. Within the callback routine,
user code can access the data by calling read() or take() on the DataReader. This method is
the way for your application to receive data with the least amount of latency.
Your application can wait for new data by using Conditions and a WaitSet, then calling wait
(). Connext DDS will block your application’s thread until the criteria (such as the arrival of
DDS samples, or a specific status) set in the Condition becomes true. Then your application
resumes and can access the data with read() or take().

The DataReader’s read() operation gives your application a copy of the data and leaves the data in
the DataReader’s receive queue. The DataReader’s take() operation removes data from the
receive queue before giving it to your application.

437

7.1 Preview: Steps to Receiving Data

See Using DataReaders to Access Data (Read & Take) (Section 7.4 on page 491) for details on using
DataReaders to access received data.
See Conditions and WaitSets (Section 4.6 on page 187) for details on using Conditions and WaitSets.
To prepare to receive data, create and configure the required Entities:
1. Create a DomainParticipant.
2. Register user data types1 with the DomainParticipant. For example, the ‘FooDataType’.
3. Use the DomainParticipant to create a Topic with the registered data type.
4. Optionally2, use the DomainParticipant to create a Subscriber.
5. Use the Subscriber or DomainParticipant to create a DataReader for the Topic.
6. Use a type-safe method to cast the generic DataReader created by the Subscriber to a type-specific
DataReader. For example, ‘FooDataReader’.
Then use one of the following mechanisms to receive data.
l

l

To receive DDS data samples by polling for new data:
l Using a FooDataReader, use the read() or take() operations to access the DDS data samples
that have been received and stored for the DataReader. These operations can be invoked at
any time, even if the receive queue is empty.
To receive DDS data samples asynchronously:
l Install a Listener on the DataReader or Subscriber that will be called back by an internal Connext DDS thread when new DDS data samples arrive for the DataReader.

1. Create a DDSDataReaderListener for the FooDataReader or a DDSSubscriberListener for Subscriber. In C++, C++/CLI, C# and Java, you must derive your own Listener class from those base
classes. In C, you must create the individual functions and store them in a structure.
If you created a DDSDataReaderListener with the on_data_available() callback enabled: on_
data_available() will be called when new data arrives for that DataReader.
If you created a DDSSubscriberListener with the on_data_on_readers() callback enabled: on_
data_on_readers() will be called when data arrives for any DataReader created by the Subscriber.

1Type registration is not required for built-in types (see Registering Built-in Types (Section 3.2.1 on page 30)).
2You are not required to explicitly create a Subscriber; instead, you can use the 'implicit Subscriber' created from the

DomainParticipant. See Creating Subscribers Explicitly vs. Implicitly (Section 7.2.1 on page 444).

438

7.1 Preview: Steps to Receiving Data

2. Install the Listener on either the FooDataReader or Subscriber.
For the DataReader, the Listener should be installed to handle changes in the DATA_
AVAILABLE status.
For the Subscriber, the Listener should be installed to handle changes in the DATA_ON_
READERS status.
3. Only 1 Listener will be called back when new data arrives for a DataReader.
Connext DDS will call the Subscriber’s Listener if it is installed. Otherwise, the DataReader’s
Listener is called if it is installed. That is, the on_data_on_readers() operation takes precedence
over the on_data_available() operation.
If neither Listeners are installed or neither Listeners are enabled to handle their respective statuses,
then Connext DDS will not call any user functions when new data arrives for the DataReader.
4. In the on_data_available() method of the DDSDataReaderListener, invoke read() or take() on the
FooDataReader to access the data.
If the on_data_on_readers() method of the DDSSubscriberListener is called, the code can invoke
read() or take() directly on the Subscriber’s DataReaders that have received new data. Alternatively, the code can invoke the Subscriber’s notify_datareaders() operation. This will in turn call
the on_data_available() methods of the DataReaderListeners (if installed and enabled) for each of
the DataReaders that have received new DDS data samples.
To wait (block) until DDS data samples arrive:
1. Use the DataReader to create a ReadCondition that describes the DDS samples for which you want
to wait. For example, you can specify that you want to wait for never-before-seen DDS samples
from DataReaders that are still considered to be ‘alive.’
Alternatively, you can create a StatusCondition that specifies you want to wait for the ON_DATA_
AVAILABLE status.
2. Create a WaitSet.
3. Attach the ReadCondition or StatusCondition to the WaitSet.
4. Call the WaitSet’s wait() operation, specifying how long you are willing to wait for the desired DDS
samples. When wait() returns, it will indicate that it timed out, or that the attached Condition become
true (and therefore the desired DDS samples are available).
5. Using a FooDataReader, use the read() or take() operations to access the DDS data samples that
have been received and stored for the DataReader.

439

7.2 Subscribers

7.2 Subscribers
An application that intends to subscribe to information needs the following Entities: DomainParticipant,
Topic, Subscriber, and DataReader. All Entities have a corresponding specialized Listener and a set of
QosPolicies. The Listener is how Connext DDS notifies your application of status changes relevant to the
Entity. The QosPolicies allow your application to configure the behavior and resources of the Entity.
l
l

l

l

The DomainParticipant defines the DDS domain on which the information will be available.
The Topic defines the name of the data to be subscribed, as well as the type (format) of the data
itself.
The DataReader is the Entity used by the application to subscribe to updated values of the data. The
DataReader is bound at creation time to a Topic, thus specifying the named and typed data stream to
which it is subscribed. The application uses the DataWriter’s read() or take() operation to access
DDS data samples received for the Topic.
The Subscriber manages the activities of several DataReader entities. The application receives data
using a DataReader that belongs to a Subscriber. However, the Subscriber will determine when the
data received from applications is actually available for access through the DataReader. Depending
on the settings of various QosPolicies of the Subscriber and DataReader, data may be buffered until
DDS data samples for associated DataReaders are also received. By default, the data is available to
the application as soon as it is received.

For more information, see Creating Subscribers Explicitly vs. Implicitly (Section 7.2.1 on page 444).
The UML diagram in Subscription Module (Section Figure 7.1 on the facing page) shows how these Entities are related as well as the methods defined for each Entity.
Subscribers are used to perform the operations listed in Table 7.1 Subscriber Operations. For details such
as formal parameters and return codes, please see the API Reference HTML documentation. Otherwise,
you can find more information about the operations by looking in the section listed under the Reference
(Section on page 442) column.

440

7.2 Subscribers

Figure 7.1 Subscription Module

Note: Some operations cannot be used within a listener callback, see Restricted Operations in Listener
Callbacks (Section 4.5.1 on page 185).

441

7.2 Subscribers

Table 7.1 Subscriber Operations
Working
Operation
with ...

DataReaders

DataReaders
cont'd

Description

begin_access

Indicates that the application is about to access the DDS data samples in
the DataReaders of the Subscriber.

create_
datareader

Creates a DataReader.

Reference
Beginning and Ending GroupOrdered Access (Section 7.2.5 on
page 453)

Creating DataReaders (Section
7.3.1 on page 463)

create_
datareader_
with_profile

Creates a DataReader with QoS from a specified QoS profile.

copy_from_
topic_qos

Copies relevant QosPolicies from a Topic into a DataReaderQoS
structure.

Subscriber QoS-Related
Operations (Section 7.2.4.6 on
page 453)

delete_
contained_
entities

Deletes all the DataReaders that were created by the Subscriber. Also
deletes the corresponding ReadConditions created by the contained
DataReaders.

Deleting Contained DataReaders
(Section 7.2.3.1 on page 447)

delete_
datareader

Deletes a specific DataReader.

Deleting DataReaders (Section
7.3.3 on page 466)

end_access

Indicates that the application is done accessing the DDS data samples in
the DataReaders of the Subscriber.

Beginning and Ending GroupOrdered Access (Section 7.2.5 on
page 453)

get_all_
datareaders

Retrieves all the DataReaders created from this Subscriber.

Getting All DataReaders (Section
7.3.2 on page 465)

get_
datareaders

Returns a list of DataReaders that contain DDS samples with the
specified sample_states, view_states and instance_states.

Getting DataReaders with Specific
DDS Samples (Section 7.2.7 on
page 456)

get_default_
datareader_
qos

Copies the Subscriber’s default DataReaderQos values into a
DataReaderQos structure.

Setting Subscriber QosPolicies
(Section 7.2.4 on page 447)

442

7.2 Subscribers

Table 7.1 Subscriber Operations
Working
Operation
with ...

DataReaders
cont'd

Libraries
and Profiles

Participants

Description

Reference

get_status_
changes

Gets all status changes.

Getting Status and Status Changes
(Section 4.1.4 on page 157)

lookup_
datareader

Retrieves a DataReader previously created for a specific Topic.

Finding a Subscriber’s Related
Entities (Section 7.2.8 on page
457)

notify_
datareaders

Invokes the on_data_available() operation for attached Listeners of
DataReaders that have new DDS data samples.

Setting Up SubscriberListeners
(Section 7.2.6 on page 454)

set_default_
datareader_
qos

Sets or changes the Subscriber’s default DataReaderQoS values.

Setting Subscriber QosPolicies
(Section 7.2.4 on page 447)

get_default_
library

Gets the Subscriber’s default QoS profile library.

get_default_
profile

Gets the Subscriber’s default QoS profile.

get_default_
profile_
library

Gets the library that contains the Subscriber’s default QoS profile.

set_default_
library

Sets the default library for a Subscriber.

set_default_
profile

Sets the default profile for a Subscriber.

get_
participant

Gets the Subscriber’s DomainParticipant.

Getting and Settings Subscriber’s
Default QoS Profile and Library
(Section 7.2.4.4 on page 451)

Finding a Subscriber’s Related
Entities (Section 7.2.8 on page
457)

443

7.2.1 Creating Subscribers Explicitly vs. Implicitly

Table 7.1 Subscriber Operations
Working
Operation
with ...

Description

Reference

enable

Enables the Subscriber.

Enabling DDS Entities (Section
4.1.2 on page 154)

equals

Compares two Subscriber’s QoS structures for equality.

Comparing QoS Values (Section
7.2.4.2 on page 450)

get_listener

Gets the currently installed Listener.

Setting Up SubscriberListeners
(Section 7.2.6 on page 454)

get_qos

Gets the Subscriber’s current QosPolicy settings. This is most often
used in preparation for calling set_qos.

Changing QoS Settings After
Subscriber Has Been Created
(Section 7.2.4.3 on page 450)

set_listener

Sets the Subscriber’s Listener. If you created the Subscriber without a
Listener, you can use this operation to add one later.

Setting Up SubscriberListeners
(Section 7.2.6 on page 454)

set_qos

Sets the Subscriber’s QoS. You can use this operation to change the
values for the Subscriber’s QosPolicies. Note, however, that not all
QosPolicies can be changed after the Subscriber has been created.

Changing QoS Settings After
Subscriber Has Been Created
(Section 7.2.4.3 on page 450)

set_qos_
with_profile

Sets the Subscriber’s QoS based on a QoS profile.

Changing QoS Settings After
Subscriber Has Been Created
(Section 7.2.4.3 on page 450)

Subscribers

7.2.1 Creating Subscribers Explicitly vs. Implicitly
To receive data, your application must have a Subscriber. However, you are not required to explicitly create a Subscriber. If you do not create one, the middleware will implicitly create a Subscriber the
first time you create a DataReader using the DomainParticipant’s operations. It will be created with
default QoS (DDS_SUBCRIBER_QOS_DEFAULT) and no Listener. The 'implicit Subscriber' can be
accessed using the DomainParticipant’s get_implicit_subscriber() operation (see Getting the Implicit Publisher or Subscriber (Section 8.3.9 on page 569)).You can use this ‘implicit Subscriber’ just like any other
Subscriber (it has the same operations, QosPolicies, etc.). So you can change the mutable QoS and set a
Listener if desired.
A Subscriber (implicit or explicit) gets its own default QoS and the default QoS for its child DataReaders
from the DomainParticipant. These default QoS are set when the Subscriber is created. (This is true for
Publishers and DataWriters, too.)
DataReaders are created by calling create_datareader() or create_datareader_with_profile()—these
operations exist for DomainParticipants and Subscribers1. If you use the DomainParticipant to create a
1In the Modern C++ API, you always use a DataReader constructor.
444

7.2.2 Creating Subscribers

DataReader, it will belong to the implicit Subscriber. If you use a Subscriber to create a DataReader, it
will belong to that Subscriber.
The middleware will use the same implicit Subscriber for all DataReaders that are created using the
DomainParticipant’s operations.
Having the middleware implicitly create a Subscriber allows you to skip the step of creating a Subscriber.
However, having all your DataReaders belong to the same Subscriber can reduce the concurrency of the
system because all the read operations will be serialized.

7.2.2 Creating Subscribers
Before you can explicitly create a Subscriber, you need a DomainParticipant (DomainParticipants (Section 8.3 on page 547)). To create a Subscriber, use the DomainParticipant’s create_subscriber() or create_subscriber_with_profile() operation.
A QoS profile is way to use QoS settings from an XML file or string. With this approach, you can change
QoS settings without recompiling the application. For details, see Configuring QoS with XML (Section
Chapter 17 on page 791).
Note: the Modern C++ API provides Subscriber constructors whose first, and only required argument is
the DomainParticipant.
DDSSubscriber* create_subscriber(
const DDS_SubscriberQos &qos,
DDSSubscriberListener * listener,
DDS_StatusMask mask)
DDSSubscriber* create_subscriber_with_profile (
const char * library_name,
const char * profile_name,
DDSSubscriberListener * listener,
DDS_StatusMask mask )

Where:
qos

If you want the default QoS settings (described in the API Reference HTML documentation),
use DDS_SUBSCRIBER_QOS_DEFAULT for this parameter (see Figure 7.2 Creating a
Subscriber with Default QosPolicies on the next page ). If you want to customize any of the
QosPolicies, supply a QoS structure (see Creating a Subscriber with Non-Default QosPolicies
(not from a profile) (Section Figure 7.3 on page 449)). The QoS structure for a Subscriber is
described in Subscriber QosPolicies (Section 7.5 on page 510).

Note : If you use DDS_SUBSCRIBER_QOS_DEFAULT, it is not safe to create the Subscriber
while another thread may be simultaneously calling set_default_subscriber_qos().

445

7.2.3 Deleting Subscribers

listener

Listeners are callback routines. Connext DDS uses them to notify your application when specific
events (new DDS data samples arrive and status changes) occur with respect to the Subscriber or
the DataReaders created by the Subscriber. The listener parameter may be set to NULL if you
do not want to install a Listener. If you use NULL, the Listener of the DomainParticipant to
which the Subscriber belongs will be used instead (if it is set). For more information on
SubscriberListeners, see Setting Up SubscriberListeners (Section 7.2.6 on page 454).

mask

This bit-mask indicates which status changes will cause the Subscriber’s Listener to be invoked.
The bits set in the mask must have corresponding callbacks implemented in the Listener. If you
use NULL for the Listener, use DDS_STATUS_MASK_NONE for this parameter. If the
Listener implements all callbacks, use DDS_STATUS_MASK_ALL. For information on Status,
see Listeners (Section 4.4 on page 177).
This bit-mask indicates which status changes will cause the Subscriber’s Listener to be invoked.
The bits set in the mask must have corresponding callbacks implemented in the Listener. If you
use NULL for the Listener, use DDS_STATUS_MASK_NONE for this parameter. If the
Listener implements all callbacks, use DDS_STATUS_MASK_ALL. For information on Status,
see Listeners (Section 4.4 on page 177).

library_name A QoS Library is a named set of QoS profiles. See URL Groups (Section 17.8 on page 814).
profile_name A QoS profile groups a set of related QoS, usually one per entity. See URL Groups (Section
17.8 on page 814).

Figure 7.2 Creating a Subscriber with Default QosPolicies

// create the subscriber
DDSSubscriber* subscriber =
participant->create_subscriber(
DDS_SUBSCRIBER_QOS_DEFAULT,
NULL, DDS_STATUS_MASK_NONE);
if (subscriber == NULL) {
// handle error
}

For more examples, see Configuring QoS Settings when the Subscriber is Created (Section 7.2.4.1 on
page 448).
After you create a Subscriber, the next step is to use the Subscriber to create a DataReader for each Topic,
see Creating DataReaders (Section 7.3.1 on page 463). For a list of operations you can perform with a
Subscriber, see Table 7.1 Subscriber Operations.

7.2.3 Deleting Subscribers
(Note: in the Modern C++ API, Entities are automatically destroyed, see Creating and Deleting DDS Entities (Section 4.1.1 on page 153))

446

7.2.3.1 Deleting Contained DataReaders

This section applies to both implicitly and explicitly created Subscribers.
To delete a Subscriber:
1. You must first delete all DataReaders that were created with the Subscriber. Use the Subscriber’s
delete_datareader() operation (Creating DataReaders (Section 7.3.1 on page 463)) to delete them
one at a time, or use the delete_contained_entities() operation (Deleting Contained DataReaders
(Section 7.2.3.1 below)) to delete them all at the same time.
DDS_ReturnCode_t delete_datareader (DDSDataReader *a_datareader)

2. Delete the Subscriber by using the DomainParticipant’s delete_subscriber() operation ().
Note: A Subscriber cannot be deleted within a listener callback, see Restricted Operations in Listener Callbacks (Section 4.5.1 on page 185).
7.2.3.1 Deleting Contained DataReaders
The Subscriber’s delete_contained_entities() operation deletes all the DataReaders that were created by
the Subscriber. It also deletes the ReadConditions created by each contained DataReader.
DDS_ReturnCode_t DDSSubscriber::delete_contained_entities ()

After this operation returns successfully, the application may delete the Subscriber (see Deleting Subscribers (Section 7.2.3 on the previous page)).
The operation will return PRECONDITION_NOT_MET if any of the contained entities cannot be
deleted. This will occur, for example, if a contained DataReader cannot be deleted because the application
has called read() but has not called the corresponding return_loan() operation to return the loaned DDS
samples.

7.2.4 Setting Subscriber QosPolicies
A Subscriber’s QosPolicies control its behavior. Think of the policies as the configuration and behavior
‘properties’ for the Subscriber. The DDS_SubscriberQos structure has the following format:
struct DDS_SubscriberQos {
DDS_PresentationQosPolicy
DDS_PartitionQosPolicy
DDS_GroupDataQosPolicy
DDS_EntityFactoryQosPolicy
DDS_ExclusiveAreaQosPolicy
DDS_EntityNameQosPolicy
};

presentation;
partition;
group_data;
entity_factory;
exclusive_area;
subscriber_name;

447

7.2.4.1 Configuring QoS Settings when the Subscriber is Created

Note: set_qos() cannot always be used by a Listener, see Restricted Operations in Listener Callbacks (Section 4.5.1 on page 185).
Table 7.2 Subscriber QosPolicies summarizes the meaning of each policy. Subscribers have the same set
of QosPolicies as Publishers; they are described in detail in Publisher/Subscriber QosPolicies (Section 6.4
on page 312). For information on why you would want to change a particular QosPolicy, see the referenced section. For defaults and valid ranges, please refer to the API Reference HTML documentation
for each policy.
Table 7.2 Subscriber QosPolicies
QosPolicy
ENTITYFACTORY QosPolicy (Section 6.4.2 on page 315)

Description
Whether or not new entities created from this entity will start out as ‘enabled.’

ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9 on
Assigns a name and role_name to a Subscriber.
page 374)
EXCLUSIVE_AREA QosPolicy (DDS Extension) (Section
6.4.3 on page 318)

Whether or not the entity uses a multi-thread safe region with deadlock protection.

GROUP_DATA QosPolicy (Section 6.4.4 on page 320)

A place to pass group-level information among applications. Usage is application-dependent.

PARTITION QosPolicy (Section 6.4.5 on page 323)

Set of strings that introduces a logical partition among Topics visible by Publisher/Subscriber.

PRESENTATION QosPolicy (Section 6.4.6 on page 330)

The order in which instance changes are presented to the Subscriber. By
default, no order is used.

7.2.4.1 Configuring QoS Settings when the Subscriber is Created
As described in Creating Subscribers (Section 7.2.2 on page 445), there are different ways to create a Subscriber, depending on how you want to specify its QoS (with or without a QoS Profile).
l

l

In Creating Subscribers (Section 7.2.2 on page 445) is an example of how to explicitly create a Subscriber with default QosPolicies. It used the special constant, DDS_SUBSCRIBER_QOS_
DEFAULT, which indicates that the default QoS values for a Subscriber should be used. The
default Subscriber QosPolicies are configured in the DomainParticipant; you can change them with
the DomainParticipant’s set_default_subscriber_qos() or set_default_subscriber_qos_with_profile() operation (see Getting and Setting Default QoS for Child Entities (Section 8.3.6.5 on page
568)).
To create a Subscriber with non-default QoS settings, without using a QoS profile, see Figure 7.3
Creating a Subscriber with Non-Default QosPolicies (not from a profile) on the facing page. It uses
the DomainParticipant’s get_default_subscriber_qos() method to initialize a DDS_Sub-

448

7.2.4.1 Configuring QoS Settings when the Subscriber is Created

scriberQos structure. Then the policies are modified from their default values before the QoS structure is passed to create_subscriber().
l

l

You can also create a Subscriber and specify its QoS settings via a QoS Profile. To do so, call create_subscriber_with_profile(), as seen in Figure 7.4 Creating a Subscriber with a QoS Profile
below.
If you want to use a QoS profile, but then make some changes to the QoS before creating the Subscriber, call get_subscriber_qos_from_profile(), modify the QoS and use the modified QoS structure when calling create_subscriber(), as seen in Figure 7.5 Getting QoS Values from a Profile,
Changing QoS Values, Creating a Subscriber with Modified QoS Values on the next page.

For more information, see Creating Subscribers (Section 7.2.2 on page 445) and Configuring QoS with
XML (Section Chapter 17 on page 791).
Figure 7.3 Creating a Subscriber with Non-Default QosPolicies (not from a profile)
DDS_SubscriberQos subscriber_qos;1
// get defaults
if (participant->get_default_subscriber_qos(subscriber_qos) !=
DDS_RETCODE_OK){
// handle error
}
// make QoS changes here. for example, this changes the ENTITY_FACTORY QoS
subscriber_qos.entity_factory.autoenable_created_entities=DDS_BOOLEAN_FALSE;
// create the subscriber
DDSSubscriber * subscriber = participant->create_subscriber(subscriber_qos,
NULL, DDS_STATUS_MASK_NONE);
if (subscriber == NULL) {
// handle error
}

Figure 7.4 Creating a Subscriber with a QoS Profile

// create the subscriber with QoS profile
DDSSubscriber * subscriber = participant->create_subscriber_with_profile(
“MySubscriberLibary”, “MySubscriberProfile”, NULL, DDS_STATUS_MASK_NONE);
if (subscriber == NULL) {
// handle error
}

1Note: In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
449

7.2.4.2 Comparing QoS Values

Figure 7.5 Getting QoS Values from a Profile, Changing QoS Values, Creating a Subscriber
with Modified QoS Values
DDS_SubscriberQos subscriber_qos;1
// Get subscriber QoS from profile
retcode = factory->get_subscriber_qos_from_profile(subscriber_qos,
“SubscriberLibrary”, “SubscriberProfile”);
if (retcode != DDS_RETCODE_OK) {
// handle error
}
// Makes QoS changes here
// for example, this changes the ENTITY_FACTORY QoS
subscriber_qos.entity_factory.autoenable_created_entities = DDS_BOOLEAN_TRUE;
// create the subscriber with modified QoS
DDSPublisher* subscriber = participant->create_subscriber(
“Example Foo”, type_name, subscriber_qos,
NULL, DDS_STATUS_MASK_NONE);
if (subscriber == NULL) {
// handle error
}

7.2.4.2 Comparing QoS Values
The equals() operation compares two Subscriber’s DDS_SubscriberQoS structures for equality. It takes
two parameters for the two Subscriber’s QoS structures to be compared, then returns TRUE is they are
equal (all values are the same) or FALSE if they are not equal.
7.2.4.3 Changing QoS Settings After Subscriber Has Been Created
There are 2 ways to change an existing Subscriber’s QoS after it is has been created—again depending on
whether or not you are using a QoS Profile.
l

l

To change an existing Subscriber’s QoS programmatically (that is, without using a QoS profile),
get_qos() and set_qos(). See the example code in Figure 7.6 Changing the Qos of an Existing Subscriber on the facing page. It retrieves the current values by calling the Subscriber’s get_qos() operation. Then it modify the value and call set_qos() to apply the new value. Note, however, that some
QosPolicies cannot be changed after the Subscriber has been enabled—this restriction is noted in the
descriptions of the individual QosPolicies.
You can also change a Subscriber’s (and all other Entities’) QoS by using a QoS Profile and calling
set_qos_with_profile(). For an example, see Figure 7.7 Changing the QoS of an Existing

1Note: In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
450

7.2.4.4 Getting and Settings Subscriber’s Default QoS Profile and Library

Subscriber with a QoS Profile on the facing page. For more information, see Configuring QoS with
XML (Section Chapter 17 on page 791).
Figure 7.6 Changing the Qos of an Existing Subscriber

DDS_SubscriberQos subscriber_qos;
// Get current QoS. subscriber points to an existing DDSSubscriber.
if (subscriber->get_qos(subscriber_qos) != DDS_RETCODE_OK) {
// handle error
}
// make changes
// New entity_factory autoenable_created_entities will be true
subscriber_qos.entity_factory.autoenable_created_entities =
DDS_BOOLEAN_TRUE;
// Set the new QoS
if (subscriber->set_qos(subscriber_qos) != DDS_RETCODE_OK ) {
// handle error
}

Figure 7.7 Changing the QoS of an Existing Subscriber with a QoS Profile

retcode = subscriber->set_qos_with_profile(
“SubscriberProfileLibrary”,”SubscriberProfile”);
if (retcode != DDS_RETCODE_OK) {
// handle error
}

7.2.4.4 Getting and Settings Subscriber’s Default QoS Profile and Library
You can retrieve the default QoS profile used to create Subscribers with the get_default_profile() operation. You can also get the default library for Subscribers, as well as the library that contains the Subscriber’s default profile (these are not necessarily the same library); these operations are called get_
default_library() and get_default_library_profile(), respectively. These operations are for informational
purposes only (that is, you do not need to use them as a precursor to setting a library or profile.) For more
information, see Configuring QoS with XML (Section Chapter 17 on page 791).
virtual const char * get_default_library ()
const char * get_default_profile ()
const char * get_default_profile_library ()

There are also operations for setting the Subscriber’s default library and profile:
DDS_ReturnCode_t
const char *
DDS_ReturnCode_t
const char *
const char *

set_default_library (
library_name)
set_default_profile (
library_name,
profile_name)

451

7.2.4.5 Getting and Setting Default QoS for DataReaders

These operations only affect which library/profile will be used as the default the next time a default Subscriber library/profile is needed during a call to one of this Subscriber’s operations.
When calling a Subscriber operation that requires a profile_name parameter, you can use NULL to refer
to the default profile. (This same information applies to setting a default library.)
If the default library/profile is not set, the Subscriber inherits the default from the DomainParticipant.
set_default_profile() does not set the default QoS for DataReaders created by the Subscriber; for this
functionality, use the Subscriber’s set_default_datareader_qos_with_profile(), see Getting and Setting
Default QoS for DataReaders (Section 7.2.4.5 below) (you may pass in NULL after having called the
Subscriber’s set_default_profile()).
set_default_profile() does not set the default QoS for newly created Subscribers; for this functionality, use
the DomainParticipant’s set_default_subscriber_qos_with_profile() operation, see Getting and Setting
Default QoS for Child Entities (Section 8.3.6.5 on page 568).
7.2.4.5 Getting and Setting Default QoS for DataReaders
These operations set the default QoS that will be used for new DataReaders if create_datareader() is
called with DDS_DATAREADER_QOS_DEFAULT as the ‘qos’ parameter:
DDS_ReturnCode_t set_default_datareader_qos (const DDS_DataReaderQos &qos)
DDS_ReturnCode_t set_default_datareader_qos_with_profile (
const char *library_name, const char *profile_name)

The above operations may potentially allocate memory, depending on the sequences contained in some
QoS policies.
To get the default QoS that will be used for creating DataReaders if create_datareader() is called with
DDS_DATAREADER_QOS_DEFAULT as the ‘qos’ parameter:
DDS_ReturnCode_t get_default_datareader_qos (DDS_DataReaderQos & qos)

The above operation gets the QoS settings that were specified on the last successful call to set_default_
datareader_qos() or set_default_datareader_qos_with_profile(), or if the call was never made, the
default values listed in DDS_DataReaderQos.
Note: It is not safe to set the default DataReader QoS values while another thread may be simultaneously
calling get_default_datareader_qos(), set_default_datareader_qos() or create_datareader() with
DDS_DATAREADER_QOS_DEFAULT as the qos parameter. It is also not safe to get the default
DataReader QoS values while another thread may be simultaneously calling set_default_datareader_
qos().

452

7.2.4.6 Subscriber QoS-Related Operations

7.2.4.6 Subscriber QoS-Related Operations
l

Copying a Topic’s QoS into a DataReader’s QoS
This method is provided as a convenience for setting the values in a DataReaderQos structure
before using that structure to create a DataReader. As explained in Setting Topic QosPolicies (Section 5.1.3 on page 204), most of the policies in a TopicQos structure do not apply directly to the
Topic itself, but to the associated DataWriters and DataReaders of that Topic. The TopicQos serves
as a single container where the values of QosPolicies that must be set compatibly across matching
DataWriters and DataReaders can be stored.
Thus instead of setting the values of the individual QosPolicies that make up a DataReaderQos
structure every time you need to create a DataReader for a Topic, you can use the Subscriber’s
copy_from_topic_qos() operation to “import” the Topic’s QosPolicies into a DataReaderQos structure. This operation copies the relevant policies in the TopicQos to the corresponding policies in the
DataReaderQos.

l

This copy operation will often be used in combination with the Subscriber’s get_default_
datareader_qos() and the Topic’s get_qos() operations. The Topic’s QoS values are merged on top
of the Subscriber’s default DataReader QosPolicies with the result used to create a new
DataReader, or to set the QoS of an existing one (see Setting DataReader QosPolicies (Section
7.3.8 on page 482)).
Copying a Subscriber’s QoS

l

In the C API users should use the DDS_SubscriberQos_copy() operation rather than using structure assignment when copying between two QoS structures. The copy() operation will perform a
deep copy so that policies that allocate heap memory such as sequences are copied correctly. In
C++, C++/CLI, C# and Java, a copy constructor is provided to take care of sequences automatically.
Clearing QoS-Related Memory
Some QosPolicies contain sequences that allocate memory dynamically as they grow or shrink. The
C API’s DDS_SubscriberQos_finalize() operation frees the memory used by sequences but otherwise leaves the QoS unchanged. C users should call finalize() on all DDS_SubscriberQos
objects before they are freed, or for QoS structures allocated on the stack, before they go out of
scope. In C++, C++/CLI, C# and Java, the memory used by sequences is freed in the destructor.

7.2.5 Beginning and Ending Group-Ordered Access
The Subscriber’s begin_access() operation indicates that the application is about to access the DDS data
samples in any of the DataReaders attached to the Subscriber.
If the Subscriber’s access_scope (in the PRESENTATION QosPolicy (Section 6.4.6 on page 330)) is
GROUP or HIGHEST_OFFERED and ordered_access (also in the PRESENTATION QosPolicy

453

7.2.6 Setting Up SubscriberListeners

(Section 6.4.6 on page 330)) is TRUE, the application is required to use this operation to access the DDS
samples in order across DataWriters of the same group (Publisher with access_scope GROUP).
In the above case, begin_access() must be called prior to calling any of the sample-accessing operations:
get_datareaders() on the Subscriber, and read(), take(), read_w_condition(), and take_w_condition()
on any DataReader.
Once the application has finished accessing the DDS data samples, it must call end_access().
The application is not required to call begin_access() and end_access() to access the DDS samples in
order if the Publisher’s access_scope is something other than GROUP. In this case, calling begin_access
() and end_access() is not considered an error and has no effect.
Calls to begin_access() and end_access() may be nested and must be balanced. That is, end_access()
close a previous call to begin_access().

7.2.6 Setting Up SubscriberListeners
Like all Entities, Subscribers may optionally have Listeners. Listeners are user-defined objects that implement a DDS-defined interface (i.e. a pre-defined set of callback functions). Listeners provide the means for
Connext DDS to notify applications of any changes in Statuses (events) that may be relevant to it. By writing the callback functions in the Listener and installing the Listener into the Subscriber, applications can be
notified to handle the events of interest. For more general information on Listeners and Statuses, see Listeners (Section 4.4 on page 177).
Some operations cannot be used within a listener callback, see Restricted Operations in Listener Callbacks (Section 4.5.1 on page 185).
Note:

As illustrated in Subscription Module (Section Figure 7.1 on page 441), the SubscriberListener interface
extends the DataReaderListener interface. In other words, the SubscriberListener interface contains all the
functions in the DataReaderListener interface. In addition, a SubscriberListener has an additional function:
on_data_on_readers(), corresponding to the Subscriber’s DATA_ON_READERS status. This is the
only status that is specific to a Subscriber. This status is closely tied to the DATA_AVAILABLE status
(DATA_AVAILABLE Status (Section 7.3.7.1 on page 471)) of DataReaders.
The Subscriber’s DATA_ON_READERS status is set whenever the DATA_AVAILABLE status is set
for any of the DataReaders created by the Subscriber. This implies that one of its DataReaders has
received new DDS data samples. When the DATA_ON_READERS status is set, the
SubscriberListener’s on_data_on_readers() method will be invoked.
The DATA_ON_READERS status of a Subscriber takes precedence over the DATA_AVAILABLE
status of any of its DataReaders. Thus, when data arrives for a DataReader, the on_data_on_readers()
operation of the SubscriberListener will be called instead of the on_data_available() operation of the
DataReaderListener—assuming that the Subscriber has a Listener installed that is enabled to handle
changes in the DATA_ON_READERS status. (Note however, that in the SubscriberListener’s on_

454

7.2.6 Setting Up SubscriberListeners

data_on_readers() operation, you may choose to call notify_datareaders(), which in turn may cause the
DataReaderListener’s on_data_available() operation to be called.)
All of the other methods of a SubscriberListener will be called back for changes in the Statuses of Subscriber’s DataReaders only if the DataReader is not set up to handle the statuses itself.
If you want a Subscriber to handle status events for its DataReaders, you can set up a SubscriberListener
during the Subscriber’s creation or use the set_listener() method after the Subscriber is created. The last
parameter is a bit-mask with which you should set which Status events that the SubscriberListener will
handle. For example,
DDS_StatusMask mask =
DDS_REQUESTED_DEADLINE_MISSED_STATUS |
DDS_REQUESTED_INCOMPATIBLE_QOS_STATUS;
subscriber = participant->create_subscriber(
DDS_SUBSCRIBER_QOS_DEFAULT, listener, mask);

or
DDS_StatusMask mask =
DDS_REQUESTED_DEADLINE_MISSED_STATUS |
DDS_REQUESTED_INCOMPATIBLE_QOS_STATUS;
subscriber->set_listener(listener, mask);

As previously mentioned, the callbacks in the SubscriberListener act as ‘default’ callbacks for all the
DataReaders contained within. When Connext DDS wants to notify a DataReader of a relevant Status
change (for example, SUBSCRIPTION_MATCHED), it first checks to see if the DataReader has the
corresponding DataReaderListener callback enabled (such as the on_subscription_matched() operation).
If so, Connext DDS dispatches the event to the DataReaderListener callback. Otherwise, Connext DDS
dispatches the event to the corresponding SubscriberListener callback.
NOTE, the reverse is true for the DATA_ON_READERS/DATA_AVAILABLE status. When
DATA_AVAILABLE changes for any DataReaders of a Subscriber, Connext DDS first checks to see if
the SubscriberListener has DATA_ON_READERS enabled. If so, Connext DDS will invoke the on_
data_on_readers() callback. Otherwise, Connext DDS dispatches the event to the Listener (on_data_
available()) of the DataReader whose DATA_AVAILABLE status actually changed.
A particular callback in a DataReader is not enabled if either:
l

l

The application installed a NULL DataReaderListener (meaning there are no callbacks for the
DataReader at all).
The application has disabled the callback for a DataReaderListener. This is done by turning off the
associated status bit in the mask parameter passed to the set_listener() or create_datareader() call

455

7.2.7 Getting DataReaders with Specific DDS Samples

when installing the DataReaderListener on the DataReader. For more information on DataReaderListener, see Setting Up DataReaderListeners (Section 7.3.4 on page 466).
Similarly, the callbacks in the DomainParticipantListener act as ‘default’ callbacks for all the Subscribers
that belong to it. For more information on DomainParticipantListeners, see Setting Up DomainParticipantListeners (Section 8.3.5 on page 560).
The Subscriber also provides an operation called notify_datareaders() that can be used to invoke the on_
data_available() callbacks of DataReaders who have new DDS data samples in their receive queues.
Often notify_datareaders() will be used in the on_data_on_readers() callback to pass off the real processing of data from the SubscriberListener to the individual DataReaderListeners.
Calling notify_datareaders() causes the DATA_ON_READERS status to be reset.
Simple SubscriberListener (Section Figure 7.8 below) shows a SubscriberListener that simply notifies its
DataReaders when new data arrives.
Figure 7.8 Simple SubscriberListener

class MySubscriberListener : public DDSSubscriberListener {
public:
void on_data_on_readers(DDSSubscriber *);
/* For this example we take no action other operations */
};
void MySubscriberListener::on_data_on_readers (DDSSubscriber *subscriber)
{
// do global processing
...
// now dispatch data arrival event to specific DataReaders
subscriber->notify_datareaders();
}

7.2.7 Getting DataReaders with Specific DDS Samples
The Subscriber’s get_datareaders() operation retrieves a list of DataReaders that have DDS samples
with specific sample_states, view_states, and instance_states.
If the application is outside a begin_access()/end_access() block, or if the Subscriber’s access_scope (in
the PRESENTATION QosPolicy (Section 6.4.6 on page 330)) is INSTANCE or TOPIC, or ordered_
access (also in the PRESENTATION QosPolicy (Section 6.4.6 on page 330)) is FALSE, the returned collection is a 'set' containing each DataReader at most once, in no specified order.
If the application is within a begin_access()/end_access() block, and the Subscriber’s access_scope is
GROUP or HIGHEST_OFFERED, and ordered_access is TRUE, the returned collection is a 'list' of
DataReaders, where a DataReader may appear more than one time.

456

7.2.8 Finding a Subscriber’s Related Entities

To retrieve the DDS samples in the order in which they were published across DataWriters of the same
group (a Publisher configured with GROUP access_scope), the application should read()/take() from
each DataReader in the same order as appears in the output sequence. The application will move to the
next DataReader when the read()/take() operation fails with NO_DATA.
DDS_ReturnCode_t get_datareaders (DDSDataReaderSeq & readers,
DDS_SampleStateMask
sample_states,
DDS_ViewStateMask
view_states,
DDS_InstanceStateMask instance_states)

For more information, see The SampleInfo Structure (Section 7.4.6 on page 504).

7.2.8 Finding a Subscriber’s Related Entities
These Subscriber operations are useful for obtaining a handle to related entities:
l
l

get_participant(): Gets the DomainParticipant with which a Subscriber was created.
lookup_datareader(): Finds a DataReader created by the Subscriber with a Topic of a particular
name. Note that if multiple DataReaders were created by the same Subscriber with the same Topic,
any one of them may be returned by this method.
You can use this operation on a built-in Subscriber to access the built-in DataReaders for the builtin topics. The built-in DataReader is created when this operation is called on a built-in topic for the
first time.
If you are going to modify the transport properties for the built-in DataReaders, do so before using
this operation. Built-in transports are implicitly registered when the DomainParticipant is enabled or
the first DataWriter/DataReader is created. To ensure that built-in DataReaders receive all the discovery traffic, you should lookup the DataReader before the DomainParticipant is enabled. Therefore the suggested sequence when looking up built-in DataReaders is:
1. Create a disabled DomainParticipant (see ENTITYFACTORY QosPolicy (Section 6.4.2
on page 315)).
2. If you want to use non-default values, modify the built-in transport properties (see Setting
Builtin Transport Properties of Default Transport Instance—get/set_builtin_transport_properties() (Section 15.5 on page 746)).
3. Call get_builtin_subscriber() (see Built-in DataReaders (Section 16.2 on page 773)).
4. Call lookup_datareader().
5. Call enable() on the DomainParticipant (see Enabling DDS Entities (Section 4.1.2 on page
154)).

l

DDS_Subscriber_as_Entity(): This method is provided for C applications and is necessary when
invoking the parent class Entity methods on Subscribers. For example, to call the Entity method get_

457

7.2.9 Statuses for Subscribers

status_changes() on a Subscriber, my_sub, do the following:
DDS_Entity_get_status_changes(DDS_Subscriber_as_Entity(my_sub))

l

DDS_Subscriber_as_Entity() is not provided in the C++, C++/CLI, C# and Java APIs because
the object-oriented features of those languages make it unnecessary.

7.2.9 Statuses for Subscribers
The status indicators for a Subscriber are the same as those available for its DataReaders, with one additional status: DATA_ON_READERS (DATA_ON_READERS Status (Section 7.2.9.1 below)). The
following statuses can be monitored by the SubscriberListener.
l

DATA_ON_READERS Status (Section 7.2.9.1 below)

l

DATA_AVAILABLE Status (Section 7.3.7.1 on page 471)

l

LIVELINESS_CHANGED Status (Section 7.3.7.4 on page 475)

l

REQUESTED_DEADLINE_MISSED Status (Section 7.3.7.5 on page 476)

l

REQUESTED_INCOMPATIBLE_QOS Status (Section 7.3.7.6 on page 477)

l

SAMPLE_LOST Status (Section 7.3.7.7 on page 478)

l

SAMPLE_REJECTED Status (Section 7.3.7.8 on page 479)

l

SUBSCRIPTION_MATCHED Status (Section 7.3.7.9 on page 482)

You can access Subscriber status by using a SubscriberListener or its inherited get_status_changes() operation (see Getting Status and Status Changes (Section 4.1.4 on page 157)), which can be used to explicitly
poll for the DATA_ON_READERS status of the Subscriber.
7.2.9.1 DATA_ON_READERS Status
The DATA_ON_READERS status, like the DATA_AVAILABLE status for DataReaders, is a read
communication status, which makes it somewhat different from other plain communication statuses. (See
Types of Communication Status (Section 4.3.1 on page 170) for more information on statuses and the difference between read and plain statuses.) In particular, there is no status-specific data structure; the status
is either changed or not, there is no additional associated information.
The DATA_ON_READERS status indicates that there is new data available for one or more DataReaders that belong to this Subscriber. The DATA_AVAILABLE status for each such DataReader will also
be updated.
The DATA_ON_READERS status is reset (the corresponding bit in the bitmask is turned off) when you
call read(), take(), or one of their variations on any of the DataReaders that belong to the Subscriber. This

458

7.3 DataReaders

is true even if the DataReader on which you call read/take is not the same DataReader that caused the
DATA_ON_READERS status to be set in the first place. This status is also reset when you call notify_
datareaders() on the Subscriber, or after on_data_on_readers() is invoked.
If a SubscriberListener has both on_data_on_readers() and on_data_available() callbacks enabled (by
turning on both status bits), only on_data_on_readers() is called.

7.3 DataReaders
To create a DataReader, you need a DomainParticipant, a Topic, and optionally, a Subscriber. You need
at least one DataReader for each Topic whose DDS data samples you want to receive.
After you create a DataReader, you will be able to use the operations listed in Table 7.3 DataReader Operations. You are likely to use many of these operations from within your DataReader’s Listener, which is
invoked when there are status changes or new DDS data samples. For more details on all operations, see
the API reference HTML documentation. The DataReaderListener is described in Setting Up DataReaderListeners (Section 7.3.4 on page 466).
DataReaders are created by using operations on a DomainParticipant or a Subscriber, as described in
Creating Subscribers Explicitly vs. Implicitly (Section 7.2.1 on page 444). If you use the DomainParticipant’s operations, the DataReader will belong to an implicit Subscriber that is automatically created by
the middleware. If you use a Subscriber’s operations, the DataReader will belong to that Subscriber. So
either way, the DataReader belongs to a Subscriber.
Some operations cannot be used within a listener callback, see Restricted Operations in Listener Callbacks (Section 4.5.1 on page 185).
Note:

Table 7.3 DataReader Operations
Purpose

Configuring the
DataReader

Operation

Description

Reference

enable

Enables the DataReader.

Enabling DDS Entities
(Section 4.1.2 on page 154)

equals

Compares two DataReader’s QoS structures for
equality.

Comparing QoS Values
(Section 7.3.8.2 on page 487)

get_qos

Gets the QoS.

set_qos

Modifies the QoS.

set_qos_with_
profile

Modifies the QoS based on a QoS profile.

get_listener

Gets the currently installed Listener.

set_listener

Replaces the Listener.

Setting DataReader
QosPolicies (Section 7.3.8
on page 482)

Setting Up
DataReaderListeners (Section
7.3.4 on page 466)

459

7.3 DataReaders

Table 7.3 DataReader Operations
Purpose

Operation

Description

Reference

read

Reads (copies) a collection of DDS data samples
from the DataReader.

Accessing DDS Data
Samples with Read or Take
(Section 7.4.3 on page 493)

read_instance

Identical to read, but all DDS samples returned
belong to a single instance, which you specify as a
parameter.

read_instance and take_
instance (Section 7.4.3.4 on
page 497)

Accessing DDS Data
Samples with “Read”

read_instance_
w_condition

Identical to read_instance, but all DDS samples
returned belong to a single instance and satisfy a
specific ReadCondition.

(Use
FooData-Reader, see Accessing
DDS Data Samples with Read or
Take (Section 7.4.3 on page
493))

read_instance_w_condition
and take_instance_w_
condition (Section 7.4.3.7 on
page 500)

read_next_
instance

Similar to read_instance, but the actual instance is not
directly specified as a parameter. Instead, the DDS
samples will all belong to instance ordered after the
one previously read.

read_next_instance and take_
next_instance (Section 7.4.3.5
on page 498)

read_next_
instance_w_
condition

Accesses a collection of DDS data samples of the
next instance that match a specific set of
ReadConditions, from the DataReader.

read_next_instance_w_
condition and take_next_
instance_w_condition
(Section 7.4.3.8 on page 501)

read_next_
sample

Reads the next not-previously-accessed data value
from the DataReader.

read_next_sample and take_
next_sample (Section 7.4.3.3
on page 497)

read_w_
condition

Accesses a collection of DDS data samples from the
DataReader that match specific ReadCondition
criteria.

read_w_condition and take_
w_condition (Section 7.4.3.6
on page 500)

460

7.3 DataReaders

Table 7.3 DataReader Operations
Purpose

Accessing DDS Data
Samples with “Take”
(Use
FooData-Reader, see Accessing
DDS Data Samples with Read or
Take (Section 7.4.3 on page
493))

Working with DDS Data
Samples and FooData-Reader
(Use FooData-Reader, see
Accessing DDS Data Samples
with Read or Take (Section 7.4.3
on page 493))

Operation

Description

Reference

take

Accessing DDS Data
Like read, but the DDS samples are removed from the
Samples with Read or Take
DataReader’s receive queue.
(Section 7.4.3 on page 493)

take_instance

read_instance and take_
Identical to take, but all DDS samples returned belong
instance (Section 7.4.3.4 on
to a single instance, which you specify as a parameter.
page 497)

take_instance_
w_condition

Identical to take_instance, but all DDS samples
returned belong to a single instance and satisfy a
specific ReadCondition.

read_instance_w_condition
and take_instance_w_
condition (Section 7.4.3.7 on
page 500)

take_next_
instance

Like read_next_instance, but the DDS samples are
removed from the DataReader’s receive queue.

read_next_instance and take_
next_instance (Section 7.4.3.5
on page 498)

take_next_
instance_w_
condition

Accesses (and removes) a collection of DDS data
samples of the next instance that match a specific set
of ReadConditions, from the DataReader.

read_next_instance_w_
condition and take_next_
instance_w_condition
(Section 7.4.3.8 on page 501)

take_next_
sample

Like read_next_sample, but the DDS samples are
removed from the DataReader’s receive queue.

read_next_sample and take_
next_sample (Section 7.4.3.3
on page 497)

take_w_
condition

Accesses (and removes) a collection of DDS data
samples from the DataReader that match specific
ReadCondition criteria.

read_w_condition and take_
w_condition (Section 7.4.3.6
on page 500)

narrow

A type-safe way to cast a pointer. This takes a
Using a Type-Specific
DDSDataReader pointer and ‘narrows’ it to a
DataReader (FooDataReader)
‘FooDataReader’ where ‘Foo’ is the related data type. (Section 7.4.1 on page 491)

return_loan

Loaning and Returning Data
Returns buffers loaned in a previous read or take call. and SampleInfo Sequences
(Section 7.4.2 on page 492)

get_key_value

Gets the key for an instance handle.

Getting the Key Value for an
Instance (Section 7.3.9.5 on
page 491)

lookup_
instance

Gets the instance handle that corresponds to an
instance key.

Looking Up an Instance
Handle (Section 7.3.9.4 on
page 490)

461

7.3 DataReaders

Table 7.3 DataReader Operations
Purpose

Operation

Description

acknowledge_
all

Acknowledge all previously accessed DDS samples.

acknowledge_
sample

Acknowledge a single DDS sample.

get_liveliness_
changed_
status

Gets LIVELINESS_CHANGED_STATUS
status.

get_requested_
deadline_
missed_status

Gets REQUESTED_DEADLINE_
MISSED_STATUS status.

get_requested_
incompatible_
qos_status

Gets REQUESTED_INCOMPATIBLE_
QOS_STATUS status.

get_sample_
lost_status

Gets SAMPLE_LOST_STATUS status.

get_sample_
rejected_
status

Gets SAMPLE_REJECTED_STATUS status.

get_
subscription_
matched_
status

Gets SUBSCRIPTION_MATCHED_STATUS
status.

get_status_
changes

Getting Status and Status
Gets a list of statuses that changed since last time the
Changes (Section 4.1.4 on
application read the status or the listeners were called.
page 157)

Acknowledging DDS Samples

Checking Status

Reference

Acknowledging DDS
Samples (Section 7.4.4 on
page 502)

Statuses for DataReaders
(Section 7.3.7 on page 470)

get_datareader_
cache_
Gets DATA_READER_CACHE_STATUS status.
status
get_datareader_
Gets DATA_READER_PROTOCOL_
protocol_
STATUS status.
status
get_matched_
publication_
datareader_
protocol_
status

Checking DataReader Status
and StatusConditions (Section
7.3.5 on page 468)
Statuses for DataReaders
(Section 7.3.7 on page 470)

Get the protocol status for this DataReader, per
matched publication identified by the publication_
handle.

462

7.3.1 Creating DataReaders

Table 7.3 DataReader Operations
Purpose

Navigating Relationships

Operation

Description

get_instance_
handle

Returns the DDS_InstanceHandle_t associated with
the Entity.

get_matched_
publication_
data

Gets information on a publication with a matching
Topic and compatible QoS.

get_matched_
publications

Gets a list of publications that have a matching Topic
and compatible QoS. These are the publications
currently associated with the DataReader.

get_matched_
Gets information on a DomainParticipant of a
publication_
matching publication.
participant_data
get_subscriber

Gets the Subscriber that created the DataReader.

get_
Gets the Topic associated with the DataReader.
topicdescription

Working with
Conditions

Waiting for Historical Data

Reference
Getting an Entity’s Instance
Handle (Section 4.1.3 on
page 157)

Finding Matching
Publications (Section 7.3.9.1
on page 489)

Finding the Matching
Publication’s
ParticipantBuiltinTopicData
(Section 7.3.9.2 on page 490)
Finding a DataReader’s
Related Entities (Section
7.3.9.3 on page 490)

create_
querycondition

Creates a QueryCondition.

create_
readcondition

Creates a ReadCondition.

delete_
readcondition

Deletes a ReadCondition/QueryCondition attached to
the DataReader.

delete_
contained_
entities

Deletes all the ReadConditions/QueryConditions that
were created by means of the "create" operations on
the DataReader.

Deleting Contained
ReadConditions (Section
7.3.3.1 on page 466)

get_
statuscondition

Gets the StatusCondition associated with the Entity.

StatusConditions (Section
4.6.8 on page 197)

wait_for_
historical_data

Waits until all "historical" (previously sent) data is
received. Only valid for Reliable DataReaders with
non-VOLATILE DURABILITY.

Waiting for Historical Data
(Section 7.3.6 on page 469)

ReadConditions and
QueryConditions (Section
4.6.7 on page 195)

7.3.1 Creating DataReaders
Before you can create a DataReader, you need a DomainParticipant and a Topic.
DataReaders are created by calling create_datareader() or create_datareader_with_profile()—these
operations exist for DomainParticipants and Subscribers. If you use the DomainParticipant to create a

463

7.3.1 Creating DataReaders

DataReader, it will belong to the implicit Subscriber described in Creating Subscribers Explicitly vs. Implicitly (Section 7.2.1 on page 444). If you use a Subscriber’s operations to create a DataReader, it will
belong to that Subscriber.
A QoS profile is way to use QoS settings from an XML file or string. With this approach, you can change
QoS settings without recompiling the application. For details, see Configuring QoS with XML (Section
Chapter 17 on page 791).
Note: In the Modern C++ API, DataReaders provide constructors whose first argument is a Subscriber.
The only required arguments are the subscriber and the topic.
DDSDataReader* create_datareader(
DDSTopicDescription *topic,
const DDS_DataReaderQos &qos,
DDSDataReaderListener *listener,
DDS_StatusMask mask);
DDSDataReader * create_datareader_with_profile (
DDSTopicDescription * topic,
const char * library_name,
const char * profile_name,
DDSDataReaderListener * listener,
DDS_StatusMask mask)

Where:
topic

The Topic to which the DataReader is subscribing. This must have been previously created by
the same DomainParticipant.

qos

If you want the default QoS settings (described in the API Reference HTML documentation),
use DDS_DATAREADER_QOS_DEFAULT for this parameter (see Creating a DataReader
with Default QosPolicies (Section Figure 7.9 on the facing page)). If you want to customize
any of the QosPolicies, supply a QoS structure (see Setting DataReader QosPolicies (Section
7.3.8 on page 482)).

Note: If you use DDS_DATAREADER_QOS_DEFAULT for the qos parameter, it is not safe
to create the DataReader while another thread may be simultaneously calling the Subscriber’s
set_default_datareader_qos() operation.
listener

A DataReader’sListener is where you define the callback routine that will be notified when new
DDS data samples arrive. Connext DDS also uses this Listener to notify your application of
specific events (status changes) that may occur with respect to the DataReader. For more
information, see Setting Up DataReaderListeners (Section 7.3.4 on page 466) and Statuses
for DataReaders (Section 7.3.7 on page 470).
The listener parameter is optional; you may use NULL instead. In that case, the Subscriber’s
Listener (or if that is NULL, the DomainParticipant’s Listener) will receive the notifications
instead. See Setting Up DataReaderListeners (Section 7.3.4 on page 466) for more on
DataReaderListeners.

464

7.3.2 Getting All DataReaders

mask

This bit mask indicates which status changes will cause the Listener to be invoked. The bits set
in the mask must have corresponding callbacks implemented in the Listener. If you use NULL
for the Listener, use DDS_STATUS_MASK_NONE for this parameter. If the Listener
implements all callbacks, use DDS_STATUS_MASK_ALL. For information on statuses, see
Listeners (Section 4.4 on page 177).

library_name A QoS Library is a named set of QoS profiles. See URL Groups (Section 17.8 on page 814).
profile_name A QoS profile groups a set of related QoS, usually one per entity. See URL Groups (Section
17.8 on page 814).

After you create a DataReader, you can use it to retrieve received data. See Using DataReaders to Access
Data (Read & Take) (Section 7.4 on page 491).
Note: When a DataReader is created, only those transports already registered are available to the
DataReader. The built-in transports are implicitly registered when (a) the DomainParticipant is enabled,
(b) the first DataReader is created, or (c) you lookup a built-in DataReader, whichever happens first.
Creating a DataReader with Default QosPolicies (Section Figure 7.9 below) shows an example of how to
create a DataReader with default QosPolicies.
Figure 7.9 Creating a DataReader with Default QosPolicies
// MyReaderListener is user defined, extends DDSDataReaderListener
DDSDataReaderListener *reader_listener = new MyReaderListener();
DataReader* reader = subscriber->create_datareader(topic,
DDS_DATAREADER_QOS_DEFAULT,
reader_listener, DDS_STATUS_MASK_ALL);
if (reader == NULL) {
// ... error
}
// narrow it into your specific data type
FooDataReader* foo_reader = FooDataReader::narrow(reader);

For more examples on how to create a DataWriter, see Configuring QoS Settings when the DataReader is
Created (Section 7.3.8.1 on page 485)

7.3.2 Getting All DataReaders
To retrieve all the DataReaders created by the Subscriber, use the Subscriber’s get_all_datareaders()
operation:
DDS_ReturnCode_t get_all_datareaders(
DDS_Subscriber* self,
struct DDS_DataReaderSeq* readers);

In the Modern C++ API, use the freestanding function rti::sub::find_datareaders().

465

7.3.3 Deleting DataReaders

7.3.3 Deleting DataReaders
(Note: in the Modern C++ API, Entities are automatically destroyed, see Creating and Deleting DDS Entities (Section 4.1.1 on page 153))
To delete a DataReader:
Delete any ReadConditions and QueryConditions that were created with the DataReader. Use the
DataReader’s delete_readcondition() operation to delete them one at a time, or use the delete_contained_entities() operation (Deleting Contained ReadConditions (Section 7.3.3.1 below)) to delete them
all at the same time.
DDS_ReturnCode_t delete_readcondition

(DDSReadCondition *condition)

Delete the DataReader by using the Subscriber’s delete_datareader() operation (Deleting Subscribers
(Section 7.2.3 on page 446)).
Note: A DataReader cannot be deleted within its own reader listener callback, see Restricted Operations
in Listener Callbacks (Section 4.5.1 on page 185).
To delete all of a Subscriber’s DataReaders, use the Subscriber’s delete_contained_entities() operation
(see Deleting Contained DataReaders (Section 7.2.3.1 on page 447)).
7.3.3.1 Deleting Contained ReadConditions
The DataReader’s delete_contained_entities() operation deletes all the ReadConditions and QueryConditions (ReadConditions and QueryConditions (Section 4.6.7 on page 195)) that were created by the
DataReader.
DDS_ReturnCode_t delete_contained_entities ()

After this operation returns successfully, the application may delete the DataReader (see Deleting
DataReaders (Section 7.3.3 above)).

7.3.4 Setting Up DataReaderListeners
DataReaders may optionally have Listeners. A DataReaderListener is a collection of callback methods;
these methods are invoked by Connext DDS when DDS data samples are received or when there are
status changes for the DataReader.
Some operations cannot be used within a listener callback, see Restricted Operations in Listener Callbacks (Section 4.5.1 on page 185).
Note:

If you do not implement a DataReaderListener, the associated Subscriber’s Listener is used instead. If that
Subscriber does not have a Listener either, then the DomainParticipant’s Listener is used if one exists (see
Setting Up SubscriberListeners (Section 7.2.6 on page 454) and Setting Up DomainParticipantListeners
(Section 8.3.5 on page 560)).

466

7.3.4 Setting Up DataReaderListeners

If you do not require asynchronous notification of data availability or status changes, you do not need to
set a Listener for the DataReader. In that case, you will need to periodically call one of the read() or take
() operations described in Using DataReaders to Access Data (Read & Take) (Section 7.4 on page 491) to
access the data that has been received.
Listeners are typically set up when the DataReader is created (see Creating DataReaders (Section 7.3.1
on page 463)). You can also set one up after creation by using the DataReader’s get_listener() and set_
listener() operations. Connext DDS will invoke a DataReader’s Listener to report the status changes listed
in Table 7.4 DataReaderListener Callbacks (if the Listener is set up to handle the particular status, see Setting Up DataReaderListeners (Section 7.3.4 on the previous page)).
Table 7.4 DataReaderListener Callbacks
This DataReaderListener callback...

...is triggered by a change in this status:

on_data_available()

DATA_AVAILABLE Status (Section 7.3.7.1 on page 471)

on_liveliness_changed()

LIVELINESS_CHANGED Status (Section 7.3.7.4 on page 475)

on_requested_deadline_missed()

REQUESTED_DEADLINE_MISSED Status (Section 7.3.7.5 on page 476)

on_requested_incompatible_qos()

REQUESTED_INCOMPATIBLE_QOS Status (Section 7.3.7.6 on page 477)

on_sample_lost()

SAMPLE_LOST Status (Section 7.3.7.7 on page 478)

on_sample_rejected()

SAMPLE_REJECTED Status (Section 7.3.7.8 on page 479)

on_subscription_matched()

SUBSCRIPTION_MATCHED Status (Section 7.3.7.9 on page 482)

Note that the same callbacks can be implemented in the SubscriberListener or DomainParticipantListener
instead. There is only one SubscriberListener callback that takes precedence over a DataReaderListener’s.
An on_data_on_readers() callback in the SubscriberListener (or DomainParticipantListener) takes precedence over the on_data_available() callback of a DataReaderListener.
If the SubscriberListener implements an on_data_on_readers() callback, it will be invoked instead of the
DataReaderListener’s on_data_available() callback when new data arrives. The on_data_on_readers()
operation can in turn cause the on_data_available() method of the appropriate DataReaderListener to be
invoked by calling the Subscriber’s notify_datareaders() operation. For more information on status and
Listeners, see Listeners (Section 4.4 on page 177).
Simple DataReaderListener (Section Figure 7.10 on the next page) shows a DataReaderListener that
simply prints the data it receives.

467

7.3.5 Checking DataReader Status and StatusConditions

Figure 7.10 Simple DataReaderListener

class MyReaderListener : public DDSDataReaderListener {
public:
virtual void on_data_available(DDSDataReader* reader);
// don’t do anything for the other callbacks
};
void MyReaderListener::on_data_available(DDSDataReader* reader)
{
FooDataReader *Foo_reader = NULL;
FooSeq data_seq; // In C, sequences have to be initialized
DDS_SampleInfoSeq info_seq; // before use, see The Sequence Data Structure (Section
7.4.5 on page 502)
DDS_ReturnCode_t retcode;
int i;
// Must cast generic reader into reader of specific type
Foo_reader = FooDataReader::narrow(reader);
if (Foo_reader == NULL) {
printf("DataReader narrow error\n");
return;
}
retcode = Foo_reader->take(data_seq, info_seq,
DDS_LENGTH_UNLIMITED, DDS_ANY_SAMPLE_STATE,
DDS_ANY_VIEW_STATE, DDS_ANY_INSTANCE_STATE);
if (retcode == DDS_RETCODE_NO_DATA) {
return;
} else if (retcode != DDS_RETCODE_OK) {
printf("take error %d\n", retcode);
return;
}
for (i = 0; i < data_seq.length(); ++i) {
// the data may not be valid if the DDS sample is
// meta information about the creation or deletion
// of an instance
if (info_seq[i].valid_data) {
FooTypeSupport::print_data(&data_seq[i]);
}
}
// Connext DDS gave a pointer to internal memory via
// take(), must return the memory when finished processing the data
retcode = Foo_reader->return_loan(data_seq, info_seq);
if (retcode != DDS_RETCODE_OK) {
printf("return loan error %d\n", retcode);
}
}

7.3.5 Checking DataReader Status and StatusConditions
You can access individual communication status for a DataReader with the operations shown in Table 7.5
DataReader Status Operations.
468

7.3.6 Waiting for Historical Data

Table 7.5 DataReader Status Operations
Use this operation...
get_datareader_cache_status

...to retrieve this status:
DATA_READER_CACHE_STATUS (Section 7.3.7.2 on page 471)

get_datareader_protocol_status
get_matched_publication_
datareader_protocol_status

DATA_READER_PROTOCOL_STATUS (Section 7.3.7.3 on page 472)

get_liveliness_changed_status

LIVELINESS_CHANGED Status (Section 7.3.7.4 on page 475)

get_sample_lost_status

SAMPLE_LOST Status (Section 7.3.7.7 on page 478)

get_sample_rejected_status

SAMPLE_REJECTED Status (Section 7.3.7.8 on page 479)

get_requested_deadline_missed_status

REQUESTED_DEADLINE_MISSED Status (Section 7.3.7.5 on page 476)

get_requested_incompatible_qos_status

REQUESTED_INCOMPATIBLE_QOS Status (Section 7.3.7.6 on page 477)

get_subscription_match_status

SUBSCRIPTION_MATCHED Status (Section 7.3.7.9 on page 482)

get_status_changes

All of the above

get_statuscondition

See StatusConditions (Section 4.6.8 on page 197)

These methods are useful in the event that no Listener callback is set to receive notifications of status
changes. If a Listener is used, the callback will contain the new status information, in which case calling
these methods is unlikely to be necessary.
The get_status_changes() operation provides a list of statuses that have changed since the last time the
status changes were ‘reset.’ A status change is reset each time the application calls the corresponding get_
*_status(), as well as each time Connext DDS returns from calling the Listener callback associated with
that status.
For more on status, see Setting Up DataReaderListeners (Section 7.3.4 on page 466), Statuses for
DataReaders (Section 7.3.7 on the next page), and Listeners (Section 4.4 on page 177).

7.3.6 Waiting for Historical Data
The wait_for_historical_data() operation waits (blocks) until all "historical" data is received from
matched DataWriters. "Historical" data means DDS samples that were written before the DataReader
joined the DDS domain.
This operation is intended only for DataReaders that have:

469

7.3.7 Statuses for DataReaders

l

l

DURABILITY QosPolicy (Section 6.5.7 on page 368) kind set to TRANSIENT_LOCAL (not
VOLATILE)
RELIABILITY QosPolicy (Section 6.5.19 on page 400) kind set to RELIABLE

Calling wait_for_historical_data() on a non-reliable DataReader will always return immediately, since
Connext DDS will never deliver historical data to non-reliable DataReaders.
As soon as an application enables a non-VOLATILE DataReader, it will start receiving both "historical"
data as well as any new data written by matching DataWriters. If you want the subscribing application to
wait until all "historical" data is received, use this operation:
DDS_ReturnCode_t wait_for_historical_data (const DDS_Duration_t &

max_wait)

The wait_for_historical_data() operation blocks the calling thread until either all "historical" data is
received or the duration specified by the max_wait parameter elapses, whichever happens first. A return
value of OK indicates that all the "historical" data was received; a return value of TIMEOUT indicates that
max_wait elapsed before all the data was received.
wait_for_historical_data() will return immediately if no DataWriters have been discovered at the time the
operation is called. Therefore it is advisable to make sure at least one DataWriter has been discovered
before calling this operation; one way to do this is to use get_subscription_matched_status(), like this:
while (1) {
DDS_SubscriptionMatchedStatus status;
MyType_reader->get_subscription_matched_status(status);
if (status.current_count > 0) { break; }
NDDSUtility::sleep(sleep_period);
}

7.3.7 Statuses for DataReaders
There are several types of statuses available for a DataReader. You can use the get_*_status() operations
(Checking DataReader Status and StatusConditions (Section 7.3.5 on page 468)) to access and reset them,
use a DataReaderListener (Setting Up DataReaderListeners (Section 7.3.4 on page 466)) to listen for
changes in their values (for those statuses that have Listeners), or use a StatusCondition and a WaitSet
(StatusConditions (Section 4.6.8 on page 197)) to wait for changes. Each status has an associated data
structure and is described in more detail in the following sections.
l

DATA_AVAILABLE Status (Section 7.3.7.1 on the facing page)

l

DATA_READER_CACHE_STATUS (Section 7.3.7.2 on the facing page)

l

DATA_READER_PROTOCOL_STATUS (Section 7.3.7.3 on page 472)

l

LIVELINESS_CHANGED Status (Section 7.3.7.4 on page 475)

l

REQUESTED_DEADLINE_MISSED Status (Section 7.3.7.5 on page 476)

470

7.3.7.1 DATA_AVAILABLE Status

l

REQUESTED_INCOMPATIBLE_QOS Status (Section 7.3.7.6 on page 477)

l

SAMPLE_LOST Status (Section 7.3.7.7 on page 478)

l

SAMPLE_REJECTED Status (Section 7.3.7.8 on page 479)

l

SUBSCRIPTION_MATCHED Status (Section 7.3.7.9 on page 482)

7.3.7.1 DATA_AVAILABLE Status
This status indicates that new data is available for the DataReader. In most cases, this means that one new
DDS sample has been received. However, there are situations in which more than one DDS samples for
the DataReader may be received before the DATA_AVAILABLE status changes. For example, if the
DataReader has the DURABILITY QosPolicy (Section 6.5.7 on page 368) set to be non-VOLATILE,
then the DataReader may receive a batch of old DDS data samples all at once. Or if data is being received
reliably from DataWriters, Connext DDS may present several DDS samples of data simultaneously to the
DataReader if they have been originally received out of order.
A change to this status also means that the DATA_ON_READERS status is changed for the
DataReader’s Subscriber. This status is reset when you call read(), take(), or one of their variations.
Unlike most other statuses, this status (as well as DATA_ON_READERS for Subscribers) is a read communication status. See Statuses for Subscribers (Section 7.2.9 on page 458) and Types of Communication
Status (Section 4.3.1 on page 170) for more information on read communication statuses.
The DataReaderListener’s on_data_available() callback is invoked when this status changes, unless the
SubscriberListener (Setting Up SubscriberListeners (Section 7.2.6 on page 454)) or DomainParticipantListener (Setting Up DomainParticipantListeners (Section 8.3.5 on page 560)) has implemented an
on_data_on_readers() callback. In that case, on_data_on_readers() will be invoked instead.
7.3.7.2 DATA_READER_CACHE_STATUS
This status keeps track of the number of DDS samples in the reader's cache.
This status does not have an associated Listener. You can access this status by calling the DataReader’s
get_datareader_cache_status() operation, which will return the status structure described in Table 7.6
DDS_DataReaderCacheStatus; this operation will also reset the status so it is no longer considered
“changed.”
Table 7.6 DDS_DataReaderCacheStatus
Type

Field
Name

sample_
DDS_
count_
Long
peak

Description

Highest number of DDS samples in the DataReader’s queue over the lifetime of the DataReader.

471

7.3.7.3 DATA_READER_PROTOCOL_STATUS

Table 7.6 DDS_DataReaderCacheStatus
Type

Field
Name

Description
Current number of DDS samples in the DataReader’s queue.

DDS_ sample_
Long count

Includes DDS samples that may not yet be available to be read or taken by the user due to DDS samples being received
out of order or settings in the PRESENTATION QosPolicy (Section 6.4.6 on page 330).

7.3.7.3 DATA_READER_PROTOCOL_STATUS
The status of a DataReader’s internal protocol related metrics (such as the number of DDS samples
received, filtered, rejected) and the status of wire protocol traffic. The structure for this status appears in
Table 7.7 DDS_DataReaderProtocolStatus.
This status does not have an associated Listener. You can access this status by calling the following operations on the DataReader (which return the status structure described in Table 7.7 DDS_DataReaderProtocolStatus):
get_datareader_protocol_status() returns the sum of the protocol status for all the matched publications
for the DataReader.
get_matched_publication_datareader_protocol_status() returns the protocol status of a particular
matched publication, identified by a publication_handle.
The get_*_status() operations also reset the related status so it is no longer considered “changed.”
Note: Status for a remote entity is only kept while the entity is alive. Once a remote entity is no longer
alive, its status is deleted. If you try to get the matched subscription status for a remote entity that is no
longer alive, the ‘get status’ call will return an error.

472

7.3.7.3 DATA_READER_PROTOCOL_STATUS

Table 7.7 DDS_DataReaderProtocolStatus
Type

Field
Name

Description

received_
The number of DDS samples from a remote DataWriter received for the first time by a local
sample_count DataReader.
received_
sample_
count_
change

The incremental change in the number of DDS samples from a remote DataWriter received for the
first time by a local DataReader since the last time the status was read.

received_
sample_bytes

The number of bytes of DDS samples from a remote DataWriter received for the first time by a local
DataReader.

received_
sample_
bytes_
change

The incremental change in the number of bytes of DDS samples from a remote DataWriter received
for the first time by a local DataReader since the last time the status was read.

DDS_LongLong

duplicate_
The number of DDS samples from a remote DataWriter received, not for the first time, by a local
sample_count DataReader.
duplicate_
sample_
count_
change

The incremental change in the number of DDS samples from a remote DataWriter received, not for
the first time, by a local DataReader since the last time the status was read.

duplicate_
sample_bytes

The number of bytes of DDS samples from a remote DataWriter received, not for the first time, by a
local DataReader.

duplicate_
sample_
bytes_
change

The incremental change in the number of bytes of DDS samples from a remote DataWriter received,
not for the first time, by a local DataReader since the last time the status was read.

DDS_LongLong

filtered_
The number of DDS samples filtered by the local DataReader due to ContentFilteredTopics or Timesample_count Based Filter.
filtered_
sample_
count_
change

The incremental change in the number of DDS samples filtered by the local DataReader due to
Content-FilteredTopics or Time-Based Filter since the last time the status was read.

filtered_
sample_bytes

The number of bytes of DDS samples filtered by the local DataReader due to ContentFilteredTopics
or Time-Based Filter.

filtered_
sample_
bytes_
change

The incremental change in the number of bytes of DDS samples filtered by the local DataReader due
to ContentFilteredTopics or Time-Based Filter since the last time the status was read.

DDS_LongLong

473

7.3.7.3 DATA_READER_PROTOCOL_STATUS

Table 7.7 DDS_DataReaderProtocolStatus
Type

Field
Name

Description

received_
heartbeat_
count

The number of Heartbeats from a remote DataWriter received by a local DataReader.

received_
heartbeat_
count_
change

The incremental change in the number of Heartbeats from a remote DataWriter received by a local
DataReader since the last time the status was read.

received_
heartbeat_
bytes

The number of bytes of Heartbeats from a remote DataWriter received by a local DataReader.

received_
heartbeat_
bytes_
change

The incremental change in the number of bytes of Heartbeats from a remote DataWriter received by a
local DataReader since the last time the status was read.

sent_ack_
count

The number of ACKs sent from a local DataReader to a matching remote DataWriter.

DDS_LongLong

sent_ack_
The incremental change in the number of ACKs sent from a local DataReader to a matching remote
count_change DataWriter since the last time the status was read.
DDS_LongLong
sent_ack_
bytes

The number of bytes of ACKs sent from a local DataReader to a matching remote DataWriter.

sent_ack_
bytes_change

The incremental change in the number of bytes of ACKs sent from a local DataReader to a matching
remote DataWriter since the last time the status was read.

sent_nack_
count

The number of NACKs sent from a local DataReader to a matching remote DataWriter.

sent_nack_
The incremental change in the number of NACKs sent from a local DataReader to a matching remote
count_change DataWriter since the last time the status was read.
DDS_LongLong
sent_nack_
bytes

The number of bytes of NACKs sent from a local DataReader to a matching remote DataWriter.

sent_nack_
bytes_change

The incremental change in the number of bytes of NACKs sent from a local DataReader to a
matching remote DataWriter since the last time the status was read.

474

7.3.7.4 LIVELINESS_CHANGED Status

Table 7.7 DDS_DataReaderProtocolStatus
Type

Field
Name

Description

received_gap_
The number of GAPs received from remote DataWriter to this DataReader.
count
received_gap_ The incremental change in the number of GAPs received from remote DataWriter to this DataReader
count_change since the last time the status was read.
DDS_LongLong
received_gap_
The number of bytes of GAPs received from remote DataWriter to this DataReader.
bytes
received_gap_ The incremental change in the number of bytes of GAPs received from remote DataWriter to this
bytes_change DataReader since the last time the status was read.
rejected_
The number of times a DDS sample is rejected for unanticipated reasons in the receive path.
sample_count
DDS_LongLong

rejected_
The incremental change in the number of times a DDS sample is rejected for unanticipated reasons in
sample_
the receive path since the last time the status was read.
count_change
first_
available_
sample_
sequence_
number

Sequence number of the first available DDS sample in a matched DataWriter's reliability queue.
Applicable only when retrieving matched DataWriter statuses.

last_available_
sample_
Sequence number of the last available DDS sample in a matched DataWriter's reliability queue.
DDS_
Applicable only when retrieving matched DataWriter statuses.
SequenceNumber_ sequence_
number
t
last_
committed_
sample_
sequence_
number

DDS_Long

Sequence number of the last committed DDS sample (i.e. available to be read or taken) in a matched
DataWriter's reliability queue. Applicable only when retrieving matched DataWriter statuses.
For best-effort DataReaders, this is the sequence number of the latest DDS sample received.
For reliable DataReaders, this is the sequence number of the latest DDS sample that is available to be
read or taken from the DataReader's queue.

uncommitted_ Number of received DDS samples that are not yet available to be read or taken due to being received
sample_count out of order. Applicable only when retrieving matched DataWriter statuses.

7.3.7.4 LIVELINESS_CHANGED Status
This status indicates that the liveliness of one or more matched DataWriters has changed (i.e., one or more
DataWriters has become alive or not alive). The mechanics of determining liveliness between a
DataWriter and a DataReader is specified in their LIVELINESS QosPolicy (Section 6.5.13 on page
382).

475

7.3.7.5 REQUESTED_DEADLINE_MISSED Status

The structure for this status appears in Table 7.8 DDS_LivelinessChangedStatus.
Table 7.8 DDS_LivelinessChangedStatus
Type

Field Name

DDS_Long

DDS_
InstanceHandle_t

Description

alive_count

Number of matched DataWriters that are currently alive.

not_alive_count

Number of matched DataWriters that are not currently alive.

alive_count_change

The change in the alive_count since the last time the Listener was called or the status was
read.

not_alive_count_
change

The change in the not_alive_count since the last time the Listener was called or the status
was read.

last_publication_
handle

A handle to the last DataWriter to change its liveliness.

The DataReaderListener’s on_liveliness_changed() callback may be called for the following reasons:
l

Liveliness is truly lost—a DDS sample has not been received within the time-frame specified in the
LIVELINESS QosPolicy (Section 6.5.13 on page 382) lease_duration.

l

Liveliness is recovered after being lost.

l

A new matching entity has been discovered.

l

A QoS has changed such that a pair of matching entities are no longer matching (such as a change
to the PartitionQosPolicy). In this case, the middleware will no longer keep track of the entities’ liveliness. Furthermore:
l If liveliness was maintained: alive_count will decrease and not_alive_count will remain the
same.
l

If liveliness had been lost: alive_count will remain the same and not_alive_count will
decrease.

You can also retrieve the value by calling the DataReader’s get_liveliness_changed_status() operation;
this will also reset the status so it is no longer considered “changed.”
This status is reciprocal to the RELIABLE_READER_ACTIVITY_CHANGED Status (DDS Extension) (Section 6.3.6.9 on page 281) for a DataWriter.
7.3.7.5 REQUESTED_DEADLINE_MISSED Status
This status indicates that the DataReader did not receive a new DDS sample for an data-instance within
the time period set in the DataReader’s DEADLINE QosPolicy (Section 6.5.5 on page 363). For non-

476

7.3.7.6 REQUESTED_INCOMPATIBLE_QOS Status

keyed Topics, this simply means that the DataReader did not receive data within the DEADLINE period.
For keyed Topics, this means that for one of the data-instances that the DataReader was receiving, it has
not received a new DDS sample within the DEADLINE period. For more information about keys and
instances, see DDS Samples, Instances, and Keys (Section 2.3.1 on page 14).
The structure for this status appears in Table 7.9 DDS_RequestedDeadlineMissedStatus.
Table 7.9 DDS_RequestedDeadlineMissedStatus
Type

Field Name
total_count

DDS_Long

Description
Cumulative number of times that the deadline was violated for any instance read by the
DataReader.

total_count_change The change in total_count since the last time the Listener was called or the status was read.
DDS_
InstanceHandle_t

last_instance_
handle

Handle to the last data-instance in the DataReader for which a requested deadline was missed.

The DataReaderListener’s on_requested_deadline_missed() callback is invoked when this status
changes. You can also retrieve the value by calling the DataReader’s get_requested_deadline_missed_
status() operation; this will also reset the status so it is no longer considered “changed.”
7.3.7.6 REQUESTED_INCOMPATIBLE_QOS Status
A change to this status indicates that the DataReader discovered a DataWriter for the same Topic, but that
DataReader had requested QoS settings incompatible with this DataWriter’s offered QoS.
The structure for this status appears in Table 7.10 DDS_RequestedIncompatibleQosStatus .
Table 7.10 DDS_RequestedIncompatibleQosStatus
Type

Field
Name

Description

DDS_Long

total_
count

Cumulative number of times the DataReader discovered a DataWriter for the same Topic with an offered
QoS that is incompatible with that requested by the DataReader.

DDS_Long

total_
count_
change

The change in total_count since the last time the Listener was called or the status was read.

DDS_QosPolicyId_ last_
The ID of the QosPolicy that was found to be incompatible the last time an incompatibility was detected.
t
policy_id (Note: if there are multiple incompatible policies, only one of them is reported here.)
DDS_
policies
QosPolicyCountSeq

A list containing—for each policy—the total number of times that the DataReader discovered a
DataWriter for the same Topic with a offered QoS that is incompatible with that requested by the
DataReader.

477

7.3.7.7 SAMPLE_LOST Status

The DataReaderListener’s on_requested_incompatible_qos() callback is invoked when this status
changes. You can also retrieve the value by calling the DataReader’s get_requested_incompatible_qos_
status() operation; this will also reset the status so it is no longer considered “changed.”
7.3.7.7 SAMPLE_LOST Status
This status indicates that one or more DDS samples written by a matched DataWriter have failed to be
received.
For a DataReader, when there are insufficient resources to accept incoming DDS samples of data, DDS
samples may be dropped by the receiving application. Those DDS samples are considered to be
REJECTED (see SAMPLE_REJECTED Status (Section 7.3.7.8 on the facing page)). But DataWriters
are limited in the number of published DDS data samples that they can store, so that if a DataWriter continues to publish DDS data samples, new data may overwrite old data that have not yet been received by
the DataReader. The DDS samples that are overwritten can never be resent to the DataReader and thus
are considered to be lost.
This status applies to reliable and best-effort DataReaders, see the RELIABILITY QosPolicy (Section
6.5.19 on page 400).
The structure for this status appears in Table 7.11 DDS_SampleLostStatus.
Table 7.11 DDS_SampleLostStatus
Type

Field Name

Description

total_count

Cumulative count of all the DDS samples that have been lost, across all instances of data
written for the Topic.

total_count_
change

The incremental number of DDS samples lost since the last time the Listener was called or the
status was read.

last_reason

The reason the last DDS sample was lost. See Table 7.12 DDS_SampleLostStatusKind.

DDS_Long

DDS_
SampleLostStatusKind

The reason the DDS sample was lost appears in the last_reason field. The possible values are listed in
Table 7.12 DDS_SampleLostStatusKind.
Table 7.12 DDS_SampleLostStatusKind
Reason Kind

Description

NOT_LOST

The DDS sample was not lost.

LOST_BY_AVAILABILITY_
WAITING_TIME

AvailabilityQosPolicy’s max_data_availability_waiting_time expired.

478

7.3.7.8 SAMPLE_REJECTED Status

Table 7.12 DDS_SampleLostStatusKind
Reason Kind

Description

LOST_BY_INCOMPLETE_
COHERENT_SET

A DDS sample is lost because it is part of an incomplete coherent set.

LOST_BY_INSTANCES_
LIMIT

A resource limit on the number of instances was reached.

LOST_BY_LARGE_
COHERENT_SET

A DDS sample is lost because it is part of a large coherent set.

LOST_BY_REMOTE_
WRITER_SAMPLES_
PER_VIRTUAL_QUEUE_
LIMIT"

A resource limit on the number of DDS samples published by a remote writer on behalf of a virtual
writer that a DataReader may store was reached.

LOST_BY_REMOTE_
WRITERS_PER_
INSTANCE_LIMIT

A resource limit on the number of remote writers for a single instance from which a DataReader may
read was reached.

LOST_BY_REMOTE_
WRITERS_PER_
SAMPLE_LIMIT

A resource limit on the number of remote writers per DDS sample was reached.

LOST_BY_SAMPLES_PER_
REMOTE_
WRITER_LIMIT

A resource limit on the number of DDS samples from a given remote writer that a DataReader may
store was reached.

LOST_BY_VIRTUAL_
WRITERS_LIMIT

A resource limit on the number of virtual writers from which a DataReader may read was reached.

LOST_BY_WRITER

A DataWriter removed the DDS sample before being received by the DataReader.

The DataReaderListener’s on_sample_lost() callback is invoked when this status changes. You can also
retrieve the value by calling the DataReader’s get_sample_lost_status() operation; this will also reset the
status so it is no longer considered “changed.”
7.3.7.8 SAMPLE_REJECTED Status
This status indicates that one or more DDS samples received from a matched DataWriter have been
dropped by the DataReader because a resource limit would have been exceeded. For example, if the
receive queue is full, the number of DDS samples in the queue is equal to the max_samples parameter of
the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405).
The structure for this status appears in Table 7.13 DDS_SampleRejectedStatus. The reason the DDS
sample was rejected appears in the last_reason field. The possible values are listed in Table 7.14 DDS_
SampleRejectedStatusKind.

479

7.3.7.8 SAMPLE_REJECTED Status

Table 7.13 DDS_SampleRejectedStatus
Type

Field Name

Description

total_count

Cumulative count of all the DDS samples that have been rejected by the DataReader.

total_count_
change

The incremental number of DDS samples rejected since the last time the Listener was called
or the status was read.

current_count

The current number of writers with which the DataReader is matched.

current_count_
change

The change in current_count since the last time the Listener was called or the status was
read.

DDS_Long

DDS_
last_reason
SampleRejectedStatusKind
DDS_InstanceHandle_t

Reason for rejecting the last DDS sample. See Table 7.14 DDS_
SampleRejectedStatusKind.

last_instance_
handle

Handle to the data-instance for which the last DDS sample was rejected.

Table 7.14 DDS_SampleRejectedStatusKind
Reason Kind

Description

DDS_NOT_
REJECTED

DDS sample was accepted.

DDS_
REJECTED_
BY_
INSTANCES_
LIMIT

A resource limit on the number of instances that can be handled at
the same time by the DataReader was reached.

DDS_
REJECTED_
BY_
REMOTE_
WRITERS_
LIMIT

A resource limit on the number of DataWriters from which a
DataReader may read was reached.

DDS_
REJECTED_
BY_
REMOTE_
WRITERS_
PER_
INSTANCE_
LIMIT

Related QosPolicy

RESOURCE_LIMITS QosPolicy (Section 6.5.20
on page 405)

DATA_READER_RESOURCE_LIMITS
QosPolicy (DDS Extension) (Section 7.6.2 on
page 517)
A resource limit on the number of DataWriters for a single instance
from which a DataReader may read was reached.

480

7.3.7.8 SAMPLE_REJECTED Status

Table 7.14 DDS_SampleRejectedStatusKind
Reason Kind
DDS_
REJECTED_
BY_
SAMPLES_
LIMIT

Description

A resource limit on the total number of DDS samples was reached.

RESOURCE_LIMITS QosPolicy (Section 6.5.20
on page 405)

DDS_
REJECTED_
BY_
SAMPLES_
PER_
INSTANCE_
LIMIT

A resource limit on the number of DDS samples per instance was
reached.

DDS_
REJECTED_
BY_
SAMPLES_
PER_
REMOTE_
WRITER_
LIMIT

A resource limit on the number of DDS samples that a DataReader
may store from a specific DataWriter was reached.

DDS_
REJECTED_
BY_
VIRTUAL_
WRITERS_
LIMIT

A resource limit on the number of virtual writers from which a
DataReader may read was reached.

DDS_
REJECTED_
BY_
REMOTE_
WRITERS_
PER_
SAMPLE_
LIMIT

Related QosPolicy

DATA_READER_RESOURCE_LIMITS
QosPolicy (DDS Extension) (Section 7.6.2 on
page 517)
A resource limit on the number of remote writers per DDS sample
was reached.

DDS_
REJECTED_
BY_
A resource limit on the number of DDS samples published by a remote
REMOTE_
writer on behalf of a virtual writer that a DataReader may store was
WRITER_
reached.
SAMPLES_
PER_
VIRTUAL_
QUEUE_LIMIT

481

7.3.7.9 SUBSCRIPTION_MATCHED Status

The DataReaderListener’s on_sample_rejected() callback is invoked when this status changes. You can
also retrieve the value by calling the DataReader’s get_sample_rejected_status() operation; this will also
reset the status so it is no longer considered “changed.”
7.3.7.9 SUBSCRIPTION_MATCHED Status
A change to this status indicates that the DataReader discovered a matching DataWriter. A ‘match’ occurs
only if the DataReader and DataWriter have the same Topic, same data type (implied by having the same
Topic), and compatible QosPolicies. In addition, if user code has directed Connext DDS to ignore certain
DataWriters, then those DataWriters will never be matched. See Ignoring Publications and Subscriptions
(Section 16.4.2 on page 786) for more on setting up a DomainParticipant to ignore specific DataWriters.
The structure for this status appears in Table 7.15 DDS_SubscriptionMatchedStatus.
Table 7.15 DDS_SubscriptionMatchedStatus
Type

DDS_Long

DDS_InstanceHandle_t

Field Name

Description

total_count

Cumulative number of times the DataReader discovered a "match" with a DataWriter.

total_count_change

The change in total_count since the last time the Listener was called or the status was
read.

current_count

The number of DataWriters currently matched to the concerned DataReader.

current_count_change

The change in current_count since the last time the listener was called or the status was
read.

current_count_peak

The highest value that current_count has reached until now.

last_publication_
handle

Handle to the last DataWriter that matched the DataReader causing the status to change.

The DataReaderListener’s on_subscription_matched() callback is invoked when this status changes.
You can also retrieve the value by calling the DataReader’s get_subscription_match_status() operation;
this will also reset the status so it is no longer considered “changed.”

7.3.8 Setting DataReader QosPolicies
A DataReader’s QosPolicies control its behavior. Think of QosPolicies as the ‘properties’ for the
DataReader. The DDS_DataReaderQos structure has the following format:
struct DDS_DataReaderQos {
DDS_DurabilityQosPolicy
DDS_DeadlineQosPolicy
DDS_LatencyBudgetQosPolicy
DDS_LivelinessQosPolicy
DDS_ReliabilityQosPolicy

durability;
deadline;
latency_budget;
liveliness;
reliability;

482

7.3.8 Setting DataReader QosPolicies

DDS_DestinationOrderQosPolicy
destination_order;
DDS_HistoryQosPolicy
history;
DDS_ResourceLimitsQosPolicy
resource_limits;
DDS_UserDataQosPolicy
user_data;
DDS_TimeBasedFilterQosPolicy
time_based_filter;
DDS_ReaderDataLifecycleQosPolicy
reader_data_lifecycle;
DDS_TransportPriorityQosPolicy
transport_priority;
DDS_TypeConsistencyEnforcementQosPolicy type_consistency;
// Extensions to the DDS standard:
DDS_DataReaderResourceLimitsQosPolicy reader_resource_limits;
DDS_DataReaderProtocolQosPolicy
protocol;
DDS_TransportSelectionQosPolicy
transport_selection;
DDS_TransportUnicastQosPolicy
unicast;
DDS_TransportMulticastQosPolicy
multicast;
DDS_PropertyQosPolicy
property;
DDS_ServiceQosPolicy
service;
DDS_AvailabilityQosPolicy
availability;
DDS_EntityNameQosPolicy
subscription_name;
DDS_TypeSupportQosPolicy
type_support;
};

Note: set_qos() cannot always be used within a listener callback, see Restricted Operations in Listener
Callbacks (Section 4.5.1 on page 185).
Table 7.16 DataReader QosPolicies summarizes the meaning of each policy. (They appear alphabetically
in the table.) For information on why you would want to change a particular QosPolicy, see the referenced
section. For defaults and valid ranges, please refer to the API Reference HTML documentation.
Table 7.16 DataReader QosPolicies
QosPolicy

Description
This QoS policy is used in the context of two features:

Availability

For a Collaborative DataWriter, specifies the group of DataWriters expected to collaboratively provide data
and the timeouts that control when to allow data to be available that may skip DDS samples.
For a Durable Subscription, configures a set of Durable Subscriptions on a DataWriter.
See AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on page 337)

DataReaderProtocol

This QosPolicy configures the DDS on-the-network protocol, RTPS. See DATA_READER_PROTOCOL
QosPolicy (DDS Extension) (Section 7.6.1 on page 511).

DataReaderResourceLimits

Various settings that configure how DataReaders allocate and use physical memory for internal resources.
See DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 7.6.2 on page 517).
For a DataReader, specifies the maximum expected elapsed time between arriving DDS data samples.

Deadline

For a DataWriter, specifies a commitment to publish DDS samples with no greater elapsed time between
them.
See DEADLINE QosPolicy (Section 6.5.5 on page 363).

483

7.3.8 Setting DataReader QosPolicies

Table 7.16 DataReader QosPolicies
QosPolicy

Description

DestinationOrder

Controls how Connext DDS will deal with data sent by multiple DataWriters for the same topic. Can be
set to "by reception timestamp" or to "by source timestamp". See DESTINATION_ORDER QosPolicy
(Section 6.5.6 on page 365).

Durability

Specifies whether or not Connext DDS will store and deliver data that were previously published to new
DataReaders. See DURABILITY QosPolicy (Section 6.5.7 on page 368).

EntityName

Assigns a name to a DataReader. See ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9 on
page 374).

History

Specifies how much data must to stored by Connext DDSfor the DataWriter or DataReader. This
QosPolicy affects the RELIABILITY QosPolicy (Section 6.5.19 on page 400) as well as the
DURABILITY QosPolicy (Section 6.5.7 on page 368). See HISTORY QosPolicy (Section 6.5.10 on
page 376).

LatencyBudget

Suggestion to Connext DDS on how much time is allowed to deliver data. See LATENCYBUDGET QoS
Policy (Section 6.5.11 on page 380).

Liveliness

Specifies and configures the mechanism that allows DataReaders to detect when DataWriters become
disconnected or "dead." See LIVELINESS QosPolicy (Section 6.5.13 on page 382).

Property

Stores name/value (string) pairs that can be used to configure certain parameters of Connext DDS that are
not exposed through formal QoS policies. It can also be used to store and propagate application-specific
name/value pairs, which can be retrieved by user code during discovery. See PROPERTY QosPolicy
(DDS Extension) (Section 6.5.17 on page 394).

ReaderDataLifeCycle

Controls how a DataReader manages the lifecycle of the data that it has received. See READER_DATA_
LIFECYCLE QoS Policy (Section 7.6.3 on page 523).

Reliability

Specifies whether or not Connext DDS will deliver data reliably. See RELIABILITY QosPolicy (Section
6.5.19 on page 400).

ResourceLimits

Controls the amount of physical memory allocated for entities, if dynamic allocations are allowed, and how
they occur. Also controls memory usage among different instance values for keyed topics. See
RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405).

Service

Intended for use by RTI infrastructure services. User applications should not modify its value. See
SERVICE QosPolicy (DDS Extension) (Section 6.5.21 on page 408).

TimeBasedFilter

Set by a DataReader to limit the number of new data values received over a period of time. See TIME_
BASED_FILTER QosPolicy (Section 7.6.4 on page 526).

TransportMulticast

Specifies the multicast address on which a DataReader wants to receive its data. Can specify a port number
as well as a subset of the available transports with which to receive the multicast data. See TRANSPORT_
MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on page 529).

484

7.3.8.1 Configuring QoS Settings when the DataReader is Created

Table 7.16 DataReader QosPolicies
QosPolicy

Description

TransportPriority

Set on a DataReader to tell Connext DDS that the data being sent has a different "priority" than other data.
For DataReaders, the data being sent refers to ACKNACK messages. See TRANSPORT_PRIORITY
QosPolicy (Section 6.5.22 on page 409).

TransportSelection

Allows you to select which physical transports a DataWriter or DataReader may use to send or receive its
data. See TRANSPORT_SELECTION QosPolicy (DDS Extension) (Section 6.5.23 on page 411).

TransportUnicast

Specifies a subset of transports and port number that can be used by an Entity to receive data. See
TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412).

TypeSupport

Used to attach application-specific value(s) to a DataWriter or DataReader. These values are passed to the
serialization or deserialization routine of the associated data type. See TYPESUPPORT QosPolicy (DDS
Extension) (Section 6.5.25 on page 416).

Defines rules that determine whether the type used to publish a given data stream is consistent with that
TypeConsistencyEnforcement used to subscribe to it. See TYPE_CONSISTENCY_ENFORCEMENT QosPolicy (Section 7.6.6 on page
532).

UserData

Along with Topic Data QosPolicy and Group Data QosPolicy, used to attach a buffer of bytes to Connext
DDS's discovery meta-data. See USER_DATA QosPolicy (Section 6.5.26 on page 417).

For a DataReader to communicate with a DataWriter, their corresponding QosPolicies must be compatible. For QosPolicies that apply both to the DataWriter and the DataReader, the setting in the
DataWriter is considered what the DataWriter “offers” and the setting in the DataReader is what the
DataReader “requests.” Compatibility means that what is offered by the DataWriter equals or surpasses
what is requested by the DataReader. See QoS Requested vs. Offered Compatibility—the RxO Property
(Section 4.2.1 on page 167).
Some of the policies may be changed after the DataReader has been created. This allows the application
to modify the behavior of the DataReader while it is in use. To modify the QoS of an existing
DataReader, use the get_qos() and set_qos() operations on the DataReader. This is a general pattern for
all Entities, described in more detail in Changing the QoS for an Existing Entity (Section 4.1.7.3 on page
161).
7.3.8.1 Configuring QoS Settings when the DataReader is Created
As described in Creating DataReaders (Section 7.3.1 on page 463), there are different ways to create a
DataReader, depending on how you want to specify its QoS (with or without a QoS Profile).
In Creating a DataReader with Default QosPolicies (Section Figure 7.9 on page 465), we saw an example
of how to create a DataReader with default QosPolicies by using the special constant, DDS_
DATAREADER_QOS_DEFAULT, which indicates that the default QoS values for a DataReader
should be used. The default DataReader QoS values are configured in the Publisher or

485

7.3.8.1 Configuring QoS Settings when the DataReader is Created

DomainParticipant; you can change them with set_default_datareader_qos() or set_default_
datareader_qos_with_profile(). Then any DataReaders created with the Subscriber will use the new
default values. As described in Getting, Setting, and Comparing QosPolicies (Section 4.1.7 on page 158),
this is a general pattern that applies to the construction of all Entities.
To create a DataReader with non-default QoS without using a QoS Profile, see the example code in Figure 7.11 Creating a DataReader with Modified QosPolicies (not from a profile) below. It uses the Publisher’s get_default_reader_qos() method to initialize a DDS_DataReaderQos structure. Then, the
policies are modified from their default values before the structure is used in the create_datareader()
method.
You can also create a DataReader and specify its QoS settings via a QoS Profile. To do so, you will call
create_datareader_with_profile(), as seen in Figure 7.12 Creating a DataReader with a QoS Profile on
the facing page.
If you want to use a QoS profile, but then make some changes to the QoS before creating the DataReader,
call get_datareader_qos_from_profile() and create_datareader() as seen in Figure 7.13 Getting QoS
Values from a Profile, Changing QoS Values, Creating a DataReader with Modified QoS Values on the
facing page.
For more information, see Creating DataReaders (Section 7.3.1 on page 463) and Configuring QoS with
XML (Section Chapter 17 on page 791).
Figure 7.11 Creating a DataReader with Modified QosPolicies (not from a profile)
DDS_DataReaderQos reader_qos;1
// initialize reader_qos with default values
subscriber->get_default_datareader_qos(reader_qos);
// make QoS changes here
reader_qos.history.depth = 5;
// Create the reader with modified qos
DDSDataReader * reader = subscriber->create_datareader(
topic, reader_qos, NULL, DDS_STATUS_MASK_NONE);
if (reader == NULL) {
// ... error
}
// narrow it for your specific data type
FooDataReader* foo_reader = FooDataReader::narrow(reader);

1Note: In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
486

7.3.8.2 Comparing QoS Values

Figure 7.12 Creating a DataReader with a QoS Profile

// Create the datareader
DDSDataReader * reader =
subscriber->create_datareader_with_profile(
topic, “MyReaderLibrary”, “MyReaderProfile”,
NULL, DDS_STATUS_MASK_NONE);
if (reader == NULL) {
// ... error
};
// narrow it for your specific data type
FooDataReader* foo_reader = FooDataReader::narrow(reader);

Figure 7.13 Getting QoS Values from a Profile, Changing QoS Values, Creating a
DataReader with Modified QoS Values
DDS_DataReaderQos reader_qos;1
// Get reader QoS from profile
retcode = factory->get_datareader_qos_from_profile(reader_qos,
“ReaderProfileLibrary”, “ReaderProfile”);
if (retcode != DDS_RETCODE_OK) {
// handle error
}
// Makes QoS changes here
reader_qos.history.depth = 5;
DDSDataReader * reader = subscriber->create_datareader(topic, reader_qos,
NULL, DDS_STATUS_MASK_NONE);
if (reader == NULL) {
// handle error
}

7.3.8.2 Comparing QoS Values
The equals() operation compares two DataReader’s DDS_DataWriterQoS structures for equality. It takes
two parameters for the two DataReader’s QoS structures to be compared, then returns TRUE is they are
equal (all values are the same) or FALSE if they are not equal.
7.3.8.3 Changing QoS Settings After DataReader Has Been Created
There are 2 ways to change an existing DataReader’s QoS after it is has been created—again depending
on whether or not you are using a QoS Profile.

1Note: In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
487

7.3.8.4 Using a Topic’s QoS to Initialize a DataWriter’s QoS

l

l

To change QoS programmatically (that is, without using a QoS Profile), use get_qos() and set_qos
(). See the example code in Figure 7.14 Changing the QoS of an Existing DataReader (without a
QoS Profile) below. It retrieves the current values by calling the DataReader’s get_qos() operation.
Then it modifies the value and calls set_qos() to apply the new value. Note, however, that some
QosPolicies cannot be changed after the DataReader has been enabled—this restriction is noted in
the descriptions of the individual QosPolicies.
You can also change a DataReader’s (and all other Entities’) QoS by using a QoS Profile and calling set_qos_with_profile(). For an example, see Figure 7.15 Changing the QoS of an Existing
DataReader with a QoS Profile below. For more information, see Configuring QoS with XML (Section Chapter 17 on page 791).

Figure 7.14 Changing the QoS of an Existing DataReader (without a QoS Profile)

// Get current QoS
if (datareader->get_qos(reader_qos) != DDS_RETCODE_OK) {
// handle error
}
// Makes QoS changes here
reader_qos.history.depth = 5;
// Set the new QoS
if (datareader->set_qos(reader_qos) != DDS_RETCODE_OK ) {
// handle error
}

Figure 7.15 Changing the QoS of an Existing DataReader with a QoS Profile
retcode = datareader->set_qos_with_profile(
“ReaderProfileLibrary”,”ReaderProfile”);
if (retcode != DDS_RETCODE_OK) {
// handle error
}

7.3.8.4 Using a Topic’s QoS to Initialize a DataWriter’s QoS
Several DataReader QosPolicies can also be found in the QosPolicies for Topics (see Setting Topic
QosPolicies (Section 5.1.3 on page 204)). The QosPolicies set in the Topic do not directly affect the
DataReaders (or DataWriters) that use that Topic. In many ways, some QosPolicies are a Topic-level
concept, even though the DDS standard allows you to set different values for those policies for different
DataReaders and DataWriters of the same Topic. Thus, the policies in the DDS_TopicQos structure exist
as a way to help centralize and annotate the intended or suggested values of those QoSs. Connext DDS
does not check to see if the actual policies set for a DataReader is aligned with those set in the Topic to
which it is bound.

488

7.3.9 Navigating Relationships Among Entities

There are many ways to use the QosPolicies’ values set in the Topic when setting the QosPolicies’ values
in a DataReader. The most straight forward way is to get the values of policies directly from the Topic and
use them in the policies for the DataReader. Figure 6.21 Copying Selected QoS from a Topic when Creating a DataWriter on page 307 shows an example of how to this for a DataWriter; the pattern applies to
DataReaders as well.
The Subscriber’s copy_from_topic_qos() operation can be used to copy all the common policies from the
Topic QoS to a DataReaderQoS, as illustrated in Figure 6.22 Copying all QoS from a Topic when Creating a DataWriter on page 308 for DataWriters.
The special macro, DDS_DATAREADER_QOS_USE_TOPIC_QOS, can be used to indicate that the
DataReader should be created with the QoS that results from modifying the default DataReader QoS with
the values specified by the Topic. See Figure 6.23 Combining Default Topic and DataWriter QoS (Option
1) on page 309 and Figure 6.24 Combining Default Topic and DataWriter QoS (Option 2) on page 309
for examples involving DataWriters. The same pattern applies to DataReaders. For more information on
the use and manipulation of QoS, see Getting, Setting, and Comparing QosPolicies (Section 4.1.7 on
page 158).

7.3.9 Navigating Relationships Among Entities
7.3.9.1 Finding Matching Publications
The following DataReader operations can be used to get information about the DataWriters that will send
data to this DataReader.
l

get_matched_publications()

l

get_matched_publication_data()

The get_matched_publications() operation will return a sequence of handles to matched DataWriters.
You can use these handles in the get_matched_publication_data() method to get information about the
DataWriter such as the values of its QosPolicies.
Note that DataWriter that have been ignored using the DomainParticipant’s ignore_publication() operation are not considered to be matched even if the DataWriter has the same Topic and compatible
QosPolicies. Thus, they will not be included in the list of DataWriters returned by get_matched_publications(). See Ignoring Publications and Subscriptions (Section 16.4.2 on page 786) for more on
ignore_publication().
You can also get the DATA_READER PROTOCOL_STATUS for matching publications with get_
matched_publication_datareader_protocol_status() (see DATA_READER_PROTOCOL_STATUS
(Section 7.3.7.3 on page 472)).

489

7.3.9.2 Finding the Matching Publication’s ParticipantBuiltinTopicData

Note:
l

Status/data for a matched publication is only kept while the matched publication is alive. Once a
matched publication is no longer alive, its status is deleted. If you try to get the status/data for a
matched publication that is no longer alive, the 'get data' or 'get status' call will return an error.

See also: Finding the Matching Publication’s ParticipantBuiltinTopicData (Section 7.3.9.2 below)
7.3.9.2 Finding the Matching Publication’s ParticipantBuiltinTopicData
get_matched_publication_participant_data() allows you to get the DDS_ParticipantBuiltinTopicData
(see Table 16.1 Participant Built-in Topic’s Data Type (DDS_ParticipantBuiltinTopicData)) of a matched
publication using a publication handle.
This operation retrieves the information on a discovered DomainParticipant associated with the publication that is currently matching with the DataReader.
The publication handle passed into this operation must correspond to a publication currently associated
with the DataReader. Otherwise, the operation will fail with RETCODE_BAD_PARAMETER. The
operation may also fail with RETCODE_PRECONDITION_NOT_MET if the publication handle corresponds to the same DomainParticipant to which the DataReader belongs.
Use get_matched_publications() (see Finding Matching Publications (Section 7.3.9.1 on the previous
page)) to find the publications that are currently matched with the DataReader.
Note: This operation does not retrieve the ParticipantBuiltinTopicData_property. This information is available through the on_data_available() callback (if a DataReaderListener is installed on the PublicationBuiltinTopicDataDataReader.
7.3.9.3 Finding a DataReader’s Related Entities
These DataReader operations are useful for obtaining a handle to various related entities:
l

get_subscriber()

l

get_topicdescription()

The get_subscriber() operation returns the Subscriber that created the DataReader. get_topicdescription
() returns the Topic with which the DataReader is associated.
7.3.9.4 Looking Up an Instance Handle
Some operations, such as read_instance() and take_instance(), take an instance_handle parameter. If
you need to get such as handle, you can call the lookup_instance() operation, which takes an instance as a
parameter and returns a handle to that instance.
490

7.3.9.5 Getting the Key Value for an Instance

7.3.9.5 Getting the Key Value for an Instance
If you have a handle to a data-instance, you can use the FooDataReader’s get_key_value() operation to
retrieve the key for that instance. The value of the key is decomposed into its constituent fields and
returned in a Foo structure. For information on keys and keyed data types, please see DDS Samples,
Instances, and Keys (Section 2.3.1 on page 14).

7.4 Using DataReaders to Access Data (Read & Take)
For user applications to access the data received for a DataReader, they must use the type-specific derived
class or set of functions in the C API. Thus for a user data type ‘Foo’, you must use methods of the
FooDataReader class. The type-specific class or functions are automatically generated if you use RTI
Code Generator. Else, you will have to create them yourself, see Type Codes for Built-in Types (Section
3.8.4.1 on page 143) for more details.

7.4.1 Using a Type-Specific DataReader (FooDataReader)
This section doesn't apply to the Modern C++ API, where a DataReader's data type is part of its
template definition: DataReader.
Using a Subscriber you will create a DataReader associating it with a specific data type, for example
‘Foo’. Note that the Subscriber’s create_datareader() method returns a generic DataReader. When your
code is ready to access DDS data samples received for the DataReader, you must use type-specific operations associated with the FooDataReader, such as read() and take().
To cast the generic DataReader returned by create_datareader() into an object of type FooDataReader,
you should use the type-safe narrow() method of the FooDataReader class. narrow() will make sure
that the generic DataReader passed to it is indeed an object of the FooDataReader class before it makes
the cast. Else, it will return NULL. Simple SubscriberListener (Section Figure 7.8 on page 456) shows an
example:
Foo_reader = FooDataReader::narrow(reader);

Table 7.3 DataReader Operations lists type-specific operations using a FooDataReader. Also listed are
generic, non-type specific operations that can be performed using the base class object DDSDataReader
(or DDS_DataReader in C). In C, you must pass a pointer to a DDS_DataReader to those generic functions.

491

7.4.2 Loaning and Returning Data and SampleInfo Sequences

7.4.2 Loaning and Returning Data and SampleInfo Sequences
7.4.2.1 C, Traditional C++, Java and .NET
The read() and take() operations (and their variations) return information to your application in two
sequences:
l Received DDS data samples in a sequence of the data type
l

Corresponding information about each DDS sample in a SampleInfo sequence

These sequences are parameters that are passed by your code into the read() and take() operations.
If you use empty sequences (sequences that are initialized but have a maximum length of 0), Connext DDS will fill those sequences with memory directly loaned from the receive queue itself. There
is no copying of the data or of SampleInfo when the contents of the sequences are loaned. This is
certainly the most efficient way for your code to retrieve the data.
However when you do so, your code must return the loaned sequences back to Connext DDS so
that they can be reused by the receive queue. If your code does not return the loan by calling the
FooDataReader’s return_loan() method, then Connext DDS will eventually run out of memory to
store DDS data samples received from the network for that DataReader. See Using Loaned
Sequences in read() and take() (Section Figure 7.16 below) for an example of borrowing and returning loaned sequences.
DDS_ReturnCode_t return_loan(
FooSeq &received_data, DDS_SampleInfoSeq &info_seq);

Figure 7.16 Using Loaned Sequences in read() and take()
// In C++ and Java, sequences are automatically initialized
// to be empty
FooSeq data_seq;1
DDS_SampleInfoSeq info_seq;
DDS_ReturnCode_t retcode;
...
// with empty sequences, a take() or read() will return loaned
// sequence elements
retcode = Foo_reader->take(data_seq, info_seq,
DDS_LENGTH_UNLIMITED, DDS_ANY_SAMPLE_STATE,
DDS_ANY_VIEW_STATE, DDS_ANY_INSTANCE_STATE);

1For the C API, you must use the FooSeq_initialize() and DDS_SampleInfoSeq_initialize() operations or

the macro DDS_SEQUENCE_INITIALIZER to initialize the FooSeq and DDS_SampleInfoSeq to be
empty. For example, DDS_SampleInfoSeq infoSeq; DDS_SampleInfoSeq_initialize(&infoSeq); or
FooSeq fooSeq = DDS_SEQUENCE_INITIALIZER;
492

7.4.2.2 Modern C++

... // process the returned data
// must return the loaned sequences when done processing
Foo_reader->return_loan(data_seq, info_seq);
...

If your code provides its own sequences to the read/take operations, then Connext DDS will copy
the data from the receive queue. In that case, you do not have to call return_loan() when you are
finished with the data. However, you must make sure the following is true, or the read/take operation will fail with a return code of DDS_RETCODE_PRECONDITION_NOT_MET:
l The received_data of type FooSeq and info_seq of type DDS_SampleInfoSeq passed in as
parameters have the same maximum size (length).
l

The maximum size (length) of the sequences are less than or equal to the passed in parameter,
max_samples.

7.4.2.2 Modern C++
The read() and take() operations (and their variations) return LoanedSamples, an iterable collection
of loaned, read-only samples each containing the actual data and meta-information about the sample.
A LoanedSamples collection automatically returns the loan to the middleware in its destructor. You
can also explicitly call LoanedSamples::return_loan().
Figure 7.17 Using LoanedSamples to read data
dds::sub::LoanedSamples samples = reader.take();
for (auto sample : samples) { // process the data
if (sample.info().valid()) {
std::cout << sample.data() << std::endl;
}
}

7.4.3 Accessing DDS Data Samples with Read or Take
To access the DDS data samples that Connext DDS has received for a DataReader, you must invoke the
read() or take() methods. These methods return a list (sequence) of DDS data samples and additional
information about the DDS samples in a corresponding list (sequence) of SampleInfo structures. The contents of SampleInfo are described in The SampleInfo Structure (Section 7.4.6 on page 504).
Calling read(), take(), or one of their variations resets the DATA_AVAILABLE status.
The way Connext DDS builds the collection of DDS samples depends on QoS policies set on the
DataReader and Subscriber, the source_timestamp of the DDS samples, and the sample_states, view_
states, and instance_states parameters passed to the read/take operation.

493

7.4.3.1 Read vs. Take

In read() and take(), you may enter parameters so that Connext DDS selectively returns DDS data
samples currently stored in the DataReader’s receive queue. You may want Connext DDS to return all of
the data in a single list or only a subset of the available DDS samples as configured using the sample_
states, view_states, and instance_states masks. The SampleInfo Structure (Section 7.4.6 on page 504)
describes how these masks are used to determine which DDS data samples should be returned.
7.4.3.1 Read vs. Take
The difference between read() and take() is how Connext DDS treats the data that is returned. With take
(), Connext DDS will remove the data from the DataReader’s receive queue. The data returned by Connext DDS is no longer stored by Connext DDS. With read(), Connext DDS will continue to store the data
in the DataReader’s receive queue. The same data may be read again until it is taken in subsequent take()
calls. Note that the data stored in the DataReader’s receive queue may be overwritten, even if it has not
been read, depending on the setting of the HISTORY QosPolicy (Section 6.5.10 on page 376).
The read() and take() operations are non-blocking calls, so that they may return no data (DDS_
RETCODE_NO_DATA) if the receive queue is empty or has no data that matches the criteria specified
by the StateMasks.
The read_w_condition() and take_w_condition() operations take a ReadCondition as a parameter
instead of DDS sample, view or instance states. The only DDS samples returned will be those for which
the ReadCondition is TRUE. These operations, in conjunction with ReadConditions and a WaitSet, allow
you to perform ‘waiting reads.’ For more information, see ReadConditions and QueryConditions (Section
4.6.7 on page 195).
As you will see, read and take have the same parameters:
DDS_ReturnCode_t read( FooSeq &received_data_seq,
DDS_SampleInfoSeq &info_seq,
DDS_Long max_samples,
DDS_SampleStateMask sample_states,
DDS_ViewStateMask view_states,
DDS_InstanceStateMask instance_states);
DDS_ReturnCode_t take( FooSeq &received_data_seq,
DDS_SampleInfoSeq &info_seq,
DDS_Long max_samples,
DDS_SampleStateMask sample_states,
DDS_ViewStateMask view_states,
DDS_InstanceStateMask instance_states);

Note: These operations may loan internal Connext DDS memory, which must be returned with return_
loan(). See Loaning and Returning Data and SampleInfo Sequences (Section 7.4.2 on page 492).

494

7.4.3.1 Read vs. Take

Both operations return an ordered collection of DDS data samples (in the received_data_seq parameter)
and information about each DDS sample (in the info_seq parameter). Exactly how they are ordered
depends on the setting of the PRESENTATION QosPolicy (Section 6.4.6 on page 330) and the
DESTINATION_ORDER QosPolicy (Section 6.5.6 on page 365). For more details please see the API
Reference HTML documentation for read() and take().
In read() and take(), you can use the sample_states, view_states, and instance_states parameters to specify properties that are used to select the actual DDS samples that are returned by those methods. With different combinations of these three parameters, you can direct Connext DDS to return all DDS data
samples, DDS data samples that you have not accessed before, the DDS data samples of instances that you
have not seen before, DDS data samples of instances that have been disposed, etc. The possible values for
the different states are described both in the API Reference HTML documentation and in The SampleInfo
Structure (Section 7.4.6 on page 504).
Table 7.17 Read and Take Operations lists the variations of the read() and take() operations.
Table 7.17 Read and Take Operations
Read
Take
Operations Operations

Modern
C++1

Description

Reference

reader.read()
or
read

take

reader.select()

Reads/takes a collection of DDS data samples from the
DataReader.

.state(...)

Can be used for both keyed and non-keyed data types.

Accessing DDS Data
Samples with Read or
Take (Section 7.4.3 on
page 493)

.read()
reader.select()
read_instance

take_instance

.instance(...)
.read()
reader.select()

read_
instance_
w_condition

take_
instance_
w_condition

.instance()
.condition(...)
.read()

read_next_
instance

take_next_
instance

Identical to read() and take(), but all returned DDS samples read_instance and take_
belong to a single instance, which you specify as a parameter. instance (Section 7.4.3.4
on page 497)
Can only be used with keyed data types.
read_instance_w_
Identical to read_instance() and take_instance(), but all
condition and take_
returned DDS samples belong to the single specified instance instance_w_condition
and satisfy the specified ReadCondition.
(Section 7.4.3.7 on page
500)

Similar to read_instance() and take_instance(), but the
reader.select
actual instance is not directly specified as a parameter.
().next_instance Instead, the DDS samples will all belong to instance ordered
(...).read()
after the instance that is specified by the previous_handle
parameter.

read_next_instance and
take_next_instance
(Section 7.4.3.5 on page
498)

1For the Modern C++, only the read() operation is shown; the take() variant is parallel.

495

7.4.3.2 General Patterns for Accessing Data

Table 7.17 Read and Take Operations
Read
Take
Operations Operations

Modern
C++1

Description

Reference

reader.select()
read_next_
instance_
w_condition

take_next_
instance_
w_condition

.next_instance
(...)
.condition(...)

read_next_instance_w_
Accesses a collection of DDS data samples of the next
condition and take_next_
instance that match a specific set of ReadConditions, from the instance_w_condition
DataReader.
(Section 7.4.3.8 on page
501)

.read()
reader.select()
read_next_
sample

take_next_
sample

read_w_
condition

take_w_
condition

read_next_sample and
Provides a convenient way to access the next DDS DDS
take_next_sample
.state
(DataState::not_ sample in the receive queue that has not been accessed before. (Section 7.4.3.3 on the
facing page)
read())

reader.select()
.condition(...)

Accesses a collection of DDS data samples from the
DataReader that match specific ReadCondition criteria.

read_w_condition and
take_w_condition
(Section 7.4.3.6 on page
500)

7.4.3.2 General Patterns for Accessing Data
Once the DDS data samples are available to the data readers, the DDS samples can be read or taken by the
application. The basic rule is that the application may do this in any order it wishes. This approach is very
flexible and allows the application ultimate control.
To access data coherently, or in order, the PRESENTATION QosPolicy (Section 6.4.6 on page 330)
must be set properly.
Accessing DDS samples If No Order or Coherence Is Required
Simply access the data by calling read/take on each DataReader in any order you want.
You do not have to call begin_access() and end_access(). However, doing so is not an error and it will
have no effect.
You can call the Subscriber’sget_datareaders() operation to see which DataReaders have data to be
read, but you do not need to read all of them or read them in a particular order. The get_datareaders()
operation will return a logical 'set' in the sense that the same DataReader will not appear twice. The order
of the DataReaders returned is not specified.
Accessing DDS samples within a SubscriberListener

1For the Modern C++, only the read() operation is shown; the take() variant is parallel.
496

7.4.3.3 read_next_sample and take_next_sample

This case describes how to access the data inside the listener's on_data_on_readers() operation (regardless of the PRESENTATION QoS policy settings).
To do so, you can call read/take on each DataReader in any order. You can also delegate accessing of the
data to the DataReaderListeners by calling the Subscriber’snotify_datareaders() operation.
Similar to the previous case, you can still call the Subscriber’s get_datareaders() operation to determine
which DataReaders have data to be read, but you do not have to read all of them, or read them in a particular order. get_datareaders() will return a logical 'set.'
You do not have to call begin_access() and end_access(). However, doing so is not an error and it will
have no effect.
7.4.3.3 read_next_sample and take_next_sample
The read_next_sample() or take_next_sample() operation is used to retrieve the next DDS sample that
hasn’t already been accessed. It is a simple way to 'read' DDS samples and frees your application from
managing sequences and specifying DDS sample, instance or view states. It behaves the same as calling
read() or take() with max_samples = 1, sample_states = NOT_READ, view_states = ANY_VIEW_
STATE, and instance_states = ANY_INSTANCE_STATE.
DDS_ReturnCode_t read_next_sample(
Foo & received_data, DDS_SampleInfo & sample_info);
DDS_ReturnCode_t take_next_sample(
Foo & received_data, DDS_SampleInfo & sample_info);

It copies the next, not-previously-accessed data value from the DataReader. It also copies the DDS
sample’s corresponding DDS_SampleInfo structure.
If there is no unread data in the DataReader, the operation will return DDS_RETCODE_NO_DATA
and nothing is copied.
Since this operation copies both the DDS data sample and the SampleInfo into user-provided storage, it
does not allocate nor loan memory. You do not have to call return_loan() after this operation.
Note: If the received_data parameter references a structure that contains a sequence and that sequence
has not been initialized, the operation will return DDS_RETCODE_ERROR.
7.4.3.4 read_instance and take_instance
The read_instance() and take_instance() operations are identical to read() and take(), but they are used
to access DDS samples for just a specific instance (key value). The parameters are the same, except you
must also supply an instance handle. These functions can only be used when the DataReader is tied to a

497

7.4.3.5 read_next_instance and take_next_instance

keyed type, see DDS Samples, Instances, and Keys (Section 2.3.1 on page 14) for more about keyed data
types.
These operations may return BAD_PARAMETER if the instance handle does not correspond to an existing data-object known to the DataReader.
The handle to a particular data instance could have been cached from a previous read() operation (value
taken from the SampleInfo struct) or created by using the DataReader’s lookup_instance() operation.
DDS_ReturnCode_t read_instance(
FooSeq &received_data,
DDS_SampleInfoSeq &info_seq,
DDS_Long max_samples,
const DDS_InstanceHandle_t &a_handle,
DDS_SampleStateMask sample_states,
DDS_ViewStateMask view_states,
DDS_InstanceStateMask instance_states);

Note: This operation may loan internal Connext DDS memory, which must be returned with return_loan
(). See Loaning and Returning Data and SampleInfo Sequences (Section 7.4.2 on page 492).
7.4.3.5 read_next_instance and take_next_instance
The read_next_instance() and take_next_instance() operations are similar to read_instance() and take_
instance() in that they return DDS samples for a specific data instance (key value). The difference is that
instead of passing the handle of the data instance for which you want DDS data samples, instead you pass
the handle to a ‘previous’ instance. The returned DDS samples will all belong to the 'next' instance, where
the ordering of instances is explained below.
DDS_ReturnCode_t read_next_instance(
FooSeq &received_data,
DDS_Long max_samples,
const DDS_InstanceHandle_t &previous_handle
DDS_SampleStateMask sample_states,
DDS_ViewStateMask view_states,
DDS_InstanceStateMask instance_states)

Connext DDS orders all instances relative to each other.1 This ordering depends on the value of the key as
defined for the data type associated with the Topic. For the purposes of this discussion, it is 'as if' each

aThe ordering of the instances is specific to each implementation of the DDS standard; to maximize the

portability of your code, do not assume any particular order. In the case of Connext DDS (and likely other
DDS implementations as well), the order is not likely to be meaningful to you as a developer; it is simply
important that some ordering exists.
498

7.4.3.5 read_next_instance and take_next_instance

instance handle is represented by a unique integer and thus different instance handles can be ordered by
their value.
This operation will return values for the next instance handle that has DDS data samples stored in the
receive queue (that meet the criteria specified by the StateMasks). The next instance handle will be
ordered after the previous_handle that is passed in as a parameter.
The special value DDS_HANDLE_NIL can be passed in as the previous_handle. Doing so, you will
receive values for the “smallest” instance handle that has DDS data samples stored in the receive queue
that you have not yet accessed.
You can call the read_next_instance() operation with a previous_handle that does not correspond to an
instance currently managed by the DataReader. For example, you could use this approach to iterate
though all the instances, take all the DDS samples with a NOT_ALIVE_NO_WRITERS instance_state,
return the loans (at which point the instance information may be removed, and thus the handle becomes
invalid), and then try to read the next instance.
The example below shows how to use take_next_instance() iteratively to process all the data received for
an instance, one instance at a time. We always pass in DDS_HANDLE_NIL as the value of previous_
handle. Each time through the loop, we will receive DDS samples for a different instance, since the previous time through the loop, all of the DDS samples of the previous instance were returned (and thus
accessed).
FooSeq received_data;1
DDS_SampleInfoSeq info_seq;
while (retcode = reader->take_next_instance(received_data, info_seq,
DDS_LENGTH_UNLIMITED, DDS_HANDLE_NIL,
DDS_ANY_SAMPLE_STATE, DDS_ANY_VIEW_STATE,
DDS_ANY_INSTANCE_STATE)
!= DDS_RETCODE_NO_DATA) {
// the data samples returned in received_data will all
// be for a single instance
// process the data
// now return the loaned sequences
if (reader->return_loan(received_data, info_seq)
!= DDS_RETCODE_OK) {
// handle error
}
}

1In the C API, you must use the FooSeq_initialize() and DDS_SampleInfoSeq_initialize() operations or

the macro DDS_SEQUENCE_INITIALIZER to initialize the FooSeq and DDS_SampleInfoSeq to be
empty. For example, DDS_SampleInfoSeq infoSeq; DDS_SampleInfoSeq_initialize(&infoSeq); or
FooSeq fooSeq = DDS_SEQUENCE_INITIALIZER;
499

7.4.3.6 read_w_condition and take_w_condition

Note: This operation may loan internal Connext DDS memory, which must be returned with return_loan
(). See Loaning and Returning Data and SampleInfo Sequences (Section 7.4.2 on page 492).
7.4.3.6 read_w_condition and take_w_condition
The read_w_condition() and take_w_condition() operations are identical to read() and take(), but
instead of passing in the sample_states, view_states, and instance_states mask parameters directly, you
pass in a ReadCondition (which specifies these masks).
DDS_ReturnCode_t read_w_condition (
FooSeq &received_data,
DDS_SampleInfoSeq &info_seq,
DDS_Long max_samples,
DDSReadCondition *condition)

Note: This operation may loan internal Connext DDS memory, which must be returned with return_loan
(). See Loaning and Returning Data and SampleInfo Sequences (Section 7.4.2 on page 492).
7.4.3.7 read_instance_w_condition and take_instance_w_condition
The read_instance_w_condition() and take_instance_w_condition() operations are similar to read_
instance() and take_instance(), respectively, except that the returned DDS samples must also satisfy a specified ReadCondition.
DDS_ReturnCode_t read_instance_w_condition(
FooSeq & received_data,
DDS_SampleInfoSeq & info_seq,
DDS_Long max_samples,
const DDS_InstanceHandle_t & a_handle,
DDSReadCondition * condition);

The behavior of read_instance_w_condition() and take_instance_w_condition() follows the same rules
as read() and take() regarding pre-conditions and post-conditions for the received_data and sample_info
parameters.
These functions can only be used when the DataReader is tied to a keyed type, see DDS Samples,
Instances, and Keys (Section 2.3.1 on page 14) for more about keyed data types.
Similar to read(), these operations must be provided on the specialized class that is generated for the particular application data-type that is being accessed.
Note: These operations may loan internal Connext DDS memory, which must be returned with return_
loan(). See Loaning and Returning Data and SampleInfo Sequences (Section 7.4.2 on page 492).

500

7.4.3.8 read_next_instance_w_condition and take_next_instance_w_condition

7.4.3.8 read_next_instance_w_condition and take_next_instance_w_condition
The read_next_instance_w_condition() and take_next_instance_w_condition() operations are identical
to read_next_instance() and take_next_instance(), but instead of passing in the sample_states, view_
states, and instance_states mask parameters directly, you pass in a ReadCondition (which specifies these
masks).
DDS_ReturnCode_t read_next_instance_w_condition (
FooSeq &received_data,
DDS_SampleInfoSeq &info_seq,
DDS_Long max_samples,
const DDS_InstanceHandle_t &previous_handle,
DDSReadCondition *condition)

Note: This operation may loan internal Connext DDS memory, which must be returned with return_loan
(). See Loaning and Returning Data and SampleInfo Sequences (Section 7.4.2 on page 492).
7.4.3.9 The select() API (Modern C++)
The Modern C++ API combines all the previous ways to read data into a single operation: reader.select
().This call is followed by one or more calls to functions that configure the query and always ends in a call
to read() or take(). These are the functions that configure a select():
Function

Description

Default

max_
samples()

Specifies the maximum number of samples to read or take in this call

Up to the value specified in max_
samples_per_read (Section on page
518)

instance()

Specifies an instance to read or take

All instances

next_
instance()

Indicates that read or take should return samples for the instance that follows the one
being passed (Note: both next_instance() and instance() can't be specified at the same
time)

All instances

state()

Specifies the sample state, view state and instance state

All samples

content()

Specifies a query on the data values to read

All samples

Specifies a condition (see read_w_condition()). If condition() is specified state() and
content()cannot be specified.
condition()

All samples
When running a query more than once on the same DataReader, it is more efficient to
create a QueryCondition and pass it to condition() rather than using content().

To read or take using the default options, simply call reader.read() or reader.take() with no arguments.
The following example shows how to call select():

501

7.4.4 Acknowledging DDS Samples

dds::sub::LoanedSamples samples =
reader.select()
.max_samples(20)
.state(dds::sub::status::DataState::new_instance())
.content(dds::sub::Query(reader, "x > 10"))
.instance(my_instance_handle)
.take();

7.4.4 Acknowledging DDS Samples
DDS samples can be acknowledged one at a time, or as a group.
To explicitly acknowledge a single DDS sample:
DDS_ReturnCode_t acknowledge_sample (
const DDS_SampleInfo & sample_info);
DDS_ReturnCode_t acknowledge_sample (
const DDS_SampleInfo & sample_info,
const DDS_AckResponseData_t & response_data);

Or you may acknowledge all previously accessed DDS samples by calling:
DDS_ReturnCode_t DDSDataReader::acknowledge_all ()
DDS_ReturnCode_t DDSDataReader::acknowledge_all (
const DDS_AckResponseData_t & response_data)

Where:
sample_info

is of type DDS_SampleInfo, identifying the DDS sample being acknowledged

response_data

is response data sent to the DataWriter upon acknowledgment

These operations can only be used when the DataReader’sRELIABILITY QosPolicy (Section 6.5.19 on
page 400) has an acknowledgment_kind set to DDS_APPLICATION_EXPLICIT_
ACKNOWLEDGMENT_MODE. You must also set max_app_ack_response_length (in the DATA_
READER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 7.6.2 on page 517)) to a value
greater than zero.
See also: Application Acknowledgment (Section 6.3.12 on page 288) and Guaranteed Delivery of Data
(Section Chapter 13 on page 695).

7.4.5 The Sequence Data Structure
(This section doesn't apply to the Modern C++ API)
The DDS specification uses sequences whenever a variable-length array of elements must be passed
through the API. This includes passing QosPolicies into Connext DDS, as well as retrieving DDS data

502

7.4.5 The Sequence Data Structure

samples from Connext DDS. A sequence is an ordered collection of elements of the same type. The type
of a sequence containing elements of type “Foo” (whether “Foo” is one of your types or a built-in Connext DDS type) is typically called “FooSeq.”
In all APIs except Java, FooSeq contains deep copies of Foo elements; in Java, which does not provide
direct support for deep copy semantics, FooSeq contains references to Foo objects. In Java, sequences
implement the java.util.List interface, and thus support all of the collection APIs and idioms familiar to
Java programmers.
A sequence is logically composed of three things: an array of elements, a maximum number of elements
that the array may contain (i.e. its allocated size), and a logical length indicating how many of the allocated
elements are valid. The length may vary dynamically between 0 and the maximum (inclusive); it is not permissible to access an element at an index greater than or equal to the length.
A sequence may either “own” the memory associated with it, or it may “borrow” that memory. If a
sequence owns its own memory, then the sequence itself will allocate the its memory and is permitted to
grow and shrink that memory (i.e. change its maximum) dynamically.
You can also loan a sequence of memory using the sequence-specific operations loan_contiguous() or
loan_discontiguous(). This is useful if you want Connext DDS to copy the received DDS data samples
directly into data structures allocated in user space.
Please do not confuse (a) the user loaning memory to a sequence with (b) Connext DDS loaning internal
memory from the receive queue to the user code via the read() or take() operations. For sequences of user
data, these are complementary operations. read() and take() loan memory to the user, passing in a
sequence that has been loaned memory with loan_contiguous() or loan_discontinguous().
A sequence with loaned of memory may not change its maximum size.
For C developers:
In C, because there is no concept of a constructor, sequences must be initialized before they are used. You
can either set a sequence equal to the macro DDS_SEQUENCE_INITIALIZER or use a sequence-specific method, Seq_initialize(), to initialize sequences.
For C++, C++/CLI, and C# developers:
C++ sequence classes overload the [ ] operators to allow you to access their elements as if the sequence
were a simple array. However, for code portability reasons, Connext DDS’s implementation of sequences
does not use the Standard Template Library (STL).
For Java developers:
In Java, sequences implement the List interface, and typically, a List must contain Objects; it cannot contain primitive types directly. This restriction makes Lists of primitives types less efficient because each
type must be wrapped and unwrapped into and from an Object as it is added to and removed from the
List.

503

7.4.6 The SampleInfo Structure

Connext DDS provides a more efficient implementation for sequences of primitive types. In Connext
DDS, primitive sequence types (e.g., IntSeq, FloatSeq, etc.) are implemented as wrappers around arrays
of primitive types. The wrapper also provides the usual List APIs; however, these APIs manipulate
Objects that store the primitive type.
More efficient APIs are also provided that manipulate the primitive types directly and thus avoid unnecessary memory allocations and type casts. These additional methods are named according to the pattern
; for example, the IntSeq class defines methods addInt() and getInt()
that correspond to the List APIs add() and get(). addInt() and getInt() directly manipulate int values
while add() and get() manipulate Objects that contain a single int.
For more information on sequence APIs in all languages, please consult the API Reference HTML documentation (from the main page, select Modules, RTI Connext DDS API Reference, Infrastructure
Module, Sequence Support).

7.4.6 The SampleInfo Structure
When you invoke the read/take operations, for every DDS data sample that is returned, a corresponding
SampleInfo is also returned. SampleInfo structures provide you with additional information about the
DDS data samples received by Connext DDS.
Table 7.18 DDS_SampleInfo Structure shows the format of the SampleInfo structure.
Table 7.18 DDS_SampleInfo Structure
Type

Field
Name

Description

DDS_
SampleStateKind

sample_state

See Sample States (Section 7.4.6.2 on page 506)

DDS_
ViewStateKind

view_state

See View States (Section 7.4.6.3 on page 506)

DDS_
InstanceStateKind

instance_state See Instance States (Section 7.4.6.4 on page 507)

DDS_Time_t

source_
timestamp

Time stored by the DataWriter when the DDS sample was written.

DDS_
InstanceHandle_t

instance_
handle

Handle to the data-instance corresponding to the DDS sample.

DDS_
InstanceHandle_t

publication_
handle

Local handle to the DataWriter that modified the instance. This is the same instance handle returned by
get_matched_publications(). You can use this handle when calling get_matched_publication_data
().

504

7.4.6 The SampleInfo Structure

Table 7.18 DDS_SampleInfo Structure
Type

Field
Name

Description

disposed_
generation_
count
no_writers_
generation_
count
DDS_Long

sample_rank

See Generation Counts and Ranks (Section 7.4.6.5 on page 508).

generation_
rank
absolute_
generation_
rank
DDS_Boolean

valid_data

Indicates whether the DDS data sample includes valid data. See Valid Data Flag (Section 7.4.6.6 on
page 510).

DDS_Time_t

reception_
timestamp

Time stored when the DDS sample was committed by the DataReader. See Reception Timestamp
(Section 7.4.6.1 on the next page).

DDS_
publication_
SequenceNumber_ sequence_
t
number

Publication sequence number assigned when the DDS sample was written by the DataWriter.

DDS_
reception_
SequenceNumber_ sequence_
t
number

Reception sequence number assigned when the DDS sample was committed by the DataReader. See
Reception Timestamp (Section 7.4.6.1 on the next page).

struct DDS_
GUID_t

original_
publication_
virtual_guid

original_
struct DDS_
publication_
SequenceNumber_ virtual_
t
sequence_
number

DDS_Long

flag

Original publication virtual GUID.
If the Publisher’s access_scope is GROUP, this field contains the Publisher virtual GUID that
uniquely identifies the DataWriter group.

Original publication virtual sequence number.
If the Publisher’s access_scope is GROUP, this field contains the Publisher virtual sequence number
that uniquely identifies a DDS sample within the DataWriter group.

Flags associated with the DDS sample; set by using the flag field in DDS_WriteParams_t when
writing a DDS sample with FooDataWriter_write_w_params() (see Writing Data (Section 6.3.8 on
page 283)).

505

7.4.6.1 Reception Timestamp

Table 7.18 DDS_SampleInfo Structure
Type

Field
Name

Description

struct DDS_
GUID_t

source_guid

The application logical data source associated with the sample.

struct DDS_
GUID_t

related_
source_guid

The application logical data source that is related to the sample.

struct DDS_
GUID_t

related_
subscription_ The related_reader_guid associated with the sample.
guid

7.4.6.1 Reception Timestamp
In reliable communication, if DDS data samples are received out received of order, Connext DDS will not
deliver them until all the previous DDS data samples have been received. For example, if DDS sample 2
arrives before DDS sample 1, DDS sample 2 cannot be delivered until DDS sample 1 is received. The
reception_timestamp is the time when all previous DDS samples has been received—the time at which
the DDS sample is committed. If DDS samples are all received in order, the committed time will be same
as reception time. However, if DDS samples are lost on the wire, then the committed time will be later
than the initial reception time.
7.4.6.2 Sample States
For each DDS sample received, Connext DDS keeps a sample_state relative to each DataReader. The
sample_state can be either:
l

READ: The DataReader has already accessed that DDS sample by means of read().

l

NOT_READ: The DataReader has never accessed that DDS sample before.

The DDS samples retrieved by a read() or take() need not all have the same sample_state.
7.4.6.3 View States
For each instance (identified by a unique key value), Connext DDS keeps a view_state relative to each
DataReader. The view_state can be either:
l

NEW: Either this is the first time the DataReader has ever accessed DDS samples of the instance, or
the DataReader has accessed previous DDS samples of the instance, but the instance has since been
reborn (i.e. become not-alive and then alive again). These two cases are distinguished by examining

506

7.4.6.4 Instance States

the disposed_generation_count and the no_writers_generation_count (see Generation Counts
and Ranks (Section 7.4.6.5 on the next page)).
l

NOT_NEW:The DataReader has already accessed DDS samples of the same instance and the
instance has not been reborn since.

The view_state in the SampleInfo structure is really a per-instance concept (as opposed to the sample_
state which is per DDS sample). Thus all DDS data samples related to the same instance that are returned
by read() or take() will have the same value for view_state.
7.4.6.4 Instance States
As seen in Instance States (Section Figure 7.18 on the next page), Connext DDS keeps an instance_state
for each instance; it can be:
l

l

l

ALIVE: The following are all true: (a) DDS samples have been received for the instance, (b) there
are live DataWriters writing the instance, and (c) the instance has not been explicitly disposed (or
more DDS samples have been received after it was disposed).
NOT_ALIVE_DISPOSED: The instance was explicitly disposed by a DataWriter by means of the
dispose() operation.
NOT_ALIVE_NO_WRITERS: The instance has been declared as not-alive by the DataReader
because it has determined that there are no live DataWriter entities writing that instance.

The events that cause the instance_state to change can depend on the setting of the OWNERSHIP
QosPolicy (Section 6.5.15 on page 389):
l

l

If OWNERSHIP QoS is set to EXCLUSIVE, the instance_state becomes NOT_ALIVE_
DISPOSED only if the DataWriter that currently “owns” the instance explicitly disposes it. The
instance_state will become ALIVE again only if the DataWriter that owns the instance writes it.
Note that ownership of the instance is determined by a combination of the OWNERSHIP and
OWNERSHIP_STRENGTH QosPolicies. Ownership of an instance can dynamically change.
If OWNERSHIP QoS is set to SHARED, the instance_state becomes NOT_ALIVE_
DISPOSED if any DataWriter explicitly disposes the instance. The instance_state becomes
ALIVE as soon as any DataWriter writes the instance again.

507

7.4.6.5 Generation Counts and Ranks

Figure 7.18 Instance States

Since the instance_state in the SampleInfo structure is a per-instance concept, all DDS data samples
related to the same instance that are returned by read() or take() will have the same value for instance_
state.
7.4.6.5 Generation Counts and Ranks
Generation counts and ranks allow your application to distinguish DDS samples belonging to different
‘generations’ of the instance. It is possible for an instance to become alive, be disposed and become notalive, and then to cycle again from alive to not-alive states during the operation of an application. Each
time an instance becomes alive defines a new generation for the instance.
It is possible that an instance may cycle through alive and not-alive states multiple times before the application accesses the DDS data samples for the instance. This means that the DDS data samples returned by
508

7.4.6.5 Generation Counts and Ranks

read() and take() may cross generations. That is, some DDS samples were published when the instance
was alive in one generation and other DDS samples were published when the instance transitioned
through the non-alive state into the alive state again. It may be important to your application to distinguish
the DDS data samples by the generation in which they were published.
Each DataReader keeps two counters for each new instance it detects (recall that instances are distinguished by their key values):
l

l

disposed_generation_count: Counts how many times the instance_state of the corresponding
instance changes from NOT_ALIVE_DISPOSED to ALIVE. The counter is reset when the
instance resource is reclaimed.
no_writers_generation_count: Counts how many times the instance_state of the corresponding
instance changes from NOT_ALIVE_NO_WRITERS to ALIVE. The counter is reset when the
instance resource is reclaimed.

The disposed_generation_count and no_writers_generation_count fields in the SampleInfo structure
capture a snapshot of the corresponding counters at the time the corresponding DDS sample was received.
The sample_rank and generation_rank in the SampleInfo structure are computed relative to the
sequence of DDS samples returned by read() or take():
l

l

l

sample_rank: Indicates how many DDS samples of the same instance follow the current one in the
sequence. The DDS samples are always time-ordered, thus the newest DDS sample of an instance
will have a sample_rank of 0. Depending on what you have configured read() and take() to
return, a sample_rank of 0 may or may not be the newest DDS sample that was ever received. It is
just the newest DDS sample in the sequence that was returned.
generation_rank: Indicates the difference in ‘generations’ between the DDS sample and the newest
DDS sample of the same instance as returned in the sequence. If a DDS sample belongs to the same
generation as the newest DDS sample in the sequence returned by read() and take(), then generation_rank will be 0.
absolute_generation_rank: Indicates the difference in ‘generations’ between the DDS sample and
the newest DDS sample of the same instance ever received by the DataReader. Recall that the data
sequence returned by read() and take() may not contain all of the data in the DataReader’s receive
queue. Thus, a DDS sample that belongs to the newest generation of the instance will have an absolute_generation_rank of 0.

Like the ‘generation count’ values, the ‘rank’ values are also reset to 0 if the instance resource is reclaimed.
By using the sample_rank, generation_rank and absolute_generation_rank information in the
SampleInfo structure, your application can determine exactly what happened to the instance and thus
make appropriate decisions of what to do with the DDS data samples received for the instance. For
example:

509

7.4.6.6 Valid Data Flag

l

l

l

A DDS sample with sample_rank = 0 is the newest DDS sample of the instance in the returned
sequence.
DDS samples that belong to the same generation will have the same generation_rank (as well as
absolute_generation_rank).
DDS samples with absolute_generation_rank = 0 belong to the newest generation for the instance
received by the DataReader.

7.4.6.6 Valid Data Flag
The SampleInfo structure’s valid_data flag indicates whether the DDS sample contains data or is only
used to communicate a change in the instance_state of the instance.
Normally, each DDS sample contains both a SampleInfo structure and some data. However, there are situations in which the DDS sample only contains the SampleInfo and does not have any associated data. This
occurs when Connext DDS notifies the application of a change of state for an instance that was caused by
some internal mechanism (such as a timeout) for which there is no associated data. An example is
whenConnext DDS detects that an instance has no writers and changes the corresponding instance_state
to NOT_ALIVE_NO_WRITERS.
If this flag is TRUE, then the DDS sample contains valid Data. If the flag is FALSE, the Dds Sample contains no data.
To ensure correctness and portability, your application must check the valid_data flag prior to accessing
the data associated with the DDS sample, and only access the data if it is TRUE.

7.5 Subscriber QosPolicies
Subscribers have the same set of QosPolicies as Publishers; see Publisher/Subscriber QosPolicies (Section
6.4 on page 312).
l

ENTITYFACTORY QosPolicy (Section 6.4.2 on page 315)

l

EXCLUSIVE_AREA QosPolicy (DDS Extension) (Section 6.4.3 on page 318)

l

GROUP_DATA QosPolicy (Section 6.4.4 on page 320)

l

PARTITION QosPolicy (Section 6.4.5 on page 323)

l

PRESENTATION QosPolicy (Section 6.4.6 on page 330)

7.6 DataReader QosPolicies
This section describes the QosPolicies that are strictly for DataReaders (not for DataWriters). For a complete list of QosPolicies that apply to DataReaders, see Table 7.16 DataReader QosPolicies .

510

7.6.1 DATA_READER_PROTOCOL QosPolicy (DDS Extension)

l
l

DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 below)
DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 7.6.2 on page
517)

l

READER_DATA_LIFECYCLE QoS Policy (Section 7.6.3 on page 523)

l

TIME_BASED_FILTER QosPolicy (Section 7.6.4 on page 526)

l

TRANSPORT_MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on page 529)

l

TYPE_CONSISTENCY_ENFORCEMENT QosPolicy (Section 7.6.6 on page 532)

7.6.1 DATA_READER_PROTOCOL QosPolicy (DDS Extension)
The DATA_READER_PROTOCOL QosPolicy applies only to DataReaders that are set up for reliable
operation (see RELIABILITY QosPolicy (Section 6.5.19 on page 400)). This policy allows the application to fine-tune the reliability protocol separately for each DataReader. For details of the reliable protocol used by Connext DDS, see Reliable Communications (Section Chapter 10 on page 629).
Connext DDS uses a standard protocol for packet (user and meta data) exchange between applications.
The DataReaderProtocol QosPolicy gives you control over configurable portions of the protocol, including the configuration of the reliable data delivery mechanism of the protocol on a per DataReader basis.
These configuration parameters control timing and timeouts, and give you the ability to trade off between
speed of data loss detection and repair, versus network and CPU bandwidth used to maintain reliability.
It is important to tune the reliability protocol on a per DataReader basis to meet the requirements of the
end-user application so that data can be sent between DataWriters and DataReaders in an efficient and
optimal manner in the presence of data loss.
You can also use this QosPolicy to control how DDS responds to "slow" reliable DataReaders or ones
that disconnect or are otherwise lost.
See the RELIABILITY QosPolicy (Section 6.5.19 on page 400) for more information on the perDataReader/DataWriter reliability configuration. The HISTORY QosPolicy (Section 6.5.10 on page 376)
and RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) also play an important role in the
DDS reliability protocol.
This policy includes the members presented in Table 7.19 DDS_DataReaderProtocolQosPolicy and Table
7.20 DDS_RtpsReliableReaderProtocol_t. For defaults and valid ranges, please refer to the API Reference
HTML documentation.
When setting the fields in this policy, the following rule applies. If this is false, Connext DDS returns
DDS_RETCODE_INCONSISTENT_POLICY when setting the QoS:
max_heartbeat_response_delay >= min_heartbeat_response_delay

511

7.6.1 DATA_READER_PROTOCOL QosPolicy (DDS Extension)

Table 7.19 DDS_DataReaderProtocolQosPolicy
Type

Field
Name

Description
The virtual GUID (Global Unique Identifier) is used to uniquely identify the same DataReader across
multiple incarnations. In other words, this value allows Connext DDS to remember information about a
DataReader that may be deleted and then recreated.
This value is used to provide durable reader state.

DDS_GUID_t

virtual_guid

For more information, see Durability and Persistence Based on Virtual GUIDs (Section 12.2 on page
680).
By default, Connext DDS will assign a virtual GUID automatically. If you want to restore the
DataReader’s state after a restart, you can get the DataReader's virtual GUID using its get_qos()
operation, then set the virtual GUID of the restarted DataReader to the same value.
Determines the DataReader’s RTPS object ID, according to the DDS-RTPS Interoperability Wire
Protocol.

DDS_
UnsignedLong

rtps_object_
id

Only the last 3 bytes are used; the most significant byte is ignored.
The rtps_host_id, rtps_app_id, rtps_instance_id in the WIRE_PROTOCOL QosPolicy (DDS
Extension) (Section 8.5.9 on page 610), together with the 3 least significant bytes in rtps_object_id, and
another byte assigned by Connext DDS to identify the entity type, forms the BuiltinTopicKey in
SubscriptionBuiltinTopicData.
Specifies whether this DataReader expects inline QoS with every DDS sample.
DataReaders usually rely on the discovery process to propagate QoS changes for matched DataWriters.
Another way to get QoS information is to have it sent inline with a DDS sample.

DDS_
Boolean

expects_
inline_qos

With Connext DDS, DataWriters and DataReaders cache discovery information, so sending inline QoS
is typically unnecessary. The use of inline QoS is only needed for stateless implementations of DDS in
which DataReaders do not cache Discovery information.
The complete set of QoS that a DataWriter may send inline is specified by the Real-Time PublishSubscribe (RTPS) Wire Interoperability Protocol.
Note: The use of inline QoS creates an additional wire-payload, consuming extra bandwidth and
serialization/deserialization time.
Determines whether the DataReader sends positive acknowledgements (ACKs) to matching DataWriters.

DDS_
Boolean

When TRUE. the matching DataWriter will keep DDS samples in its queue for this DataReader for a
disable_
minimum keep duration (see Disabling Positive Acknowledgements (Section 6.5.3.3 on page 354)).
positive_acks
When strict-reliability is not required and NACK-based reliability is sufficient, setting this field reduces
overhead network traffic.

512

7.6.1 DATA_READER_PROTOCOL QosPolicy (DDS Extension)

Table 7.19 DDS_DataReaderProtocolQosPolicy
Type

Field
Name

Description
Indicates whether or not an instance can move to the DDS_NOT_ALIVE_DISPOSED_INSTANCE_
STATE state without being in the DDS_ALIVE_INSTANCE_STATE state.

DDS_Boolean

propagate_
dispose_
of_
unregistered_
instances

When set to TRUE, the DataReader will receive dispose notifications even if the instance is not alive.
This field only applies to keyed DataReaders.
To make sure the key is available to the FooDataReader’s get_key_value() operation, use this option in
combination with setting the DataWriter’s serialize_key_with_dispose field (in the DATA_WRITER_
PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347)) to TRUE.
See Propagating Serialized Keys with Disposed-Instance Notifications (Section 6.5.3.5 on page 356).

DDS_Boolean

Indicates whether or not an instance can move to the DDS_NOT_ALIVE_NO_WRITERS_INSTANCE_
propagate_
STATE state without being in the DDS_ALIVE_INSTANCE_STATE state.
unregister_
of_disposed_ When set to TRUE, the DataReader will receive unregister notifications even if the instance is not alive.
instances
This field only applies to keyed DataReaders.

DDS_Rtpsrtps_reliable_
ReliableReaderSee Table 7.20 DDS_RtpsReliableReaderProtocol_t
reader
Protocol_t

513

7.6.1 DATA_READER_PROTOCOL QosPolicy (DDS Extension)

Table 7.20 DDS_RtpsReliableReaderProtocol_t
Type

Field
Name

Description
How long additionally received heartbeats are suppressed.

heartbeat_
When a reliable DataReader receives consecutive heartbeats within a short duration, this may trigger redundant
suppression_ NACKs. To prevent the DataReader from sending redundant NACKs, the DataReader may ignore the latter
duration
heartbeat(s) for this amount of time.
See How Often Heartbeats are Resent (heartbeat_period) (Section 10.3.4.1 on page 645).
min_
heartbeat_
DDS_
Duration_ response_
delay
t

Minimum delay between when the DataReader receives a heartbeat and when it sends an ACK/NACK.

max_
heartbeat_
response_
delay

Maximum delay between when the DataReader receives a heartbeat and when it sends an ACK/NACK.
Increasing this value helps prevent NACK storms, but increases latency.

nack_period

Rate at which to send negative acknowledgements to new DataWriters. See Example (Section 7.6.1.3 on page
516).

DDS_
Long

receive_
The number of received out-of-order DDS samples a reader can keep at a time. See Receive Window Size
window_size (Section 7.6.1.1 on the facing page)

DDS_
round_trip_
Duration_
time
t

The duration from sending a NACK to receiving a repair of a DDS sample. See Round-Trip Time For Filtering
Redundant NACKs (Section 7.6.1.2 on page 516)
The period at which application-level acknowledgment messages are sent.

DDS_
app_ack_
Duration_
period
t

A DataReader sends application-level acknowledgment messages to a DataWriter at this periodic rate, and will
continue sending until it receives a message from the DataWriter that it has received and processed the
acknowledgment.

514

7.6.1.1 Receive Window Size

Table 7.20 DDS_RtpsReliableReaderProtocol_t
Type

Field
Name

Description
The minimum number of DDS samples acknowledged by one application-level acknowledgment message.
This setting applies only when the RELIABILITY QosPolicy (Section 6.5.19 on page 400) acknowledgment_
kind is set to APPLICATION_EXPLICIT or APPLICATION_AUTO.
A DataReader will immediately send an application-level acknowledgment message when it has at least this
many DDS samples that have been acknowledged. It will not send an acknowledgment message until it has at
least this many DDS samples pending acknowledgment.

DDS_
Boolean

samples_
For example, calling the DataReader’s acknowledge_sample() this many times consecutively will trigger the
per_app_ack sending of an acknowledgment message. Calling the DataReader’s acknowledge_all() may trigger the sending
of an acknowledgment message, if at least this many DDS samples are being acknowledged at once. See
Acknowledging DDS Samples (Section 7.4.4 on page 502).
This is independent of the DDS_RtpsReliableReaderProtocol_t’s app_ack_period, where a DataReader will
send acknowledgment messages at the periodic rate regardless.
When this is set to DDS_LENGTH_UNLIMITED, acknowledgment messages are sent only periodically, at the
rate set by DDS_RtpsReliableReaderProtocol_t’s app_ack_period.
Minimum duration for which application-level acknowledgment response data is kept.

min_app_
DDS_
ack_
Duration_ response_
t
keep_
duration

The user-specified response data of an explicit application-level acknowledgment (called by DataReader’s
acknowledge_sample() or acknowledge_all() operations) is cached by the DataReader for the purpose of
reliably resending the data with the acknowledgment message. After this duration has passed from the time of
the first acknowledgment, the response data is dropped from the cache and will not be resent with future
acknowledgments for the corresponding DDS sample(s).

7.6.1.1 Receive Window Size
A reliable DataReader presents DDS samples it receives to the user in-order. If it receives DDS samples
out-of-order, it stores them internally until the other missing DDS samples are received. For example, if the
DataWriter sends DDS samples 1 and 2, if the DataReader receives 2 first, it will wait until it receives 1
before passing the DDS samples to the user.
The number of out-of-order DDS samples that a DataReader can keep is set by the receive_window_size.
A larger window allows more out-of-order DDS samples to be kept. When the window is full, any subsequent out-of-order DDS samples received will be dropped, and such drops would necessitate NACK
repairs that would degrade throughput. So, in network environments where out-of-order samples are more
probable or where NACK repairs are costly, this window likely should be increased.
By default, the window is set to 256, which is the maximum number of DDS samples a single NACK
submessage can request.

515

7.6.1.2 Round-Trip Time For Filtering Redundant NACKs

7.6.1.2 Round-Trip Time For Filtering Redundant NACKs
When a DataReader requests for a DDS sample to be resent, there is a delay from when the NACK is
sent, to when it receives the resent DDS sample. During that delay, the DataReader may receive
HEARTBEATs that normally would trigger another NACK for the same DDS sample. Such redundant
repairs waste bandwidth and degrade throughput.
The round_trip_time is a user-configured estimate of the delay between sending a NACK to receiving a
repair. A DataReader keeps track of when a DDS sample has been NACK'd, and will prevent subsequent
NACKs from redundantly requesting for the same DDS sample, until the round trip time has passed.
Note that the default value of 0 seconds means that the DataReader does not filter for redundant NACKs.
7.6.1.3 Example
For many applications, changing these values will not be necessary. However, the more nodes that your
distributed application uses, and the greater the amount of network traffic it generates, the more likely it is
that you will want to consider experimenting with these values.
When a reliable DataReader receives a heartbeat from a DataWriter, it will send an ACK/NACK packet
back to the DataWriter. Instead of sending the packet out immediately, the DataReader can choose to
send it after a delay. This policy sets the minimum and maximum time to delay; the actual delay will be a
random value in between. (For more on heartbeats and ACK/NACK messages, see Discovery (Section
Chapter 14 on page 709).)
Why is a delay useful? For DataWriters that have multiple reliable DataReaders, an efficient way of heartbeating all of the DataReaders is to send a single heartbeat via multicast. In that case, all of the DataReaders will receive the heartbeat (approximately) simultaneously. If all DataReaders immediately respond
with a ACK/NACK packet, the network may be flooded. While the size of a ACK/NACK packet is relatively small, as the number of DataReaders increases, the chance of packet collision also increases. All of
these conditions may lead to dropped packets which forces the DataWriter to send out additional heartbeats that cause more simultaneous heartbeats to be sent, ultimately resulting a network packet storm.
By forcing each DataReader to wait for a random amount of time, bounded by the minimum and maximum values
in this policy, before sending an ACK/NACK response to a heartbeat, the use of the network is spread out over a
period of time, decreasing the peak bandwidth required as well as the likelihood of dropped packets due to collisions.
This can increase the overall performance of the reliable connection while avoiding a network storm.

When a reliable DataReader first matches a reliable DataWriter, the DataReader sends periodic NACK
messages at the specified period to pull historical data from the DataWriter. The DataReader will stop
sending periodic NACKs when it has received all historical data available at the time that it matched the
DataWriter. The DataReader ensures that at least one NACK is sent per period; for example, if, within a
NACK period, the DataReader responds to a HEARTBEAT message with a NACK, then the
DataReader will not send another periodic NACK.

516

7.6.1.4 Properties

7.6.1.4 Properties
This QosPolicy cannot be modified after the DataReader is created.
It only applies to DataReaders, so there are no restrictions for setting it compatibly with respect to
DataWriters.
7.6.1.5 Related QosPolicies
l

DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347)

l

RELIABILITY QosPolicy (Section 6.5.19 on page 400)

7.6.1.6 Applicable Dds Entities
l

DataReaders (Section 7.3 on page 459)

7.6.1.7 System Resource Considerations
Changing the values in this policy requires making tradeoffs between minimizing latency (decreasing
min_heartbeat_response_delay), maximizing determinism (decreasing the difference between min_
heartbeat_response_delay and max_heartbeat_response_delay), and minimizing network collisions/spreading out the ACK/NACK packets across a time interval (increasing the difference between
min_heartbeat_response_delay and max_heartbeat_response_delay and/or shifting their values
between different DataReaders).
If the values are poorly chosen with respect to the characteristics and requirements of a given application,
the latency and/or throughput of the application may suffer.

7.6.2 DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension)
The DATA_READER_RESOURCE_LIMITS QosPolicy extends your control over the memory allocated by Connext DDS for DataReaders beyond what is offered by the RESOURCE_LIMITS QosPolicy
(Section 6.5.20 on page 405). RESOURCE_LIMITS controls memory allocation with respect to the
DataReader itself: the number of DDS samples that it can store in the receive queue and the number of
instances that it can manage simultaneously. DATA_READER_RESOURCE_LIMITS controls memory
allocation on a per matched-DataWriter basis. The two are orthogonal.
This policy includes the members in Table 7.21 DDS_DataReaderResourceLimitsQosPolicy. For defaults
and valid ranges, please refer to the API Reference HTML documentation.

517

7.6.2 DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 7.21 DDS_DataReaderResourceLimitsQosPolicy
Type Field Name
max_remote_
writers

Description
Maximum number of DataWriters from which a DataReader may receive DDS data samples, among all
instances.
For unkeyed Topics: max_remote_writers must = max_remote_writers_per_instance

max_remote_
writers_
per_instance

Maximum number of DataWriters from which a DataReader may receive DDS data samples for a single
instance.
For unkeyed Topics: max_remote_writers must = max_remote_writers_per_instance

max_samples_ Maximum number of DDS samples received out-of-order that a DataReader can store from a single reliable
DataWriter.
per_remote_
writer
max_samples_per_remote_writer must be <= RESOURCE_LIMITS::max_samples
Maximum number of DDS_SampleInfo structures that a DataReader can allocate.
max_infos
max_infos must be >= RESOURCE_LIMITS::max_samples

DDS_
Long

initial_remote_ Initial number of DataWriters from which a DataReader may receive DDS data samples, including all instances.
writers
For unkeyed Topics: initial_remote_writers must = initial_remote_writers_per_instance
initial_remote_ Initial number of DataWriters from which a DataReader may receive DDS data samples for a single instance.
writers_per_
For unkeyed Topics: initial_remote_writers must = initial_remote_writers_per_instance
instance
initial_infos

Initial number of DDS_SampleInfo structures that a DataReader will allocate.

initial_
outstanding_
reads

Initial number of times in which memory can be concurrently loaned via read/take calls without being returned
with return_loan().

max_
outstanding_
reads

Maximum number of times in which memory can be concurrently loaned via read/take calls without being
returned with return_loan().

max_samples_
per_
Maximum number of DDS samples that can be read/taken on a DataReader.
read
disable_
Determines whether the DataReader can receive fragmented DDS samples.
DDS_
fragmentation_
Boolean
When fragmentation support is not needed, disabling fragmentation support will save some memory resources.
support

518

7.6.2 DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 7.21 DDS_DataReaderResourceLimitsQosPolicy
Type Field Name

Description
The maximum number of DDS samples for which the DataReader may store fragments at a given point in time.

max_
fragmented_
samples

At any given time, a DataReader may store fragments for up to max_fragmented_samples DDS samples
while waiting for the remaining fragments. These DDS samples need not have consecutive sequence numbers
and may have been sent by different DataWriters. Once all fragments of a DDS sample have been received, the
DDS sample is treated as a regular DDS sample and becomes subject to standard QoS settings, such as max_
samples. Connext DDS will drop fragments if the max_fragmented_samples limit has been reached.
For best-effort communication, Connext DDS will accept a fragment for a new DDS sample, but drop the
oldest fragmented DDS sample from the same remote writer.
For reliable communication, Connext DDS will drop fragments for any new DDS samples until all fragments
for at least one older DDS sample from that writer have been received.
Only applies if disable_fragmentation_support is FALSE.

DDS_
Long

initial_
fragmented_
samples
max_
fragmented_
samples_per_
remote_
writer
max_
fragments_
per_
sample

The initial number of DDS samples for which a DataReader may store fragments.
Only applies if disable_fragmentation_support is FALSE.

The maximum number of DDS samples per remote writer for which a DataReader may store fragments. This is
a logical limit, so a single remote writer cannot consume all available resources.
Only applies if disable_fragmentation_support is FALSE.

Maximum number of fragments for a single DDS sample.
Only applies if disable_fragmentation_support is FALSE.

By default, the middleware does not allocate memory upfront, but instead allocates memory from the heap upon
receiving the first fragment of a new sample. The amount of memory allocated equals the amount of memory
needed to store all fragments in the sample. Once all fragments of a sample have been received, the sample is
deserialized and stored in the regular receive queue. At that time, the dynamically allocated memory is freed
again.
dynamically_
DDS_ allocate_
Boolean fragmented_
samples

This QoS setting is useful for large, but variable-sized data types where up-front memory allocation for multiple
samples based on the maximum possible sample size may be expensive. The main disadvantage of not preallocating memory is that one can no longer guarantee the middleware will have sufficient resources at run-time.
If dynamically_allocate_fragmented_samples is FALSE, the middleware will allocate memory up-front for
storing fragments for up to initial_fragmented_samples samples. This memory may grow up to max_
fragmented_samples if needed.
Only applies if disable_fragmentation_support is FALSE.

DDS_
Long

max_total_
instances

Maximum number of instances for which a DataReader will keep state.
See max_total_instances and max_instances (Section 7.6.2.1 on page 522)

519

7.6.2 DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 7.21 DDS_DataReaderResourceLimitsQosPolicy
Type Field Name

Description
The maximum number of virtual writers (identified by a virtual GUID) from which a DataReader may read,
including all instances.

DDS_
Long

max_remote_
virtual_
writers

DDS_
Long

initial_remote_
virtual_
The initial number of virtual writers from which a DataReader may read, including all instances.
writers

When the Subscriber’s access_scope is GROUP, this value determines the maximum number of DataWriter
groups supported by the Subscriber. Since the Subscriber may contain more than one DataReader, only the
setting of the first applies.

Maximum number of virtual remote writers that can be associated with an instance.
For unkeyed types, this value is ignored.
The features of Durable Reader State and MultiChannel DataWriters, as well as Persistence Servicea, require
Connext DDS to keep some internal state per virtual writer and instance that is used to filter duplicate DDS
samples. These duplicate DDS samples could be coming from different DataWriter channels or from multiple
executions of Persistence Service.
DDS_
Long

max_remote_
virtual_
writers_per_
instance

Once an association between a remote virtual writer and an instance is established, it is permanent—it will not
disappear even if the physical writer incarnating the virtual writer is destroyed.
If max_remote_virtual_writers_per_instance is exceeded for an instance, Connext DDS will not associate
this instance with new virtual writers. Duplicate DDS samples coming from these virtual writers will not be
filtered on the reader.
If you are not using Durable Reader State, MultiChannel DataWriters or Persistence Service, you can set this
property to 1 to optimize resources.
For additional information about the virtual writers see Mechanisms for Achieving Information Durability and
Persistence (Section Chapter 12 on page 675).

DDS_
Long

initial_remote_
Initial number of virtual remote writers per instance.
virtual_
writers_per_
For unkeyed types, this value is ignored.
instance

DDS_
Long

max_remote_
writers_
per_sample

Maximum number of remote writers that are allowed to write the same DDS sample.
One scenario in which two DataWriters may write the same DDS sample is when using Persistence Service. The
DataReader may receive the same DDS sample from the original DataWriter and from an Persistence Service
DataWriter.

aPersistence Service is included with the Connext DDS Professional, Evaluation, and Basic package types. It saves DDS

data samples so they can be delivered to subscribing applications that join the system at a later time (see Introduction
to RTI Persistence Service (Section Chapter 26 on page 933)).

520

7.6.2 DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 7.21 DDS_DataReaderResourceLimitsQosPolicy
Type Field Name

Description
This value determines the maximum number of unique query condition content filters that a reader may create.

DDS_
Long

max_query_
condition_
filters

DDS_
Long

Maximum length of application-level acknowledgment response data.
max_app_ack_
response_
The maximum length of response data in an application-level acknowledgment.
length
When set to zero, no response data is sent with application-level acknowledgments.

keep_
DDS_ minimum_
Boolean state_for_
instances

Each query condition content filter is comprised of both its query_expression and query_parameters. Two
query conditions that have the same query_expression will require unique query condition filters if their
query_parameters differ. Query conditions that differ only in their state masks will share the same query
condition filter.

Determines whether the DataReader keeps a minimum instance state for up to max_total_instances. The
minimum state is useful for filtering samples in certain scenarios. See max_total_instances and max_instances
(Section 7.6.2.1 on the next page)

DataReaders must allocate internal structures to handle: the maximum number of DataWriters that may
connect to it; whether or not a DataReader handles data fragmentation and how many data fragments that
it may handle (for DDS data samples larger than the MTU of the underlying network transport); how
many simultaneous outstanding loans of internal memory holding DDS data samples can be provided to
user code; as well as others.
Most of these internal structures start at an initial size and, by default, will grow as needed by dynamically
allocating additional memory. You may set fixed, maximum sizes for these internal structures if you want
to bound the amount of memory that can be used by a DataReader. Setting the initial size to the maximum
size will prevent Connext DDS from dynamically allocating any memory after the DataReader is created.
This policy also controls how the allocated internal data structure may be used. For example, DataReaders
need data structures to keep track of all of the DataWriters that may be sending it DDS data samples. The
total number of DataWriters that it can keep track of is set by the initial_remote_writers and max_
remote_writers values. For keyed Topics, initial_remote_writers_per_instance and max_remote_
writers_per_instance control the number of DataWriters allowed by the DataReader to modify the value
of a single instance.
By setting the max value to be less than max_remote_writers, you can prevent instances with many
DataWriters from using up the resources and starving other instances. Once the resources for keeping
track of DataWriters are used up, the DataReader will not be able to accept “connections” from new
DataWriters. The DataReader will not be able to receive data from new matching DataWriters which
would be ignored.
In the reliable protocol used by Connext DDS to support a RELIABLE setting for the RELIABILITY
QosPolicy (Section 6.5.19 on page 400), the DataReader must temporarily store DDS data samples that
have been received out-of-order from a reliable DataWriter. The storage of out-of-order DDS samples is

521

7.6.2.1 max_total_instances and max_instances

allocated from the DataReader’s receive queue and shared among all reliable DataWriters. The parameter
max_samples_per_remote_writer controls the maximum number of out-of-order data DDS samples that
the DataReader is allowed to store for a single DataWriter. This value must be less than the max_samples
value set in the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405).
max_samples_per_remote_writer allows Connext DDS to share the limited resources of the
DataReader equitably so that a single DataWriter is unable to use up all of the storage of the DataReader
while missing DDS data samples are being resent.
When setting the values of the members, the following rules apply:
l
l

l

max_remote_writers >= initial_remote_writers
max_remote_writers_per_instance >= initial_remote_writers_per_instance
max_remote_writers_per_instance <= max_remote_writers
max_infos >= initial_infos
max_infos >= RESOURCE_LIMITS::max_samples

l

max_outstanding_reads >= initial_outstanding_reads

l

max_remote_writers >= max_remote_writers_per_instance

l

max_samples_per_remote_writer <= RESOURCE_LIMITS::max_samples

If any of the above are false, Connext DDS returns the error code DDS_RETCODE_
INCONSISTENT_POLICY when setting the DataReader’s QoS.
7.6.2.1 max_total_instances and max_instances
The features Durable Reader State (Section 12.4 on page 686), Multi-channel DataWriters (Section
Chapter 18 on page 824), and Persistence Service (Part 6: RTI Persistence Service (Section on page 932))
require Connext DDS to keep some internal state even for instances without DataWriters or DDS samples
in the DataReader’s queue or that have been purged due to a dispose. The additional state is used to filter
duplicate DDS samples that could be coming from different DataWriter channels or from multiple executions of Persistence Service. The total maximum number of instances that will be managed by the middleware, including instances without associated DataWriters or DDS samples or that have been purged
due to a dispose, is determined by max_total_instances. This additional state will only be kept for up to
max_total_instances if keep_minimum_state_for_instances is TRUE, otherwise the additional state
will not be kept for any instances.
7.6.2.2 Example
The max_samples_per_remote_writer value affects sharing and starvation. max_samples_per_remote_
writer can be set to less than the RESOURCE_LIMITS QosPolicy’s max_samples to prevent a single

522

7.6.2.3 Properties

DataWriter from starving others. This control is especially important for Topics that have their
OWNERSHIP QosPolicy (Section 6.5.15 on page 389) set to SHARED.
In the case of EXCLUSIVE ownership, a lower-strength remote DataWriter can "starve" a higherstrength remote DataWriter by making use of more of the DataReader's resources, an undesirable condition. In the case of SHARED ownership, a remote DataWriter may starve another remote DataWriter,
making the sharing not really equal.
7.6.2.3 Properties
This QosPolicy cannot be modified after the DataReader is created.
It only applies to DataReaders, so there are no restrictions for setting it compatibly on the DataWriter.
7.6.2.4 Related QosPolicies
l

RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)

l

OWNERSHIP QosPolicy (Section 6.5.15 on page 389)

7.6.2.5 Applicable Dds Entities
l

DataReaders (Section 7.3 on page 459)

7.6.2.6 System Resource Considerations
Increasing any of the “initial” values in this policy will increase the amount of memory allocated by Connext DDS when a new DataReader is created. Increasing any of the “max” values will not affect the initial
memory allocated for a new DataReader, but will affect how much additional memory may be allocated
as needed over the DataReader’s lifetime.
Setting a max value greater than an initial value thus allows your application to use memory more dynamically and efficiently in the event that the size of the application is not well-known ahead of time.
However, Connext DDS may dynamically allocate memory in response to network communications.

7.6.3 READER_DATA_LIFECYCLE QoS Policy
This policy controls the behavior of the DataReader with regards to the lifecycle of the data instances it
manages, that is, the data instances that have been received and for which the DataReader maintains some
internal resources.
When a DataReader receives data, it is stored in a receive queue for the DataReader. The user application
may either take the data from the queue or leave it there. This QoS controls whether or not Connext DDS
will automatically remove data from the receive queue (so that user applications cannot access it afterwards) when Connext DDS detects that there are no more DataWriters alive for that data.

523

7.6.3 READER_DATA_LIFECYCLE QoS Policy

DataWriters may also call dispose() on its data, informing DataReaders that the data no longer exists. This
QosPolicy also controls whether or not Connext DDS automatically removes disposed data from the
receive queue.
For keyed Topics, the consideration of removing DDS data samples from the receive queue is done on a
per instance (key) basis. Thus when Connext DDS detects that there are no longer DataWriters alive for a
certain key value for a Topic (an instance of the Topic), it can be configured to remove all DDS data
samples for a certain instance (key). DataWriters also can dispose its data on a per instance basis. Only the
DDS data samples of disposed instances would be removed by Connext DDS if so configured.
This policy helps purge untaken DDS samples from not-alive-instances and thus may prevent a
DataReader from reclaiming resources. With this policy, the untaken DDS samples from not-aliveinstances are purged and treated as if the DDS samples were taken after the specified amount of time.
The DataReader internally maintains the DDS samples that have not been taken by the application, subject to the constraints imposed by other QoS policies such as HISTORY QosPolicy (Section 6.5.10 on
page 376) and RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405).
The DataReader also maintains information regarding the identity, view-state, and instance-state of data
instances, even after all DDS samples have been ‘taken’ (see Accessing DDS Data Samples with Read or
Take (Section 7.4.3 on page 493)). This is needed to properly compute the states when future DDS
samples arrive.
Under normal circumstances, a DataReader can only reclaim all resources for instances for which there
are no DataWriters and for which all DDS samples have been ‘taken.’ The last DDS sample taken by the
DataReader for that instance will have an instance state of NOT_ALIVE_NO_WRITERS or NOT_
ALIVE_DISPOSED_INSTANCE (depending on whether or not the instance was disposed by the last
DataWriter that owned it.) If you are using the default (infinite) values for this QosPolicy, this behavior
can cause problems if the application does not ‘take’ those DDS samples for some reason. The ‘untaken’
DDS samples will prevent the DataReader from reclaiming the resources and they would remain in the
DataReader indefinitely.
A DataReader can also reclaim all resources for instances that have an instance state of NOT_ALIVE_
DISPOSED and for which all DDS samples have been 'taken'. DataReaders will only reclaim resources
in this situation when autopurge_disposed_instances_delay has been set to zero.
It includes the members in Table 7.22 DDS_ReaderDataLifecycleQosPolicy.
Table 7.22 DDS_ReaderDataLifecycleQosPolicy
Type

Field Name

DDS_
autopurge_
Duration_ nowriter_
t
samples_delay

Description
How long the DataReader maintains information about an instance once its instance_state becomes NOT_
ALIVE_NO_WRITERS.

524

7.6.3.1 Properties

Table 7.22 DDS_ReaderDataLifecycleQosPolicy
Type

Field Name

Description

DDS_
autopurge_
Duration_ disposed_
t
samples_delay

How long the DataReader maintains information about an instance once its instance_state becomes NOT_
ALIVE_DISPOSED.

DDS_
autopurge_
Duration_ disposed_
t
instances_delay

How long the DataReader maintains information about and instance once its instance_state becomes
NOT_ALIVE_DISPOSED. (Note: only values of 0 or INFINITE are currently supported).

autopurge_nowriter_samples_delay: This defines the minimum duration for which the DataReader will
maintain information regarding an instance once its instance_state becomes NOT_ALIVE_NO_
WRITERS. After this time elapses, the DataReader will purge all internal information regarding the
instance, any untaken DDS samples will also be lost.
autopurge_disposed_samples_delay: This defines the minimum duration for which the DataReader will
maintain DDS samples of an instance once its instance_state becomes NOT_ALIVE_DISPOSED.
After this time elapses, the DataReader will purge all internal information regarding the instance; any
untaken DDS samples will also be lost.
autopurge_disposed_instances_delay: This defines the minimum duration for which the DataReader
will maintain DDS samples of an instance once its instance_state becomes NOT_ALIVE_DISPOSED.
After this time elapses, the DataReader will purge all internal information regarding the instance.
7.6.3.1 Properties
This QoS policy can be modified after the DataReader is enabled.
It only applies to DataReaders, so there are no RxO restrictions for setting it compatibly on the
DataWriter.
7.6.3.2 Related QoS Policies
l

HISTORY QosPolicy (Section 6.5.10 on page 376)

l

LIVELINESS QosPolicy (Section 6.5.13 on page 382)

l

OWNERSHIP QosPolicy (Section 6.5.15 on page 389)

l

RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)

l

WRITER_DATA_LIFECYCLE QoS Policy (Section 6.5.27 on page 419)

7.6.3.3 Applicable Dds Entities
l

DataReaders (Section 7.3 on page 459)

525

7.6.3.4 System Resource Considerations

7.6.3.4 System Resource Considerations
None.

7.6.4 TIME_BASED_FILTER QosPolicy
The TIME_BASED_FILTER QosPolicy allows you to specify that data should not be delivered more
than once per specified period for data-instances of a DataReader—regardless of how fast DataWriters are
publishing new DDS samples of the data-instance.
This QoS policy allows you to optimize resource usage (CPU and possibly network bandwidth) by only
delivering the required amount of data to different DataReaders.
DataWriters may send data faster than needed by a DataReader. For example, a DataReader of sensor
data that is displayed to a human operator in a GUI application does not need to receive data updates faster
than a user can reasonably perceive changes in data values. This is often measure in tenths (0.1) of a
second up to several seconds. However, a DataWriter of sensor information may have DataReaders that
are processing the sensor information to control parts of the system and thus need new data updates in
measures of hundredths (0.01) or thousandths (0.001) of a second.
With this QoS policy, different DataReaders can set their own time-based filters, so that data published
faster than the period set by a DataReader will be dropped by the middleware and not delivered to the
DataReader. Note that all filtering takes place on the reader side.
It includes the member in Table 7.23 DDS_TimeBasedFilterQosPolicy. For the default and valid range,
please refer to the API Reference HTML documentation.
Table 7.23 DDS_TimeBasedFilterQosPolicy
Type

Field Name

Description
Minimum separation time between DDS samples of the same instance.

DDS_Duration_t

minimum_separation
Must be <= DEADLINE::period

As seen in Accepting Data for DataReaders (Section Figure 7.19 on the facing page), it is inconsistent to
set a DataReader’s minimum_separation longer than its DEADLINE QosPolicy (Section 6.5.5 on page
363) period.

526

7.6.4 TIME_BASED_FILTER QosPolicy

Figure 7.19 Accepting Data for DataReaders

DDS data samples for a DataReader can be filtered out using the TIME_BASED_FILTER QoS (minimum_separation).
Once a DDS sample for an instance has been received, Connext DDS will accept but drop any new data samples for the
same instance that arrives within the time specified by minimum_separation. After the minimum_separation, a new DDS
sample that arrives is accepted and stored in the receive queue, and the timer starts again. If no DDS samples arrive by
the DEADLINE, the REQUESTED_DEADLINE_MISSED status will be changed and Listeners called back if installed.

This QosPolicy allows a DataReader to subsample the data being published for a data instance by
DataWriters. If a user application only needs new DDS samples for a data instance to be received at a specified period, then there is no need for Connext DDS to deliver data faster than that period. However,
whether or not data being published by a DataWriter at a faster rate than set by the TIME_BASED_
FILTER QoS is sent on the wire depends on several factors, including whether the DataReader is receiving the data reliably and if the data is being sent via multicast for multiple DataReaders.
For best effort data delivery, if the data type is unkeyed and the DataWriter has an infinite liveliness lease_
duration (LIVELINESS QosPolicy (Section 6.5.13 on page 382)), Connext DDS will only send as
many packets to a DataReader as required by the TIME_BASED_FILTER, no matter how fast the
DataWriter’s write() function is called.
For multicast data delivery to multiple DataReaders, the DataReader with the lowest TIME_BASED_
FILTER minimum_separation determines the DataWriter's send rate. For example, if a DataWriter
sends multicast to two DataReaders, one with minimum_separation of 2 seconds and one with minimum_separation of 1 second, the DataWriter will send every 1 second.
Other configurations (for example, when the DataWriter is reliable, or the data type is keyed, or the
DataWriter has a finite liveliness lease_duration) must send all data published by the DataWriter. On
reception, only the data that passes the TIME_BASED_FILTER will be stored in the DataReader’s

527

7.6.4.1 Example

receive queue. Extra data will be accepted but dropped. Note that filtering is only applied on ‘alive’ DDS
samples (that is, DDS samples that have not been disposed/unregistered).
7.6.4.1 Example
The purpose of this QosPolicy is to prevent fast DataWriters from overwhelming a DataReader that cannot process the data at the rate the data is being published. In certain configurations, the number of packets
sent by Connext DDS can also be reduced thus minimizing the consumption of network bandwidth.
You may want to change the minimum_separation between DDS data samples for one or more of the following reasons:
l

l

l

The DataReader is connected to the network via a low-bandwidth connection that is unable to sustain the amount of traffic generated by the matched DataWriter(s).
The rate at which the matched DataWriter(s) can generate DDS samples is faster than the rate at
which the DataReader can process them. Or faster than needed by the DataReader. For example, a
graphical user interface seldom needs to be updated faster than 30 times a second, even if new data
values are available much faster.
The resource limits of the DataReader are constrained relative to the number of DDS samples that
could be generated by the matched DataWriter(s). Too many packets coming at once will cause
them to be exhausted before the DataReader has time to process them.

7.6.4.2 Properties
This QosPolicy can be modified at any time.
It only applies to DataReaders, so there are no restrictions for setting it compatibly on the DataWriter.
7.6.4.3 Related QosPolicies
l

RELIABILITY QosPolicy (Section 6.5.19 on page 400)

l

DEADLINE QosPolicy (Section 6.5.5 on page 363)

l

TRANSPORT_MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on the facing page)

7.6.4.4 Applicable Dds Entities
l

DataReaders (Section 7.3 on page 459)

7.6.4.5 System Resource Considerations
Depending on the values of other QosPolicies such as RELIABILITY and TRANSPORT_
MULTICAST, this policy may be able to decrease the usage of network bandwidth and CPU by preventing unneeded packets from being sent and processed.

528

7.6.5 TRANSPORT_MULTICAST QosPolicy (DDS Extension)

7.6.5 TRANSPORT_MULTICAST QosPolicy (DDS Extension)
This QosPolicy specifies the multicast address on which a DataReader wants to receive its data. It can also
specify a port number as well as a subset of the available transports with which to receive the multicast
data.
By default, DataWriters will send individually addressed packets for each DataReader that subscribes to
the topic of the DataWriter—this is known as unicast delivery. Thus, as many copies of the data will be
sent over the network as there are DataReaders for the data. The network bandwidth used by a
DataWriter will thus increase linearly with the number of DataReaders.
Multicast is a concept supported by some transports, most notably UDP/IP, so that a single packet on the
network can be addressed such that it is received by multiple nodes. This is more efficient when the same
data needs to be sent to multiple nodes. By using multicast, the network bandwidth usage will be constant,
independent of the number of DataReaders.
Coordinating the multicast address specified by DataReaders can help optimize network bandwidth usage
in systems where there are multiple DataReaders for the same Topic.
The QosPolicy structure includes the members in Table 7.24 DDS_TransportMulticastQosPolicy.
Table 7.24 DDS_TransportMulticastQosPolicy
Type
DDS_TransportMulticastSettingSeq
(A sequence of the type shown in
Table 7.25 DDS_
TransportMulticastSetting_t)

Field
Name

value

Description

A sequence of multicast locators. (See Locator Format (Section 14.2.1.1 on page 714).)

This field can be set to one of the following two values: DDS_AUTOMATIC_
TRANSPORT_MULTICAST_QOS or DDS_UNICAST_ONLY_TRANSPORT_
MULTICAST_QOS.
If it is set to DDS_AUTOMATIC_TRANSPORT_MULTICAST_QOS, the behavior
depends on the content of DDS_TransportMulticastQosPolicy::value:

DDS_TransportMulticastKind

kind

If DDS_TransportMulticastQosPolicy::value does not have any elements, multicast will
not be used.
If DDS_TransportMulticastQosPolicy::value first element has an empty address, the
address will be obtained from DDS_TransportMulticastMappingQosPolicy.
If none of the elements in DDS_TransportMulticastQosPolicy::value are empty, and at
least one element has a valid address, then that address will be used.
If it is set to DDS_UNICAST_ONLY_TRANSPORT_MULTICAST_QOS, then
multicast will not be used.

529

7.6.5 TRANSPORT_MULTICAST QosPolicy (DDS Extension)

Table 7.25 DDS_TransportMulticastSetting_t
Type

Field
Name

Description

DDS_
A sequence of transport aliases that specifies which transports should be used to receive multicast messages for this
transports
StringSeq
DataReader.
char *

receive_
address

A multicast group address to which the DataWriter should send data for this DataReader.

DDS_
Long

receive_
port

The port that should be used in the addressing of multicast messages destined for this DataReader. A value of 0
will cause Connext DDS to use a default port number based on domain ID. See Ports Used for Discovery
(Section 14.5 on page 738).

To take advantage of multicast, the value of this QosPolicy must be coordinated among all of the applications on a network for DataReaders of the same Topic. For a DataWriter to send a single packet that will
be received by all DataReaders simultaneously, the same multicast address must be used.
To use this QosPolicy, you will also need to specify a port number. A port number of 0 will cause Connext DDS to automatically use a default value. As explained in Ports Used for Discovery (Section 14.5 on
page 738), the default port number for multicast addresses is based on the domain ID. Should you choose
to use a different port number, then for every unique port number used by Entities in your application,
depending on the transport, Connext DDS may create a thread to process messages received for that port
on that transport. See Connext DDS Threading Model (Section Chapter 19 on page 837) for more about
threads.
Threads are created on a per-transport basis, so if this QosPolicy specifies multiple transports for a
receive_port, then a thread may be created for each transport for that unique port. Some transports may be
able to share a single thread for different ports, others can not. Note that different Entities can share the
same port number, and thus, the same thread will process all of the data for all of the Entities sharing the
same port number for a transport.
Also note that if the port number specified by this QoS is the same as a port number specified by a
TRANSPORT_UNICAST QoS, then the transport may choose to process data received both via multicast and unicast with a single thread. Whether or not a transport must use different threads to process data
received via multicast or unicast for the same port number depends on the implementation of the transport.
Notes:
l
l

The same multicast address can be used by DataReaders of different Topics.
Even though the TRANSPORT_MULTICAST QoS allows you to specify multiple multicast
addresses for a DataReader, Connext DDS currently only uses one multicast address (the first in the
sequence) per DataReader.

530

7.6.5.1 Example

l

If a DataWriter is using the MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on
page 386), the multicast addresses specified in the TRANSPORT_MULTICAST QosPolicy are
ignored by that DataWriter. The DataWriter will not publish DDS samples on those locators.

7.6.5.1 Example
In an airport, there may be many different monitors that display current flight information. Assuming each
monitor is controlled by a networked application, network bandwidth would be greatly reduced if flight
information was published using multicast.
Figure 7.20 Setting Up a Multicast DataReader below shows an example of how to set this QosPolicy.
Figure 7.20 Setting Up a Multicast DataReader
...
DDS_DataReaderQos reader_qos;
reader_listener = new HelloWorldListener();
if (reader_listener == NULL) {
// handle error
}
// Get default data reader QoS to customize
retcode = subscriber->get_default_datareader_qos(reader_qos);
if (retcode != DDS_RETCODE_OK) {
// handle error
}
// Set up multicast reader
reader_qos.multicast.value.ensure_length(1,1);
reader_qos.multicast.value[0].receive_address =
DDS_String_dup("239.192.0.1");
reader = subscriber->create_datareader(
topic,reader_qos,
reader_listener, DDS_STATUS_MASK_ALL);

7.6.5.2 Properties
This QosPolicy cannot be modified after the Entity is created.
For compatibility between DataWriters and DataReaders, the DataWriter must be able to send to the multicast address that the DataReader has specified.
7.6.5.3 Related QosPolicies
l

MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386)

l

TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412)

l

TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section 8.5.7 on page 606)

531

7.6.5.4 Applicable DDS Entities

7.6.5.4 Applicable DDS Entities
l

DomainParticipants (Section 8.3 on page 547)

l

DataReaders (Section 7.3 on page 459)

7.6.5.5 System Resource Considerations
On Ethernet-based systems, the number of multicast addresses that can be “listened” to by the network
interface card is usually limited. The exact number of multicast addresses that can be monitored simultaneously by a NIC depends on its manufacturer. Setting a multicast address for a DataReader will use
up one of the multicast-address slots of the NIC.
What happens if the number of different multicast addresses used by different DataReaders across different applications on the same node exceeds the total number supported by a NIC depends on the specific
operating system. Some will prevent you from configuring too many multicast addresses to be monitored.
Many operating systems will accommodate the extra multicast addresses by putting the NIC in promiscuous mode. This means that the NIC will pass every Ethernet packet to the operating system, and the
operating system will pass the packets with the specified multicast addresses to the application(s). This results in extra CPU usage. We recommend that your applications do not use more multicast addresses on a
single node than the NICs on that node can listen to simultaneously in hardware.
Depending on the implementation of a transport, Connext DDS may need to create threads to receive and
process data on a unique-port-number basis. Some transports can share the same thread to process data
received for different ports; others like UDPv4 must have different threads for different ports. In addition,
if the same port is used for both unicast and multicast, the transport implementation will determine whether
or not the same thread can be used to process both unicast and multicast data. For UDPv4, only one thread
is needed per port–independent of whether the data was received via unicast or multicast data. See Receive
Threads (Section 19.3 on page 839) for more information.

7.6.6 TYPE_CONSISTENCY_ENFORCEMENT QosPolicy
The TypeConsistencyEnforcementQosPolicy defines the rules that determine whether the type used to publish a given topic is consistent with the type used to subscribe to it.
The QosPolicy structure includes the member in Table 7.26 DDS_TypeConsistencyEnforcementQosPolicy.

532

7.6.6 TYPE_CONSISTENCY_ENFORCEMENT QosPolicy

Table 7.26 DDS_TypeConsistencyEnforcementQosPolicy
Type

Field Name

Description
Can be either:

DDS_TypeConsistencyKind

kind

l

DISALLOW_TYPE_COERCION

l

ALLOW_TYPE_COERCION (default)

See Values for TypeConsistencyKind (Section below) for details.

The type-consistency enforcement rules consist of two steps:
1. If both the DataWriter and DataReader specify a TypeObject, it is considered first. If the
DataReader allows type coercion, then its type must be assignable from the DataWriter’s type. If
the DataReader does not allow type coercion, then its type must be structurally identical to the type
of the DataWriter.
2. If either the DataWriter or the DataReader does not provide a TypeObject definition, then the
registered type names are examined. The DataReader’s and DataWriter’s registered type names
must match exactly.
If either Step 1 or Step 2 fails, the Topics associated with the DataReader and DataWriter are considered
to be inconsistent and the INCONSISTENT_TOPIC Status (Section 5.3.1 on page 211) is updated.
The default enforcement kind is DDS_ALLOW_TYPE_COERCION. However, when the middleware
is introspecting the built-in topic data declaration of a remote DataReader in order to determine whether it
can match with a local DataWriter, if it observes that no TypeConsistencyEnforcementQosPolicy value is
provided (as would be the case when communicating with a Service implementation not in conformance
with this specification), it assumes a kind of DDS_DISALLOW_TYPE_COERCION.
Values for TypeConsistencyKind
l

DISALLOW_TYPE_COERCION

With this setting, the DataWriter and DataReader must support the same data type in order for them to
communicate. (This is the degree of enforcement required by the OMG DDS Specification prior to
theOMG ‘Extensible and Dynamic Topic Types for DDS’ Specification.)
When Connext DDS is introspecting the built-in topic data declaration of a remote DataWriter or
DataReader, if no TypeConsistencyEnforcementQosPolicy value is provided (as would be the case when
communicating with an implementation not in conformance with the Extensible and Dynamic Topic
Types for DDS" (DDS-XTypes) specification), Connext DDS shall assume a kind of DISALLOW_
TYPE_COERCION.

533

7.6.6.1 Properties

l

ALLOW_TYPE_COERCION (default)

With this setting, the DataWriter and the DataReader need not support the same data type in order for
them to communicate, as long as the DataReader’s type is assignable from the DataWriter’s type.
For example, the following two extensible types will be assignable to each other since MyDerivedType
contains all the members of MyBaseType (member_1) plus an additional element (member_2).
struct MyBaseType {
long member_1;
};
struct MyDerivedType: MyBaseType {
long member_2;
};

Even if MyDerivedType was not explicitly inherited from MyBaseType, the types would still be
assignable. For example:
struct MyBaseType {
long member_1;
};
struct MyDerivedType {
long member_1;
long member_2;
};

For more information, see the RTI Connext DDS Core Libraries Getting Started Guide Addendum for
Extensible Types and the OMG ‘Extensible and Dynamic Topic Types for DDS’ Specification.
7.6.6.1 Properties
This QosPolicy cannot be modified after the DataReader is enabled.
It only applies to DataReaders, so there is no requirement that the publishing and subscribing sides use
compatible values.
7.6.6.2 Related QoS Policies
l

None.

7.6.6.3 Applicable Entities
l

DataReaders (Section 7.3 on page 459)

534

7.6.6.4 System Resource Considerations

7.6.6.4 System Resource Considerations
None.

535

Chapter 8 Working with DDS Domains
This section discusses how to use DomainParticipants. It describes the types of operations that are
available for them and their QosPolicies.
This section includes:
The goal of this section is to help you become familiar with the objects you need for setting up
your Connext DDS application. For specific details on any mentioned operations, see the API
Reference HTML documentation.

8.1 Fundamentals of DDS Domains and DomainParticipants
DomainParticipants are the focal point for creating, destroying, and managing other Connext DDS
objects. A DDS domain is a logical network of applications: only applications that belong to the
same DDS domain may communicate using Connext DDS. A DDS domain is identified by a
unique integer value known as a domain ID. An application participates in a DDS domain by creating a DomainParticipant for that domain ID.

536

8.1 Fundamentals of DDS Domains and DomainParticipants

Figure 8.1 Relationship between Applications and DDS Domains

Applications can belong to multiple DDS domains—A belongs to DDS domains 1 and 2. Applications in the same DDS
domain can communicate with each other, such as A and B, or A and C. Applications in different DDS domains, such as
B and C, are not even aware of each other and will not exchange messages.

As seen in Figure 8.1 Relationship between Applications and DDS Domains above, a single application
can participate in multiple DDS domains by creating multiple DomainParticipants with different domain
IDs. DomainParticipants in the same DDS domain form a logical network; they are isolated from
DomainParticipants of other DDS domains, even those running on the same set of physical computers
sharing the same physical network. DomainParticipants in different DDS domains will never exchange
messages with each other. Thus, a DDS domain establishes a “virtual network” linking all DomainParticipants that share the same domain ID.
An application that wants to participate in a certain DDS domain will need to create a DomainParticipant.
As seen in Figure 8.2 DDS Domain Module on the facing page, a DomainParticipant object is a container
for all other Entities that belong to the same DDS domain. It acts as factory for the Publisher, Subscriber,
and Topic entities. (As seen in Sending Data (Section Chapter 6 on page 242) and Receiving Data (Section Chapter 7 on page 437), in turn, Publishers are factories for DataWriters and Subscribers are factories for DataReaders.) DomainParticipants cannot contain other DomainParticipants.
Like all Entities, DomainParticipants have QosPolicies and Listeners. The DomainParticipant entity also
allows you to set ‘default’ values for the QosPolicies for all the entities created from it or from the entities
that it creates (Publishers, Subscribers, Topics, DataWriters, and DataReaders).

537

8.1 Fundamentals of DDS Domains and DomainParticipants

Figure 8.2 DDS Domain Module

Note: MultiTopics are not supported.

538

8.2 DomainParticipantFactory

8.2 DomainParticipantFactory
l

C, Traditional C++, Java and .NET APIs:
The main purpose of a DomainParticipantFactory is to create and destroy DomainParticipants.
In C++ terms, this is a singleton class; that is, you will only have a single DomainParticipantFactory in an application—no matter how many DomainParticipants the application may
create. Figure 8.3 Instantiating a DomainParticipantFactory below shows how to instantiate a
DomainParticipantFactory. Notice that there are no parameters to specify. Alternatively, in C++,
C++/CLI, and C#, the predefined macro, DDSTheParticipantFactory,1 can also be used to
retrieve the singleton factory.
Unlike the other Entities that you create, the DomainParticipantFactory does not have an associated
Listener. However, it does have associated QosPolicies, see Setting DomainParticipantFactory
QosPolicies (Section 8.2.1 on page 543). You can change them using the factory’s get_qos() and
set_qos() operations. The DomainParticipantFactory also stores the default QoS settings that can
be used when a DomainParticipant is created. These default settings can be changed as well, see
Getting and Setting Default QoS for Child Entities (Section 8.3.6.5 on page 568).
Figure 8.3 Instantiating a DomainParticipantFactory
DDSDomainParticipantFactory* factory = NULL;
factory = DDSDomainParticipantFactory::get_instance();
if (factory == NULL) {
// ... error
}

1In C, the macro is DDS_TheParticipantFactory . In Java, use the static class method DomainPar-

ticipantFactory.TheParticipantFactory.

539

8.2 DomainParticipantFactory

l

Modern C++ API:
In the Modern C++ API, there isn’t a explicit DomainParticipantFactory. DomainParticipants are
created using their constructors andare automatically destroyed as a reference type (See Creating and
Deleting DDS Entities (Section 4.1.1 on page 153)).
The operations to set and get the default DomainParticipantQos are static functions in DomainParticipant: DomainParticipant::default_participant_qos(). The operations to look up participants
are freestanding functions in the dds::domain and rti::domain namespaces: dds::domain::find(),
rti::domain::find_participant_by_name(), and rti::domain::find_participants(). The class
QosProvider is responsible for managing QoS profiles (see How to Load XML-Specified QoS Settings (Section 17.5 on page 810)).
There is a DomainParticipantFactoryQos, but it only contains the ENTITY_FACTORY to indicate
if a DomainParticipant should be enabled in its constructor or by calling enable(), and SYSTEM_
RESOURCE_LIMITS. The DomainParticipantFactoryQos getter and setter are static functions in
DomainParticipant: DomainParticipant::participant_factory_qos().
Another static function in DomainParticipant allows finalizing the implicit DomainParticipantFactory singleton: DomainParticipant::finalize_participant_factory().

Once you have a DomainParticipantFactory, you can use it to perform the operations listed in Table 8.1
DomainParticipantFactory Operations. The most important one is create_participant(), described in Creating a DomainParticipant (Section 8.3.1 on page 556). For more details on all operations, see the API
Reference HTML documentation as well as the section of the manual listed in the Reference column.

540

8.2 DomainParticipantFactory

Table 8.1 DomainParticipantFactory Operations
Working
Operation
with ...
create_
participant

Reference

Creates a DomainParticipant.
Creating a DomainParticipant
(Section 8.3.1 on page 556)

create_
participant_
with_
profile

Creates a DomainParticipant based on a QoS profile.

delete_
participant

Deletes a DomainParticipant.

Deleting DomainParticipants
(Section 8.3.2 on page 558)

get_default_
participant_
qos

Gets the default QoS for DomainParticipants.

Getting and Setting Default QoS
for DomainParticipants (Section
8.2.2 on page 545)

get_
Domainparticipants
Participants

The
Factory’s
Instance

Description

Returns a sequence of pointers to all the DomainParticipants within the
DomainParticipantFactory.

lookup_
participant

Finds a specific DomainParticipant, based on a domain ID.

lookup_
participant_
by_name

Finds a specific DomainParticipant, based on a domain name.

set_default_
participant_
qos

Sets the default QoS for DomainParticipants.

set_default_
participant_
qos_
with_profile

Sets the default QoS for DomainParticipants based on a QoS profile.

get_instance

Gets the singleton instance of this class.

finalize_
instance

Destroys the singleton instance of this class.

Looking Up DomainParticipants
(Section 8.2.4 on page 546)

Getting and Setting Default QoS
for DomainParticipants (Section
8.2.2 on page 545)

Freeing Resources Used by the
DomainParticipantFactory
(Section 8.2.3 on page 546)

get_qos
The
Factory’s
Own QoS

Gets/sets the DomainParticipantFactory’s QoS.
set_qos
equals

Getting, Setting, and Comparing
QosPolicies (Section 4.1.7 on
page 158)

Compares two DomainParticipantFactory’s QoS structures for equality.

541

8.2 DomainParticipantFactory

Table 8.1 DomainParticipantFactory Operations
Working
Operation
with ...
set_thread_
factory

Description

Reference

Specifies a ThreadFactory implementation that DomainParticipants will use User-Managed Threads (Section
to create and delete all threads.
19.7 on page 844)
Frees all resources related to a thread.

Threads
unregister_
thread

This function is intended to be used at the end of any user-created threads
that invoke Connext DDS APIs (not all users will have this situation). The
best approach is to call it immediately before exiting such a thread, after all
Connext DDS APIs have been called.

542

8.2.1 Setting DomainParticipantFactory QosPolicies

Table 8.1 DomainParticipantFactory Operations
Working
Operation
with ...
get_default_
library

Profiles &
Libraries

Description

Reference

Gets the default library for a DomainParticipantFactory.

get_default_
profile

Gets the default QoS profile for a DomainParticipantFactory.

get_default_
profile_
library

Gets the library that contains the default QoS profile for a
DomainParticipantFactory.

get_
_
qos_from_
profile

Gets the  QoS values associated with a specified QoS profile.
 may be topic, datareader, datawriter, subscriber, publisher, or
participant.

Getting and Setting the
DomainParticipantFactory’s
Default QoS Profile and Library
(Section 8.2.1.1 on the facing
page)

Getting QoS Values from a QoS
Profile (Section 8.2.5 on page
547)

get_
_
qos_from_
profile_w_
topic_name

Like get__qos_from_profile(), but this operation allows you to
specify a topic name associated with the entity. The topic filter expressions
in the profile will be evaluated on the topic name.

get_qos_
profiles

Gets the names of all XML QoS profiles associated with a specified XML
QoS profile library.

Configuring QoS with XML
(Section 17.4 on page 803)

get_qos_
profile_
libraries

Gets the names of all XML QoS profile libraries associated with the
DomainParticipantFactory.

Retrieving a List of Available
Libraries (Section 17.10.1 on
page 823)

Explicitly loads or reloads the QoS profiles.

Loading, Reloading and
Unloading Profiles (Section
17.5.1 on page 811)

 may be topic, datareader, or datawriter.

load_profiles
reload_
profiles
set_default_
profile

Sets the default QoS profile for a DomainParticipantFactory.

set_default_
library

Sets the default library for a DomainParticipantFactory.

unload_
profiles

Frees the resources associated with loading QoS profiles.

Getting and Setting the
DomainParticipantFactory’s
Default QoS Profile and Library
(Section 8.2.1.1 on the facing
page)
Loading, Reloading and
Unloading Profiles (Section
17.5.1 on page 811)

8.2.1 Setting DomainParticipantFactory QosPolicies
The DDS_DomainParticipantFactoryQos structure has the following format:

543

8.2.1.1 Getting and Setting the DomainParticipantFactory’s Default QoS Profile and Library

struct DDS_DomainParticipantFactoryQos {
DDS_EntityFactoryQosPolicy
DDS_SystemResourceLimitsQosPolicy
DDS_ProfileQosPolicy
DDS_LoggingQosPolicy
};

entity_factory;
resource_limits;
profile;
logging;

For information on why you would want to change a particular QosPolicy, see the section referenced in
Table 8.2 DomainParticipantFactory QoS.
Table 8.2 DomainParticipantFactory QoS
QosPolicy

Description

EntityFactory

Controls whether or not child entities are created in the enabled state. See ENTITYFACTORY QosPolicy (Section 6.4.2
on page 315).

Logging

Configures the properties associated with Connext DDS logging. See LOGGING QosPolicy (DDS Extension) (Section
8.4.1 on page 572).

Profile

Configures the way that XML documents containing QoS profiles are loaded by RTI. See PROFILE QosPolicy (DDS
Extension) (Section 8.4.2 on page 573).

Configures DomainParticipant-independent resources used by Connext DDS. Mainly used to change the maximum
SystemResourcenumber of DomainParticipants that can be created within a single process (address space). See SYSTEM_
Limits
RESOURCE_LIMITS QoS Policy (DDS Extension) (Section 8.4.3 on page 575).

8.2.1.1 Getting and Setting the DomainParticipantFactory’s Default QoS Profile and Library
You can retrieve the default QoS profile for the DomainParticipantFactory with the get_default_profile()
operation. You can also get the default library for the DomainParticipantFactory, as well as the library that
contains the DomainParticipantFactory’s default profile (these are not necessarily the same library); these
operations are called get_default_library() and get_default_library_profile(), respectively. These operations are for informational purposes only (that is, you do not need to use them as a precursor to setting a
library or profile.) For more information, see Configuring QoS with XML (Section Chapter 17 on page
791).
virtual const char * get_default_library ()
const char * get_default_profile ()
const char * get_default_profile_library ()

There are also operations for setting the DomainParticipantFactory’s default library and profile:
DDS_ReturnCode_t set_default_library (const char * library_name)
DDS_ReturnCode_t set_default_profile (const char * library_name,
const char * profile_name)

544

8.2.2 Getting and Setting Default QoS for DomainParticipants

set_default_profile() specifies the profile that will be used as the default the next time a default
DomainParticipantFactory profile is needed during a call to a DomainParticipantFactory operation.
When calling a DomainParticipantFactory operation that requires a profile_name parameter, you can use
NULL to refer to the default profile. (This same information applies to setting a default library.)
set_default_profile() does not set the default QoS for the DomainParticipant that can be created by the
DomainParticipantFactory. To set the default QoS using a profile, use the DomainParticipantFactory’s
set_default_participant_qos_with_profile() operation (see Getting and Setting Default QoS for
DomainParticipants (Section 8.2.2 below)).

8.2.2 Getting and Setting Default QoS for DomainParticipants
To get the default QoS that will be used for creating DomainParticipants if create_participant() is called
with DDS_PARTICIPANT_QOS_DEFAULT as the qos parameter, use this DomainParticipantFactory
operation:
DDS_ReturnCode_t get_default_participant_qos (DDS_DomainParticipantQos & qos)

This operation gets the QoS settings that were specified on the last successful call to set_default_participant_qos() or set_default_participant_qos_with_profile(), or if the call was never made, the default
values listed in DDS_DomainParticipantQos.
To set the default QoS that will be used for new DomainParticipants, use the following operations. Then
these default QoS will be used if create_participant() is called with DDS_PARTICIPANT_QOS_
DEFAULT as the ‘qos’ parameter.
DDS_ReturnCode_t set_default_participant_qos (
const DDS_DomainParticipantQos &qos)

or
DDS_ReturnCode_t set_default_participant_qos_with_profile (
const char *library_name, const char *profile_name)

Notes:
l

l

These operations may potentially allocate memory, depending on the sequences contained in some
QoS policies.
It is not safe to set the default DomainParticipant QoS values while another thread may be simultaneously calling get_default_participant_qos(), set_default_participant_qos(), or create_participant() with DDS_PARTICIPANT_QOS_DEFAULT as the qos parameter. It is also not safe to

545

8.2.3 Freeing Resources Used by the DomainParticipantFactory

get the default DomainParticipant QoS values while another thread may be simultaneously calling
set_default_participant_qos().

8.2.3 Freeing Resources Used by the DomainParticipantFactory
The finalize_instance() operation explicitly reclaims resources used by the participant factory singleton
(including resources use for QoS profiles).
On many operating systems, these resources are automatically reclaimed by the OS when the program terminates. However, some memory-check tools will flag those resources as unreclaimed. This method
provides a way to clean up all the memory used by the participant factory.
Before calling finalize_instance() on a DomainParticipantFactory, all of the participants created by the
factory must have been deleted. For a DomainParticipant to be successfully deleted, all Entities created by
the participant or by the Entities that the participant created must have been deleted. In essence, the
DomainParticipantFactory cannot be deleted until all other Entities have been deleted in an application.
Except for Linux systems: get_instance() and finalize_instance() are UNSAFE on the FIRST call. It is
not safe for two threads to simultaneously make the first call to get or finalize the factory instance. Subsequent calls are thread safe.

8.2.4 Looking Up DomainParticipants
The DomainParticipantFactory has these useful operations for retrieving its DomainParticipants:
l

get_participants() returns a sequence of pointers to all the DomainParticipants within the
DomainParticipantFactory.
DDS_ReturnCode_t
get_participants (DDSDomainParticipantSeq & participants)

l

lookup_participant() locates an existing DomainParticipant based on its domain ID.
DDSDomainParticipant *
lookup_participant (DDS_DomainId_t domainId)

l

lookup_participant_by_name () locates an existing DomainParticipant based on its name.
DDSDomainParticipant *
lookup_participant_by_name(const char * participant_name)

Note: in the Modern C++ API these operations are freestanding functions rti::domain::find_participants(),
dds::domain::find(), and rti::domain::find_participant_by_name() respectively.

546

8.2.5 Getting QoS Values from a QoS Profile

8.2.5 Getting QoS Values from a QoS Profile
A QoS Profile may include configuration settings for all types of Entities. If you just want the settings for a
specific type of Entity, call get__qos_from_profile() (where  may be participant, publisher, subscriber, datawriter, datareader, or topic). This is useful if you want to get the QoS values
from the profile in a structure, make some changes, and then use that structure to create an entity.
DDS_ReturnCode_t get__qos_from_profile (
DDS_Qos &qos,
const char *library_name,
const char *profile_name)

For an example, see Getting QoS Values from a Profile, Changing QoS Values, Creating a Publisher with
Modified QoS Values (Section Figure 6.5 on page 254).
The get__qos_from_profile() operations do not take into account the topic_filter attributes that
may be set for DataWriter, DataReader, or Topic QoSs in profiles (see Topic Filters (Section 17.3.4 on
page 799)). If there is a topic name associated with an entity, you can call get__qos_from_profile_w_topic_name() (where  can be datawriter, datareader, or topic) and the topic filter expressions in the profile will be evaluated on the topic name.
DDS_ReturnCode_t get__qos_from_profile_w_topic_name(
DDS_Qos &qos,
const char *library_name,
const char *profile_name,
const char *topic_name)

get__qos_from_profile() and get__qos_from_profile_w_topic_name() may allocate
memory, depending on the sequences contained in some QoS policies.
Note: in the Modern C++ API, the class QosProvider provides the functionality described in thi section.
Please see the API Reference HTML documentation: Modules, RTI Connext DDS API Reference, Configuring QoS Profiles with XML, QosProvider.

8.3 DomainParticipants
A DomainParticipant is a container for Entity objects that all belong to the same DDS domain. Each
DomainParticipant has its own set of internal threads and internal data structures that maintain information
about the Entities created by itself and other DomainParticipants in the same DDS domain. A DomainParticipant is used to create and destroy Publishers, Subscribers and Topics.
Once you have a DomainParticipant, you can use it to perform the operations listed in Table 8.3
DomainParticipant Operations. For more details on all operations, see the API Reference HTML documentation. Some of the first operations you’ll be interested in are create_topic(), create_subscriber(),
and create_publisher().

547

8.3 DomainParticipants

Note: Some operations cannot be used within a listener callback, see Restricted Operations in Listener
Callbacks (Section 4.5.1 on page 185).
Table 8.3 DomainParticipant Operations
Working
with ...
Builtin
Subscriber

Operation
get_builtin_
subscriber

Description

Returns the builtin Subscriber.

Reference
Built-in DataReaders (Section
16.2 on page 773)

548

8.3 DomainParticipants

Table 8.3 DomainParticipant Operations
Working
with ...

DomainParticipants

Operation

Description

Reference

add_peer

Adds an entry to the peer list.

Adding and Removing Peers
List Entries (Section 8.5.2.3 on
page 581)

enable

Enables the DomainParticipant.

Enabling DDS Entities (Section
4.1.2 on page 154)

equals

Compares two DomainParticipant’s QoS structures for equality.

Comparing QoS Values
(Section 8.3.6.2 on page 565)

get_discovered_
participant_data

Provides the ParticipantBuiltinTopicData for a discovered
DomainParticipant.

get_discovered_
participants

Provides a list of DomainParticipants that have been discovered.

Learning about Discovered
DomainParticipants (Section
8.3.11 on page 571)

get_domain_id

Gets the domain ID of the DomainParticipant.

Choosing a Domain ID and
Creating Multiple DDS
Domains (Section 8.3.4 on page
559)

get_listener

Gets the currently installed DomainParticipantListener.

Setting Up
DomainParticipantListeners
(Section 8.3.5 on page 560)

get_qos

Gets the DomainParticipant QoS.

Setting DomainParticipant
QosPolicies (Section 8.3.6 on
page 562)

ignore_participant

Rejects the connection to a remote DomainParticipant.

Restricting Communication—
Ignoring Entities (Section 16.4
on page 784)

remove_peer

Removes an entry from the peer list.

Adding and Removing Peers
List Entries (Section 8.5.2.3 on
page 581)

set_listener

Replaces the DomainParticipantListener.

Setting Up
DomainParticipantListeners
(Section 8.3.5 on page 560)

set_qos

Sets the DomainParticipant QoS.

set_qos_with_
profile

Sets the DomainParticipant QoS based on a QoS profile.

Setting DomainParticipant
QosPolicies (Section 8.3.6 on
page 562)

549

8.3 DomainParticipants

Table 8.3 DomainParticipant Operations
Working
with ...

Operation

Description

Reference

create_
contentfilteredtopic
Creates a ContentFilteredTopic that can be used to process content- Creating ContentFilteredTopics
based subscriptions.
(Section 5.4.3 on page 214)
create_
contentfilteredtopic_
with_filter
ContentFilteredTopics

delete_
contentfilteredtopic

Deletes a ContentFilteredTopic.

Deleting ContentFilteredTopics
(Section 5.4.4 on page 218)

register_
contentfilter

Registers a new content filter.

Registering a Custom Filter
(Section 5.4.8.2 on page 234)

unregister_
contentfilter

Unregisters a new content filter.

Unregistering a Custom Filter
(Section 5.4.8.3 on page 236)

lookup_contentfilter Gets a previously registered content filter.

create_datareader

DataReaders

Retrieving a ContentFilter
(Section 5.4.8.4 on page 237)

Creates a DataReader with a given DataReaderListener, and an
implicit Subscriber.

create_datareader_
with_
profile

Creates a DataReader based on a QoS profile, with a given
DataReaderListener, and an implicit Subscriber.

delete_datareader

Deletes a DataReader that belongs to the ‘implicit Subscriber.’

get_default_
datareader_qos

Copies the default DataReaderQoS values into the provided
structure.

Creating DataReaders (Section
7.3.1 on page 463)

Deleting DataReaders (Section
7.3.3 on page 466)

ignore_subscription Rejects the connection to a DataReader
set_default_
datareader_qos

Sets the default DataReaderQos values.

set_default_
datareader_
qos_with_profile

Sets the default DataReaderQos using values from a QoS profile.

Getting and Setting Default QoS
for Child Entities (Section
8.3.6.5 on page 568)

550

8.3 DomainParticipants

Table 8.3 DomainParticipant Operations
Working
with ...

Operation

create_datawriter

DataWriters

Description

Reference

Creates a DataWriter with a given DataWriterListener, and an
implicit Publisher.
Creating Publishers (Section
6.2.2 on page 249)

create_datawriter_
with_
profile

Creates a DataWriter based on a QoS profile, with a given
DataWriterListener, and an implicit Publisher.

delete_datawriter

Deletes a DataWriter that belongs to the ‘implicit Publisher.’

Deleting Publishers (Section
6.2.3 on page 250)

ignore_publication

Rejects the connection to a DataWriter.

Restricting Communication—
Ignoring Entities (Section 16.4
on page 784)

get_default_
datawriter_qos

Copies the default DataWriterQos values into the provided
DataWriterQos structure.

set_default_
datawriter_qos

Sets the default DataWriterQoS values.

set_default_
datawriter_
qos _with_profile

Sets the default DataWriterQos using values from a profile.

Getting and Setting Default QoS
for Child Entities (Section
8.3.6.5 on page 568)

551

8.3 DomainParticipants

Table 8.3 DomainParticipant Operations
Working
with ...

Publishers

Operation

Description

Reference

create_publisher

Creates a Publisher and a PublisherListener.

create_publisher_
with_
profile

Creates a Publisher based on a QoS profile, and a
PublisherListener.

delete_publisher

Deletes a Publisher.

Deleting Publishers (Section
6.2.3 on page 250)

get_default_
publisher_qos

Copies the default PublisherQos values into the provided
PublisherQos structure.

Getting and Setting Default QoS
for Child Entities (Section
8.3.6.5 on page 568)

get_implicit_
publisher

Gets the Publisher that is implicitly created by the
DomainParticipant.

Getting the Implicit Publisher or
Subscriber (Section 8.3.9 on
page 569)

get_publishers

Provides a list of all Publishers owned by the DomainParticipant.

Getting All Publishers and
Subscribers (Section 8.3.13.3
on page 572)

set_default_
publisher_qos

Sets the default PublisherQos values.

set_default_
publisher_qos_
with_profile

Sets the default PublisherQos using values from a QoS profile.

Creating Publishers (Section
6.2.2 on page 249)

Getting and Setting Default QoS
for Child Entities (Section
8.3.6.5 on page 568)

552

8.3 DomainParticipants

Table 8.3 DomainParticipant Operations
Working
with ...

Subscribers

Operation

Description

create_subscriber

Creates a Subscriber and a SubscriberListener.

create_subscriber_
with_
profile

Creates a Subscriber based on a QoS profile, and a
SubscriberListener.

delete_subscriber

Deletes a Subscriber.

Deleting Subscribers (Section
7.2.3 on page 446)

get_default_
subscriber_qos

Copies the default SubscriberQos values into the provided
SubscriberQos structure.

Getting and Setting Default QoS
for Child Entities (Section
8.3.6.5 on page 568)

get_implicit_
subscriber

Gets the Subscriber that is implicitly created by the
DomainParticipant.

Getting the Implicit Publisher or
Subscriber (Section 8.3.9 on
page 569)

get_subscribers

Provides a list of all Subscribers owned by the
DomainParticipant.

Getting All Publishers and
Subscribers (Section 8.3.13.3
on page 572)

set_default_
subscriber_qos

Sets the default SubscriberQos values.

set_default_
subscriber_qos_
with_profile

Sets the default SubscriberQos values using values from a QoS
profile.

delete_durable_
subscription

Deletes an existing Durable Subscription. The quorum of the
existing DDS samples will be considered satisfied.
Creates a Durable Subscription that will receive all DDS samples
published on a Topic, including those published while a
DataReader is inactive or before it may be created.

Durable
Subscriptions

Reference

register_durable_
subscription

RTI Persistence Service will ensure that all the DDS samples on
that Topic are retained until they are acknowledged by at least N
DataReaders belonging to the Durable Subscription, where N is
the quorum count.

Creating Subscribers (Section
7.2.2 on page 445)

Getting and Setting Default QoS
for Child Entities (Section
8.3.6.5 on page 568)

Configuring Durable
Subscriptions in Persistence
Service (Section 27.9 on page
955)

If the same Durable Subscription is created on a different Topic,
RTI Persistence Service will implicitly delete the previous Durable
Subscription and create a new one on the new Topic.

553

8.3 DomainParticipants

Table 8.3 DomainParticipant Operations
Working
with ...

Topics

Operation

Description

Reference

create_topic

Creates a Topic and a TopicListener.

create_topic _with_
profile

Creates a Topic based on a QoS profile, and a TopicListener.

delete_topic

Deletes a Topic.

get_default_topic_
qos

Copies the default TopicQos values into the provided TopicQos
structure.

get_discovered_
topic_data

Retrieves the BuiltinTopicData for a discovered Topic.

get_discovered_
topics

Returns a list of all (non-ignored) discovered Topics.

ignore_topic

Rejects a remote topic.

Restricting Communication—
Ignoring Entities (Section 16.4
on page 784)

lookup_
topicdescription

Gets an existing locally-created TopicDescription (Topic).

Looking up Topic Descriptions
(Section 8.3.7 on page 568)

set_default_topic_
qos

Sets the default TopicQos values.

set_default_topic_
qos_with_profile

Sets the default TopicQos values using values from a profile.

find_topic

Finds an existing Topic, based on its name.

Creating Topics (Section 5.1.1
on page 202)

Getting and Setting Default QoS
for Child Entities (Section
8.3.6.5 on page 568)

Learning about Discovered
Topics (Section 8.3.12 on page
571)

Getting and Setting Default QoS
for Child Entities (Section
8.3.6.5 on page 568)

Finding a Topic (Section 8.3.8
on page 569)

554

8.3 DomainParticipants

Table 8.3 DomainParticipant Operations
Working
with ...

FlowControllers

Operation

Description

create_
flowcontroller

Creates a custom FlowController object.

delete_
flowcontroller

Deletes a custom FlowController object.

get_default_
flowcontroller_
property

Gets the default properties used when a new FlowController is
created.

set_default_
flowcontroller_
property

Sets the default properties used when a new FlowController is
created.

lookup_
flowcontroller

Finds a FlowController, based on its name.

get_default_library

Gets the default library.

get_default_profile

Gets the default profile.

Libraries and get_default_profile_
Gets the library that contains the default profile.
Profiles
library
set_default_profile

Sets the default QoS profile.

set_default_library

Sets the default library.

create_multitopic

Creates a MultiTopic that can be used to subscribe to multiple
topics and combine/filter the received data into a resulting type.

MultiTopics
delete_multitopic

Reference

Creating and Deleting
FlowControllers (Section 6.6.6
on page 433)

Getting/Setting Default
FlowController Properties
(Section 6.6.7 on page 434)

Other FlowController
Operations (Section 6.6.10 on
page 435)

Getting and Setting
DomainParticipant’s Default
QoS Profile and Library
(Section 8.3.6.4 on page 567)

Currently not supported.

Deletes a MultiTopic.

555

8.3.1 Creating a DomainParticipant

Table 8.3 DomainParticipant Operations
Working
with ...

Other

Operation

Description

Reference

assert_liveliness

Manually asserts the liveliness of this DomainParticipant.

Getting the Implicit Publisher or
Subscriber (Section 8.3.9 on
page 569)

delete_contained_
entities

Recursively deletes all the entities that were created using the
"create" operations on the DomainParticipant and its children.

Deleting Contained Entities
(Section 8.3.3 on page 559)

contains_entity

Confirms if an entity belongs to the DomainParticipant or not.

Verifying Entity Containment
(Section 8.3.13.1 on page 571)

get_current_time

Gets the current time used by Connext DDS.

Getting the Current Time
(Section 8.3.13.2 on page 571)

get_status_changes

Gets a list of statuses that have changed since the last time the
application read the status or the Listeners were called.

Getting Status and Status
Changes (Section 4.1.4 on page
157)

8.3.1 Creating a DomainParticipant
Typically, you will only need to create one DomainParticipant per DDS domain per application.
(Although unusual, you can create multiple DomainParticipants for the same DDS domain in an application.)
To create a DomainParticipant, use the DomainParticipantFactory’s create_participant() or create_participant_with_profile() operation:
A QoS profile is way to use QoS settings from an XML file or string. With this approach, you can change
QoS settings without recompiling the application. For details, see Configuring QoS with XML (Section
Chapter 17 on page 791).
Note: In the Modern C++ API, you will use the DomainParticipant constructors.
DDSDomainParticipant * create_participant(
DDS_DomainId_t domainId,
const DDS_DomainParticipantQos &qos,
DDSDomainParticipantListener *listener,
DDS_StatusMask mask)
DDSDomainParticipant * create_participant_with_profile (
DDS_DomainId_t domainId,
const char * library_name,
const char *profile_name,
DDSDomainParticipantListener *listener,
DDS_StatusMask mask)

556

8.3.1 Creating a DomainParticipant

Where:
domainId

The domain ID uniquely identifies the DDS domain that the DomainParticipant is in. It controls
with which other DomainParticipants it will communicate. See Choosing a Domain ID and
Creating Multiple DDS Domains (Section 8.3.4 on page 559) for more information on domain
IDs.

qos

If you want the default QoS settings (described in the API Reference HTML documentation),
use DDS_PARTICIPANT_QOS_DEFAULT for this parameter (see Creating a
DomainParticipant with Default QosPolicies (Section Figure 8.4 on the facing page)). If you
want to customize any of the QosPolicies, supply a DomainParticipantQos structure that is
described in Setting DomainParticipant QosPolicies (Section 8.3.6 on page 562).

Note: If you use DDS_PARTICIPANT_QOS_DEFAULT, it is not safe to create the
DomainParticipant while another thread may simultaneously be calling the
DomainParticipantFactory’s set_default_participant_qos() operation.
listener

Listeners are callback routines. Connext DDS uses them to notify your application of specific
events (status changes) that may occur. The listener parameter may be set to NULL if you do
not want to install a Listener. The DomainParticipant’s Listener is a catchall for all of the
events of all of its Entities. If an event is not handled by an Entity ’s Listener, then the
DomainParticipantListener may be called in response to the event. For more information, see
Setting Up DomainParticipantListeners (Section 8.3.5 on page 560).

mask

This bit mask indicates which status changes will cause the Listener to be invoked. The bits set
in the mask must have corresponding callbacks implemented in the Listener. If you use NULL
for the Listener, use DDS_STATUS_MASK_NONE for this parameter. If the Listener
implements all callbacks, use DDS_STATUS_MASK_ALL. For information on statuses, see
Listeners (Section 4.4 on page 177).

library_name A QoS Library is a named set of QoS profiles. See URL Groups (Section 17.8 on page 814).
profile_name

A QoS profile groups a set of related QoS, usually one per entity. See URL Groups (Section
17.8 on page 814).
After you create a DomainParticipant, the next step is to register the data types that will be used
by the application, see Using RTI Code Generator (rtiddsgen) (Section 3.6 on page 138). Then
you will need to create the Topics that the application will publish and/or subscribe, see Creating
Topics (Section 5.1.1 on page 202). Finally, you will use the DomainParticipant to create
Publishers and/or Subscribers, see Creating Publishers (Section 6.2.2 on page 249) and
Creating Subscribers (Section 7.2.2 on page 445).

Note: It is not safe to create one DomainParticipant while another thread may simultaneously
be looking up (Looking Up DomainParticipants (Section 8.2.4 on page 546)) or deleting
(Deleting DomainParticipants (Section 8.3.2 on the facing page)) the same
DomainParticipant.
For more examples, see Configuring QoS Settings when DomainParticipant is Created
(Section 8.3.6.1 on page 564).

557

8.3.2 Deleting DomainParticipants

Figure 8.4 Creating a DomainParticipant with Default QosPolicies
DDS_DomainId_t domain_id = 10;
// MyDomainParticipantListener is user defined and
// extends DDSDomainParticipantListener
MyDomainParticipantListener* participant_listener =
new MyDomainParticipantListener(); // or = NULL
// Create the participant
DDSDomainParticipant* participant = factory->create_participant(
domain_id, DDS_PARTICIPANT_QOS_DEFAULT,
participant_listener, DDS_STATUS_MASK_ALL);
if (participant == NULL) {
// ... error
};

8.3.2 Deleting DomainParticipants
If the application is no longer interested in communicating in a certain DDS domain, the DomainParticipant can be deleted. A DomainParticipant can be deleted only after all the Entities that were created by
the DomainParticipant have been deleted (see Deleting Contained Entities (Section 8.3.3 on the next
page)).
To delete a DomainParticipant:
You must first delete all Entities (Publishers, Subscribers, ContentFilteredTopics, and Topics) that were
created with the DomainParticipant. Use the DomainParticipant’s delete_() operations to delete
them one at a time, or use the delete_contained_entities() operation (Deleting Contained Entities (Section
8.3.3 on the next page)) to delete them all at the same time.
DDS_ReturnCode_t delete_publisher (DDSPublisher *p)
DDS_ReturnCode_t delete_subscriber (DDSSubscriber *s)
DDS_ReturnCode_t delete_contentfilteredtopic
(DDSContentFilteredTopic *a_contentfilteredtopic)
DDS_ReturnCode_t delete_topic (DDSTopic *topic)

Delete the DomainParticipant by using the DomainParticipantFactory’s delete_participant() operation.
DDS_ReturnCode_t delete_participant
(DDSDomainParticipant *a_participant)

Note: A DomainParticipant cannot be deleted within its Listener callback, see Restricted Operations in
Listener Callbacks (Section 4.5.1 on page 185).
After a DomainParticipant has been deleted, all of the participant’s internal Connext DDS threads and
allocated memory will have been deleted. You should delete the DomainParticipantListener only after the
DomainParticipant itself has been deleted.
Note: In the Modern C++ API, Entities are automatically destroyed.

558

8.3.3 Deleting Contained Entities

8.3.3 Deleting Contained Entities
The DomainParticipant’s delete_contained_entities() operation deletes all the Publishers (including an
implicitly created one, if it exists), Subscribers (including an implicitly created one, if it exists), ContentFilteredTopics, MultiTopics, and Topics that have been created by the DomainParticipant.
DDS_ReturnCode_t delete_contained_entities( )

Prior to deleting each contained entity, this operation recursively calls the corresponding delete_contained_entities() operation on each contained entity (if applicable). This pattern is applied recursively.
Therefore, delete_contained_entities() on the DomainParticipant will end up deleting all the entities
recursively contained in the DomainParticipant, that is also the DataWriter, DataReader, as well as the
QueryCondition and ReadCondition objects belonging to the contained DataReader.
If delete_contained_entities() returns successfully, the application may delete the DomainParticipant
knowing that it has no contained entities (see Deleting DomainParticipants (Section 8.3.2 on the previous
page)).

8.3.4 Choosing a Domain ID and Creating Multiple DDS Domains
A domain ID identifies the DDS domain in which the DomainParticipant is communicating. DomainParticipants with the same domain ID are on the same communication “channel”. DomainParticipants with
different domain IDs are completely isolated from each other.
The domain ID is a purely arbitrary value; you can use any integer 0 or higher, provided it does not violate
the guidelines for the DDS_RtpsWellKnownPorts_t structure (Ports Used for Discovery (Section 8.5.9.3
on page 613)). Domain IDs are typically between 0 and 232. Please see the API Reference HTML documentation for the DDS_RtpsWellKnownPorts_t structure and in particular, DDS_INTEROPERABLE_
RTPS_WELL_KNOWN_PORTS.
Most distributed systems can use a single DDS domain for all of its applications. Thus a single domain ID
is sufficient. Some systems may need to logically partition nodes to prevent them from communicating
with each other directly, and thus will need to use multiple DDS domains. However, even in systems that
only use a single DDS domain, during the testing and development phases, one may want to assign different users/testers different domain IDs for running their applications so that their tests do not interfere
with each other.
To run multiple applications on the same node with the same domain ID, Connext DDS uses a participant
ID to distinguish between the different DomainParticipants in the different applications. The participant
ID is simply an integer value that must be unique across all DomainParticipants created on the same node
that use the same domain ID. The participant_id is part of the WIRE_PROTOCOL QosPolicy (DDS
Extension) (Section 8.5.9 on page 610).
Although usually those DomainParticipants have been created in different applications, the same application can also create multiple DomainParticipants with the same domain ID. For optimal results, the par-

559

8.3.5 Setting Up DomainParticipantListeners

ticipant_id should be assigned sequentially to the different DomainParticipants, starting from the default
value of 0.
Once you have a DomainParticipant, you can retrieve its domain ID with the get_domain_id() operation.
The domain ID and participant ID are mapped to port numbers that are used by transports for discovery
traffic. For information on how port numbers are calculated, see Ports Used for Discovery (Section 14.5
on page 738). How DomainParticipants discover each other is discussed in Discovery (Section Chapter
14 on page 709).

8.3.5 Setting Up DomainParticipantListeners
DomainParticipants may optionally have Listeners. Listeners are essentially callback routines and are how
Connext DDS will notify your application of specific events (changes in status) for entities Topics, Publishers, Subscribers, DataWriters, and DataReaders. Each Entity may have a Listener installed and
enabled to process the events for itself and all of the sub-Entities created from it. If an Entity does not have
a Listener installed or is not enabled to listen for a particular event, then Connext DDS will propagate the
event to the Entity’s parent. If the parent Entity does not process the event, Connext DDS will continue to
propagate the event up the object hierarchy until either a Listener is invoked or the event is dropped.
The DomainParticipantListener is the last chance that an event can be processed for the Entities descended from a DomainParticipant. The DomainParticipantListener is used only if an event is not handled
by any of the Entities contained by the participant.
A Listener is typically set up when the DomainParticipant is created (see Creating a DomainParticipant
(Section 8.3.1 on page 556)). You can also set one up after creation time by using the set_listener() operation, as illustrated in Setting up DomainParticipantListener (Section Figure 8.5 below). The get_listener()
operation can be used to retrieve the current DomainParticipantListener.
Figure 8.5 Setting up DomainParticipantListener
// MyDomainParticipantListener only handles PUBLICATION_MATCHED and
// SUBSCRIPTION_MATCHED status for DomainParticipant Entities
class MyDomainParticipantListener :
public DDSDomainParticipantListener {
public:
virtual void on_publication_matched(DDSDataWriter *writer,
const DDS_PublicationMatchedStatus &status);
virtual void on_subscription_matched(DDSDataReader *reader,
const DDS_SubscriptionMatchedStatus &status);
};
void MyDomainParticipantListener::on_publication_matched(
DDSDataWriter *writer,
const DDS_PublicationMatchedStatus &status)
{
const char *name = writer->get_topic()->get_name();
printf(“Number of matching DataReaders for Topic %s is %d\n”,
name, status.current_count);

560

8.3.5 Setting Up DomainParticipantListeners

};
void MyDomainParticipantListener::on_subscription_matched(
DDSDataReader *reader,
const DDS_SubscriptionMatchedStatus &status)
{
const char *name =
reader->get_topicdescription()->get_name();
printf(“Number of matching DataWriters for Topic %s is %d\n”,
name, status.current_count);
};
// Set up participant listener
MyDomainParticipantListener* participant_listener =
new MyDomainParticipantListener();
if (participant_listener == NULL) {
// ... handle error
}
// Create the participant with a listener
DDSDomainParticipant* participant = factory->create_participant(
domain_id, participant_qos, participant_listener,
DDS_PUBLICATION_MATCHED_STATUS |
DDS_SUBSCRIPTION_MATCHED_STATUS );
if (participant == NULL) {
// ... handle error
}

If a Listener is set for a DomainParticipant, the Listener needs to exist as long as the DomainParticipant
exists. It is unsafe to destroy the Listener while it is attached to a participant. However, you may remove
the DomainParticipantListener from a DomainParticipant by calling set_listener() with a NULL value.
Once the Listener has been removed from the participant, you may safely destroy it (see Types of Listeners (Section 4.4.1 on page 177)).
Notes:
l

l

Due to a thread-safety issue, the destruction of a DomainParticipantListener from an enabled
DomainParticipant should be avoided—even if the DomainParticipantListener has been removed
from the DomainParticipant. (This limitation does not affect the Java API.)
It is possible for multiple internal Connext DDS threads to call the same method of a DomainParticipantListener simultaneously. You must write the methods of a DomainParticipantListener to be
multithread safe and reentrant. The methods of the Listener of other Entities do not have this constraint and are guaranteed to have single threaded access.

See also:
l

Setting Up TopicListeners (Section 5.1.5 on page 208)

l

Setting Up PublisherListeners (Section 6.2.5 on page 257)

561

8.3.6 Setting DomainParticipant QosPolicies

l

Setting Up DataWriterListeners (Section 6.3.4 on page 269)

l

Setting Up SubscriberListeners (Section 7.2.6 on page 454)

l

Setting Up DataReaderListeners (Section 7.3.4 on page 466)

8.3.6 Setting DomainParticipant QosPolicies
A DomainParticipant’s QosPolicies are used to configure discovery, database sizing, threads, information
sent to other DomainParticipants, and the behavior of the DomainParticipant when acting as a factory for
other Entities.
Note: set_qos() cannot always be used in a listener callback; see Restricted Operations in Listener Callbacks (Section 4.5.1 on page 185).
The DDS_DomainParticipantQos structure has the following format:
struct DDS_DomainParticipantQos {
DDS_UserDataQosPolicy
user_data;
DDS_EntityFactoryQosPolicy
entity_factory;
DDS_WireProtocolQosPolicy
wire_protocol;
DDS_TransportBuiltinQosPolicy
transport_builtin;
DDS_TransportUnicastQosPolicy
default_unicast;
DDS_DiscoveryQosPolicy
discovery;
DDS_DomainParticipantResourceLimitsQosPolicy resource_limits;
DDS_EventQosPolicy
event;
DDS_ReceiverPoolQosPolicy
receiver_pool;
DDS_DatabaseQosPolicy
database;
DDS_DiscoveryConfigQosPolicy
discovery_config;
DDS_PropertyQosPolicy
property;
DDS_EntityNameQosPolicy
participant_name;
DDS_TransportMulticastMappingQosPolicy multicast_mapping;
DDS_TypeSupportQosPolicy
type_support;
};

Table 8.4 DomainParticipant QosPolicies summarizes the meaning of each policy (listed alphabetically).
For information on why you would want to change a particular QosPolicy, see the section referenced in
the table.
Table 8.4 DomainParticipant QosPolicies
QosPolicy

Description

Database

Various settings and resource limits used by Connext DDS to control its internal database. See
DATABASE QosPolicy (DDS Extension) (Section 8.5.1 on page 577).

Discovery

Configures the mechanism used by Connext DDS to automatically discover and connect with new
remote applications. See DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580).

562

8.3.6 Setting DomainParticipant QosPolicies

Table 8.4 DomainParticipant QosPolicies
QosPolicy
DiscoveryConfig

Description
Controls the amount of delay in discovering entities in the system and the amount of discovery traffic in
the network. See DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585).

Various settings that configure how DomainParticipants allocate and use physical memory for internal
DomainParticipantResourceLimits resources, including the maximum sizes of various properties. See DOMAIN_PARTICIPANT_
RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593).
EntityFactory

Controls whether or not child entities are created in the enabled state. See ENTITYFACTORY
QosPolicy (Section 6.4.2 on page 315).

EntityName

Assigns a name to a DomainParticipant. See ENTITY_NAME QosPolicy (DDS Extension) (Section
6.5.9 on page 374).

Event

Configures the DomainParticipant’s internal thread that handles timed events. See EVENT QosPolicy
(DDS Extension) (Section 8.5.5 on page 602).

Property

Stores name/value(string) pairs that can be used to configure certain parameters of Connext DDS that
are not exposed through formal QoS policies. It can also be used to store and propagate applicationspecific name/value pairs, which can be retrieved by user code during discovery. See PROPERTY
QosPolicy (DDS Extension) (Section 6.5.17 on page 394).

ReceiverPool

Configures threads used by Connext DDS to receive and process data from transports (for example,
UDP sockets). See RECEIVER_POOL QosPolicy (DDS Extension) (Section 8.5.6 on page 604).

TransportBuiltin

Specifies which built-in transport plugins are used. See TRANSPORT_BUILTIN QosPolicy (DDS
Extension) (Section 8.5.7 on page 606).

TransportMulticastMapping

Specifies the automatic mapping between a list of topic expressions and multicast address that can be
used by a DataReader to receive data for a specific topic. See TRANSPORT_MULTICAST_
MAPPING QosPolicy (DDS Extension) (Section 8.5.8 on page 608).

TransportUnicast

Specifies a subset of transports and port number that can be used by an Entity to receive data. See
TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412).

TypeSupport

Used to attach application-specific value(s) to a DataWriter or DataReader. These values are passed to
the serialization or deserialization routine of the associated data type. See TYPESUPPORT QosPolicy
(DDS Extension) (Section 6.5.25 on page 416).

UserData

Along with Topic Data QosPolicy and Group Data QosPolicy, used to attach a buffer of bytes to
Connext DDS's discovery meta-data. See USER_DATA QosPolicy (Section 6.5.26 on page 417).

WireProtocol

Specifies IDs used by the RTPS wire protocol to create globally unique identifiers. See WIRE_
PROTOCOL QosPolicy (DDS Extension) (Section 8.5.9 on page 610).

563

8.3.6.1 Configuring QoS Settings when DomainParticipant is Created

8.3.6.1 Configuring QoS Settings when DomainParticipant is Created
As described in Creating a DomainParticipant (Section 8.3.1 on page 556), there are different ways to create a DomainParticipant, depending on how you want to specify its QoS (with or without a QoS Profile).
l

l

l

l

Figure 8.4 Creating a DomainParticipant with Default QosPolicies on page 558 has an example of
how to create a DomainParticipant with default QosPolicies by using the special constant, DDS_
PARTICIPANT_QOS_DEFAULT, which indicates that the default QoS values for a DomainParticipant should be used. The default DomainParticipant QoS values are configured in the
DomainParticipantFactory; you can change them with set_default_participant_qos() or set_
default_participant_qos_with_profile() (see Getting and Setting Default QoS for DomainParticipants (Section 8.2.2 on page 545)). Then any DomainParticipants created with the DomainParticipantFactory will use the new default values. As described in Getting, Setting, and Comparing
QosPolicies (Section 4.1.7 on page 158), this is a general pattern that applies to the construction of
all Entities.
To create a DomainParticipant with non-default QoS without using a QoS Profile, see the example
code in Figure 8.6 Creating DomainParticipant with Modified QosPolicies (not from profile) below.
It uses the DomainParticipantFactory’s get_default_participant_qos() method to initialize a
DDS_ParticipantQos structure. Then, the policies are modified from their default values before the
structure is used in the create_participant() method.
You can also create a DomainParticipant and specify its QoS settings via a QoS Profile. To do so,
you will call create_participant_with_profile(), as seen in Figure 8.7 Creating DomainParticipant
with QoS Profile on the next page.
If you want to use a QoS profile, but then make some changes to the QoS before creating the
DomainParticipant, call get_participant_qos_from_profile() and create_participant() as seen in
Figure 8.8 Getting QoS from Profile, Creating DomainParticipant with Modified QoS Values on the
next page.

For more information, see Creating a DomainParticipant (Section 8.3.1 on page 556) and Configuring
QoS with XML (Section Chapter 17 on page 791).
Figure 8.6 Creating DomainParticipant with Modified QosPolicies (not from profile)
DDS_DomainId_t domain_id = 10;
DDS_DomainParticipantQos participant_qos;1
// initialize participant_qos with default values
factory->get_default_participant_qos(participant_qos);
// make QoS changes here
participant_qos.wire_protocol.participant_id = 2;

1In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
564

8.3.6.2 Comparing QoS Values

// Create the participant with modified qos
DDSDomainParticipant* participant = factory->create_participant(
domain_id, participant_qos, NULL, DDS_STATUS_MASK_NONE);
if (participant == NULL) {
// ... error
}

Figure 8.7 Creating DomainParticipant with QoS Profile
DDS_DomainId_t domain_id = 10;
// MyDomainParticipantListener is user defined and
// extends DDSDomainParticipantListener
MyDomainParticipantListener* participant_listener
= new MyDomainParticipantListener(); // or = NULL
// Create the participant
DDSDomainParticipant* participant =
factory->create_participant_with_profile(domain_id,
“MyDomainLibrary”, “MyDomainProfile”,
participant_listener, DDS_STATUS_MASK_ALL);
if (participant == NULL) {
// ... error
};

Figure 8.8 Getting QoS from Profile, Creating DomainParticipant with Modified QoS Values
DDS_DomainParticipantQos participant_qos;1
// Get DomainParticipant QoS from profile
retcode = factory->get_participant_qos_from_profile( participant_qos,
“DomainParticipantProfileLibrary”, “DomainParticipantProfile”);
if (retcode != DDS_RETCODE_OK) {
// handle error
}
// Makes QoS changes here
participant_qos.entity_factory.autoenable_created_entities = DDS_BOOLEAN_FALSE;
// create participant with modified QoS
DDSDomainParticipant* participant = factory->create_participant(domain_id,
participant_qos, NULL, DDS_STATUS_MASK_NONE);
if (participant == NULL) {
// handle error
}

8.3.6.2 Comparing QoS Values
The equals() operation compares two DomainParticipant’s DDS_DomainParticipantQoS structures for
equality. It takes two parameters for the two DomainParticipant’s QoS structures to be compared, then
returns TRUE is they are equal (all values are the same) or FALSE if they are not equal.

1In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
565

8.3.6.3 Changing QoS Settings After DomainParticipant Has Been Created

8.3.6.3 Changing QoS Settings After DomainParticipant Has Been Created
There are two ways to change an existing DomainParticipant’s QoS after it is has been created—again
depending on whether or not you are using a QoS Profile.
l

l

To change QoS programmatically (that is, without using a QoS Profile), use get_qos() and set_qos
(). See the example code in Figure 8.9 Changing QoS of Existing Participant (without QoS Profile)
below. It retrieves the current values by calling the DomainParticipant’s get_qos() operation. Then it
modifies the value and calls set_qos() to apply the new value. Note, however, that some
QosPolicies cannot be changed after the DomainParticipant has been enabled—this restriction is
noted in the descriptions of the individual QosPolicies.
You can also change a DomainParticipant’s (and all other Entities’) QoS by using a QoS Profile
and calling set_qos_with_profile(). For an example, see Figure 8.10 Changing QoS of Existing Participant with QoS Profile below. For more information, see Configuring QoS with XML (Section
Chapter 17 on page 791).

Figure 8.9 Changing QoS of Existing Participant (without QoS Profile)
DDS_DomainParticipantQos participant_qos;1
// Get current QoS
//participant points to an existing DDSDomainParticipant
if (participant->get_qos(participant_qos) != DDS_RETCODE_OK) {
// handle error
}
// Make QoS changes
participant_qos.entity_factory.autoenable_created_entities =
DDS_BOOLEAN_FALSE;
// Set the new QoS
if (participant->set_qos(participant_qos) != DDS_RETCODE_OK ) {
// handle error
}

Figure 8.10 Changing QoS of Existing Participant with QoS Profile
DDS_DomainParticipantQos participant_qos;2
// Get current QoS
//participant points to an existing DDSDomainParticipant
if (participant->get_qos(participant_qos) != DDS_RETCODE_OK) {
// handle error
}

1In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
2In C, you must initialize the QoS structures before they are used, see Special QosPolicy Handling

Considerations for C (Section 4.2.2 on page 168).
566

8.3.6.4 Getting and Setting DomainParticipant’s Default QoS Profile and Library

// Make QoS changes
participant_qos.entity_factory.autoenable_created_entities =
DDS_BOOLEAN_FALSE;
// Set the new QoS
if (participant->set_qos(participant_qos) != DDS_RETCODE_OK ) {
// handle error
}

8.3.6.4 Getting and Setting DomainParticipant’s Default QoS Profile and Library
You can get the default QoS profile for the DomainParticipant with the get_default_profile() operation.
You can also get the default library for the DomainParticipant, as well as the library that contains the
DomainParticipant’s default profile (these are not necessarily the same library); these operations are called
get_default_library() and get_default_library_profile(), respectively. These operations are for informational purposes only (that is, you do not need to use them as a precursor to setting a library or profile.) For
more information, see Configuring QoS with XML (Section Chapter 17 on page 791).
virtual const char * get_default_library ()
const char * get_default_profile ()
const char * get_default_profile_library ()

There are also operations for setting the DomainParticipant’s default library and profile:
DDS_ReturnCode_t set_default_library (
const char * library_name)
DDS_ReturnCode_t set_default_profile (
const char * library_name,
const char * profile_name)

If the default profile/library is not set, the DomainParticipant inherits the default from the DomainParticipantFactory.
l

l

l

set_default_profile() specifies the profile that will be used as the default the next time a default
DomainParticipant profile is needed during a call to one of this DomainParticipant’s operations.
When calling a DomainParticipant operation that requires a profile_name parameter, you can use
NULL to refer to the default profile. (This same information applies to setting a default library.)
set_default_profile() does not set the default QoS for entities created by the DomainParticipant;
for this functionality, use the DomainParticipant’s set_default__qos_with_profile() operation (you may pass in NULL after having called set_default_profile(), see Getting and Setting
Default QoS for Child Entities (Section 8.3.6.5 on the facing page)).
set_default_profile() does not set the default QoS for newly created DomainParticipants; for this
functionality, use the DomainParticipantFactory’s set_default_participant_qos_with_profile(), see
Getting and Setting Default QoS for DomainParticipants (Section 8.2.2 on page 545)).

567

8.3.6.5 Getting and Setting Default QoS for Child Entities

8.3.6.5 Getting and Setting Default QoS for Child Entities
The set_default__qos() and set_default__qos_with_profile() operations set the default
QoS that will be used for newly created entities (where  may be publisher, subscriber,
datawriter, datareader, or topic). The new QoS settings will only be used if DDS__QOS_
DEFAULT is specified as the qos parameter when create_() is called. For example, for a Publisher, you can use either:
DDS_ReturnCode_t set_default_publisher_qos (
const DDS_PublisherQos &qos)
DDS_ReturnCode_t set_default_publisher_qos_with_profile (
const char *library_name,
const char *profile_name)

The following operation gets the default QoS that will be used for creating Publishers if DDS_
PUBLISHER_QOS_DEFAULT is specified as the ‘qos’ parameter when create_publisher() is called:
DDS_ReturnCode_t get_default_publisher_qos (
DDS_PublisherQos & qos)

There are similar operations for Subscribers, DataWriters, DataReaders and Topics. These operations,
get_default__qos(), get the QoS settings that were specified on the last successful call to set_
default__qos() or set_default__qos_with_profile(), or if the call was never made, the
default values listed in DDS_Qos. They may potentially allocate memory depending on the
sequences contained in some QoS policies.

Note: It is not safe to set default QoS values for an entity while another thread may be simultaneously getting or setting them, or using the QOS_DEFAULT constant to create the entity.

8.3.7 Looking up Topic Descriptions
The lookup_topicdescription() operation allows you to access a locally created DDSTopicDescription
based on the Topic’s name.
DDSTopicDescription* lookup_topicdescription(const char *topic_name)

DDSTopicDescription is the base class for Topics, MultiTopics1 and ContentFilteredTopics. You can narrow the DDSTopicDescription returned from lookup_topicdescription() to a Topic or ContentFilteredTopic as appropriate.
Unlike find_topic() (see Finding a Topic (Section 8.3.8 on the next page)), which logically returns a new
Topic that must be independently deleted, this operation returns a reference to the original local object.

1Multitopics are not supported.

568

8.3.8 Finding a Topic

If no TopicDescription has been created yet with the given Topic name, this method will return a NULL
value.
The DomainParticipant does not have to be enabled when you call lookup_topicdescription().
Note: It is not safe to create or delete a topic while another thread is calling lookup_topicdescription() for
that same topic.

8.3.8 Finding a Topic
The find_topic() operation finds an existing (or ready to exist) Topic, based on its name. This call can be
used to block for a specified duration to wait for the Topic to be created.
DDSTopic* DDSDomainParticipant::find_topic (const char * topic_name,
const DDS_Duration_t & timeout)

If the requested Topic already exists, it is returned. Otherwise, find_topic() waits until either another
thread creates it, or returns when the specified timeout occurs.
find_topic() is useful when multiple threads are concurrently creating and looking up topics. In that case,
one thread can call find_topic() and, if another thread has not yet created the topic being looked up, it can
wait for some period of time for it to do so. In almost all other cases, it is more straightforward to call
lookup_topicdescription() (see Looking up Topic Descriptions (Section 8.3.7 on the previous page)).
The DomainParticipant must be enabled when you call find_topic().
Note: Each

DDSTopic obtained by find_topic() must also be deleted by calling the DomainParticipant’s
delete_topic() operation (see Deleting Topics (Section 5.1.2 on page 204)).

8.3.9 Getting the Implicit Publisher or Subscriber
The get_implicit_publisher() operation allows you to access the DomainParticipant’s implicit Publisher.
If one does not already exist, this operation creates an implicit Publisher.
There is a similar operation for implicit Subscribers:
DDSPublisher

* get_implicit_publisher ()

DDSSubscriber * get_implicit_subscriber()

There can only be one implicit Publisher and one implicit Subscriber per DomainParticipant. They are created with default QoS values (DDS_PUBLISHER_QOS_DEFAULT) and no Listener. For more information, see Creating Publishers Explicitly vs. Implicitly (Section 6.2.1 on page 248). You can use an
implicit Publisher or implicit Subscriber just like an explicitly created one.
An implicit Publisher/Subscriber is deleted automatically when delete_contained_entities() is called. It
can also be deleted by calling delete_publisher/subscriber() with the implicit Publisher/Subscriber as a
parameter.

569

8.3.10 Asserting Liveliness

When a DomainParticipant is deleted, if there are no attached DataReaders that belong to the implicit Subscriber or no attached DataWriters that belong to the implicit Publisher, any implicit Publisher/Subscriber
will be deleted by the middleware implicitly.
Note: It is not safe to create an implicit Publisher/Subscriber while another thread may be simultaneously
calling set_default_[publisher/subscriber]_qos().
How to get the implicit Publisher/Subscriber. (For simplicity, error handling is not shown.)
using namespace DDS;
...
Publisher * publisher = NULL;
Subscriber * subscriber = NULL;
PublisherQos publisher_qos;
SubscriberQos subscriber_qos;
...
publisher = participant->get_implicit_publisher();
/* Change implicit publisher QoS */
publisher->get_qos(publisher_qos);
publisher_qos.partition.name.maximum(3);
publisher_qos.partition.name.length(3);
publisher_qos.partition.name[0] = DDS_String_dup("partition_A");
publisher_qos.partition.name[1] = DDS_String_dup("partition_B");
publisher_qos.partition.name[2] = DDS_String_dup("partition_C");
publisher->set_qos(publisher_qos);
/* Get implicit subscriber */
subscriber = participant->get_implicit_subscriber();
/* Change implicit subscriber QoS */
subscriber_qos.partition.name.maximum(3);
subscriber _qos.partition.name.length(3);
subscriber _qos.partition.name[0] = DDS_String_dup("partition_A");
subscriber _qos.partition.name[1] = DDS_String_dup("partition_B");
subscriber _qos.partition.name[2] = DDS_String_dup("partition_C");
subscriber->set_qos(subscriber_qos);

8.3.10 Asserting Liveliness
The assert_liveliness() operation manually asserts the liveliness of all the DataWriters created by this
DomainParticipant that has LIVELINESS QosPolicy (Section 6.5.13 on page 382) kind set to
MANUAL_BY_PARTICIPANT. When assert_liveliness() is called, then for those DataWriters who
have their LIVELINESS set to MANUAL_BY_PARTICIPANT, Connext DDS will send a packet to all
matched DataReaders that indicates that the DataWriter is still alive.
However, the LIVELINESS contract of periodically sending liveliness packets to DataReaders is also fulfilled when the write(), assert_liveliness(), unregister_instance() and dispose() operations on a
DataWriter itself is called. Those calls will also cause Connext DDS to send packets that indicate the liveliness of the DataWriter. Therefore, it is necessary for the application to call assert_liveliness() on the
DomainParticipant only if those operations on a DataWriter are not being invoked within the period specified by the LIVELINESS QosPolicy (Section 6.5.13 on page 382)

570

8.3.11 Learning about Discovered DomainParticipants

8.3.11 Learning about Discovered DomainParticipants
The get_discovered_participants() operation provides you with a list of DomainParticipants that have
been discovered in the DDS domain (except any that you have said to ignore via the ignore_participant()
operation (see Restricting Communication—Ignoring Entities (Section 16.4 on page 784))).
Once you have a list of discovered DomainParticipants, you can get more information about them by calling the get_discovered_participant_data() operation. This operation can only be used on DomainParticipants that are in the same DDS domain and have not been marked as ‘ignored.’ Otherwise, the
operation will fail and return DDS_RETCODE_PRECONDITION_NOT_MET. The returned information is of type DDS_ParticipantBuiltinTopicData, described in Table 16.1 Participant Built-in Topic’s
Data Type (DDS_ParticipantBuiltinTopicData).

8.3.12 Learning about Discovered Topics
The get_discovered_topics() operation provides you with a list of Topics that have been discovered in the
DDS domain (except any that you have said to ignore via the ignore_topic() operation (see Restricting
Communication—Ignoring Entities (Section 16.4 on page 784))).
Once you have a list of discovered Topics, you can get more information about them by calling the get_discovered_topic_data() operation. This operation can only be used on Topics that have been created by a
DomainParticipant in the same DDS domain as the participant on which this operation is invoked and
must not have been "ignored" by means of the DomainParticipant ignore_topic() operation. Otherwise,
the operation will fail and return DDS_RETCODE_PRECONDITION_NOT_MET. The returned information is of type DDS_TopicBuiltinTopicData, described in Table 16.4 Topic Built-in Topic’s Data Type
(DDS_TopicBuiltinTopicData) .

8.3.13 Other DomainParticipant Operations
8.3.13.1 Verifying Entity Containment
If you have a handle to an Entity, and want to see if that Entity was created from your DomainParticipant
(or any of its Publishers or Subscribers), use the contains_entity() operation, which returns a boolean.
An Entity’s instance handle may be obtained from built-in topic data (see Built-In Topics (Section Chapter
16 on page 772)), various statuses, or from the get_instance_handle() operation (see Getting an Entity’s
Instance Handle (Section 4.1.3 on page 157)).
8.3.13.2 Getting the Current Time
The get_current_time() operation returns the current time value from the same time-source (clock) that
Connext DDS uses to timestamp the data published by DataWriters (source_timestamp of the SampleInfo
structure, see The SampleInfo Structure (Section 7.4.6 on page 504)). The time-sources used by Connext
DDS do not have to be synchronized nor are they synchronized by Connext DDS.

571

8.3.13.3 Getting All Publishers and Subscribers

See also: Clock Selection (Section 8.6 on page 619).
8.3.13.3 Getting All Publishers and Subscribers
The get_publishers() and get_subscribers() operations will provide you with a list of the DomainParticipant’s Publishers and Subscribers, respectively.

8.4 DomainParticipantFactory QosPolicies
This section describes QosPolicies that are strictly for the DomainParticipantFactory (not the DomainParticipant). For a complete list of QosPolicies that apply to DomainParticipantFactory, see Table 8.2
DomainParticipantFactory QoS.
l

LOGGING QosPolicy (DDS Extension) (Section 8.4.1 below)

l

PROFILE QosPolicy (DDS Extension) (Section 8.4.2 on the next page)

l

SYSTEM_RESOURCE_LIMITS QoS Policy (DDS Extension) (Section 8.4.3 on page 575)

8.4.1 LOGGING QosPolicy (DDS Extension)
This QosPolicy configures the properties associated with the Connext DDS logging facility.
This QosPolicy includes the members in Table 8.5 DDS_LoggingQosPolicy. For defaults and valid
ranges, please refer to the API Reference HTML documentation.
See also: Controlling Messages from Connext DDS (Section 21.2 on page 865) and Configuring Logging
via XML (Section 21.2.2 on page 871).
Table 8.5 DDS_LoggingQosPolicy
Type

Field Name

Description

NDDS_ConfigLogVerbosity

verbosity

Specifies the verbosity at which Connext DDS diagnostic information will be logged.

NDDS_Config_LogCategory

category

Specifies the category for which logging needs to be enabled.

NDDS_Config_LogPrintFormat print_format

Specifies the format to be used to output the Connext DDS diagnostic information.

char *

Specifies the file to which the logged output is redirected.

output_file

8.4.1.1 Example
DSDomainParticipantFactory *factory =
DDSDomainParticipantFactory::get_instance();
DDS_DomainParticipantFactoryQos factoryQos;
DDS_ReturnCode_t retcode = factory->get_qos(factoryQos);

572

8.4.1.2 Properties

if (retcode != DDS_RETCODE_OK) {
// error
}
factoryQos.logging.output_file = DDS_String_dup(“myOutput.txt”);
factoryQos.logging.verbosity = NDDS_CONFIG_LOG_VERBOSITY_STATUS_LOCAL;
factory->set_qos(factoryQos);

8.4.1.2 Properties
This QosPolicy can be changed at any time.
Since it is only configuring logging, there are no compatibility restrictions for how it is set on the publishing and subscribing sides.
8.4.1.3 Related QosPolicies
l

None

8.4.1.4 Applicable DDS Entities
l

DomainParticipantFactory (Section 8.2 on page 539)

8.4.1.5 System Resource Considerations
Because the output_file will be freed by Connext DDS, you should use DDS_String_dup() to allocate
the string.when providing an output_file.

8.4.2 PROFILE QosPolicy (DDS Extension)
This QosPolicy determines the way that XML documents containing QoS profiles are loaded.
All QoS values for Entities can be configured with QoS profiles defined in XML documents. XML documents can be passed to Connext DDS in string form, or more likely, through files found on a file system.
This QoS configures how a DomainParticipantFactory loads the QoS profiles defined in XML. QoS profiles may be stored in this QoS as XML documents as a string. The location of XML files defining QoS
profiles may be configured via this QoS. There are also default locations where the DomainParticipantFactory will look for files to load QoS profiles. You may disable any or all of these default locations using the Profile QoS. For more information about QoS profiles and libraries, please see
Configuring QoS with XML (Section Chapter 17 on page 791).
This QosPolicy includes the members in Table 8.6 DDS_ProfileQosPolicy. For the defaults and valid
ranges, please refer to the API Reference HTML documentation.

573

8.4.2.1 Example

Table 8.6 DDS_ProfileQosPolicy
Type

Field
Name

Description
Sequence of strings (empty by default) containing a XML document to load.

DDS_
StringSeq

string_profile The concatenation of the strings in this sequence must be a valid XML document according to the XML QoS
profile schema.
A sequence of URL groups (empty by default) containing a set of XML documents to load.
url_profile
See URL Groups (Section 17.8 on page 814).
ignore_user_ When TRUE, the QoS profiles contained in the file USER_QOS_PROFILES.xml in the current working
profile
directory will be ignored.

DDS_
Boolean

ignore_
environment_ When TRUE, the value of the environment variable NDDS_QOS_PROFILES will be ignored.
profile

ignore_
resource_
profile

When TRUE, the QoS profiles in the file $NDDSHOME/resource/xml/NDDS_QOS_PROFILES.xml will
be ignored.
NDDS_QOS_PROFILES.xml does not exist by default. However, NDDS_QOS_PROFILES.example.xml
is shipped with the host bundle of the product; you can copy it to NDDS_QOS_PROFILES.xml and modify it
for your own use.

In the Modern C++ API, there is not a PROFILE QosPolicy, because the class that manages QoS profiles
is dds::core::QosProvider—not the DomainParticipantFactory. A QosProvider can receive a QosProviderParams instance, which encapsulates the fields described before.
8.4.2.1 Example
Traditional C++:
DDSDomainParticipantFactory *factory =
DDSDomainParticipantFactory::get_instance();
DDS_DomainParticipantFactoryQos factoryQos;
DDS_ReturnCode_t retcode = factory->get_qos(factoryQos);
if (retcode != DDS_RETCODE_OK) {
// error
}
const char *url_profiles[2] = {
"file://usr/local/default_dds.xml",
"file://usr/local/alternative_default_dds.xml" };
factoryQos.profile.url_profile.from_array(url_profiles, 2);
factoryQos.profile.ignore_resource_profile = DDS_BOOLEAN_TRUE;
factory->set_qos(factoryQos);
rti::core::QosProviderParams params =

574

8.4.2.2 Properties

dds::core::QosProvider::Default()->default_provider_params();
std::vector url_profiles = {
"file://usr/local/default_dds.xml",
"file://usr/local/alternative_default_dds.xml" };
params.url_profile(url_profiles);
params.ignore_resource_profile(true);
dds::core::QosProvider::Default()->default_provider_params(params);

8.4.2.2 Properties
This QosPolicy can be changed at any time.
Since it is only for the DomainParticipantFactory, there are no compatibility restrictions for how it is set on
the publishing and subscribing sides.
8.4.2.3 Related QosPolicies
l

None

8.4.2.4 Applicable Entities
l

DomainParticipantFactory (Section 8.2 on page 539)

8.4.2.5 System Resource Considerations
Once the QoS profiles are loaded, the DomainParticipantFactory will keep one copy of each QoS in the
QoS profiles in memory.
You can free the memory associated with the XML QoS profiles by calling the DomainParticipantFactory’s unload_profiles() operation.

8.4.3 SYSTEM_RESOURCE_LIMITS QoS Policy (DDS Extension)
The SYSTEM_RESOURCE_LIMITS QosPolicy configures DomainParticipant-independent resources
used by Connext DDS. Its main use is to change the maximum number of DomainParticipants that can be
created within a single process (address space).
It contains the single member as shown in Table 8.7 DDS_SystemResourceLimitsQosPolicy. For the
default and valid range, please refer to the API Reference HTML documentation.

575

8.4.3.1 Example

Table 8.7 DDS_SystemResourceLimitsQosPolicy
Type
DDS_
Long

Field Name
max_objects_per_
thread

Description
Sizes the thread storage that is allocated on a per-thread basis when the thread calls Connext DDS
APIs.

The only parameter that you can set, max_objects_per_thread, controls the size of thread-specific storage
that is allocated by Connext DDS for every thread that invokes a Connext DDS API. This storage is used
to cache objects that have to be created on a per-thread basis when a thread traverses different portions of
Connext DDS internal code.
Thus instead of dynamically creating and destroying the objects as a thread enters and leaves different
parts of the code, Connext DDS caches the objects by storing them in thread-specific storage. We assume
that a thread will repeatedly call Connext DDS APIs so that the objects cached will be needed again and
again.
The number of objects that will be stored in the cache depends the number of APIs (sections of Connext
DDS code) that a thread invokes. It also depends on the number of different DomainParticipants with
which the thread interacts. For a single DomainParticipant, the maximum number of objects that could be
stored is a constant–independent of the number of Entities created in or by the participant. A safe number
to use is 200 objects per DomainParticipant.
A user thread that only interacts with a single DomainParticipant or the Entities thereof, would never have
more than 200 objects stored in its cache. However, if the same thread invokes Connext DDS APIs on
other Entities of other DomainParticipants, the maximum number of objects that may be stored will
increase with the number of participants involved.
The default setting of this resource should work for most user applications. However, if your application
uses more than 4 DomainParticipants, you may need to increase the value of max_objects_per_thread.
8.4.3.1 Example
Say an application uses 10 DomainParticipants. If a single thread was used to create all 10 DomainParticipants, or a single thread is used to call write() on DataWriters belonging to all 10 participants, it is possible to run out of thread-specific storage. Either the creation of the participant or the write() will fail.
In that case, you will need to increase the value of max_objects_per_thread.
8.4.3.2 Properties
This QoS policy cannot be modified after the DomainParticipantFactory is used to create the first
DomainParticipant or WaitSet in an application.
This QoS can be set differently in different applications.

576

8.4.3.3 Related QoS Policies

8.4.3.3 Related QoS Policies
There are no interactions with other QosPolicies.
8.4.3.4 Applicable Dds Entities
l

DomainParticipantFactory (Section 8.2 on page 539)

8.4.3.5 System Resource Considerations
Increasing the value of max_objects_per_thread will increase the amount of memory allocated by Connext
DDS for every thread that access Connext DDS code. This includes internal Connext DDS threads as well
as user threads. Each object uses about 32 bytes of memory.

8.5 DomainParticipant QosPolicies
This section describes the QosPolicies that are strictly for DomainParticipants (and no other types of Entities). For a complete list of QosPolicies that apply to DomainParticipant, see Table 8.4 DomainParticipant
QosPolicies.
l

DATABASE QosPolicy (DDS Extension) (Section 8.5.1 below)

l

DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580)

l

DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585)

l

DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
on page 593)

l

EVENT QosPolicy (DDS Extension) (Section 8.5.5 on page 602)

l

RECEIVER_POOL QosPolicy (DDS Extension) (Section 8.5.6 on page 604)

l

TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section 8.5.7 on page 606)

l

l

TRANSPORT_MULTICAST_MAPPING QosPolicy (DDS Extension) (Section 8.5.8 on page
608)
WIRE_PROTOCOL QosPolicy (DDS Extension) (Section 8.5.9 on page 610)

8.5.1 DATABASE QosPolicy (DDS Extension)
The Database QosPolicy configures how Connext DDS manages its internal database, including how
often it cleans up, the priority of the database thread, and limits on resources that may be allocated by the
database. RTI uses an internal in-memory database to store information about entities created locally as
well as remote entities found during the discovery process. This database uses a background thread to
garbage-collect records related to deleted entities. When the DomainParticipant that maintains this database is deleted, it shuts down this thread..

577

8.5.1 DATABASE QosPolicy (DDS Extension)

It includes the members in Table 8.8 DDS_DatabaseQosPolicy. For defaults and valid ranges, please refer
to the API Reference HTML documentation.
Table 8.8 DDS_DatabaseQosPolicy
Type

Field
Name

Description

thread.mask Thread settings for the database thread used by Connext DDS to periodically remove deleted records
DDS_
thread.priority from the database. The values used for these settings are OS-dependent; see the RTI Connext DDS Core
ThreadSettings_
Libraries Platform Notes for details.
thread.stack_
t
size
Note: thread.cpu_list and thread.cpu_rotation are not relevant in this QoS policy.

DDS_
Duration_t

shutdown_
timeout

The maximum time that the DomainParticipant will wait for the database thread to terminate when the
participant is destroyed.

DDS_
Duration_t

cleanup_
period

The period at which the database thread wakes up to removed deleted records.

DDS_
Duration_t

shutdown_
cleanup_
period

The period at which the database thread wakes up to removed deleted records when the
DomainParticipant is being destroyed.

DDS_Long

initial_records

The number of records that is initially created for the database. These records hold information for both
local and remote entities that are dynamically created or discovered.
This is a performance tuning parameter that optimizes the time it takes to search the database for a record.
A ‘Skip List’ is an algorithm for maintaining a list that is faster to search than a binary tree.

DDS_Long

max_skiplist_
level

This value should be set to log2(N), where N is the maximum number of elements that will be stored in a
single list. The list that stores the records for remote DataReaders or the one for remote DataWriters tend
to have the most entries. So, the number of DataWriters or DataReaders in a system across all
DomainParticipants in a single DDS domain, which ever is greater, can be used to set this parameter.
This parameter sets the maximum number of entries in the weak reference table. Weak references are used
as a technique for ensuring that unreferenced objects are deleted.
The actual number of weak references is permitted to grow from the value set by initial_weak_references
to this maximum.

DDS_Long

max_weak_
references

To prevent Connext DDS from allocating memory for weak references after initialization, you should
set the initial and maximum weak references to the same value.
However, it is difficult to calculate how many weak references an application will use. To allow
Connext DDS to grow the weak reference table as needed, and thus dynamically allocate memory, you
should set the value of this field to DDS_LENGTH_UNLIMITED, the default setting.
The initial number of entries in the weak reference table.

DDS_Long

initial_weak_
references

See max_weak_references.

Connext DDS may decide to use a larger initial value if initial_weak_references is set too small. If you
access this parameter after a DomainParticipant has been created, you will see the actual value used.

578

8.5.1.1 Example

You may be interested in modifying the shutdown_timeout and shutdown_cleanup_period parameters
to decrease the time it takes to delete a DomainParticipant when your application is shutting down.
The DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on
page 593) controls the memory allocation for elements stored in the database.
Real-time programmers will probably want to adjust the priorities of all of the threads created by Connext
DDS relative to each other as well as relative to non-Connext DDS threads in their applications. Connext
DDS Threading Model (Section Chapter 19 on page 837), EVENT QosPolicy (DDS Extension) (Section
8.5.5 on page 602), and RECEIVER_POOL QosPolicy (DDS Extension) (Section 8.5.6 on page 604)
discuss the other threads that are created by Connext DDS.
A record in the database can be deleted only when no threads are using it. Connext DDS uses a thread that
periodically checks the database if records that have been marked for deletion can be removed. This period
is set by cleanup_period. When a DomainParticipant is being destroyed, the thread will wake up faster at
the shutdown_cleanup_period as other threads delete and release records in preparation for shutting down.
On Windows and VxWorks systems, the thread that is destroying the DomainParticipant may block up to
shutdown_timeout seconds while waiting for the database thread to finish removing all records and terminating. On other operating systems, the thread destroying the DomainParticipant will block as long as
required for the database thread to terminate.
The default values for those and the rest of the parameters in this QosPolicy should be sufficient for most
applications.
8.5.1.1 Example
The priority of the database thread should be set to the lowest priority among all threads in a real-time system. Although, the database thread should not be permitted to starve, the work that it performs is non-timecritical.
8.5.1.2 Properties
This QosPolicy cannot be modified after the DomainParticipant is created.
It can be set differently on the publishing and subscribing sides.
8.5.1.3 Related QosPolicies
l

DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
on page 593)

l

EVENT QosPolicy (DDS Extension) (Section 8.5.5 on page 602)

l

RECEIVER_POOL QosPolicy (DDS Extension) (Section 8.5.6 on page 604)

579

8.5.1.4 Applicable Dds Entities

8.5.1.4 Applicable Dds Entities
l

DomainParticipants (Section 8.3 on page 547)

8.5.1.5 System Resource Considerations
Setting the thread parameters correctly on a real-time operating system is usually critical to the proper overall functionality of the applications on that system. Larger values for the thread.stack_size parameter will
use up more memory.
Smaller values for the cleanup_period and shutdown_cleanup_period will cause the database thread to
wake up more frequently using more CPU.
Connext DDS is permitted to use up more memory for larger values of max_skiplist_level and max_
weak_references. Whether or not more memory is actually used depends on actual operating conditions.

8.5.2 DISCOVERY QosPolicy (DDS Extension)
The DISCOVERY QoS configures how DomainParticipants discover each other on the network. It identifies where on the network this application can potentially discover other applications with which to communicate. The middleware will periodically send network packets to these locations, announcing itself to
any remote applications that may be present, and will listen for announcements from those applications.
The discovery process is described in detail in Discovery (Section Chapter 14 on page 709).
This QosPolicy includes the members in Table 8.9 DDS_DiscoveryQosPolicy. For defaults and valid
ranges, please refer to the API Reference HTML documentation.
Table 8.9 DDS_DiscoveryQosPolicy
Type

Field Name

Description

DDS_
enabled_
StringSeq transports

Transports available for use by the discovery process. See Transports Used for Discovery (Section 8.5.2.1 on
the next page).

DDS_
initial_peers
StringSeq

Unicast locators (address/indices) of potential participants with which this DomainParticipant will attempt to
establish communications. See Setting the ‘Initial Peers’ List (Section 8.5.2.2 on the next page).

multicast_
DDS_
receive_
StringSeq
addresses

List of multicast addresses on which Discovery-related messages can be received by the DomainParticipant.
See Configuring Multicast Receive Addresses (Section 8.5.2.4 on page 582).

DDS_
Long

metatraffic_
transport_
priority

Transport priority to be used for sending Discovery messages. See Meta-Traffic Transport Priority (Section
8.5.2.5 on page 583).

DDS_
Boolean

accept_
unknown_
peers

Whether to accept a participant discovered via unicast that is not in the initial_peers list. See Controlling
Acceptance of Unknown Peers (Section 8.5.2.6 on page 583).

580

8.5.2.1 Transports Used for Discovery

Table 8.9 DDS_DiscoveryQosPolicy
Type
DDS_
Boolean

Field Name
enable_
endpoint_
discovery

Description
Whether endpoint discovery will automatically occur with discovered DomainParticipants. See Supervising
Endpoint Discovery (Section 16.4.5 on page 788).

8.5.2.1 Transports Used for Discovery
The enabled_transports field allows you to specify the set of installed and enabled transports that can be
used to discover other DomainParticipants. This field is a sequence of strings where each string specifies
an alias of a registered (and thus installed and enabled) transport. Please see the API Reference HTML documentation (select Modules, RTI Connext DDS API Reference, Pluggable Transports) for more
information.
8.5.2.2 Setting the ‘Initial Peers’ List
When a DomainParticipant is created, it needs to find other participants in the same DDS domain—this is
known as the ‘discovery process’ which is discussed in Discovery (Section Chapter 14 on page 709). One
way to do so is to use this QosPolicy to specify a list of potential participants. This is the role of the parameter initial_peers. The strings containing peer descriptors are stored in the initial_peers string sequence.
The format of a string discussed in Peer Descriptor Format (Section 14.2.1 on page 713).
The peers stored in initial_peers are merely potential peers—there is no requirement that the peer
DomainParticipants are actually up and running or even will eventually exist. The Connext DDS discovery process will try to contact all potential peer participants in the list periodically using unicast transports (as configured by the DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page
585)).
The initial_peers parameter can be modified in source code or it can be initialized from an environment
variable, NDDS_DISCOVERY_PEERS or from a text file, see Configuring the Peers List Used in Discovery (Section 14.2 on page 711).
8.5.2.3 Adding and Removing Peers List Entries
The DomainParticipant’s add_peer() operation adds a peer description to the internal peer list that was initialized by the initial_peer field of the DISCOVERY QosPolicy.
DDS_ReturnCode_t DDSDomainParticipant::add_peer (
const char* peer_desc)

The peer_desc string must be formatted as specified in Peer Descriptor Format (Section 14.2.1 on page
713).

581

8.5.2.4 Configuring Multicast Receive Addresses

You can call this operation any time after the DomainParticipant has been enabled. An attempt will be
made to contact the new peer immediately.
Adding peers with this operation has no effect on the initial_peers list. After a DomainParticipant has
been created, the contents of the initial_peers field merely shows what the internal peer list was initialized
to be. Therefore, initial_peers may not reflect the actual potential peer list used by a DomainParticipant.
Furthermore, if you call get_qos(), the returned list of peers will not include the added peer—get_qos()
will only show you what is set in the initial_peers list.
A peer added with add_peer() is not considered to be “unknown.” (That is, you may have accept_
unknown_peers (Controlling Acceptance of Unknown Peers (Section 8.5.2.6 on the next page)) set to
FALSE and still use add_peer().)
You can remove an entry from the list with remove_peer().
You can ignore data from a participant by using the ignore_participant() operation described in Restricting Communication—Ignoring Entities (Section 16.4 on page 784).
8.5.2.4 Configuring Multicast Receive Addresses
The multicast_receive_addresses field in the DISCOVERY QosPolicy is a sequence of strings that specifies a set of multicast group addresses on which the DomainParticipant will listen for discovery metatraffic. Each string must have a valid multicast address in either IPv4 dot notation or IPv6 presentation
format. Please look at publicly available documentation of the IPv4 and IPv6 standards for the definition
and valid address ranges for multicast.
The multicast_receive_addresses field can be initialized from multicast addresses that appear in the
NDDS_DISCOVERY_PEERS environment variable or text file, see Configuring the Peers List Used in
Discovery (Section 14.2 on page 711). A multicast address found in the environment variable or text file
will be added both to the initial_peers and multicast_receive_addresses fields. Note that the addresses in initial_peers are ones in which the DomainParticipant will send discovery meta-traffic, and the ones in multicast_receive_addresses are used for receiving discovery meta-traffic.
If NDDS_DISCOVERY_PEERS does not contain a multicast address, then multicast_receive_
addresses is cleared and the RTI discovery process will not listen for discovery messages via multicast.
If NDDS_DISCOVERY_PEERS contains one or more multicast addresses, the addresses are stored in
multicast_receive_addresses, starting at element 0. They will be stored in the order in which they appear
in NDDS_DISCOVERY_PEERS.
Note: Currently, Connext DDS will only listen for discovery traffic on the first multicast address (element
0) in multicast_receive_addresses.
If you want to send discovery meta-traffic on a different set of multicast addresses than you want to receive
discovery meta-traffic, set initial_peers and multicast_receive_addresses via the QosPolicy API.

582

8.5.2.5 Meta-Traffic Transport Priority

8.5.2.5 Meta-Traffic Transport Priority
The metatraffic_transport_priority field is used to specify the transport priority to be used for sending all
discovery meta-traffic. See the TRANSPORT_PRIORITY QosPolicy (Section 6.5.22 on page 409) for
details on how transport priorities may be used.
8.5.2.6 Controlling Acceptance of Unknown Peers
The accept_unknown_peers field controls whether or not a DomainParticipant is allowed to communicate
with other DomainParticipants found via unicast transport that are not in its peers list (which is the combination of the initial_peers list and any peers added with the add_peer() operation described in Adding
and Removing Peers List Entries (Section 8.5.2.3 on page 581)).
Suppose Participant A is included in Participant B’s initial peers list, but Participant B is not in Participant
A’s list. When Participant B contacts Participant A by sending it a unicast discovery packet, then Participant A has a choice:
l

l

If accept_unknown_peers is DDS_BOOLEAN_TRUE, then Participant A will reply to Participant B, and communications will be established.
If accept_unknown_peers is DDS_BOOLEAN_FALSE, then Participant A will ignore Participant B, and A and B will never talk.

Note that Participants do not exchange peer lists. So if Participant A knows about Participant B, and Participant B knows about Participant C, Participant A will not discover Participant C.
Note: If accept_unknown_peers is false and shared memory is disabled, applications on the same node
will not communicate if only ‘localhost’ is specified in the peer list. If shared memory is disabled or
‘shmem://’ is not specified in the peer list, if you want to communicate with other applications on the same
node through the loopback interface, you must put the actual node address or hostname in NDDS_
DISCOVERY_PEERS.
8.5.2.7 Example
You will always use this policy to set the participant_id when you want to run more than one DomainParticipant in the same DDS domain on the same host.
The easiest way to set the initial peers list is to use the NDDS_DISCOVERY_PEERS environment variable. However, should you want asymmetric multicast addresses for sending or receiving meta-traffic, you
will need to use this QosPolicy directly.
A reason to use asymmetric multicast addresses is to take advantage of the efficiency provided by using
multicast, while at the same time preventing all participants from discovering each other. For example, suppose you have a system in which you have a single server node and a hundred client nodes. The client

583

8.5.2.8 Properties

nodes do not publish or subscribe to each other’s data and thus never need to know about each others existence.
If we did not use multicast, we would have to populate the server application’s peer list with 100 peer
descriptors for each of the client nodes. Each client application would only need to have the server application in its peer list. The maintenance of the list is unwieldy, especially if nodes are constantly reconfigured
and addresses changed. In addition, the server will send out discovery packets on a per client basis since
the peer list essentially holds 100 unicast addresses.
Instead, if we used a single multicast address in the NDDS_DISCOVERY_PEERS environment variable,
the server and all of the clients would discover each other. Certainly, the list is easier to maintain, but the
total amount of traffic has actually increased since the clients are now exchanging packets with each other
uselessly.
To keep the list maintainable, as well as to minimize discovery traffic, we can have the server send out
packets on a multicast address by modifying its initial_peer field. The clients would have their multicast_
receive_addresses field set to the same address used by the server. The initial_peers of the clients would
only need the single unicast peer descriptor of the server as before.
Now, the server can send a single packet that will be received by all of the clients, but the clients will not
discover each other because they never send out a multicast packet themselves.
8.5.2.8 Properties
This QosPolicy cannot be modified after the DomainParticipant is created.
It can be set differently on the publishing and subscribing sides.
8.5.2.9 Related QosPolicies
l

DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on the next page)

l

TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section 8.5.7 on page 606)

8.5.2.10 Applicable Entities
l

DomainParticipants (Section 8.3 on page 547)

8.5.2.11 System Resource Considerations
For every entry in the initial_peers list, Connext DDS will periodically send a discovery packet to see if
that participant exists. If the list has many potential participants that are never started, then CPU and network bandwidth may be wasted in sending out packets that will never be received.

584

8.5.3 DISCOVERY_CONFIG QosPolicy (DDS Extension)

8.5.3 DISCOVERY_CONFIG QosPolicy (DDS Extension)
The DISCOVERY_CONFIG QosPolicy is used to tune the discovery process. It controls how often to
send discovery packets, how to determine when participants are alive or dead, and resources used by the
discovery mechanism.
The amount of network traffic required by the discovery process can vary widely based on how your
application has chosen to configure the middleware's network addressing (e.g. unicast vs. multicast, multicast TTL, etc.), the size of the system, whether all applications are started at the same time or whether
start times are staggered, and other factors. Your application can use this policy to make trade-offs between
discovery completion time and network bandwidth utilization. In addition, you can introduce random
back-off periods into the discovery process to decrease the probability of network contention when many
applications start simultaneously.
This QosPolicy includes the members in Table 8.10 DDS_DiscoveryConfigQosPolicy. Many of these
members are described in Discovery (Section Chapter 14 on page 709). For defaults and valid ranges,
please refer to the API Reference HTML documentation.
Table 8.10 DDS_DiscoveryConfigQosPolicy
Type

Field Name Description

DDS_Duration_t

participant_
liveliness_
lease_duration

The time period after which other DomainParticipants can consider this one
dead if they do not receive a liveliness packet from this DomainParticipant.

DDS_Duration_t

participant_
liveliness_
assert_period

The period of time at which this DomainParticipant will send out packets
asserting that it is alive.

DDS_RemoteParticipantPurgeKind

remote_
participant_
purge_kind

Controls the DomainParticipant's behavior for purging records of remote
participants (and their contained entities) with which discovery communication
has been lost. See Controlling Purging of Remote Participants (Section 8.5.3.2
on page 591).

DDS_Duration_t

max_
liveliness_
loss_
detection_
period

The maximum amount of time between when a remote entity stops maintaining
its liveliness and when the matched local entity realizes that fact.

DDS_Long

initial_
Sets how many initial liveliness announcements the DomainParticipant will
participant_
send when it is first enabled, or after discovering a new remote participant.
announcements

585

8.5.3 DISCOVERY_CONFIG QosPolicy (DDS Extension)

Table 8.10 DDS_DiscoveryConfigQosPolicy
Type

DDS_Duration_t

DDS_Duration_t

Field Name Description
min_initial_
participant_
Sets the minimum and maximum times between liveliness announcements.
announcement_
When a participant is first enabled, or after discovering a new remote
period
participant, Connext DDS sends initial_paricipant_annoucements number of
discovery messages. These messages are sent with a sleep period between
max_initial_
them that is a random duration between min_initial_participant_
participant_
announcement_ announcement_period and max_initial_participant_announcement_period.
period

DDS_
BuiltinTopicReaderResourceLimits_t
(Section Table 8.11 on page 589)

participant_
Configures the resource for the built-in DataReaders used to access discovery
reader_
information; see Resource Limits for Builtin-Topic DataReaders (Section
resource_limits 8.5.3.1 on page 589) and Built-In Topics (Section Chapter 16 on page 772).

DDS_RtpsReliableReaderProtocol_t
(Section Table 7.20 on page 514)

publication_
reader

DDS_
BuiltinTopicReaderResourceLimits_t
(Section Table 8.11 on page 589)

publication_
Configures the resource for the built-in DataReaders used to access discovery
reader_
information; see Resource Limits for Builtin-Topic DataReaders (Section
resource_limits 8.5.3.1 on page 589) and Built-In Topics (Section Chapter 16 on page 772).

Configures the RTPS reliable protocol parameters for a built-in publication
reader.

Configures the RTPS reliable protocol parameters for a built-in subscription
reader.
DDS_RtpsReliableReaderProtocol_t
(Section Table 7.20 on page 514)

subscription_
reader

DDS_
BuiltinTopicReaderResourceLimits_t
(Section Table 8.11 on page 589)

subscription_
Configures the resource for the built-in DataReaders used to access discovery
reader_
information; see Resource Limits for Builtin-Topic DataReaders (Section
resource_limits 8.5.3.1 on page 589) and Built-In Topics (Section Chapter 16 on page 772).

Built-in subscription readers receive discovery information reliably from
DomainParticipants that were dynamically discovered (see Discovery (Section
Chapter 14 on page 709)).

Configures the RTPS reliable protocol parameters for the writer side of a
reliable connection.
DDS_RtpsReliableWriterProtocol_t
(Section Table 6.37 on page 350)

publication_
writer

WRITER_DATA_LIFECYCLE QoS
Policy (Section 6.5.27 on page 419)

publication_
writer_data_
lifecycle

Built-in DataWriters send reliable discovery information to
DomainParticipants that were dynamically discovered (see Discovery (Section
Chapter 14 on page 709)).
Configures writer data-lifecycle settings for a built-in publication writer.
(DDS_WriterDataLifecycleQosPolicy::
autodispose_unregistered_instances will always be TRUE.)
Configures the RTPS reliable protocol parameters for the writer side of a
reliable connection.

DDS_RtpsReliableWriterProtocol_t
(Section Table 6.37 on page 350)

subscription_
writer

Built-in DataWriters send reliable discovery information to
DomainParticipants that were dynamically discovered (see Discovery (Section
Chapter 14 on page 709)).

586

8.5.3 DISCOVERY_CONFIG QosPolicy (DDS Extension)

Table 8.10 DDS_DiscoveryConfigQosPolicy
Type

Field Name Description

WRITER_DATA_LIFECYCLE QoS
Policy (Section 6.5.27 on page 419)

subscription_
writer_data_
lifecycle

Configures writer data-lifecycle settings for a built-in subscription writer.
(DDS_WriterDataLifecycleQosPolicy::autodispose_unregistered_instances
will always be TRUE.)
The kind mask for selecting built-in discovery plugins:
l

builtin_
DDS_
discovery_
DiscoveryConfigBuiltinPluginKindMask
plugins

Simple Discovery Protocol:
DDS_DISCOVERYCONFIG_BUILTIN_SDP
l

Enterprise Discovery Service:
DDS_DISCOVERYCONFIG_BUILTIN_EDS
(Requires a separate component, RTI Enterprise Discovery Service.)
The period at which a participant will announce itself to the default DDS
domain 0 using the default UDPv4 multicast group address for discovery
traffic on that DDS domain.
For DDS domain 0, the default discovery multicast address is
239.255.0.1:7400.
To disable announcement to the default DDS domain, set this to
DURATION_INFINITE.

DDS_Duration_t

default_
When this period is set to a value other than
domain_
announcement_ DURATION_INFINITE and ignore_default_domain_announcements (see
period
below) is FALSE, you can get information about participants running in
different DDS domains by creating a participant in DDS domain 0 and
implementing the on_data_available callback (see DATA_AVAILABLE
Status (Section 7.3.7.1 on page 471)) in the ParticipantBuiltinTopicData builtin DataReader's listener (see Built-in DataReaders (Section 16.2 on page
773)).
You can learn the domain ID associated with a participant by looking at the
domain_id (Section on page 774) in the ParticipantBuiltinTopicData.
When TRUE, ignores the announcements received by a participant on the
default DDS domain 0 corresponding to participants running on domains
IDs other than 0.

DDS_Boolean

This setting only applies to participants running on the default DDS domain
ignore_default_ 0 and using the default port mapping.
domain_
When TRUE, a participant running on the default DDS domain 0 will ignore
announcements announcements from participants running on different DDS domain IDs.
When FALSE, a participant running on the default DDS domain 0 will
provide announcements from participants running on different DDS domain
IDs to the application via the ParticipantBuiltinTopicData built-in DataReader
(see Built-in DataReaders (Section 16.2 on page 773)).

587

8.5.3 DISCOVERY_CONFIG QosPolicy (DDS Extension)

Table 8.10 DDS_DiscoveryConfigQosPolicy
Type

Field Name Description

DDS_RtpsReliableReaderProtocol_t
(Section Table 7.20 on page 514)

participant_
message_
reader

DDS_ReliabilityQosPolicyKind
See Table 6.59 DDS_
ReliabilityQosPolicy

DDS_RtpsReliableWriterProtocol_t
(Section Table 6.37 on page 350)

participant_
message_
reader_
reliability_kind
participant_
message_
writer

RTPS protocol-related configuration settings for a built-in participant message
reader.

Reliability kind configuration setting
for a built-in participant message reader (default: best-effort).

RTPS protocol-related configuration settings for a built-in participant message
writer.

PUBLISH_MODE QosPolicy (DDS
publication_
Extension) (Section 6.5.18 on page 397) writer_
publish_mode

Determines whether the Discovery built-in publication DataWriter publishes
data synchronously or asynchronously and how.

PUBLISH_MODE QosPolicy (DDS
subscription_
Extension) (Section 6.5.18 on page 397) writer_
publish_mode

Determines whether the Discovery built-in subscription DataWriter publishes
data synchronously or asynchronously and how.

ASYNCHRONOUS_PUBLISHER
QosPolicy (DDS Extension) (Section
6.4.1 on page 313)

asynchronous_ Asynchronous publishing settings for the Discovery Publisher and all entities
publisher
that are created by it.

A DomainParticipant needs to send a message periodically to other DomainParticipants to let the other
participants know that it is still alive. These liveliness messages are sent to all peers in the peer list that was
initialized by the initial_peers parameter of the DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2
on page 580). Peer participants on the peer list may or may not be alive themselves. The peer DomainParticipants that already know about this DomainParticipant will use the participant_liveliness_lease_duration provided by this participant to declare the participant dead, if they have not received a liveliness
message for the specified time.
The participant_liveliness_assert_period is the periodic rate at which this DomainParticipant will be sending liveliness messages. Since these liveliness messages are not sent reliably and can get dropped by the
transport, it is important to set:
participant_liveliness_assert_period < participant_liveliness_lease_duration/N
where N is the number of liveliness messages that other DomainParticipants must miss before they decide
that this DomainParticipant is dead.
DomainParticipants that receive a liveliness message from a participant that they did not know about previously will have “discovered” the participant. When one DomainParticipant discovers another, the discoverer will immediately send its own liveliness packets back. initial_participant_announcements controls

588

8.5.3.1 Resource Limits for Builtin-Topic DataReaders

how many of these initial liveliness messages are sent, and max_initial_participant_announcement_period
controls the time period in between each message.
After the initial set of liveliness messages are sent, the DomainParticipant will return to sending liveliness
packets to all peers in its peer list at the rate governed by participant_liveliness_assert_period.
For more information on the discovery process, see Discovery (Section Chapter 14 on page 709).
8.5.3.1 Resource Limits for Builtin-Topic DataReaders
The DDS_BuiltinTopicReaderResourceLimits_t structure is shown in Table 8.11 DDS_BuiltinTopicReaderResourceLimits_t. This structure contains several fields that are used to configure the
resource limits of the builtin-topic DataReaders used to receive discovery meta-traffic from other
DomainParticipants.
Table 8.11 DDS_BuiltinTopicReaderResourceLimits_t
Type Field Name

Description

initial_samples Initial number of meta-traffic DDS data samples that can be stored by a builtin-topic DataReader.
max_samples

Maximum number of meta-traffic DDS data samples that can be stored by a builtin-topic DataReader.

initial_infos

Initial number of DDS_SampleInfo structures allocated for the builtin-topic DataReader.
Maximum number of DDS_SampleInfo structures that can be allocated for the built-in topic DataReader.

max_infos
max_infos must be >= max_samples
DDS_
Long

initial_
outstanding_
reads

Initial number of times in which memory can be concurrently loaned via read/take calls on the builtin-topic
DataReader without being returned with return_loan().

max_
outstanding_
reads

Maximum number of times in which memory can be concurrently loaned via read/take calls on the builtin-topic
DataReader without being returned with return_loan().

max_samples_
Maximum number of DDS samples that can be read/taken on a same built-in topic DataReader.
per_read
disable_
Determines whether the builtin-topic DataReader can receive fragmented DDS samples.
DDS_
fragmentation_
Boolean
When fragmentation support is not needed, disabling fragmentation support will save some memory resources.
support

589

8.5.3.1 Resource Limits for Builtin-Topic DataReaders

Table 8.11 DDS_BuiltinTopicReaderResourceLimits_t
Type Field Name

Description
The maximum number of DDS samples for which the DataReader may store fragments at a given point in time.

DDS_
Long

max_
fragmented_
samples

At any given time, a DataReader may store fragments for up to max_fragmented_samples DDS samples
while waiting for the remaining fragments. These DDS samples need not have consecutive sequence numbers
and may have been sent by different DataWriters. Once all fragments of a DDS sample have been received, the
DDS sample is treated as a regular DDS sample and becomes subject to standard QoS settings, such as max_
samples. Connext DDS will drop fragments if the max_fragmented_samples limit has been reached.
For best-effort communication, Connext DDS will accept a fragment for a new DDS sample, but drop the
oldest fragmented DDS sample from the same remote writer.
For reliable communication, Connext DDS will drop fragments for any new DDS samples until all fragments
for at least one older DDS sample from that writer have been received.
Only applies if disable_fragmentation_support is FALSE.

DDS_
Long

initial_
fragmented_
samples

DDS_
Long

max_
fragmented_
samples_per_
remote_writer

DDS_
Long

max_
fragments_
per_sample

The initial number of DDS samples for which a builtin-topic DataReader may store fragments.
Only applies if disable_fragmentation_support (Section on the previous page) is FALSE.
The maximum number of DDS samples per remote writer for which a builtin-topic DataReader may store
fragments.
Logical limit so a single remote writer cannot consume all available resources.
Only applies if disable_fragmentation_support (Section on the previous page) is FALSE.
Maximum number of fragments for a single DDS sample.
Only applies if disable_fragmentation_support (Section on the previous page) is FALSE.
By default, the middleware does not allocate memory upfront, but instead allocates memory from the heap upon
receiving the first fragment of a new sample. The amount of memory allocated equals the amount of memory
needed to store all fragments in the sample. Once all fragments of a sample have been received, the sample is
deserialized and stored in the regular receive queue. At that time, the dynamically allocated memory is freed
again.

dynamically_
DDS_ allocate_
Boolean fragmented_
samples

This QoS setting is useful for large, but variable-sized data types where up-front memory allocation for multiple
samples based on the maximum possible sample size may be expensive. The main disadvantage of not preallocating memory is that one can no longer guarantee the middleware will have sufficient resources at run-time.
If dynamically_allocate_fragmented_samples is FALSE, the middleware will allocate memory up-front for
storing fragments for up to initial_fragmented_samples samples. This memory may grow up to max_
fragmented_samples if needed.
Only applies if disable_fragmentation_support (Section on the previous page) is FALSE.

There are builtin-topics for exchanging data about DomainParticipants, for publications
(Publisher/DataWriter combination) and for subscriptions (Subscriber/DataReader combination). The
DataReaders for the publication and subscription builtin-topics are reliable. The DataReader for the participant builtin-topic is best effort.

590

8.5.3.2 Controlling Purging of Remote Participants

You can set listeners on these DataReaders that are created automatically when a DomainParticipant is
created. With these listeners, your code can be notified when remote DomainParticipants,
Publishers/DataWriters, and Subscriber/DataReaders are discovered. You can always check the receive
queues of those DataReaders for the same information about discovered entities at any time. Please see
Built-In Topics (Section Chapter 16 on page 772) for more details.
The initial_samples and max_samples, and related initial_infos and max_infos, fields size the amount of
declaration messages can be stored in each builtin-topic DataReader.
8.5.3.2 Controlling Purging of Remote Participants
When discovery communication with a remote participant has been lost, the local participant must make a
decision about whether to continue attempting to communicate with that participant and its contained entities. The remote_participant_purge_kind is used to select the desired behavior.
This does not pertain to the situation in which a remote participant has been gracefully deleted and notification of that deletion has been successfully received by its peers. In that case, the local participant will
immediately stop attempting to communicate with those entities and will remove the associated remote
entity records from its internal database.
The remote_participant_purge_kind can be set to the following values:
DDS_LIVELINESS_BASED_REMOTE_PARTICIPANT_PURGE
This value causes Connext DDS to keep the state of a remote participant and its contained entities for as
long as the participant maintains its liveliness contract (as specified by its participant_liveliness_lease_
duration in the DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585)).
A participant will maintain its own liveliness to any remote participant via inter-participant liveliness traffic
(see LIVELINESS QosPolicy (Section 6.5.13 on page 382)).
The default Simple Discovery Protocol described in Discovery (Section Chapter 14 on page 709) automatically maintains this liveliness, whereas other discovery mechanisms may or may not.
DDS_NO_REMOTE_PARTICIPANT_PURGE
With this value, Connext DDS will never purge the records of a remote participant with which discovery
communication has been lost.
l
l

If the remote participant is later rediscovered, the records that remain in the database will be re-used.
If the remote participant is not rediscovered, the records will continue to take up space in the database for as long as the local participant remains in existence.

In most cases, you will not need to change this value from its default, DDS_LIVELINESS_BASED_
REMOTE_PARTICIPANT_PURGE.

591

8.5.3.3 Controlling the Reliable Protocol Used by Builtin-Topic DataWriters/DataReaders

However, DDS_NO_REMOTE_PARTICIPANT_PURGE may be a good choice if the following conditions apply:
Discovery communication with a remote participant may be lost while data communication remains intact.
This will not be the typical case if discovery takes place over the Simple Discovery Protocol, but may
occur if you are using RTI Enterprise Discovery Service. 1
Extensive and prolonged lack of discovery communication between participants is not expected to be common, either because loss of the participant will be rare, or because participants may be lost sporadically but
will typically return again.
Maintaining inter-participant liveliness is problematic, perhaps because a participant has no writers with the
appropriate LIVELINESS QosPolicy (Section 6.5.13 on page 382) kind.
8.5.3.3 Controlling the Reliable Protocol Used by Builtin-Topic DataWriters/DataReaders
The connection between the DataWriters and DataReaders for the publication and subscription builtin-topics are reliable. The publication_writer, subscription_writer, publication_reader, and subscription_reader
parameters of the DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585) configure the reliable messaging protocol used by Connext DDS for those topics. Connext DDS’s reliable
messaging protocol is discussed in Reliable Communications (Section Chapter 10 on page 629).
See also:
l

DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347)

l

DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511).

8.5.3.4 Example
Users will be most interested in setting the participant_liveliness_lease_duration and participant_liveliness_
assert_period values for their DomainParticipants. Basically, the lease duration governs how fast an application realizes another application dies unexpectedly. The shorter the periods, the quicker a DomainParticipant can determine that a remote participant is dead and act accordingly by declaring all of the remote
DataWriters and DataReaders of that participant dead as well.
However, you should realize that the shorter the period the more liveliness packets will sent by the
DomainParticipant. How many packets is also determined by the number of peers in the peer list of the
participant–whether or not the peers on the list are actually alive.

1RTI Enterprise Discovery Service is an optional package that provides participant-matching services for Connext DDS

applications.

592

8.5.3.5 Properties

8.5.3.5 Properties
This QosPolicy cannot be modified after the DomainParticipant is created.
It can be set differently on the publishing and subscribing sides.
8.5.3.6 Related QosPolicies
l
l

DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580)
DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
below)

l

WIRE_PROTOCOL QosPolicy (DDS Extension) (Section 8.5.9 on page 610)

l

DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347)

l

DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511)

l

DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 7.6.2 on page
517)

8.5.3.7 Applicable Dds Entities
l

DomainParticipants (Section 8.3 on page 547)

8.5.3.8 System Resource Considerations
Setting smaller values for time periods can increase the CPU and network bandwidth usage. Setting larger
values for maximum limits can increase the maximum memory that Connext DDS may allocate for a
DomainParticipant while increasing the initial values will increase the initial memory allocated for a
DomainParticipant.

8.5.4 DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS
Extension)
The DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy includes various settings that configure how DomainParticipants allocate and use physical memory for internal resources, including the
maximum sizes of various properties.
This QosPolicy sets maximum size limits on variable-length parameters used by the participant and its contained Entities. It also controls the initial and maximum sizes of data structures used by the participant to
store information about locally-created and remotely-discovered entities (such as DataWriters/DataReaders), as well as parameters used by the internal database to size the hash tables used by the data structures.
By default, a DomainParticipant is allowed to dynamically allocate memory as needed as users create
local Entities such as DataWriters and DataReaders or as the participant discovers new applications to
store their information. By setting fixed values for the maximum parameters in this QosPolicy, you can

593

8.5.4 DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension)

bound the memory that can be allocated by a DomainParticipant. In addition, by setting the initial values
to the maximum values, you can prevent DomainParticipants from allocating memory after the initialization period.
The maximum sizes of several variable-length parameters—such as the number of partitions that can be
stored in the PARTITION QosPolicy (Section 6.4.5 on page 323), the maximum length of data store in
the USER_DATA QosPolicy (Section 6.5.26 on page 417) and GROUP_DATA QosPolicy (Section
6.4.4 on page 320), and many others—can be changed from their defaults using this QoS. However, it is
important that all DomainParticipants that need to communicate with each other use the same set of maximum values. Otherwise, when these parameters are propagated from one DomainParticipant to another, a
DomainParticipant with a smaller maximum length may reject the parameter resulting in an error.
This QosPolicy includes the members in Table 8.12 DDS_DomainParticipantResourceLimitsQosPolicy .
For defaults and valid ranges, please refer to the API Reference HTML documentation.
Table 8.12 DDS_DomainParticipantResourceLimitsQosPolicy
Type

Field Name

Description
Each allocation structure configures how many objects of each type, _allocation, will be
allocated by the DomainParticipant.
See Configuring Resource Limits for Asynchronous DataWriters (Section 8.5.4.1 on page 600).

DDS_AllocationSettings_t

DDS_AllocationSettings_t
(see description
column)

local_writer_
allocation

{
DDS_Long initial_count;
DDS_Long max_count;
DDS_Long incremental_count;
};

See above row

local_reader_
allocation

See above row

See above row

local_
publisher_
allocation

See above row

See above row

local_
subscriber_
allocation

See above row

594

8.5.4 DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 8.12 DDS_DomainParticipantResourceLimitsQosPolicy
Type

Field Name

Description

See above row

local_topic_
allocation

See above row

remote_writer_
See above row
allocation

See above row

remote_
reader_
allocation

See above row

See above row

remote_
participant_
allocation

See above row

See above row

matching_
writer_reader_ See above row
pair_allocation

See above row

matching_
reader_writer_ See above row
pair_allocation

See above row

ignored_
entity_
allocation

See above row

See above row

content_
filtered_topic_
allocation

See above row

See above row

content_filter_
allocation

See above row

See above row

read_
condition_
allocation

See above row

See above row

query_
condition_
allocation

See above row

See above row

outstanding_
asynchronous_
See above row
sample_
allocation

See above row

595

8.5.4 DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 8.12 DDS_DomainParticipantResourceLimitsQosPolicy
Type

Field Name

Description

See above row

flow_
controller_
allocation

See above row

DDS_
DomainParticipant
ResourceLimits
IgnoredEntity
ReplacementKind

ignored_
entity_
replacement_
kind

Sets the kinds of entities allowed to be replaced when a DomainParticipant reaches ignored_entity_
allocation.max_count. See Resource Limits Considerations for Ignored Entities (Section 16.4.4 on
page 788).

DDS_Long

local_writer_
hash_buckets

Used to configure the hash tables used for database searches. If these numbers are too large then
memory is wasted. If these number are too small, searching for an object will be less efficient.

DDS_Long

local_reader_
hash_buckets

See above row

DDS_Long

local_
publisher_
hash_buckets

See above row

DDS_Long

local_
subscriber_
hash_buckets

See above row

DDS_Long

local_topic_
hash_buckets

See above row

DDS_Long

remote_writer_
See above row
hash_buckets

DDS_Long

remote_
reader_hash_
buckets

See above row

DDS_Long

remote_
participant_
hash_buckets

See above row

DDS_Long

matching_
writer_reader_
See above row
pair_
hash_buckets

DDS_Long

matching_
reader_writer_
See above row
pair_
hash_buckets

596

8.5.4 DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 8.12 DDS_DomainParticipantResourceLimitsQosPolicy
Type

Field Name

Description

DDS_Long

ignored_
entity_hash_
buckets

See above row

DDS_Long

content_
filtered_topic_
hash_buckets

See above row

DDS_Long

content_filter_
hash_buckets

See above row

DDS_Long

flow_
controller_
hash_buckets

See above row

DDS_Long

max_gather_
destinations

Configures the maximum number of destinations that a message can be addressed in a single network
send operation. Can improve efficiency if the underlying transport support can send to multiple
destinations.

DDS_Long

participant_
user_data_
max_length

Controls the maximum lengths of USER_DATA QosPolicy (Section 6.5.26 on page 417), TOPIC_
DATA QosPolicy (Section 5.2.1 on page 209) and GROUP_DATA QosPolicy (Section 6.4.4 on
page 320) for different entities.
Must be configured to be the same values on all DomainParticipants in the same DDS domain.

DDS_Long

topic_data_
max_length

See above row

DDS_Long

publisher_
group_data_
max_length

See above row

DDS_Long

subscriber_
group_data_
max_length

See above row

DDS_Long

writer_user_
data_max_
length

See above row

DDS_Long

reader_user_
data_max_
length

See above row

DDS_Long

Controls the maximum number of partitions that can be assigned to a Publisher or Subscriber with the
max_partitions PARTITION QosPolicy (Section 6.4.5 on page 323).
Must be configured to be the same value on all DomainParticipants in the same DDS domain.

597

8.5.4 DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 8.12 DDS_DomainParticipantResourceLimitsQosPolicy
Type

Field Name

Description

DDS_Long

max_partition_ Controls the maximum number of combined characters among all partition names in the PARTITION
QosPolicy (Section 6.4.5 on page 323).
cumulative_
characters
Must be configured to be the same value on all DomainParticipants in the same DDS domain.

DDS_Long

type_code_
max_
serialized_
length

Maximum size of serialized string for type code.
If your data type has an especially complex type code, you may need to increase this value. See Using
Generated Types without Connext DDS (Standalone) (Section 3.7 on page 139).
Maximum length, in bytes, that the buffer to serialize TypeObject can consume.

DDS_Long

type_object_
max_
serialized_
length

This parameter limits the size of the TypeObject that a DomainParticipant is able to propagate. Since
TypeObjects contain all of the information of a data structure, including the strings that define the
names of the members of a structure, complex data-structures can result in TypeObjects larger than the
default maximum. This field allows you to specify a larger value.
Cannot be unlimited.

DDS_Long

type_object_
max_
deserialized_
length

Maximum number of bytes that a deserialized TypeObject can consume.
This parameter limits the size of the TypeObject that a DomainParticipant is able to store.

DDS_Long

deserialized_
type_object_
dynamic_
allocation_
threshold

DDS_Long

contentfilter_
property_max_ Maximum length of all data related to ContentFilteredTopics (Section 5.4 on page 212).
length

DDS_Long

channel_seq_
max_length

DDS_Long

channel_filter_
Maximum length of a channel filter_expression in a DataWriter’sMULTI_CHANNEL QosPolicy
expression_
(DDS Extension) (Section 6.5.14 on page 386).
max_length

DDS_Long

participant_
property_list_
max_length

Maximum number of properties ((name, value) pairs) that can be stored in the DomainParticipant’s
PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394).

DDS_Long

participant_
property_
string_max_
length

Maximum cumulative length (in bytes, including the null terminating characters) of all the (name,
value) pairs in a DomainParticipant’s Property QosPolicy.

Threshold, in bytes, for dynamic memory allocation for the deserialized TypeObject. Above it, the
memory for a TypeObject is allocated dynamically. Below it, the memory is obtained from a pool of
fixed-size buffers. The size of the buffers is equal to this threshold.

Maximum number of channels that can be specified in a DataWriter’s MULTI_CHANNEL
QosPolicy (DDS Extension) (Section 6.5.14 on page 386).

598

8.5.4 DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension)

Table 8.12 DDS_DomainParticipantResourceLimitsQosPolicy
Type

Field Name

Description

DDS_Long

writer_
property_list_
max_length

Maximum number of properties ((name, value) pairs) that can be stored in a DataWriter’s Property
QosPolicy.

DDS_Long

writer_
property_
string_max_
length

Maximum cumulative length (in bytes, including the null terminating characters) of all the (name,
value) pairs in a DataWriter’s Property QosPolicy.

DDS_Long

reader_
property_list_
max_length

Maximum number of properties ((name, value) pairs) that can be stored in a DataReader’s Property
QosPolicy.

DDS_Long

reader_
property_
string_max_
length

Maximum cumulative length (in bytes, including the null terminating characters) of all the (name,
value) pairs in a DataReader’s Property QosPolicy.

max_
endpoint_
groups

Maximum number of endpoint groups allowed in an DATA_READER_PROTOCOL QosPolicy
(DDS Extension) (Section 7.6.1 on page 511) .

max_
endpoint_
group_
cumulative_
characters

Maximum number of combined role_name characters allowed in all endpoint groups in an
AvailabilityQosPolicy. The maximum number of combined characters should account for a
terminating NULL ('') character for each role_name string.

DDS_Long

When sending DomainParticipant discovery information, this value defines the maximum number of
transports whose properties will be announced to other DomainParticipants.
If a DomainParticipant has three transports installed and this value is two, the DomainParticipant
will only announce information about the first two transports.

DDS_Long

When receiving DomainParticipant information, this value defines the maximum size of the list
transport_
containing information about the transports installed in a remote DomainParticipant.
info_list_max_
length
The information about the transports installed in a DomainParticipant is made available to remote
DomainParticipants through the sequence field transport_info in the Participant Built-in Topic’s Data
(see Table 16.1 Participant Built-in Topic’s Data Type (DDS_ParticipantBuiltinTopicData)
Setting this value to 0 disables the capability of Connext DDS to detect and report transport
misconfigurations. However, it does not affect the capability of reaching a given DomainParticipant
in all transports available on that DomainParticipant.

Most of the parameters for this QosPolicy are described in the Description column of the table. However,
you may need to refer to the sections listed in the column to fully understand the context in which the parameter is used.

599

8.5.4.1 Configuring Resource Limits for Asynchronous DataWriters

An important parameter in this QosPolicy that is often changed by users is the type_code_max_serialized_length. This parameter limits the size of the type code that a DomainParticipant is able to store and
propagate for user data types. Type codes can be used by external applications to understand user data
types without having the data type predefined in compiled form. However, since type codes contain all of
the information of a data structure including the strings that define the names of the members of a structure,
complex data structures can result in type codes larger than the default maximum of 2048 bytes. Thus it is
common for users to set this parameter to a larger value. However, as with all parameters in this QosPolicy
defining maximum sizes for variable-length elements, all DomainParticipants should set the same value
for type_code_max_serialized_length.
The  hash_buckets configure the hash-table data structure that is used to efficiently search the
database. The optimal number of buckets depend on the actual number of objects that will be stored in the
hash table. So if you know how many DataWriters will be created in a DomainParticipant, you may
change the value of local_writer_hash_buckets to balance memory usage against search efficiency. A smaller value will use up less memory, but a larger value will make database lookups for the object more efficient.
If you modify any of the _data_max_length, max_partitions, or max_partition_cummulative_
characters parameters, then you must make sure that they are modified to be the same value for all
DomainParticipants in the same DDS domain for all applications. If they are different and an application
sends data that is larger than another application is configure to hold, then the two Entities, whether a
matching DataWriter/DataReader pair or even two DomainParticipants will fail to connect.
8.5.4.1 Configuring Resource Limits for Asynchronous DataWriters
When using an asynchronous Publisher, if a call to write() is blocked due to a resource limit, the block
will last until the timeout period expires, which will prevent others from freeing the resource. To avoid this
situation, make sure that the DomainParticipant’s resource_limits.outstanding_asynchronous_sample_
allocation is always greater than the sum of all asynchronous DataWriters’ resource_limits.max_samples
(see RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)).
8.5.4.2 Configuring Memory Allocation
The _allocation configures the number of ’s that can be stored in the internal
Connext DDS database. For example, local_writer_allocation configures how many local DataWriters can
be created for the DomainParticipant.
The DDS_AllocationSettings_t structure sets the initial and maximum number of each object type that can
be stored. Memory is allocated for the storage of the objects, thus initial_count will determine how much
memory is initially allocated, and max_count will determine the maximum amount of memory that Connext DDS is allowed to allocate. The incremental_count is used to allocate more memory in chunks when
the number of objects created exceed the initial_count.

600

8.5.4.3 Example

You should modify these parameters only if you want to decrease the initial memory used by Connext
DDS when a DomainParticipant is created or increase the maximum number of local and remote Entities
that can be stored in a DomainParticipant.
How Connext DDS is allowed to allocate memory for a DomainParticipant after initialization depends on
how you set these parameters.
1. Static memory allocation
No memory is allocated by Connext DDS after creation. Set initial_count = max_count. The incremental_count should be set to 0.
l Advantage: All memory allocation is done when creating the DomainParticipant; no dynamic
allocation during run-time. You know immediately if you have enough memory to run in that
configuration.
l

Disadvantage: Requires a fairly static system and/or good estimates on the number of Entities
in the distributed system. Connext DDS will fail to execute properly once the number of Entities exceed the configure bounds.

2. Dynamic, bounded allocation
Set initial_count to configure the initial amount of memory to be allocated. Set max_count to the
maximum allowable upper bound (see the API Reference HTML documentation).
l Advantage: Initial memory usage may be lower and memory is allocated as needed and only
if needed.
l

Disadvantage: Connext DDS may allocate memory dynamically which may have an impact
on performance.

If you allow Connext DDS to allocate memory dynamically, you can either:
l Use fixed-size increments (set incremental_count to the desired fixed size).
l Advantage: well known amount of memory allocated each time.
l

l

Disadvantage: may require more frequent allocations.

Double the amount of extra memory allocated each time memory is needed (set incremental_
count to -1).
l Advantage: requires fewer allocations.
l

Disadvantage: may allocate considerably more memory than is really needed.

8.5.4.3 Example
For most applications, the default values for this QosPolicy may be sufficient. However, if an application
uses the PARTITION, USER_DATA, TOPIC_DATA, or GROUP_DATA QosPolicies, the default maximum sizes of the data associated with those policies may need to be adjusted as required by the

601

8.5.4.4 Properties

application. As noted previously, you must make sure that all DomainParticipants in the same DDS
domain use the same sets of values or it is possible that Connext DDS will not successfully connect two
Entities.
8.5.4.4 Properties
This QosPolicy cannot be modified after the DomainParticipant is created.
It can be set differently on the publishing and subscribing sides.
8.5.4.5 Related QosPolicies
l

DATABASE QosPolicy (DDS Extension) (Section 8.5.1 on page 577)

l

DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585)

l

MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386)

l

USER_DATA QosPolicy (Section 6.5.26 on page 417)

l

TOPIC_DATA QosPolicy (Section 5.2.1 on page 209)

l

GROUP_DATA QosPolicy (Section 6.4.4 on page 320)

l

PARTITION QosPolicy (Section 6.4.5 on page 323)

l

PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394)

8.5.4.6 Applicable DDS Entities
l

DomainParticipants (Section 8.3 on page 547)

8.5.4.7 System Resource Considerations
Memory and CPU usage are directly affected by the values set for parameters of this QosPolicy. See the
detailed descriptions above for specifics.

8.5.5 EVENT QosPolicy (DDS Extension)
The EVENT QosPolicy configures the internal Connext DDS Event thread.
This QoS allows the you to configure thread properties such as priority level and stack size. You can also
configure the maximum number of events that can be posted to the event thread. It contains the members
in Table 8.13 DDS_EventQoSPolicy. For defaults and valid ranges, please refer to the API Reference
HTML documentation.

602

8.5.5.1 Example

Table 8.13 DDS_EventQoSPolicy
Type

Field
Name

Description

thread.mask
Thread settings for the event thread used by Connext DDS to wake up for a timed event and possibly
DDS_
thread.priority execute listener callbacks. The values used for these settings are OS-dependent; see the RTI Connext
ThreadSettings_
DDS Core Libraries Platform Notes for details.
t
thread.stack_
Note: thread.cpu_list and thread.cpu_rotation are not relevant in this QoS policy.
size
DDS_Long

initial_count

Initial number of events that can be stored simultaneously.

DDS_Long

max_count

Maximum number of events that can be stored simultaneously.

The Event thread is used to wake up and execute timed events posted to the event queue. In a DomainParticipant, different Entities may have constraints that have to be checked at periodic intervals or at specific
times. If the constraint is violated, a callback function may need to be executed. Timed events include
checking for timeouts and deadlines, and executing internal and user timeout or exception handling
routines/callbacks. A combination of a time, constraint, and callback can be considered to be an event. For
more information, see Event Thread (Section 19.2 on page 838).
For example, a DataReader may have a constraint that requires data to be received within a period of time
specified by the DEADLINE QosPolicy (Section 6.5.5 on page 363). For that DataReader, an event is
stored by the Event thread so that it will wake up periodically to check to see if data has arrived in time. If
not, the Event thread will execute the on_requested_deadline_missed() Listener callback of the
DataReader (if it was installed and enabled).
A reliable connection between a DataWriter and DataReader will also post events for sending heartbeats
used in the reliable protocol discussed in Reliable Communications (Section Chapter 10 on page 629).
This QoS configures the parameters associated with thread creation as well as the number of events that
can be simultaneously stored by the Event thread.
8.5.5.1 Example
In a real-time operating system, the priority of the Event thread should be set relative to the priority of the
events that it must handle. For example, you may want the Event thread to have a high priority if the deadlines and callbacks that it handles are time or safety critical. It may be critical that the data of a particular
DataReader arrives on time or if not, alternative action is taken with minimal latency.
If you create many Entities in a DomainParticipant with QosPolicies that will post events that check deadlines, liveliness or send heartbeats, then you may need to increase the maximum number of events that can
be stored by the Event thread.
If your application is sending a lot of reliable data, you should increase the event thread priority to be
higher than the sending thread priority.
603

8.5.5.2 Properties

8.5.5.2 Properties
This QosPolicy cannot be modified after the DomainParticipant is created.
It can be set differently on the publishing and subscribing sides.
8.5.5.3 Related QosPolicies
l

DATABASE QosPolicy (DDS Extension) (Section 8.5.1 on page 577)

l

RECEIVER_POOL QosPolicy (DDS Extension) (Section 8.5.6 below)

8.5.5.4 Applicable DDS Entities
l

DomainParticipants (Section 8.3 on page 547)

8.5.5.5 System Resource Considerations
Increasing initial_count and max_count will increase initial and maximum memory used for storing
events.
Setting the thread parameters correctly on a real-time operating system is usually critical to the proper overall functionality of the applications on that system. Larger values for the thread.stack_size parameter will
use up more memory.
By default, a DomainParticipant will dynamically allocate memory as needed for events posted to the
event thread. However, by setting an maximum value or setting the initial and maximum value to be the
same, you can either bound the amount of memory allocated for the event thread or prevent a DomainParticipant from dynamically allocating memory for the event thread after initialization.

8.5.6 RECEIVER_POOL QosPolicy (DDS Extension)
The RECEIVER_POOL QosPolicy configures the internal Connext DDS thread used to process the data
received from a transport. The Receive thread is described in detail in Receive Threads (Section 19.3 on
page 839).
This QosPolicy contains the members in Table 8.14 DDS_ReceiverPoolQoSPolicy.

604

8.5.6 RECEIVER_POOL QosPolicy (DDS Extension)

Table 8.14 DDS_ReceiverPoolQoSPolicy
Type

Field
Name

Description

thread.mask
thread.priority

Thread settings for the receive thread(s) used by Connext DDS to process data received from a
struct DDS_
thread.stack_ transport. The values used for these settings are OS-dependent; see the RTI Connext DDS Core Libraries
ThreadSettings_ size
Platform Notes for details.
t
hread.cpu_list See also: Controlling CPU Core Affinity for RTI Threads (Section 19.5 on page 842).
thread.cpu_
rotation
Size of the receive buffer in bytes. For the default and valid range, see the API Reference HTML
documentation.
buffer_size must always be at least as large as the maximum message_size_max of any installed nonzero-copy transport.1
DDS_Long

buffer_size

DDS_Long

buffer_
alignment

The buffer_size can be adjusted automatically by the middleware by configuring its value to DDS_
LENGTH_AUTO (in C/C++) or ReceiverPoolQosPolicy.LENGTH_AUTO (in .NET and Java). When
set to this AUTO default value, the effective value will automatically be set to the largest message_size_
max of all installed transports, without needing any other configuration. Therefore you should not need
to change this value.
Byte-alignment of the receive buffer. For the default and valid range, see the API Reference HTML
documentation.

This QosPolicy sets the thread properties, like priority level and stack size, for the threads used to receive
and process data from transports. Connext DDS uses a separate receive thread per port per transport plugin. To force Connext DDS to use a separate thread to process the data for a DataReader, you should set a
unique port for the TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412)
or TRANSPORT_MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on page 529) for the
DataReader.
Connext DDS creates at least one thread for every transport that is installed and enabled for use by the
DomainParticipant for receiving data. These threads are used to process data DDS samples received for
the participant’s DataReaders, as well as messages used by Connext DDS itself in support of the application discovery process discussed in Discovery (Section Chapter 14 on page 709).
The user application may configure Connext DDS to create many more threads for receiving data sent via
multicast or even to dedicate a thread to process the DDS data samples of a single DataReader received on
a particular transport. This QosPolicy is used in the creation of all receive threads.

1A “zero-copy transport” does not use the receive buffer. A transport is zero-copy if the properties_bitmap property in the

DDS_Transport_Property_t is NDDS_TRANSPORT_PROPERTY_BIT_BUFFER_ALWAYS_LOANED. The only built-in
transport that supports zero-copy is the UDPv4 transport on VxWorks platforms.

605

8.5.6.1 Example

8.5.6.1 Example
When new data arrives on a transport, the receive thread may invoke the on_data_available() of the
Listener callback of a DataReader. Thus, you may want to adjust the priority of the receive threads with
respect to the other threads in the application as appropriate for the proper operation of the system.
8.5.6.2 Properties
This QosPolicy cannot be modified after the DomainParticipant is created.
It can be set differently on the publishing and subscribing sides.
8.5.6.3 Related QosPolicies
l

DATABASE QosPolicy (DDS Extension) (Section 8.5.1 on page 577)

l

EVENT QosPolicy (DDS Extension) (Section 8.5.5 on page 602)

8.5.6.4 Applicable Dds Entities
l

DomainParticipants (Section 8.3 on page 547)

8.5.6.5 System Resource Considerations
Increasing the buffer_size will increase memory used by a receive thread.
Setting the thread parameters correctly on a real-time operating system is usually critical to the proper overall functionality of the applications on that system. Larger values for the thread.stack_size parameter will
use up more memory.

8.5.7 TRANSPORT_BUILTIN QosPolicy (DDS Extension)
Connext DDS comes with three different transport plugins built into the core libraries (for most supported
target platforms). These are plugins for UDPv4, shared memory, and UDPv6.
This QosPolicy allows you to control which built-in transport plugins are used by a DomainParticipant.
By default, only the UDPv4 and shared memory plugins are enabled (for most platforms; on some platforms, the shared memory plugin is not available). You can disable one or all of the builtin transports.
In some cases, users will disable the shared memory transport when they do not want applications to use
shared memory to communicate when running on the same node.
If one application is configured to use UDPv4 and shared memory, while another application is
only configured for UDPv4, and these two applications run on the same node, they will not
communicate. This is due to an internal optimization which will default to use shared memory

606

8.5.7.1 Example

instead of loopback. However if the other peer application does not enable shared memory, there is
no common transport; therefore they will not communicate.
It contains the member in Table 8.15 DDS_TransportBuiltinQosPolicy. For the default and valid values,
please refer to the API Reference HTML documentation.
Table 8.15 DDS_TransportBuiltinQosPolicy
Type
DDS_TransportBuiltinKindMask

Field Name
mask

Description
A mask with bits that indicate which built-in transports will be installed.

Please see the API Reference HTML documentation (select Modules, RTI Connext DDS API Reference, Pluggable Transports, Using Transport Plugins and Built-in Transport Plugins) for more
information.
Note: Currently,

Connext DDS will only listen for discovery traffic on the first multicast address (element 0)
in multicast_receive_addresses.
8.5.7.1 Example
See System Resource Considerations (Section 8.5.7.5 on the facing page) for an example of why you
may want to use this QosPolicy.
In addition, customers may wish to install and use their own custom transport plugins instead of any of the
builtin transports. In that case, this QosPolicy may be used to disable all builtin transports.
8.5.7.2 Properties
This QosPolicy cannot be modified after the DomainParticipant is created.
It can be set differently on the publishing and subscribing sides.
8.5.7.3 Related QosPolicies
l

TRANSPORT_SELECTION QosPolicy (DDS Extension) (Section 6.5.23 on page 411)

l

TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412)

l

TRANSPORT_MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on page 529)

8.5.7.4 Applicable DDS Entities
l

DomainParticipants (Section 8.3 on page 547)

607

8.5.7.5 System Resource Considerations

8.5.7.5 System Resource Considerations
You can save memory and other system resources if you disable the built-in transports that your application will not use. For example, if you only run a single application with a single DomainParticipant on
each machine in your network, then you can disable the shared memory transport since your applications
will never use it to send or receive messages.

8.5.8 TRANSPORT_MULTICAST_MAPPING QosPolicy (DDS Extension)
The multicast address on which a DataReader wants to receive its data can be explicitly configured using
the TRANSPORT_MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on page 529). However in
systems with many multicast addresses, managing the multicast configuration can become cumbersome.
The TransportMulticastMapping QosPolicy is designed to make configuration and assignment of the
DataReader's multicast addresses more manageable. When using this QosPolicy, the middleware will automatically assign a multicast receive address for a DataReader from a range by using configurable mapping
rules.
DataReaders can be assigned a single multicast receive address using the rules defined in this QosPolicy
on the DomainParticipant. This multicast receive address is exchanged during simple discovery in the
same manner used when the multicast receive address is defined explicitly. No additional configuration on
the writer side is needed.
Mapping within a range is done through a mapping function. The middleware provides a default hash
(md5) mapping function. This interface is also pluggable, so you can specify a custom mapping function to
minimize collisions.
To use this QosPolicy, you must set the kind in the TRANSPORT_MULTICAST QosPolicy
(DDS Extension) (Section 7.6.5 on page 529) to AUTOMATIC.
This QosPolicy contains the member in Table 8.16 DDS_TransportMulticastMappingQosPolicy.
Table 8.16 DDS_TransportMulticastMappingQosPolicy
Type

Field
Name

DDS_
TransportMapping value
SettingsSeq

Description

A sequence of multicast communication settings, each of which has the format shown in Table 8.17
DDS_TransportMulticastSettings_t.

608

8.5.8.1 Formatting Rules for Addresses

Table 8.17 DDS_TransportMulticastSettings_t
Type

Field
Name

Description

A string containing a comma-separated list of IP addresses or IP address ranges to be used to receive
addresses multicast traffic for the entity with a topic that matches the topic_expression.

char *

See Formatting Rules for Addresses (Section 8.5.8.1 below).
A regular expression used to map topic names to corresponding addresses.
topic_
expression See SQL Extension: Regular Expression Matching (Section 5.4.6.5 on page 228).

char *

DDS_
TransportMulticast mapping_
MappingFunction_ function
t

Optional. Defines a user-provided pluggable mapping function. See Table 8.18 DDS_
TransportMulticastMappingFunction_t.

Table 8.18 DDS_TransportMulticastMappingFunction_t
Type

Field
Name

Description
Specifies a dynamic library that contains a mapping function.

char * dll

You may specify a relative or absolute path.
If the name is specified as "foo", the library name on Linux systems will be libfoo.so; on Windows systems it will be
foo.dll.
Specifies the name of a mapping function in the library specified in the above dll.
The function must implement the following interface:

char *

function_
name

int function(const char* topic_name, int numberOfAddresses);

The function must return an integer that indicates the index of the address to use for the given topic_name. For
example, if the first address in the list should be used, it must return 0; if the second address in the list should be used, it
must return 1, etc.

8.5.8.1 Formatting Rules for Addresses
l

l

l

The string must contain IPv4 or IPv6 addresses separated by commas. For example:
"239.255.100.1,239.255.100.2,239.255.100.3"
You may specify ranges of addresses by enclosing the start and end addresses in square brackets.
For example: "[239.255.100.1,239.255.100.3]".
You may combine the two approaches.
For example: "239.255.200.1,[239.255.100.1,239.255.100.3], 239.255.200.3"

609

8.5.8.2 Example

l
l

l
l

IPv4 addresses must be specified in Dot-decimal notation.
IPv6 addresses must be specified using 8 groups of 16-bit hexadecimal values separated by colons.
For example: FF00:0000:0000:0000:0202:B3FF:FE1E:8329.
Leading zeroes can be skipped. For example: FF00:0:0:0:202:B3FF:FE1E:8329.
You may replace a consecutive number of zeroes with a double colon, but only once within an
address. For example: FF00::202:B3FF:FE1E:8329.

8.5.8.2 Example
This QoS policy configures the multicast ranges and mapping rules at the DomainParticipant level. You
can configure a large set of multicast addresses on the DomainParticipant.
In addition, you can configure a mapping between topic names and multicast addresses. For example,
topic "A" can be assigned to address 239.255.1.1 and topic "B" can be assigned to address 239.255.1.2.
This configuration is quite flexible. For example, you can specify mappings between a subset of topics to a
range of multicast addresses. For example, topics "X", "Y" and Z" can be mapped to [239.255.1.1,
239.255.1.255], or using regular expressions, "X*" and "B-Z" can be mapped to a sub-range of addresses.
See SQL Extension: Regular Expression Matching (Section 5.4.6.5 on page 228).
8.5.8.3 Properties
This QosPolicy cannot be modified after the DomainParticipant is created.
8.5.8.4 Related QosPolicies
l

TRANSPORT_MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on page 529)

8.5.8.5 Applicable DDS Entities
l

DomainParticipants (Section 8.3 on page 547)

8.5.8.6 System Resource Considerations
See System Resource Considerations (Section 7.6.5.5 on page 532).

8.5.9 WIRE_PROTOCOL QosPolicy (DDS Extension)
The WIRE_PROTOCOL QosPolicy configures some global Real-Time Publish Subscribe (RTPS) protocol-related properties for the DomainParticipant. The RTPS OMG-standard, interoperability protocol is
used by Connext DDS to format and interpret messages between DomainParticipants.
It includes the members in Table 8.19 DDS_WireProtocolQosPolicy. For defaults and valid ranges, please
refer to the API Reference HTML documentation. (The default values contain the correctly initialized wire

610

8.5.9.1 Choosing Participant IDs

protocol attributes. They should not be modified without an understanding of the underlying Real-Time
Publish Subscribe (RTPS) wire protocol.)
Table 8.19 DDS_WireProtocolQosPolicy
Type

DDS_Long

DDS_UnsignedLong

Field
Name

Description

participant_ Unique identifier for participants that belong to the same DDS domain on the same host.
id
See Choosing Participant IDs (Section 8.5.9.1 below).
rtps_host_
id

A machine/OS-specific host ID, unique in the DDS domain. See Host, App, and
Instance IDs (Section 8.5.9.2 on page 613).

rtps_app_
id

A participant-specific ID, unique within the scope of the rtps_host_id. See Host, App, and
Instance IDs (Section 8.5.9.2 on page 613).

An instance-specific ID of the DomainParticipant that, together with the rtps_app_id, is
rtps_
unique within the scope of the rtps_host_id. See Host, App, and Instance IDs (Section
instance_id
8.5.9.2 on page 613).

DDS_RtpsWellKnownPorts_t

rtps_well_
known
_ports

Determines the well-known multicast and unicast ports for discovery and user traffic. See
Ports Used for Discovery (Section 8.5.9.3 on page 613).

DDS_
RtpsReservedPortKindMask

rtps_
reserved_
ports
_mask

Specifies which well-known multicast and unicast ports to reserve when enabling the
DomainParticipant.

DDS_
rtps_auto_
WireProtocolQosPolicyAutoKind id_kind

Kind of auto mechanism used to calculate the GUID prefix.

Note that DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347) and
DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511) configure
RTPS and reliability properties on a per DataWriter and DataReader basis.
8.5.9.1 Choosing Participant IDs
When you create a DomainParticipant, you must specify a domain ID, which identifies the communication channel across the whole system. Each DomainParticipant in the same DDS domain on the
same host also needs a unique integer, known as the participant_id.
The participant_id uniquely identifies a DomainParticipant from other DomainParticipants in the same
DDS domain on the same host. You can use the same participant_id value for DomainParticipants in the
same DDS domain but running on different hosts.

611

8.5.9.1 Choosing Participant IDs

The participant_id is also used to calculate the default unicast user-traffic and the unicast meta-traffic port
numbers, as described in Ports Used for Discovery (Section 14.5 on page 738). If you only have one
DomainParticipant in the same DDS domain on the same host, you will not need to modify this value.
You can either allow Connext DDS to select a participant ID automatically (by setting participant_id to 1), or choose a specific participant ID (by setting participant_id to the desired value).
l

Automatic Participant ID Selection
The default value of participant_id is -1, which means Connext DDS will select a participant ID for
you.
Connext DDS will pick the smallest participant ID, based on the unicast ports available on the transports enabled for discovery, based on the unicast and/or multicast ports available on the transports
enabled for discovery and/or user traffic.
The rtps_reserved_ports_mask field determines which ports to check when picking the next available participant ID. The reserved ports are calculated based on the formula specified in Inbound
Ports for Meta-Traffic (Section 14.5.1 on page 739) an Inbound Ports for User Traffic (Section
14.5.2 on page 740). By default, Connext DDS will reserve the meta-traffic unicast port, the metatraffic multicast port, and the user traffic unicast port.
Connext DDS will attempt to resolve an automatic port ID either when a DomainParticipant is
enabled, or when a DataReader or a DataWriter is created. Therefore, all the transports enabled for
discovery must have been registered by this time. Otherwise, the discovery transports registered after
resolving the automatic port index may produce port conflicts when the DomainParticipant is
enabled.
To see what value Connext DDS has selected, either:
l Change the verbosity level of the NDDS_CONFIG_LOG_CATEGORY_API category to
NDDS_CONFIG_LOG_VERBOSITY_STATUS_LOCAL (see Controlling Messages
from Connext DDS (Section 21.2 on page 865)).
l

Call get_qos() and look at the participant_id value in the WIRE_PROTOCOL QosPolicy
(DDS Extension) (Section 8.5.9 on page 610) after the DomainParticipant is enabled.

612

8.5.9.2 Host, App, and Instance IDs

l

Manual Participant ID Selection
If you do have multiple DomainParticipants on the same host, you should use consecutively
numbered participant indices start from 0. This will make it easier to specify the discovery peers
using the initial_peers parameter of this QosPolicy or the NDDS_DISCOVERY_PEERS environment variable. See Configuring the Peers List Used in Discovery (Section 14.2 on page 711) for
more information.
Do not use random participant indices since this would make DISCOVERY incredibly difficult to
configure. In addition, the participant_id has a maximum value of 120 (and will be less for domain
IDs other than 0) when using an IP-based transport since the participant_id is used to create the port
number (see Ports Used for Discovery (Section 14.5 on page 738)), and for IP, a port number cannot be larger than 65536.

For details, see Ports Used for Discovery (Section 14.5 on page 738).
8.5.9.2 Host, App, and Instance IDs
The rtps_host_id, rtps_app_id, and rtps_instance_id values are used by the RTPS protocol to allow
Connext DDS to distinguish messages received from different DomainParticipants. Their combined values must be globally unique across all existing DomainParticipants in the same DDS domain. In addition,
if an application dies unexpectedly and is restarted, the IDs used by the new instance of DomainParticipants should be different than the ones used by the previous instances. A change in these values allows
other DomainParticipants to know that they are communicating with a new instance of an application, and
not the previous instance.
If the value of rtps_host_id is set to DDS_RTPS_AUTO_ID, the IPv4 address of the host is used as the
host ID. If the host does not have an IPv4 address, the host-id will be automatically set to 0x7F000001.
If the value of rtps_app_id is set to DDS_RTPS_AUTO_ID, the process (or task) ID is used. There can
be at most 256 distinct participants in a shared address space (process) with a unique rtps_app_id.
If the value of rtps_instance_id is set to DDS_RTPS_AUTO_ID, a counter is assigned that is incremented per new participant. Thus, together with rtps_app_id, there can be at most 2^64 distinct participants in a shared address space with a unique RTPS Globally Unique Identifier (GUID).
8.5.9.3 Ports Used for Discovery
The rtps_well_known_ports structure allows you to configure the ports that are used for discovery of
inbound meta-traffic (discovery data internal to Connext DDS) and user traffic (from your application).
It includes the members in Table 8.20 DDS_RtpsWellKnownPorts_t. For defaults and valid ranges, please
refer to the API Reference HTML documentation.

613

8.5.9.4 Controlling How the GUID is Set (rtps_auto_id_kind)

Table 8.20 DDS_RtpsWellKnownPorts_t
Type

Field Name
port_base

Description
The base port offset. All mapped well-known ports are offset by this value. Resulting ports must be within
the range imposed by the underlying transport.

domain_id_gain
Tunable gain parameters. See Ports Used for Discovery (Section 14.5 on page 738).
participant_id_gain

DDS_
Long

builtin_multicast_
port_offset
Additional offset for meta-traffic port. See Inbound Ports for Meta-Traffic (Section 14.5.1 on page 739).
builtin_unicast_
port_offset
user_multicast_
port_offset
Additional offset for user traffic port. See Inbound Ports for User Traffic (Section 14.5.2 on page 740).
user_unicast_port_
offset

8.5.9.4 Controlling How the GUID is Set (rtps_auto_id_kind)
In order for the discovery process to work correctly, each DomainParticipant must have a unique identifier. This QoS policy specifies how that identifier should be generated.
RTPS defines a 96-bit prefix to this identifier; each DomainParticipant must have a unique value of this
prefix relative to all other participants in its DDS domain. In order to make it easier to control how this 96bit value is generated, Connext DDS divides it into three integers: a host ID, the value of which is based
on the identity of the machine on which the participant is executing, an application ID (whose value is
based on the process or task in which the participant is contained), and an instance ID which identifies the
participant itself.
This QoS policy provides you with a choice of algorithms for generating these values automatically. In
case none of these algorithms suit your needs, you may also choose to specify some or all of them yourself.
The following three fields compose the GUID prefix and by default are set to DDS_RTPS_AUTO_ID.
The meaning of this flag depends on the value assigned to rtps_auto_id_kind.
l

rtps_host_id

l

rtps_app_id

l

rtps_instance_id

614

8.5.9.4 Controlling How the GUID is Set (rtps_auto_id_kind)

Depending on the rtps_auto_id_kind value, there are three different scenarios:
1. In the default and most common scenario, rtps_auto_id_kind is set to DDS_RTPS_AUTO_ID_
FROM_IP. Doing so, each field is interpreted as follows:
l rtps_host_id: the 32 bit value of the IPv4 of the first up and running interface of the host
machine is assigned
l

rtps_app_id: the process (or task) ID is assigned

l

rtps_instance_id: A counter is assigned that is incremented per new participant

Note: If the IP address assigned to the interface is not unique within the network (for instance, if it
is not configured), then is it possible that the GUID (specifically, the rtps_host_id portion) may also
not be unique.
2. In this scenario, Connext DDS rtps_auto_id_kind: is set to DDS_RTPS_AUTO_ID_FROM_
MAC. As the name suggests, this alternative mechanism uses the MAC address instead of the IPv4
address. Since the MAC address size is up to 64 bits, the logical mapping of the host information,
the application ID, and the instance identifiers has to change.
Note to Solaris Users: To use DDS_RTPS_AUTO_ID_FROM_MAC, you must run the Connext
DDS application while logged in as ‘root.’
Using DDS_RTPS_AUTO_ID_FROM_MAC, the default value of each field is interpreted as follows:
l rtps_host_id: the first 32 bits of the MAC address of the first up and running interface of the
host machine are assigned
l

l

rtps_app_id: the last 32 bits of the MAC address of the first up and running interface of the
host machine are assigned
rtps_instance_id: this field is split into two different parts. The process (or task) ID is
assigned to the first 24 bits. A counter is assigned to the last 8 bits. This counter is incremented per new participant. In both scenarios, you can change the value of each field independently.

If DDS_RTPS_AUTO_ID_FROM_MAC is used, the rtps_instance_id has been logically split
into two parts: 24 bits for the process/task ID and 8 bits for the per new participant counter. To give
to users the ability to manually set the two parts independently, a bit field mechanism has been introduced for the rtps_instance_id field when it is used in combination with DDS_RTPS_AUTO_ID_
FROM_MAC. If one of the two parts is set to 0, only this part will be handled by Connext DDS
and you will be able to handle the other one manually.
3. In this scenario, rtps_auto_id_kind is set to RTPS_AUTO_ID_FROM_UUID. As the name suggests, this alternative mechanism uses a unique, randomly generated UUID to fill the rtps_host_id,

615

8.5.9.4 Controlling How the GUID is Set (rtps_auto_id_kind)

rtps_app_id, or rtps_instance_id fields.
Note: RTPS_AUTO_ID_FROM_UUID is only supported on iOS architectures.
Some examples are provided to better explain the behavior of this QoSPolicy in case you want to change
the default behavior with DDS_RTPS_AUTO_ID_FROM_MAC.
1. Get the DomainParticipant QoS from the DomainParticipantFactory:
DDS_DomainParticipantFactory_get_default_participant_qos(
DDS_DomainParticipantFactory_get_instance(),
&participant_qos);

2. Change the WireProtocolQosPolicy using one of the following options.

616

8.5.9.4 Controlling How the GUID is Set (rtps_auto_id_kind)

l

Use DDS_RTPS_AUTO_ID_FROM_MAC to explicitly set just the application/task identifier portion of the rtps_instance_id field:
participant_qos.wire_protocol.rtps_auto_id_kind =
DDS_RTPS_AUTO_ID_FROM_MAC;
participant_qos.wire_protocol.rtps_host_id =
DDS_RTPS_AUTO_ID;
participant_qos.wire_protocol.rtps_app_id =
DDS_RTPS_AUTO_ID;
participant_qos.wire_protocol.rtps_instance_id =
(/* App ID */ (12 << 8) |
/* Instance ID*/ (DDS_RTPS_AUTO_ID));

l

Only set the per participant counter and let Connext DDS handle the application/task identifier:
participant_qos.wire_protocol.rtps_auto_id_kind =
DDS_RTPS_AUTO_ID_FROM_MAC;
participant_qos.wire_protocol.rtps_host_id =
DDS_RTPS_AUTO_ID;
participant_qos.wire_protocol.rtps_app_id =
DDS_RTPS_AUTO_ID;
participant_qos.wire_protocol.rtps_instance_id =
(/* App ID */ (DDS_RTPS_AUTO_ID) |
/* Instance ID*/ (12));

l

Set the entire rtps_instance_id field yourself:
participant_qos.wire_protocol.rtps_auto_id_kind =
DDS_RTPS_AUTO_ID_FROM_MAC;
participant_qos.wire_protocol.rtps_host_id =
DDS_RTPS_AUTO_ID;
participant_qos.wire_protocol.rtps_app_id =
DDS_RTPS_AUTO_ID;
participant_qos.wire_protocol.rtps_instance_id =
( /* App ID */ (12 << 8)) |
/* Instance ID */ (9) )

Note: If you are using DDS_RTPS_AUTO_ID_FROM_MAC as rtps_auto_id_kind and you
decide to manually handle the rtps_instance_id field, you must ensure that both parts are non-zero
(otherwise Connext DDS will take responsibility for them).
RTI recommends that you always specify the two parts separately in order to avoid errors.
l

Let Connext DDS handle the entire rtps_instance_id field:

617

8.5.9.5 Example

participant_qos.wire_protocol.rtps_auto_id_kind =
DDS_RTPS_AUTO_ID_FROM_MAC;
participant_qos.wire_protocol.rtps_host_id =
DDS_RTPS_AUTO_ID;
participant_qos.wire_protocol.rtps_app_id =
DDS_RTPS_AUTO_ID;
participant_qos.wire_protocol.rtps_instance_id =
DDS_RTPS_AUTO_ID;

Note: If you are using DDS_RTPS_AUTO_ID_FROM_MAC as rtps_auto_id_kind and you
decide to manually set the rtps_instance_id field, you must ensure that both parts are non-zero
(otherwise Connext DDS will take responsibility for them). RTI recommends that you always specify the two parts separately in order to clearly show the difference.
3. Create the DomainParticipant as usual using the modified QoS structure instead of the default one.
8.5.9.5 Example
On many real-time operating systems, and even on some non-real-time operating systems, when a node is
rebooted, and applications are automatically started, process ids are deterministically assigned. That is,
when the system restarts or if an application dies and is restarted, the application will be reassigned the
same process or task ID.
This means that Connext DDS’s automatic algorithm for creating unique rtps_app_id’s will produce the
same value between sequential instances of the same application. This will confuse the other DomainParticipants on the network into thinking that they are communicating with the previous instance of the application instead of a new instance. Errors usually resulting in a failure to communicate will ensue.
Thus for applications running on nodes that may be rebooted without letting the application shutdown
appropriately (destroying the DomainParticipant), especially on nodes running real-time operating systems
like VxWorks or LynxOS, you will want to set the rtps_app_id manually. We suggest that a strictly incrementing counter is stored either on a file system or in non-volatile RAM is used for the rtps_app_id.
Whatever method you use, you should make sure that the rtps_app_id is unique across all DomainParticipants running on a host as well as DomainParticipants that were recently running on the host. After a
period configured through the DISCOVERY_CONFIG QosPolicy existing applications will eventually
flush old DomainParticipants that did not properly shutdown from their databases. When that is done,
then rtps_app_id may be reused.
8.5.9.6 Properties
This QosPolicy cannot be modified after the DomainParticipant is created.
If manually set, it must be set differently for every DomainParticipant in the same DDS domain across all
applications. The value of rtps_app_id should also change between different invocations of the same
application (for example, when an application is restarted).

618

8.5.9.7 Related QosPolicies

8.5.9.7 Related QosPolicies
l

DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585)

8.5.9.8 Applicable DDS Entities
l

DomainParticipants (Section 8.3 on page 547)

8.5.9.9 System Resource Considerations
The use of this policy does not significantly impact the use of resources.

8.6 Clock Selection
Connext DDS uses clocks to measure time and generate timestamps.
The middleware uses two clocks: an internal clock and an external clock.
l
l

The internal clock measures time and handles all timing in the middleware.
The external clock is used solely to generate timestamps (such as the source timestamp and the reception timestamp), in addition to providing the time given by the DomainParticipant’s get_current_
time() operation (see Getting the Current Time (Section 8.3.13.2 on page 571)).

8.6.1 Available Clocks
Two clock implementations are generally available: the real-time clock and the monotonic clock.
The real-time clock provides the real time of the system. This clock may generally be monotonic, but may
not be guaranteed to be so. It is adjustable and may be subject to small and large changes in time. The time
obtained from this clock is generally a meaningful time, in that it is the amount of time from a known
epoch. For the purposes of clock selection, this clock can be referenced by the names "realtime" or "system"—both names map to the same real-time clock.
The monotonic clock provides times that are monotonic from a clock that is not adjustable. This clock is
not subject to changes in the system or realtime clock, which may be adjusted by the user or via time synchronization protocols. However, this clock’s time generally starts from an arbitrary point in time, such as
system start-up. Note that the monotonic clock is not available for all architectures. Please see the RTI Connext DDS Core Libraries Platform Notes for the architectures on which it is supported. For the purposes of
clock selection, this clock can be referenced by the name "monotonic".

8.6.2 Clock Selection Strategy
To configure the clock selection, use the DomainParticipant’s PROPERTY QosPolicy (DDS Extension)
(Section 6.5.17 on page 394). Table 8.21 Clock Selection Properties lists the supported properties.

619

8.7 System Properties

Table 8.21 Clock Selection Properties
Property

Description
Comma-delimited list of clocks to use for the external clock, in the order of preference.

dds.clock.external_clock
Valid clock names are “realtime”, “system”, or “monotonic”.
Comma-delimited list of clocks to use for the internal clock, in the order of preference.
dds.clock.internal_clock
Valid clock names are “realtime”, “system”, or “monotonic”.

By default, both the internal and external clocks use the realtime clock.
If you want your application to be robust to changes in the system time, you may use the monotonic clock
as the internal clock, and leave the system clock as the external clock. However, note that this may slightly
diminish performance, in that both the send and receive paths may need to get times from both clocks.
Since the monotonic clock is not available on all architectures, you may want to specify "monotonic, realtime" for the internal_clock property (see Table 8.21 Clock Selection Properties). By doing so, the middleware will attempt to use the monotonic clock if it is available, and will fall back to the realtime clock if
the monotonic clock is not available.
If you want the application to be robust to changes in the system time, you are not relying on source
timestamps, and you want to avoid obtaining times from both clocks, you may use the monotonic clock for
both the internal and external clocks.

8.7 System Properties
Connext DDS uses the DomainParticipant’s PropertyQosPolicy to maintain a set of properties that
provide system information, such as the hostname.
Unless the default the DDS_DomainParticipantQos structure (see Setting DomainParticipant QosPolicies
(Section 8.3.6 on page 562)) is overwritten, the system properties are automatically set in the DDS_
DomainParticipantQos structure that is obtained by calling the DomainParticipantFactory’s get_default_
participant_qos() operation or by using the constant DDS_PARTICIPANT_QOS_DEFAULT.
System properties are also automatically set in the DDS_DomainParticipantQos structure loaded from an
XML QoS profile unless you disable property inheritance using the attribute inherit in the XML tag .
By default, the system properties are propagated to other DomainParticipants in the system and can be
accessed through the property field in the Table 16.1 Participant Built-in Topic’s Data Type (DDS_ParticipantBuiltinTopicData).
You can disable propagation of individual properties by setting the property’s propagate flag to FALSE
or by removing the property using the PropertyQosPolicyHelper operation, remove_property() (see
Table 6.57 PropertyQoSPolicyHelper Operations).

620

8.7 System Properties

The number of system properties that are initialized for a DomainParticipant is platform specific: only process_id and os_arch are supported on all platforms.
These properties will only be created if Connext DDS can obtain the information for them; see Table 8.22
System Properties.
System properties are affected by the DomainParticipantResourceLimitsQosPolicy’s participant_property_list_max_length and participant_property_string_max_length.
Table 8.22 System Properties
Property Name

Description

dds.sys_info.creation_timestamp

Time when the executable was created.1

dds.sys_info.executable_filepath

Name and full path of the executable.2

dds.sys_info.execution_timestamp

Time when the execution started.3

dds.sys_info.hostname

Hostname4

dds.sys_info.target

Architecture for which the library was compiled (for example, x64Darwin10gcc4.2.1).

dds.sys_info.process_id

Process ID

dds.sys_info.username

Username that is running the process.5

1Only supported on Windows and Linux architectures.
2Only supported on Windows and Linux architectures.
3Only supported on Windows and Linux architectures.
4Only supported on Windows and Linux architectures.
5Only supported on Windows and Linux architectures.

621

Chapter 9 Building Applications
This chapter provides instructions on how to build Connext DDS applications for the following
platforms:
l

UNIX-Based Platforms (Section 9.3 on page 624) (including Solaris™, Red Hat® and Yellow Dog™ Linux, QNX®, and LynxOS® systems)

l

Windows Platforms (Section 9.4 on page 625)

l

Java Platforms (Section 9.5 on page 627)

While you can create applications for other operating systems, the platforms presented in this
chapter are a good starting point. We recommend that you first build and test your application on
one of these systems.
Instructions for other supported target platforms are provided in the RTI Connext DDS Core
Libraries Platform Notes.
To build a non-Java application using Connext DDS, you must specify the following items:
l

NDDSHOME environment variable

l

Connext DDS header files

l

Connext DDS libraries to link

l

Compatible system libraries

l

Compiler options

To build Java applications using Connext DDS, you must specify the following items:
l

NDDSHOME environment variable

l

Connext DDS JAR file

622

9.1 Running on a Computer Not Connected to a Network

l

Compatible Java virtual machine (JVM)

l

Compiler options

This chapter describes the basic steps you will take to build an application on the above-mentioned platforms. Specific details, such as exactly which libraries to link, compiler flags, etc. are in the RTI Connext
DDS Core Libraries Platform Notes.

9.1 Running on a Computer Not Connected to a Network
If you want to run Connext DDS applications on the same computer, and that computer is not connected
to a network, you must set NDDS_DISCOVERY_PEERS so that it will only use shared memory. For
example:
set NDDS_DISCOVERY_PEERS=4@shmem://

(The number 4 is only an example. This is the maximum participant ID.)

9.2 Connext DDS Header Files — All Architectures
You must include the appropriate Connext DDS header files, which are listed in Table 9.1 Header Files to
Include for Connext DDS (All Architectures). The header files that need to be included depend on the API
being used.
Table 9.1 Header Files to Include for Connext DDS (All Architectures)
Connext DDS API

Header Files

C

#include “ndds/ndds_c.h”

C++

#include “ndds/ndds_cpp.h”

C++/CLI, C#, Java

none

For the compiler to find the included files, the path to the appropriate include directories must be provided.
Table 9.2 Include Paths for Compilation (All Architectures) lists the appropriate include path for use with
the compiler. The exact path depends on where you installed Connext DDS. See Paths Mentioned in
Documentation (Section on page xxxviii).

623

9.3 UNIX-Based Platforms

Table 9.2 Include Paths for Compilation (All Architectures)
Connext DDS API

Include Path Directories
/include

C and C++
/include/ndds
C++/CLI, C#, Java

none

The header files that define the data types you want to use within the application also need to be included.
For example, Table 9.3 Header Files to Include for Data Types (All Architectures) lists the files to be
include for type “Foo” (these are the filenames generated by RTI Code Generator, described in Data
Types and DDS Data Samples (Section Chapter 3 on page 23)).
Table 9.3 Header Files to Include for Data Types (All Architectures)
Connext DDS API

User Data Type Header Files
#include “Foo.h”

C and C++
#include “FooSupport.h”
C++/CLI, C#, Java

none

9.3 UNIX-Based Platforms
Before building a Connext DDS application for a UNIX-based platform (including Solaris, Red Hat and
Yellow Dog Linux, QNX, and LynxOS systems), make sure that:
l

l

l

l

A supported version of your architecture is installed. See the RTI Connext DDS Core Libraries Platform Notes for supported architectures.
Connext DDS 5.x.y is installed (where 5.x.y stands for the version number of the current release).
For installation instructions, refer to the RTI Connext DDS Core Libraries Getting Started Guide.
A “make” tool is installed. RTI recommends GNU Make. If you do not have it, you may be able to
download it from your operating system vendor. Learn more at www.gnu.org/software/make/ or
download from ftpmirror.gnu.org/make as source code.
The NDDSHOME environment variable is set to the root directory of the Connext DDS installation
(such as /home/user/rti_connext_dds-5.x.y).
l To confirm, type this at a command prompt:
echo $NDDSHOME
env | grep NDDSHOME

624

9.3.1 Required Libraries

l

If it is not set or is set incorrectly, type:
setenv NDDSHOME 

To compile a Connext DDS application of any complexity, either modify the auto-generated makefile created by running RTI Code Generator or write your own makefile.

9.3.1 Required Libraries
All required system and Connext DDS libraries are listed in the RTI Connext DDS Core Libraries Platform Notes.
You must choose between dynamic (shared) and static libraries. Do not mix the different types of libraries
during linking. The benefit of linking against the dynamic libraries is that your final executables’ sizes will
be significantly smaller. You will also use less memory when you are running several Connext DDS
applications on the same node. However, shared libraries require more set-up and maintenance during
upgrades and installations.
To see if dynamic libraries are supported for your target architecture, see the RTI Connext DDS Core
Libraries Platform Notes1.

9.3.2 Compiler Flags
See the RTI Connext DDS Core Libraries Platform Notes for information on compiler flags.

9.4 Windows Platforms
Before building an application for a Microsoft Windows® platform, make sure that:
l

l

l

Supported versions of Windows and Visual Studio are installed. See the Windows section of the
RTI Connext DDS Core Libraries Platform Notes.
Connext DDS 5.x.y is installed ( where 5.x.y stands for the version numbers of the current release).
For installation instructions, refer to the RTI Connext DDS Core Libraries Getting Started Guide.
The NDDSHOME environment variable is set to the root directory of the Connext DDS installation
(such as C:\Program Files\rti_connext_dds-5.x.y). To confirm, type this at a command prompt:
echo %NDDSHOME%

l

Use the dynamic MFC Library (not static).

1In the Platform Notes, see the “Building Instructions...” table for your target architecture.

625

9.4.1 Using Visual Studio

To avoid communication problems in your Connext DDS application, use the dynamic MFC library, not the static version. (If you use the static version, your Connext DDS application may stop
receiving DDS samples once the Windows sockets are initialized.)
To compile a Connext DDS application of any complexity, use a project file in Microsoft Visual Studio.
The project settings are described below. The Windows section of the RTI Connext DDS Core Libraries
Platform Notes contains more information.

9.4.1 Using Visual Studio
1. Select the multi-threaded project setting:
a. From the Project menu, select Properties.
b. Select the C/C++ folder.
c. Select Code Generation.
d. Set the Runtime Library field to one of the options from Table 9.4 Runtime Library Settings
for Visual Studio.
2. Link against the Connext DDS libraries:
a. Select the Linker folder on the Project, Properties dialog box.
b. Select the Input properties.
c. See the Windows section of the RTI Connext DDS Core Libraries Platform Notes for a list of
required libraries. You have a choice of whether to link with Connext DDS’s static or
dynamic libraries. Decide whether or not you want debugging symbols on. In either case, be
sure to use a space as a delimiter between libraries, not a comma. Add the libraries to the
beginning of the Additional Dependencies field.
d. Select the General properties.
e. Add the following to the Additional library path field (replace  to match your
installed system):
$(NDDSHOME)\lib\

3. Specify the path to Connext DDS’s header file:
a. Select the C/C++ folder.
b. Select the General properties.
c. In the Additional include directories: field, add paths to the “include” and “include\ndds” directories.
For example: (your paths may differ, depending on where you installed Connext DDS)

626

9.5 Java Platforms

c:\Program Files\rti_connext_dds-5.x.y\include\
c:\Program Files\rti_connext_dds-5.x.y\include\ndds

Table 9.4 Runtime Library Settings for Visual Studio
If You are using this Library Format...

Set the Runtime Library field to...

Release version of static libraries

Multi-threaded (/MT)

Debug version of static libraries

Multi-threaded Debug (/MTd)

Release version of dynamic libraries

Multi-threaded DLL (/MD)

Debug version of dynamic libraries

Multi-threaded Debug DLL (/MDd)

9.5 Java Platforms
Before building an application for a Windows or UNIX Java platform, make sure that:
l
l

Connext DDS 5.x.y is installed ( where 5.x.y stands for the version numbers of the current release).
A supported version of the Java 2 software development kit (J2SDK) is installed. See the Windows
section of the RTI Connext DDS Core Libraries Platform Notes.

9.5.1 Java Libraries
Connext DDS requires that certain Java archive (JAR) files be on your classpath when running Connext
DDS applications. See the Platform Notes for more details.

9.5.2 Native Libraries
Connext DDS for Java is implemented using Java Native Interface (JNI), so it is necessary to provide your
Connext DDS distributed applications access to certain native shared libraries. See the RTI Connext DDS
Core Libraries Platform Notes for more details.

627

Part 3: Advanced Concepts
This part of the manual will guide you through some of the more advanced concepts:
l

Reliable Communications (Section Chapter 10 on page 629)

l

Collaborative DataWriters (Section Chapter 11 on page 670)

l

Mechanisms for Achieving Information Durability and Persistence (Section Chapter 12 on
page 675)

l

Guaranteed Delivery of Data (Section Chapter 13 on page 695)

l

Discovery (Section Chapter 14 on page 709)

l

Transport Plugins (Section Chapter 15 on page 743)

l

Built-In Topics (Section Chapter 16 on page 772)

l

Configuring QoS with XML (Section Chapter 17 on page 791)

l

Multi-channel DataWriters (Section Chapter 18 on page 824)

l

Connext DDS Threading Model (Section Chapter 19 on page 837)

l

l

DDS Sample-Data and Instance-Data Memory Management (Section Chapter 20 on page
846)
Troubleshooting (Section Chapter 21 on page 863)

628

Chapter 10 Reliable Communications
Connext DDS uses best-effort delivery by default. The other type of delivery that Connext DDS
supports is called reliable. This chapter provides instructions on how to set up and use reliable communication.
This chapter includes the following sections:
l

Sending Data Reliably (Section 10.1 below)

l

Overview of the Reliable Protocol (Section 10.2 on page 631)

l

Using QosPolicies to Tune the Reliable Protocol (Section 10.3 on page 635)

10.1 Sending Data Reliably
The DCPS reliability model recognizes that the optimal balance between time-determinism and
data-delivery reliability varies widely among applications and can vary among different publications within the same application. For example, individual DDS samples of signal data can
often be dropped because their value disappears when the next DDS sample is sent. However,
each DDS sample of command data must be received and it must be received in the order sent.
The QosPolicies provide a way to customize the determinism/reliability trade-off on a per Topic
basis, or even on a per DataWriter/DataReader basis.
There are two delivery models:
l

Best-effort delivery mode“I’m not concerned about missed or unordered DDS samples.”

l

Reliable delivery model“Make sure all DDS samples get there, in order.”

10.1.1 Best-effort Delivery Model
By default, Connext DDS uses the best-effort delivery model: there is no effort spent ensuring inorder delivery or resending lost DDS samples. Best-effort DataReaders ignore lost DDS samples

629

10.1.2 Reliable Delivery Model

in favor of the latest DDS sample. Your application is only notified if it does not receive a new DDS
sample within a certain time period (set in the DEADLINE QosPolicy (Section 6.5.5 on page 363)).
The best-effort delivery model is best for time-critical information that is sent continuously. For instance,
consider a DataWriter for the value of a sensor device (such as a the pressure inside a tank), and assume
the DataWriter sends DDS samples continuously. In this situation, a DataReader for this Topic is only
interested in having the latest pressure reading available—older DDS samples are obsolete.

10.1.2 Reliable Delivery Model
Reliable delivery means the DDS samples are guaranteed to arrive, in the order published.
The DataWriter maintains a send queue with space to hold the last X number of DDS samples sent. Similarly, a DataReader maintains a receive queue with space for consecutive X expected DDS samples.
The send and receive queues are used to temporarily cache DDS samples until Connext DDS is sure the
DDS samples have been delivered and are not needed anymore. Connext DDS removes DDS samples
from a publication’s send queue after the DDS sample has been acknowledged by all reliable subscriptions. When positive acknowledgements are disabled (see DATA_WRITER_PROTOCOL
QosPolicy (DDS Extension) (Section 6.5.3 on page 347) and DATA_READER_PROTOCOL
QosPolicy (DDS Extension) (Section 7.6.1 on page 511)), DDS samples are removed from the send
queue after the corresponding keep-duration has elapsed (see Table 6.37 DDS_RtpsReliableWriterProtocol_t).
If an out-of-order DDS sample arrives, Connext DDS speculatively caches it in the DataReader’s receive
queue (provided there is space in the queue). Only consecutive DDS samples are passed on to the
DataReader.
DataWriters can be set up to wait for available queue space when sending DDS samples. This will cause
the sending thread to block until there is space in the send queue. (Or, you can decide to sacrifice sending
DDS samples reliably so that the sending rate is not compromised.) If the DataWriter is set up to ignore
the full queue and sends anyway, then older cached DDS samples will be pushed out of the queue before
all DataReaders have received them. In this case, the DataReader (or its Subscriber) is notified of the missing DDS samples through its Listener and/or Conditions.
Connext DDS automatically sends acknowledgments (ACKNACKs) as necessary to maintain reliable
communications. The DataWriter may choose to block for a specified duration to wait for these acknowledgments (see Waiting for Acknowledgments in a DataWriter (Section 6.3.11 on page 288)).
Connext DDS establishes a virtual reliable channel between the matching DataWriter and all
DataReaders. This mechanism isolates DataReaders from each other, allows the application to control
memory usage, and provides mechanisms for the DataWriter to balance reliability and determinism.
Moreover, the use of send and receive queues allows Connext DDS to be implemented efficiently without
introducing unnecessary delays in the stream.

630

10.2 Overview of the Reliable Protocol

Note that a successful return code (DDS_RETCODE_OK) from write() does not necessarily mean that all
DataReaders have received the data. It only means that the DDS sample has been added to the
DataWriter’s queue. To see if all DataReaders have received the data, look at the RELIABLE_
WRITER_CACHE_CHANGED Status (DDS Extension) (Section 6.3.6.8 on page 279) to see if any
DDS samples are unacknowledged.
Suppose DataWriter A reliably publishes a Topic to which DataReaders B and C reliably subscribe. B
has space in its queue, but C does not. Will DataWriter A be notified? Will DataReader C receive any
error messages or callbacks? The exact behavior depends on the QoS settings:
l

l

If HISTORY_KEEP_ALL is specified for C, C will reject DDS samples that cannot be put into the
queue and request A to resend missing DDS samples. The Listener is notified with the on_sample_
rejected() callback (see SAMPLE_REJECTED Status (Section 7.3.7.8 on page 479)). If A has a
queue large enough, or A is no longer writing new DDS samples, A won’t notice unless it checks
the RELIABLE_WRITER_CACHE_CHANGED Status (DDS Extension) (Section 6.3.6.8 on
page 279).
If HISTORY_KEEP_LAST is specified for C, C will drop old DDS samples and accept new ones.
To A, it is as if all DDS samples have been received by C (that is, they have all been acknowledged).

10.2 Overview of the Reliable Protocol
An important advantage of Connext DDS is that it can offer the reliability and other QoS guarantees mandated by DDS on top of a very wide variety of transports, including packet-based transports, unreliable networks, multicast-capable transports, bursty or high-latency transports, etc. Connext DDS is also capable of
maintaining liveliness and application-level QoS even in the presence of sporadic connectivity loss at the
transport level, an important benefit in mobile networks. Connext DDS accomplishes this by implementing
a reliable protocol that sequences and acknowledges application-level messages and monitors the liveliness
of the link. This is called the Real-Time Publish-Subscribe (RTPS) protocol; it is an open, international
standard.1
In order to work in this wide range of environments, the reliable protocol defined by RTPS is highly configurable with a set of parameters that let the application fine-tune its behavior to trade-off latency, responsiveness, liveliness, throughput, and resource utilization. This section describes the most important features
to the extent needed to understand how the configuration parameters affect its operation.
The most important features of the RTPS protocol are:

1For a link to the RTPS specification, see the RTI website, www.rti.com.

631

10.2 Overview of the Reliable Protocol

l

Support for both push and pull operating modes

l

Support for both positive and negative acknowledgments

l

Support for high data-rate DataWriters

l

Support for multicast DataReaders

l

Support for high-latency environments

In order to support these features, RTPS uses several types of messages: Data messages (DATA), acknowledgments (ACKNACKs), and heartbeats (HBs).
l

l

l

DATA messages contain snapshots of the value of data-objects and associate the snapshot with a
sequence number that Connext DDS uses to identify them within the DataWriter’s history. These
snapshots are stored in the history as a direct result of the application calling write() on the
DataWriter. Incremental sequence numbers are automatically assigned by the DataWriter each time
write() is called. In Basic RTPS Reliable Protocol (Section Figure 10.1 on the facing page) through
Using QosPolicies to Tune the Reliable Protocol (Section 10.3 on page 635), these messages are represented using the notation DATA(, ). For example, DATA(A,1) represents a message that communicates the value ‘A’ and associates the sequence number ‘1’ with this
message. A DATA is used for both keyed and non-keyed data types.
HB messages announce to the DataReader that it should have received all snapshots up to the one
tagged with a range of sequence numbers and can also request the DataReader to send an acknowledgement back. For example, HB(1-3) indicates to the DataReader that it should have received
snapshots tagged with sequence numbers 1, 2, and 3 and asks the DataReader to confirm this.
ACKNACK messages communicate to the DataWriter that particular snapshots have been successfully stored in the DataReader’s history. ACKNACKs also tell the DataWriter which snapshots
are missing on the DataReader side. The ACKNACK message includes a set of sequence numbers
represented as a bit map. The sequence numbers indicate which ones the DataReader is missing.
(The bit map contains the base sequence number that has not been received, followed by the number
of bits in bit map and the optional bit map. The maximum size of the bit map is 256.) All numbers
up to (not including) those in the set are considered positively acknowledged. They are represented
in Figure 10.1 Basic RTPS Reliable Protocol on the facing page through Figure 10.7 Use of heartbeat_period on page 647 as ACKNACK() or ACKNACK(-). For example, ACKNACK(4) indicates that the snapshots with sequence numbers 1, 2,
and 3 have been successfully stored in the DataReader history, and that 4 has not been received.

It is important to note that Connext DDS can bundle multiple of the above messages within a single network packet. This ‘submessage bundling’ provides for higher performance communications.

632

10.2 Overview of the Reliable Protocol

Figure 10.1 Basic RTPS Reliable Protocol

Basic RTPS Reliable Protocol (Section Figure 10.1 above) illustrates the basic behavior of the protocol
when an application calls the write() operation on a DataWriter that is associated with a DataReader. As
mentioned, the RTPS protocol can bundle multiple submessages into a single network packet. In Basic
RTPS Reliable Protocol (Section Figure 10.1 above) this feature is used to piggyback a HB message to the
DATA message. Note that before the message is sent, the data is given a sequence number (1 in this case)
which is stored in the DataWriter’s send queue. As soon as the message is received by the DataReader, it
places it into the DataReader’s receive queue. From the sequence number the DataReader can tell that it
has not missed any messages and therefore it can make the data available immediately to the user (and call
the DataReaderListener). This is indicated by the “4” symbol. The reception of the HB(1) causes the
DataReader to check that it has indeed received all updates up to and including the one with
sequenceNumber=1. Since this is true, it replies with an ACKNACK(2) to positively acknowledge all messages up to (but not including) sequence number 2. The DataWriter notes that the update has been acknowledged, so it no longer needs to be retained in its send queue. This is indicated by the “4” symbol.

633

10.2 Overview of the Reliable Protocol

Figure 10.2 RTPS Reliable Protocol in the Presence of Message Loss

RTPS Reliable Protocol in the Presence of Message Loss (Section Figure 10.2 above) illustrates the behavior of the protocol in the presence of lost messages. Assume that the message containing DATA(A,1) is
dropped by the network. When the DataReader receives the next message (DATA(B,2); HB(1-2)) the
DataReader will notice that the data associated with sequence number 1 was never received. It realizes
this because the heartbeat HB(1-2) tells the DataReader that it should have received all messages up to
and including the one with sequence number 2. This realization has two consequences:

634

10.3 Using QosPolicies to Tune the Reliable Protocol

l

l

The data associated with sequence number 2 (B) is tagged with ‘X’ to indicate that it is not deliverable to the application (that is, it should not be made available to the application, because the
application needs to receive the data associated with DDS sample 1 (A) first).
An ACKNACK(1) is sent to the DataWriter to request that the data tagged with sequence number 1
be resent.

Reception of the ACKNACK(1) causes the DataWriter to resend DATA(A,1). Once the DataReader
receives it, it can ‘commit’ both A and B such that the application can now access both (indicated by the
“ü”) and call the DataReaderListener. From there on, the protocol proceeds as before for the next data
message (C) and so forth.
A subtle but important feature of the RTPS protocol is that ACKNACK messages are only sent as a direct
response to HB messages. This allows the DataWriter to better control the overhead of these ‘administrative’ messages. For example, if the DataWriter knows that it is about to send a chain of DATA messages, it can bundle them all and include a single HB at the end, which minimizes ACKNACK traffic.

10.3 Using QosPolicies to Tune the Reliable Protocol
Reliability is controlled by the QosPolicies in Table 10.1 QosPolicies for Reliable Communications. To
enable reliable delivery, read the following sections to learn how to change the QoS for the DataWriter
and DataReader:
l

Enabling Reliability (Section 10.3.1 on page 637)

l

Tuning Queue Sizes and Other Resource Limits (Section 10.3.2 on page 638)

l

l
l

Controlling Heartbeats and Retries with DataWriterProtocol QosPolicy (Section 10.3.4 on page
645)
Avoiding Message Storms with DataReaderProtocol QosPolicy (Section 10.3.5 on page 653)
Resending DDS Samples to Late-Joiners with the Durability QosPolicy (Section 10.3.6 on page
653)

Then see Use Cases (Section 10.3.7 on page 654) to explore example use cases:

635

10.3 Using QosPolicies to Tune the Reliable Protocol

Table 10.1 QosPolicies for Reliable Communications
QosPolicy

Description

Related
Entities

Reference

1

Reliability

To establish reliable communication, this QoS must be set to
DDS_RELIABLE_RELIABILITY_QOS for the DataWriter
and its DataReaders.

DW, DR

Enabling Reliability (Section
10.3.1 on the facing page),
RELIABILITY QosPolicy
(Section 6.5.19 on page 400)

ResourceLimits

This QoS determines the amount of resources each side can use to
manage instances and DDS samples of instances. Therefore it
controls the size of the DataWriter’s send queue and the
DataReader’s receive queue. The send queue stores DDS samples DW, DR
until they have been ACKed by all DataReaders. The
DataReader’s receive queue stores DDS samples for the user’s
application to access.

Tuning Queue Sizes and
Other Resource Limits
(Section 10.3.2 on page
638), RESOURCE_LIMITS
QosPolicy (Section 6.5.20
on page 405)

This QoS affects how a DataWriter/DataReader behaves when its
DW, DR
send/receive queue fills up.

Controlling Queue Depth
with the History QosPolicy
(Section 10.3.3 on page
644), HISTORY QosPolicy
(Section 6.5.10 on page 376)

DataWriterProtocol

This QoS configures DataWriter-specific protocol. The QoS can
disable positive ACKs for its DataReaders.

DW

Controlling Heartbeats and
Retries with
DataWriterProtocol
QosPolicy (Section 10.3.4
on page 645), DATA_
WRITER_PROTOCOL
QosPolicy (DDS Extension)
(Section 6.5.3 on page 347)

DataReaderProtocol

When a reliable DataReader receives a heartbeat from a
DataWriter and needs to return an ACKNACK, the DataReader
can choose to delay a while. This QoS sets the minimum and
maximum delay. It can also disable positive ACKs for the
DataReader.

DR

Avoiding Message Storms
with DataReaderProtocol
QosPolicy (Section 10.3.5
on page 653), DATA_
READER_PROTOCOL
QosPolicy (DDS Extension)
(Section 7.6.1 on page 511)

This QoS determines additional amounts of resources that the
DataReader can use to manage DDS samples (namely, the size of
DataReaderResourceLimits the DataReader’s internal queues, which cache DDS samples until DR
they are ordered for reliability and can be moved to the
DataReader’s receive queue for access by the user’s application).

Tuning Queue Sizes and
Other Resource Limits
(Section 10.3.2 on page
638),DATA_READER_
RESOURCE_LIMITS
QosPolicy (DDS Extension)
(Section 7.6.2 on page 517)

History

1DW = DataWriter, DR = DataReader

636

10.3.1 Enabling Reliability

Table 10.1 QosPolicies for Reliable Communications
QosPolicy

Description

Related
Entities

Reference

1

This QoS affects whether late-joining DataReaders will receive all
DW, DR
previously-sent data or not.

Durability

Resending DDS Samples to
Late-Joiners with the
Durability QosPolicy
(Section 10.3.6 on page
653), DURABILITY
QosPolicy (Section 6.5.7 on
page 368)

10.3.1 Enabling Reliability
You must modify the RELIABILITY QosPolicy (Section 6.5.19 on page 400) of the DataWriter and
each of its reliable DataReaders. Set the kind field to DDS_RELIABLE_RELIABILITY_QOS:
l

DataWriter
writer_qos.reliability.kind = DDS_RELIABLE_RELIABILITY_QOS;

l

DataReader
reader_qos.reliability.kind = DDS_RELIABLE_RELIABILITY_QOS;

10.3.1.1 Blocking until the Send Queue Has Space Available
The max_blocking_time property in the RELIABILITY QosPolicy (Section 6.5.19 on page 400) indicates how long a DataWriter can be blocked during a write().
If max_blocking_time is non-zero and the reliability send queue is full, the write is blocked (the DDS
sample is not sent). If max_blocking_time has passed and the DDS sample is still not sent, write() returns
DDS_RETCODE_TIMEOUT and the DDS sample is not sent.
If the number of unacknowledged DDS samples in the reliability send queue drops below max_samples
(set in the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)) before max_blocking_time,
the DDS sample is sent and write() returns DDS_RETCODE_OK.
If max_blocking_time is zero and the reliability send queue is full, write() returns DDS_RETCODE_
TIMEOUT and the DDS sample is not sent.

1DW = DataWriter, DR = DataReader

637

10.3.2 Tuning Queue Sizes and Other Resource Limits

10.3.2 Tuning Queue Sizes and Other Resource Limits
Set the HISTORY QosPolicy (Section 6.5.10 on page 376) appropriately to accommodate however many
DDS samples should be saved in the DataWriter’s send queue or the DataReader’s receive queue. The
defaults may suit your needs; if so, you do not have to modify this QosPolicy.
Set the DDS_RtpsReliableWriterProtocol_t in the DATA_WRITER_PROTOCOL QosPolicy (DDS
Extension) (Section 6.5.3 on page 347) appropriately to accommodate the number of unacknowledged
DDS samples that can be in-flight at a time from a DataWriter.
For more information, see the following sections:
l

Understanding the Send Queue and Setting its Size (Section 10.3.2.1 on the facing page)

l

Understanding the Receive Queue and Setting Its Size (Section 10.3.2.2 on page 642)

Note: The HistoryQosPolicy’s depth must be less than or equal to the ResourceLimitsQosPolicy’s max_
samples_per_instance; max_samples_per_instance must be less than or equal to the ResourceLimitsQosPolicy’s max_samples (see RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)), and
max_samples_per_remote_writer (see DATA_READER_RESOURCE_LIMITS QosPolicy (DDS
Extension) (Section 7.6.2 on page 517)) must be less than or equal to max_samples.
l

depth <= max_samples_per_instance <= max_samples

l

max_samples_per_remote_writer <= max_samples

Examples:
DataWriter
writer_qos.resource_limits.initial_instances = 10;
writer_qos.resource_limits.initial_samples = 200;
writer_qos.resource_limits.max_instances = 100;
writer_qos.resource_limits.max_samples = 2000;
writer_qos.resource_limits.max_samples_per_instance = 20;
writer_qos.history.depth = 20;

DataReader
reader_qos.resource_limits.initial_instances = 10;
reader_qos.resource_limits.initial_samples = 200;
reader_qos.resource_limits.max_instances = 100;
reader_qos.resource_limits.max_samples = 2000;
reader_qos.resource_limits.max_samples_per_instance = 20;
reader_qos.history.depth = 20;
reader_qos.reader_resource_limits.max_samples_per_remote_writer = 20;

638

10.3.2.1 Understanding the Send Queue and Setting its Size

10.3.2.1 Understanding the Send Queue and Setting its Size
A DataWriter’s send queue is used to store each DDS sample it writes. A DDS sample will be removed
from the send queue after it has been acknowledged (through an ACKNACK) by all the reliable
DataReaders. A DataReader can request that the DataWriter resend a missing DDS sample (through an
ACKNACK). If that DDS sample is still available in the send queue, it will be resent. To elicit timely
ACKNACKs, the DataWriter will regularly send heartbeats to its reliable DataReaders.
A DataWriter’s send queue size is determined by its RESOURCE_LIMITS QosPolicy (Section 6.5.20
on page 405), specifically the max_samples field. The appropriate value depends on application parameters such as how fast the publication calls write().
A DataWriter has a "send window" that is the maximum number of unacknowledged DDS samples
allowed in the send queue at a time. The send window enables configuration of the number of DDS
samples queued for reliability to be done independently from the number of DDS samples queued for history. This is of great benefit when the size of the history queue is much different than the size of the reliability queue. For example, you may want to resend a large history to late-joining DataReaders, so the
send queue size is large. However, you do not want performance to suffer due to a large send queue; this
can happen when the send rate is greater than the read rate, and the DataWriter has to resend many DDS
samples from its large historical send queue. If the send queue size was both the historical and reliability
queue size, then both these goals could not be met. Now, with the send window, having a large history
with good live reliability performance is possible.
The send window is determined by the DataWriterProtocolQosPolicy, specifically the fields min_send_
window_size and max_send_window_size within the rtps_reliable_writer field of type DDS_RtpsReliableWriterProtocol_t. Other fields control a dynamic send window, where the send window size changes
in response to network congestion to maximize the effective send rate. Like for max_samples, the appropriate values depend on application parameters.
Strict reliability:If a DataWriter does not receive ACKNACKs from one or more reliable DataReaders, it
is possible for the reliability send queue—either its finite send window, or max_samples if its send window is infinite—to fill up. If you want to achieve strict reliability, the kind field in the HISTORY
QosPolicy (Section 6.5.10 on page 376) for both the DataReader and DataWriter must be set to KEEP_
ALL, positive acknowledgments must be enabled for both the DataReader and DataWriter, and your publishing application should wait until space is available in the reliability queue before writing any more
DDS samples. Connext DDS provides two mechanisms to do this:
l

l

Allow the write() operation to block until there is space in the reliability queue again to store the
DDS sample. The maximum time this call blocks is determined by the max_blocking_time field in
the RELIABILITY QosPolicy (Section 6.5.19 on page 400) (also discussed in Blocking until the
Send Queue Has Space Available (Section 10.3.1.1 on page 637)).
Use the DataWriter’s Listener to be notified when the reliability queue fills up or empties again.

639

10.3.2.1 Understanding the Send Queue and Setting its Size

When the HISTORY QosPolicy (Section 6.5.10 on page 376) on the DataWriter is set to KEEP_LAST,
strict reliability is not guaranteed. When there are depth number of DDS samples in the queue (set in the
HISTORY QosPolicy (Section 6.5.10 on page 376), see Controlling Queue Depth with the History
QosPolicy (Section 10.3.3 on page 644)) the oldest DDS sample will be dropped from the queue when a
new DDS sample is written. Note that in such a reliable mode, when the send window is larger than
max_samples, the DataWriter will never block, but strict reliability is no longer guaranteed. If there is a
request for the purged DDS sample from any DataReaders, the DataWriter will send a heartbeat that no
longer contains the sequence number of the dropped DDS sample (it will not be able to send the DDS
sample).
Alternatively, a DataWriter with KEEP_LAST may block on write() when its send window is smaller
than its send queue. The DataWriter will block when its send window is full. Only after the blocking time
has elapsed, the DataWriter will purge a DDS sample, and then strict reliability is no longer guaranteed.
The send queue size is set in the max_samples field of the RESOURCE_LIMITS QosPolicy (Section
6.5.20 on page 405). The appropriate size for the send queue depends on application parameters (such as
the send rate), channel parameters (such as end-to-end delay and probability of packet loss), and quality of
service requirements (such as maximum acceptable probability of DDS sample loss).
The DataReader’s receive queue size should generally be larger than the DataWriter’s send queue size.
Receive queue size is discussed in Understanding the Receive Queue and Setting Its Size (Section
10.3.2.2 on page 642).
A good rule of thumb, based on a simple model that assumes individual packet drops are not correlated
and time-independent, is that the size of the reliability send queue, N, is as shown in Calculating Minimum
Send Queue Size for a Desired Level of Reliability (Section Figure 10.3 below).
Figure 10.3 Calculating Minimum Send Queue Size for a Desired Level of Reliability

N = 2RT(log(1-Q))/log(p))
Simple formula for determining the minimum size of the send queue required for strict reliability

In the above equation, R is the rate of sending DDS samples, T is the round-trip transmission time, p is the
probability of a packet loss in a round trip, and Q is the required probability that a DDS sample is eventually successfully delivered. Of course, network-transport dropouts must also be taken into account and
may influence or dominate this calculation.
Table 10.2 Required Size of the Send Queue for Different Network Parameters gives the required size of
the send queue for several common scenarios.

640

10.3.2.1 Understanding the Send Queue and Setting its Size

Table 10.2 Required Size of the Send Queue for Different Network Parameters
Q1

p2

T3

R4

N5

99%

1%

0.0016 sec

100 Hz

1

99%

1%

0.001 sec

2000 Hz

2

99%

5%

0.001 sec

100 Hz

1

99%

5%

0.001 sec

2000 Hz

4

99.99%

1%

0.001 sec

100 Hz

1

99.99%

1%

0.001 sec

2000 Hz

6

99.99%

5%

0.001 sec

100 Hz

1

99.99%

5%

0.001 sec

2000 Hz

8

Note: Packet loss on a network frequently happens in bursts, and the packet loss events are correlated.
This means that the probability of a packet being lost is much higher if the previous packet was lost
because it indicates a congested network or busy receiver. For this situation, it may be better to use a queue
size that can accommodate the longest period of network congestion, as illustrated in Calculating Minimum Send Queue Size for Networks with Dropouts (Section Figure 10.4 below).
Figure 10.4 Calculating Minimum Send Queue Size for Networks with Dropouts

N = RD (Q)
Send queue size as a function of send rate "R" and maximum dropout time D

1"Q" is the desired level of reliability measured as the probability that any data update will eventually be delivered

successfully. In other words, percentage of DDS samples that will be successfully delivered.
2"p" is the probability that any single packet gets lost in the network.
3"T" is the round-trip transport delay in the network
4"R" is the rate at which the publisher is sending updates.
5"N" is the minimum required size of the send queue to accomplish the desired level of reliability "Q".
6The typical round-trip delay for a dedicated 100 Mbit/second ethernet is about 0.001 seconds.

641

10.3.2.2 Understanding the Receive Queue and Setting Its Size

In the above equation R is the rate of sending DDS samples, D(Q) is a time such that Q percent of the dropouts are of equal or lesser length, and Q is the required probability that a DDS sample is eventually successfully delivered. The problem with the above formula is that it is hard to determine the value of D(Q)
for different values of Q.
For example, if we want to ensure that 99.9% of the DDS samples are eventually delivered successfully,
and we know that the 99.9% of the network dropouts are shorter than 0.1 seconds, then we would use N =
0.1*R. So for a rate of 100Hz, we would use a send queue of N = 10; for a rate of 2000Hz, we would use
N = 200.
10.3.2.2 Understanding the Receive Queue and Setting Its Size
DDS samples are stored in the DataReader’s receive queue, which is accessible to the user’s application.
A DDS sample is removed from the receive queue after it has been accessed by take(), as described in
Accessing DDS Data Samples with Read or Take (Section 7.4.3 on page 493). Note that read() does not
remove DDS samples from the queue.
A DataReader's receive queue size is limited by its RESOURCE_LIMITS QosPolicy (Section 6.5.20 on
page 405), specifically the max_samples field. The storage of out-of-order DDS samples for each
DataWriter is also allocated from the DataReader’s receive queue; this DDS sample resource is shared
among all reliable DataWriters. That is, max_samples includes both ordered and out-of-order DDS
samples.
A DataReader can maintain reliable communications with multiple DataWriters (e.g., in the case of the
OWNERSHIP_STRENGTH QosPolicy (Section 6.5.16 on page 393) setting of SHARED). The maximum number of out-of-order DDS samples from any one DataWriter that can occupy in the receive
queue is set in the max_samples_per_remote_writer field of the DATA_READER_RESOURCE_
LIMITS QosPolicy (DDS Extension) (Section 7.6.2 on page 517); this value can be used to prevent a
single DataWriter from using all the space in the receive queue. max_samples_per_remote_writer must
be set to be <= max_samples.
The DataReader will cache DDS samples that arrive out of order while waiting for missing DDS samples
to be resent. (Up to 256 DDS samples can be resent; this limitation is imposed by the wire protocol.) If
there is no room, the DataReader has to reject out-of-order DDS samples and request them again later
after the missing DDS samples have arrived.
The appropriate size of the receive queue depends on application parameters, such as the DataWriter’s
sending rate and the probability of a dropped DDS sample. However, the receive queue size should generally be larger than the send queue size. Send queue size is discussed in Understanding the Send Queue
and Setting its Size (Section 10.3.2.1 on page 639).
Effect of Receive-Queue Size on Performance: Large Queue Size (Section Figure 10.5 on the facing page)
and Effect of Receive Queue Size on Performance: Small Queue Size (Section Figure 10.6 on page 644)
compare two hypothetical DataReaders, both interacting with the same DataWriter. The queue on the left

642

10.3.2.2 Understanding the Receive Queue and Setting Its Size

represents an ordering cache, allocated from receive queue—DDS samples are held here if they arrive out
of order. The DataReader in Effect of Receive-Queue Size on Performance: Large Queue Size (Section
Figure 10.5 below) has a sufficiently large receive queue (max_samples) for the given send rate of the
DataWriter and other operational parameters. In both cases, we assume that all DDS samples are taken
from the DataReader in the Listener callback. (See Accessing DDS Data Samples with Read or Take (Section 7.4.3 on page 493) for information on take() and related operations.)
In Effect of Receive Queue Size on Performance: Small Queue Size (Section Figure 10.6 on the next
page), max_samples is too small to cache out-of-order DDS samples for the same operational parameters.
In both cases, the DataReaders eventually receive all the DDS samples in order. However, the
DataReader with the larger max_samples will get the DDS samples earlier and with fewer transactions.
In particular, DDS sample “4” is never resent for the DataReader with the larger queue size.
Figure 10.5 Effect of Receive-Queue Size on Performance: Large Queue Size

643

10.3.3 Controlling Queue Depth with the History QosPolicy

Figure 10.6 Effect of Receive Queue Size on Performance: Small Queue Size

10.3.3 Controlling Queue Depth with the History QosPolicy
If you want to achieve strict reliability, set the kind field in the HISTORY QosPolicy (Section 6.5.10 on
page 376) for both the DataReader and DataWriter to KEEP_ALL; in this case, the depth does not matter.
Or, for non-strict reliability, you can leave the kind set to KEEP_LAST (the default). This will provide
non-strict reliability; some DDS samples may not be delivered if the resource limit is reached.
The depth field in the HISTORY QosPolicy (Section 6.5.10 on page 376) controls how many DDS
samples Connext DDS will attempt to keep on the DataWriter’s send queue or the DataReader’s receive

644

10.3.4 Controlling Heartbeats and Retries with DataWriterProtocol QosPolicy

queue. For reliable communications, depth should be >= 1. The depth can be set to 1, but cannot be more
than the max_samples_per_instance in RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405).
Example:
l

DataWriter
writer_qos.history.depth = ;

l

DataReader
reader_qos.history.depth = ;

10.3.4 Controlling Heartbeats and Retries with DataWriterProtocol QosPolicy
In the Connext DDS reliability model, the DataWriter sends DDS data samples and heartbeats to reliable
DataReaders. A DataReader responds to a heartbeat by sending an ACKNACK, which tells the
DataWriter what the DataReader has received so far.
In addition, the DataReader can request missing DDS samples (by sending an ACKNACK) and the
DataWriter will respond by resending the missing DDS samples. This section describes some advanced
timing parameters that control the behavior of this mechanism. Many applications do not need to change
these settings. These parameters are contained in the DATA_WRITER_PROTOCOL QosPolicy (DDS
Extension) (Section 6.5.3 on page 347).
The protocol described in Overview of the Reliable Protocol (Section 10.2 on page 631) uses very simple
rules such as piggybacking HB messages to each DATA message and responding immediately to
ACKNACKs with the requested repair messages. While correct, this protocol would not be capable of
accommodating optimum performance in more advanced use cases.
This section describes some of the parameters configurable by means of the rtps_reliable_writer structure
in the DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347) and
how they affect the behavior of the RTPS protocol.
10.3.4.1 How Often Heartbeats are Resent (heartbeat_period)
If a DataReader does not acknowledge a DDS sample that has been sent, the DataWriter resends the heartbeat. These heartbeats are resent at the rate set in the DATA_WRITER_PROTOCOL QosPolicy (DDS
Extension) (Section 6.5.3 on page 347), specifically its heartbeat_period field.
For example, a heartbeat_period of 3 seconds means that if a DataReader does not receive the latest
DDS sample (for example, it gets dropped by the network), it might take up to 3 seconds before the
DataReader realizes it is missing data. The application can lower this value when it is important that recovery from packet loss is very fast.

645

10.3.4.1 How Often Heartbeats are Resent (heartbeat_period)

The basic approach of sending HB messages as a piggyback to DATA messages has the advantage of minimizing network traffic. However, there is a situation where this approach, by itself, may result in large
latencies. Suppose there is a DataWriter that writes bursts of data, separated by relatively long periods of
silence. Furthermore assume that the last message in one of the bursts is lost by the network. This is the
case shown for message DATA(B, 2) in Use of heartbeat_period (Section Figure 10.7 on the facing page).
If HBs were only sent piggybacked to DATA messages, the DataReader would not realize it missed the
‘B’ DATA message with sequence number ‘2’ until the DataWriter wrote the next message. This may be
a long time if data is written sporadically. To avoid this situation, Connext DDS can be configured so that
HBs are sent periodically as long as there are DDS samples that have not been acknowledged even if no
data is being sent. The period at which these HBs are sent is configurable by setting the rtps_reliable_
writer.heartbeat_period field in the DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).
Note that a small value for the heartbeat_period will result in a small worst-case latency if the last message in a burst is lost. This comes at the expense of the higher overhead introduced by more frequent HB
messages.
Also note that the heartbeat_period should not be less than the rtps_reliable_reader.heartbeat_suppression_duration in the DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1
on page 511); otherwise those HBs will be lost.

646

10.3.4.2 How Often Piggyback Heartbeats are Sent (heartbeats_per_max_samples)

Figure 10.7 Use of heartbeat_period

10.3.4.2 How Often Piggyback Heartbeats are Sent (heartbeats_per_max_samples)
A DataWriter will automatically send heartbeats with new DDS samples to request regular ACKNACKs
from the DataReader. These are called “piggyback” heartbeats.
A piggyback heartbeat is sent every [(current send-window size/heartbeats_per_max_samples)] number
of DDS samples written.

647

10.3.4.2 How Often Piggyback Heartbeats are Sent (heartbeats_per_max_samples)

The heartbeats_per_max_samples field is part of the rtps_reliable_writer structure in the DATA_
WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347). If heartbeats_per_
max_samples is set equal to max_send_window_size, this means that a heartbeat will be sent with each
DDS sample. A value of 8 means that a heartbeat will be sent with every 'current send-window size/8'
DDS samples. Say current send window is 1024, then a heartbeat will be sent once every 128 DDS
samples. If you set this to zero, DDS samples are sent without any piggyback heartbeat. The max_send_
window_size field is part of the DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section
6.5.3 on page 347).
Figure 10.1 Basic RTPS Reliable Protocol and Figure 10.2 RTPS Reliable Protocol in the Presence of
Message Loss seem to imply that a heartbeat (HB) is sent as a piggyback to each DATA message.
However, in situations where data is sent continuously at high rates, piggybacking a HB to each message
may result in too much overhead; not so much on the HB itself, but on the ACKNACKs that would be
sent back as replies by the DataReader.
There are two reasons to send a HB:
l

l

To request that a DataReader confirm the receipt of data via an ACKNACK, so that the DataWriter
can remove it from its send queue and therefore prevent the DataWriter’s history from filling up
(which could cause the write() operation to temporarily block1).
To inform the DataReader of what data it should have received, so that the DataReader can send a
request for missing data via an ACKNACK.

The DataWriter’s send queue can buffer many DDS data samples while it waits for ACKNACKs, and the
DataReader’s receive queue can store out-of-order DDS samples while it waits for missing ones. So it is
possible to send HB messages much less frequently than DATA messages. The ratio of piggyback HB
messages to DATA messages is controlled by the rtps_reliable_writer.heartbeats_per_max_samples
field in the DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).
A HB is used to get confirmation from DataReaders so that the DataWriter can remove acknowledged
DDS samples from the queue to make space for new DDS samples. Therefore, if the queue size is large,
or new DDS samples are added slowly, HBs can be sent less frequently.
In Use of heartbeats_per_max_samples (Section Figure 10.8 on the facing page), the DataWriter sets the
heartbeats_per_max_samples to certain value so that a piggyback HB will be sent for every three DDS
samples. The DataWriter first writes DDS sample A and B. The DataReader receives both. However,
since no HB has been received, the DataReader won’t send back an ACKNACK. The DataWriter will
still keep all the DDS samples in its queue. When the DataWriter sends DDS sample C, it will send a
piggyback HB along with the DDS sample. Once the DataReader receives the HB, it will send back an

1Note that data could also be removed from the DataWriter’s send queue if it is no longer relevant due to some other QoS

such a HISTORY KEEP_LAST (HISTORY QosPolicy (Section 6.5.10 on page 376)) or LIFESPAN (LIFESPAN QoS
Policy (Section 6.5.12 on page 381)).

648

10.3.4.3 Controlling Packet Size for Resent DDS Samples (max_bytes_per_nack_response)

ACKNACK for DDS samples up to sequence number 3, such that the DataWriter can remove all three
DDS samples from its queue.
Figure 10.8 Use of heartbeats_per_max_samples

10.3.4.3 Controlling Packet Size for Resent DDS Samples (max_bytes_per_nack_response)
A DataWriter may resend multiple missed DDS samples in the same packet. The max_bytes_per_nack_
response field in the DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on
page 347) limits the size of this ‘repair’ packet. The reliable DataWriter will include at least one sample in
the repair packet.

649

10.3.4.4 Controlling How Many Times Heartbeats are Resent (max_heartbeat_retries)

For example, if the DataReader requests 20 DDS samples, each 10K, and the max_bytes_per_nack_
response is set to 100K, the DataWriter will only send the first 10 DDS samples at most. The DataReader
will have to ACKNACK again to receive the other DDS samples.
Regardless of this setting, the maximum number of samples that can be part of a repair packet is limited to
32. This limit cannot be changed by configuration. In addition, the number of samples is limited by the
value of NDDS_Transport_Property_t’s gather_send_buffer_count_max (see Setting the Maximum
Gather-Send Buffer Count for UDPv4 and UDPv6 (Section 15.6.1 on page 763)).
10.3.4.4 Controlling How Many Times Heartbeats are Resent (max_heartbeat_retries)
If a DataReader does not respond within max_heartbeat_retries number of heartbeats, it will be dropped
by the DataWriter and the reliable DataWriter’s Listener will be called with a RELIABLE_READER_
ACTIVITY_CHANGED Status (DDS Extension) (Section 6.3.6.9 on page 281).
If the dropped DataReader becomes available again (perhaps its network connection was down temporarily), it will be added back to the DataWriter the next time the DataWriter receives some message
(ACKNACK) from the DataReader.
When a DataReader is ‘dropped’ by a DataWriter, the DataWriter will not wait for the DataReader to
send an ACKNACK before any DDS samples are removed. However, the DataWriter will still send data
and HBs to this DataReader as normal.
The max_heartbeat_retries field is part of the DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).
10.3.4.5 Treating Non-Progressing Readers as Inactive Readers (inactivate_
nonprogressing_readers)
In addition to max_heartbeat_retries, if inactivate_nonprogressing_readers is set, then not only are
non-responsive DataReaders considered inactive, but DataReaders sending non-progressing NACKs can
also be considered inactive. A non-progressing NACK is one which requests the same oldest DDS sample
as the previously received NACK. In this case, the DataWriter will not consider a non-progressing NACK
as coming from an active reader, and hence will inactivate the DataReader if no new NACKs are received
before max_heartbeat_retries number of heartbeat periods has passed.
One example for which it could be useful to turn on inactivate_nonprogressing_readers is when a
DataReader’s (keep-all) queue is full of untaken historical DDS samples. Each subsequent heartbeat
would trigger the same NACK, and nominally the DataReader would not be inactivated. A user not requiring strict-reliability could consider setting inactivate_nonprogressing_readers to allow the DataWriter to
progress rather than being held up by this non-progressing DataReader.

650

10.3.4.6 Coping with Redundant Requests for Missing DDS Samples (max_nack_response_delay)

10.3.4.6 Coping with Redundant Requests for Missing DDS Samples (max_nack_response_
delay)
When a DataWriter receives a request for missing DDS samples from a DataReader and responds by
resending the requested DDS samples, it will ignore additional requests for the same DDS samples during
the time period max_nack_response_delay.
The rtps_reliable_writer.max_nack_response_delay field is part of the DATA_WRITER_
PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).
If your send period is smaller than the round-trip delay of a message, this can cause unnecessary DDS
sample retransmissions due to redundant ACKNACKs. In this situation, an ACKNACK triggered by an
out-of-order DDS sample is not received before the next DDS sample is sent. When a DataReader
receives the next message, it will send another ACKNACK for the missing DDS sample. As illustrated in
Resending Missing Samples due to Duplicate ACKNACKs (Section Figure 10.9 below), duplicate
ACKNACK messages cause another resending of missing DDS sample “2” and lead to wasted CPU
usage on both the publication and the subscription sides.
Figure 10.9 Resending Missing Samples due to Duplicate ACKNACKs

651

10.3.4.7 Disabling Positive Acknowledgements (disable_positive_acks_min_sample_keep_duration)

While these redundant messages provide an extra cushion for the level of reliability desired, you can conserve the CPU and network bandwidth usage by limiting how often the same ACKNACK messages are
sent; this is controlled by min_nack_response_delay.
Reliable subscriptions are prevented from resending an ACKNACK within min_nack_response_delay
seconds from the last time an ACKNACK was sent for the same DDS sample. Our testing shows that the
default min_nack_response_delay of 0 seconds achieves an optimal balance for most applications on typical Ethernet LANs.
However, if your system has very slow computers and/or a slow network, you may want to consider
increasing min_nack_response_delay. Sending an ACKNACK and resending a missing DDS sample
inherently takes a long time in this system. So you should allow a longer time for recovery of the lost DDS
sample before sending another ACKNACK. In this situation, you should increase min_nack_response_
delay.
If your system consists of a fast network or computers, and the receive queue size is very small, then you
should keep min_nack_response_delay very small (such as the default value of 0). If the queue size is
small, recovering a missing DDS sample is more important than conserving CPU and network bandwidth
(new DDS samples that are too far ahead of the missing DDS sample are thrown away). A fast system can
cope with a smaller min_nack_response_delay value, and the reliable DDS sample stream can normalize
more quickly.
10.3.4.7 Disabling Positive Acknowledgements (disable_positive_acks_min_sample_keep_
duration)
When ACKNACK storms are a primary concern in a system, an alternative to tuning heartbeat and
ACKNACK response delays is to disable positive acknowledgments (ACKs) and rely just on NACKs to
maintain reliability. Systems with non-strict reliability requirements can disable ACKs to reduce network
traffic and directly solve the problem of ACK storms. ACKs can be disabled for the DataWriter and the
DataReader; when disabled for the DataWriter, none of its DataReaders will send ACKs, whereas disabling it at the DataReader allows per-DataReader configuration.
Normally when ACKs are enabled, strict reliability is maintained by the DataWriter, guaranteeing that a
DDS sample stays in its send queue until all DataReaders have positively acknowledged it (aside from relevant DURABILITY, HISTORY, and LIFESPAN QoS policies). When ACKs are disabled, strict reliability is no longer guaranteed, but the DataWriter should still keep the DDS sample for a sufficient
duration for ACK-disabled DataReaders to have a chance to NACK it. Thus, a configurable “keep-duration” (disable_postive_acks_min_sample_keep_duration) applies for DDS samples written for ACKdisabled DataReaders, where DDS samples are kept in the queue for at least that keep-duration. After the
keep-duration has elapsed for a DDS sample, the DDS sample is considered to be “acknowledged” by its
ACK-disabled DataReaders.
The keep duration should be configured for the expected worst-case from when the DDS sample is written
to when a NACK for the DDS sample could be received. If set too short, the DDS sample may no longer
be queued when a NACK requests it, which is the cost of not enforcing strict reliability.
652

10.3.5 Avoiding Message Storms with DataReaderProtocol QosPolicy

If the peak send rate is known and writer resources are available, the writer queue can be sized so that
writes will not block. For this case, the queue size must be greater than the send rate multiplied by the keep
duration.

10.3.5 Avoiding Message Storms with DataReaderProtocol QosPolicy
DataWriters send DDS data samples and heartbeats to DataReaders. A DataReader responds to a heartbeat by sending an acknowledgement that tells the DataWriter what the DataReader has received so far
and what it is missing. If there are many DataReaders, all sending ACKNACKs to the same DataWriter
at the same time, a message storm can result. To prevent this, you can set a delay for each DataReader, so
they don’t all send ACKNACKs at the same time. This delay is set in the DATA_READER_
PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511).
If you have several DataReaders per DataWriter, varying this delay for each one can avoid ACKNACK
message storms to the DataWriter. If you are not concerned about message storms, you do not need to
change this QosPolicy.
Example:
reader_qos.protocol.rtps_reliable_reader.min_heartbeat_response_delay.sec = 0;
reader_qos.protocol.rtps_reliable_reader.min_heartbeat_response_delay.nanosec = 0;
reader_qos.protocol.rtps_reliable_reader.max_heartbeat_response_delay.sec = 0;
reader_qos.protocol.rtps_reliable_reader.max_heartbeat_response_delay.nanosec =
0.5 * 1000000000UL; // 0.5 sec

As the name suggests, the minimum and maximum response delay bounds the random wait time before the
response. Setting both to zero will force immediate response, which may be necessary for the fastest recovery in case of lost DDS samples.

10.3.6 Resending DDS Samples to Late-Joiners with the Durability
QosPolicy
The DURABILITY QosPolicy (Section 6.5.7 on page 368) is also somewhat related to Reliability. Connext DDS requires a finite time to "discover" or match DataReaders to DataWriters. If an application
attempts to send data before the DataReader and DataWriter "discover" one another, then the DDS sample
will not actually get sent. Whether or not DDS samples are resent when the DataReader and DataWriter
eventually "discover" one another depends on how the DURABILITY and HISTORY QoS are set. The
default setting for the Durability QosPolicy is VOLATILE, which means that the DataWriter will not store
DDS samples for redelivery to late-joining DataReaders.
Connext DDS also supports the TRANSIENT_LOCAL setting for the Durability, which means that the
DDS samples will be kept stored for redelivery to late-joining DataReaders, as long as the DataWriter is

653

10.3.7 Use Cases

around and the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) allows. The DDS
samples are not stored beyond the lifecycle of the DataWriter.
See also: Waiting for Historical Data (Section 7.3.6 on page 469).

10.3.7 Use Cases
This section contains advanced material that discusses practical applications of the reliability related QoS.
10.3.7.1 Importance of Relative Thread Priorities
For high throughput, the Connext DDS Event thread’s priority must be sufficiently high on the sending
application. Unlike an unreliable writer, a reliable writer relies on internal Connext DDS threads: the
Receive thread processes ACKNACKs from the DataReaders, and the Event thread schedules the events
necessary to maintain reliable data flow.
l

l

When DDS samples are sent to the same or another application on the same host, the Receive thread
priority should be higher than the writing thread priority (priority of the thread calling write() on the
DataWriter). This will allow the Receive thread to process the messages as they are sent by the writing thread. A sustained reliable flow requires the reader to be able to process the DDS samples from
the writer at a speed equal to or faster than the writer emits.
The default Event thread priority is low. This is adequate if your reliable transfer is not sustained;
queued up events will eventually be processed when the writing thread yields the CPU. The Connext DDS can automatically grow the event queue to store all pending events. But if the reliable
communication is sustained, reliable events will continue to be scheduled, and the event queue will
eventually reach its limit. The default Event thread priority is unsuitable for maintaining a fast and
sustained reliable communication and should be increased through the participant_qos.event.thread.priority. This value maps directly to the OS thread priority, see EVENT QosPolicy (DDS
Extension) (Section 8.5.5 on page 602)).

The Event thread should also be increased to minimize the reliable latency. If events are processed at a
higher priority, dropped packets will be resent sooner.
Now we consider some practical applications of the reliability related QoS:
l

Aperiodic Use Case: One-at-a-Time (Section 10.3.7.2 on the facing page)

l

Aperiodic, Bursty (Section 10.3.7.3 on page 659)

l

Periodic (Section 10.3.7.4 on page 664)

654

10.3.7.2 Aperiodic Use Case: One-at-a-Time

10.3.7.2 Aperiodic Use Case: One-at-a-Time
Suppose you have aperiodically generated data that needs to be delivered reliably, with minimum latency,
such as a series of commands (“Ready,” “Aim,” “Fire”). If a writing thread may block between each DDS
sample to guarantee reception of the just-sent DDS sample on the reader’s middleware end, a smaller
queue will provide a smaller upper bound on the DDS sample delivery time. Adequate writer QoS for this
use case are presented in Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer below.
Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer
1.

qos->reliability.kind = DDS_RELIABLE_RELIABILITY_QOS;

2.

qos->history.kind = DDS_KEEP_ALL_HISTORY_QOS;

3.

qos->protocol.push_on_write = DDS_BOOLEAN_TRUE;

4.
5.

//use these hard coded value unless you use a key

6.

qos->resource_limits.initial_samples = qos->resource_limits.max_samples = 1;

7.

qos->resource_limits.max_samples_per_instance =

8.

qos->resource_limits.max_samples;

9.

qos->resource_limits.initial_instances =

10.

qos->resource_limits.max_instances = 1;

11.
12.

// want to piggyback HB w/ every sample.

13.

qos->protocol.rtps_reliable_writer.heartbeats_per_max_samples =

14.

qos->resource_limits.max_samples;

15.
16.

qos->protocol.rtps_reliable_writer.high_watermark = 1;

17.

qos->protocol.rtps_reliable_writer.low_watermark = 0;

18.

qos->protocol.rtps_reliable_writer.min_nack_response_delay.sec = 0;

19.

qos->protocol.rtps_reliable_writer.min_nack_response_delay.nanosec = 0;

20.

//consider making non-zero for reliable multicast

21.

qos->protocol.rtps_reliable_writer.max_nack_response_delay.sec = 0;

22.

qos->protocol.rtps_reliable_writer.max_nack_response_delay.nanosec = 0;

23.
24.

// should be faster than the send rate, but be mindful of OS resolution

25.

25 qos->protocol.rtps_reliable_writer.fast_heartbeat_period.sec = 0;

26.

26 qos->protocol.rtps_reliable_writer.fast_heartbeat_period.nanosec =

655

10.3.7.2 Aperiodic Use Case: One-at-a-Time

27.

alertReaderWithinThisMs * 1000000;

28.
29.

qos->reliability.max_blocking_time = blockingTime;

30.

qos->protocol.rtps_reliable_writer.max_heartbeat_retries = 7;

31.
32.

// essentially turn off slow HB period

33.

qos->protocol.rtps_reliable_writer.heartbeat_period.sec = 3600 * 24 * 7;

Line 1 (Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer on the previous page): This is
the default setting for a writer, shown here strictly for clarity.
Line 2 (Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer on the previous page): Setting
the History kind to KEEP_ALL guarantees that no DDS sample is ever lost.
Line 3 (Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer on the previous page): This is
the default setting for a writer, shown here strictly for clarity. ‘Push’ mode reliability will yield lower
latency than ‘pull’ mode reliability in normal situations where there is no DDS sample loss. (See DATA_
WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).) Furthermore, it does
not matter that each packet sent in response to a command will be small, because our data sent with each
command is likely to be small, so that maximizing throughput for this data is not a concern.
Line 5 - Line 10 (Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer on the previous
page): For this example, we assume a single writer is writing DDS samples one at a time. If we are not
using keys (see DDS Samples, Instances, and Keys (Section 2.3.1 on page 14)), there is no reason to use
a queue with room for more than one DDS sample, because we want to resolve a DDS sample completely
before moving on to the next. While this negatively impacts throughput, it minimizes memory usage. In
this example, a written DDS sample will remain in the queue until it is acknowledged by all active readers
(only 1 for this example).
Line 12 - Line 14 (Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer on the previous
page): The fastest way for a writer to ensure that a reader is up-to-date is to force an acknowledgment with
every DDS sample. We do this by appending a Heartbeat with every DDS sample. This is akin to a certified mail; the writer learns—as soon as the system will allow—whether a reader has received the letter,
and can take corrective action if the reader has not. As with certified mail, this model has significant overhead compared to the unreliable case, trading off lower packet efficiency in favor of latency and fast recovery.
Line 16-Line 17 (Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer on the previous
page): Since the writer takes responsibility for pushing the DDS samples out to the reader, a writer will go
into a “heightened alert” mode as soon as the high water mark is reached (which is when any DDS sample
is written for this writer) and only come out of this mode when the low water mark is reached (when all
DDS samples have been acknowledged for this writer). Note that the selected high and low watermarks
are actually the default values.

656

10.3.7.2 Aperiodic Use Case: One-at-a-Time

Line 18-Line 22 (Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer on page 655): When
a reader requests a lost DDS sample, we respond to the reader immediately in the interest of faster recovery. If the readers receive packets on unicast, there is no reason to wait, since the writer will eventually
have to feed individual readers separately anyway. In case of multicast readers, it makes sense to consider
further. If the writer delayed its response enough so that all or most of the readers have had a chance to
NACK a DDS sample, the writer may coalesce the requests and send just one packet to all the multicast
readers. Suppose that all multicast readers do indeed NACK within approximately 100 µsec. Setting the
minimum and maximum delays at 100 µsec will allow the writer to collect all these NACKs and send a
single response over multicast. (See DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347) for information on setting min_nack_response_delay and max_nack_
response_delay.) Note that Connext DDS relies on the OS to wait for this 100 µsec. Unfortunately, not
all operating systems can sleep for such a fine duration. On Windows systems, for example, the minimum
achievable sleep time is somewhere between 1 to 20 milliseconds, depending on the version. On VxWorks
systems, the minimum resolution of the wait time is based on the tick resolution, which is 1/system clock
rate (thus, if the system clock rate is 100 Hz, the tick resolution is 10 millisecond). On such systems, the
achievable minimum wait is actually far larger than the desired wait time. This could have an unintended
consequence due to the delay caused by the OS; at a minimum, the time to repair a packet may be longer
than you specified.
Line 24-Line 27 (Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer on page 655): If a
reader drops a DDS sample, the writer recovers by notifying the reader of what it has sent, so that the
reader may request resending of the lost DDS sample. Therefore, the recovery time depends primarily on
how quickly the writer pings the reader that has fallen behind. If commands will not be generated faster
than one every few seconds, it may be acceptable for the writer to ping the reader several hundred milliseconds after the DDS sample is sent.
l

Suppose that the round-trip time of fairly small packets between the writer and the reader application
is 50 microseconds, and that the reader does not delay response to a Heartbeat from the writer (see
DATA_READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511) for
how to change this). If a DDS sample is dropped, the writer will ping the reader after a maximum of
the OS delay resolution discussed above and alertReaderWithinThisMs (let’s say 10 ms for this
example). The reader will request the missing DDS sample immediately, and with the code set as
above, the writer will feed the missing DDS sample immediately. Neglecting the processing time on
the writer or the reader end, and assuming that this retry succeeds, the time to recover the DDS
sample from the original publication time is: alertReaderWithinThisMs + 50 µsec + 25 µsec.
If the OS is capable of micro-sleep, the recovery time can be within 100 µsec, barely noticeable to a
human operator. If the OS minimum wait resolution is much larger, the recovery time is dominated
by the wait resolution of the OS. Since ergonomic studies suggest that delays in excess of a 0.25
seconds start hampering operations that require low latency data, even a 10 ms limitation seems to
be acceptable.

657

10.3.7.2 Aperiodic Use Case: One-at-a-Time

l

What if two packets are dropped in a row? Then the recovery time would be
2 * alertReaderWithinThisMs + 2 * 50 µsec + 25 µsec. If alertReaderWithinThisMs is 100 ms,
the recovery time now exceeds 200 ms, and can perhaps degrade user experience.

Line 29-Line 30 (Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer on page 655): What if
another command (like another button press) is issued before the recovery? Since we must not drop this
new DDS sample, we block the writer until the recovery completes. If alertReaderWithinThisMs is 10
ms, and we assume no more than 7 consecutive drops, the longest time for recovery will be just above
(alertReaderWithinThisMs * max_heartbeat_retries), or 70 ms.
So if we set blockingTime to about 80 ms, we will have given enough chance for recovery. Of course, in
a dynamic system, a reader may drop out at any time, in which case max_heartbeat_retries will be
exceeded, and the unresponsive reader will be dropped by the writer. In either case, the writer can continue writing. Inappropriate values will cause a writer to prematurely drop a temporarily unresponsive (but
otherwise healthy) reader, or be stuck trying unsuccessfully to feed a crashed reader. In the unfortunate
case where a reader becomes temporarily unresponsive for a duration exceeding (alertReaderWithinThisMs * max_heartbeat_retries), the writer may issue gaps to that reader when it
becomes active again; the dropped DDS samples are irrecoverable. So estimating the worst case unresponsive time of all potential readers is critical if DDS sample drop is unacceptable.
Line 33 (Figure 10.10 QoS for an Aperiodic, One-at-a-time Reliable Writer on page 655): Since the command may not be issued for hours or even days on end, there is no reason to keep announcing the writer’s
state to the readers.
Figure 10.11 QoS for an Aperiodic, One-at-a-time Reliable Reader below shows how to set the QoS for
the reader side, followed by a line-by-line explanation.
Figure 10.11 QoS for an Aperiodic, One-at-a-time Reliable Reader
1. qos->reliability.kind = DDS_RELIABLE_RELIABILITY_QOS;
2.

qos->history.kind = DDS_KEEP_ALL_HISTORY_QOS;

3.
4.

// 1 is ok for normal use. 2 allows fast infinite loop

5.

qos->reader_resource_limits.max_samples_per_remote_writer = 2;

6.

qos->resource_limits.initial_samples = 2;

7.

qos->resource_limits.initial_instances = 1;

8.
9.

qos->protocol.rtps_reliable_reader.max_heartbeat_response_delay.sec = 0;

10.
0;

qos->protocol.rtps_reliable_reader.max_heartbeat_response_delay.nanosec =

11.

qos->protocol.rtps_reliable_reader.min_heartbeat_response_delay.sec = 0;

658

10.3.7.3 Aperiodic, Bursty

12.
0;

qos->protocol.rtps_reliable_reader.min_heartbeat_response_delay.nanosec =

Line 1-Line 2 (Figure 10.11 QoS for an Aperiodic, One-at-a-time Reliable Reader on the previous page):
Unlike a writer, the reader’s default reliability setting is best-effort, so reliability must be turned on. Since
we don’t want to drop anything, we choose KEEP_ALL history.
Line 4-Line 6 (Figure 10.11 QoS for an Aperiodic, One-at-a-time Reliable Reader on the previous page):
Since we enforce reliability on each DDS sample, it would be sufficient to keep the queue size at 1, except
in the following case: suppose that the reader takes some action in response to the command received,
which in turn causes the writer to issue another command right away. Because Connext DDS passes the
user data up to the application even before acknowledging the DDS sample to the writer (for minimum
latency), the first DDS sample is still pending for acknowledgement in the writer’s queue when the writer
attempts to write the second DDS sample, and will cause the writing thread to block until the reader completes processing the first DDS sample and acknowledges it to the writer; all are as they should be. But if
you want to run this infinite loop at full throttle, the reader should buffer one more DDS sample. Let’s follow the packets flow under a normal circumstance:
1. The sender application writes DDS sample 1 to the reader. The receiver application processes it and
sends a user-level response 1 to the sender application, but has not yet ACK’d DDS sample 1.
2. The sender application writes DDS sample 2 to the receiving application in response to response 1.
Because the reader’s queue is 2, it can accept DDS sample 2 even though it may not yet have
acknowledged DDS sample 1. Otherwise, the reader may drop DDS sample 2, and would have to
recover it later.
3. At the same time, the receiver application acknowledges DDS sample 1, and frees up one slot in the
queue, so that it can accept DDS sample 3, which it on its way.
The above steps can be repeated ad-infinitum in a continuous traffic.
Line 7 (Figure 10.11 QoS for an Aperiodic, One-at-a-time Reliable Reader on the previous page): Since
we are not using keys, there is just one instance.
Line 9-Line 12 (Use Cases (Section 10.3.7 on page 654)): We choose immediate response in the interest
of fastest recovery. In high throughput, multicast scenario, delaying the response (with event thread priority set high of course) may decrease the likelihood of NACK storm causing a writer to drop some
NACKs. This random delay reduces this chance by staggering the NACK response. But the minimum
delay achievable once again depends on the OS.
10.3.7.3 Aperiodic, Bursty
Suppose you have aperiodically generated bursts of data, as in the case of a new aircraft approaching an
airport. The data may be the same or different, but if they are written by a single writer, the challenge to
this writer is to feed all readers as quickly and efficiently as possible when this burst of hundreds or thousands of DDS samples hits the system.

659

10.3.7.3 Aperiodic, Bursty

If you use an unreliable writer to push this burst of data, some of them may be dropped over an unreliable
transport such as UDP.
If you try to shape the burst according to however much the slowest reader can process, the system
throughput may suffer, and places an additional burden of queueing the DDS samples on the sender application.
If you push the data reliably as fast they are generated, this may cost dearly in repair packets, especially to
the slowest reader, which is already burdened with application chores.
Connext DDS pull mode reliability offers an alternative in this case by letting each reader pace its own
data stream. It works by notifying the reader what it is missing, then waiting for it to request only as much
as it can handle. As in the aperiodic one-at-a-time case (Aperiodic Use Case: One-at-a-Time (Section
10.3.7.2 on page 655)), multicast is supported, but its performance depends on the resolution of the minimum delay supported by the OS. At the cost of greater latency, this model can deliver reliability while
using far fewer packets than in the push mode. The writer QoS is given in Figure 10.12 QoS for an Aperiodic, Bursty Writer below, with a line-by-line explanation below.
Figure 10.12 QoS for an Aperiodic, Bursty Writer
1. qos->reliability.kind = DDS_RELIABLE_RELIABILITY_QOS;
2.

qos->history.kind = DDS_KEEP_ALL_HISTORY_QOS;

3.

qos->protocol.push_on_write = DDS_BOOLEAN_FALSE;

4.
5.

//use these hard coded value until you use key

6.

qos->resource_limits.initial_instances =

7.

qos->resource_limits.max_instances = 1;

8.

qos->resource_limits.initial_samples = qos->resource_limits.max_samples

9.

= worstBurstInSample;

10.

qos->resource_limits.max_samples_per_instance =

11.

qos->resource_limits.max_samples;

12.
13.

// piggyback HB not used

14.

qos->protocol.rtps_reliable_writer.heartbeats_per_max_samples = 0;

15.
16.

qos->protocol.rtps_reliable_writer.high_watermark = 1;

17.

qos->protocol.rtps_reliable_writer.low_watermark = 0;

18.
19.

qos->protocol.rtps_reliable_writer.min_nack_response_delay.sec = 0;

660

10.3.7.3 Aperiodic, Bursty

20.

qos->protocol.rtps_reliable_writer.min_nack_response_delay.nanosec = 0;

21.

qos->protocol.rtps_reliable_writer.max_nack_response_delay.sec = 0;

22.

qos->protocol.rtps_reliable_writer.max_nack_response_delay.nanosec = 0;

23.

qos->reliability.max_blocking_time = blockingTime;

24.
25.

// should be faster than the send rate, but be mindful of OS resolution

26.

qos->protocol.rtps_reliable_writer.fast_heartbeat_period.sec = 0;

27.

qos->protocol.rtps_reliable_writer.fast_heartbeat_period.nanosec =

28.

alertReaderWithinThisMs * 1000000;

29.

qos->protocol.rtps_reliable_writer.max_heartbeat_retries = 5;

30.
31.
32.

// essentially turn off slow HB period
qos->protocol.rtps_reliable_writer.heartbeat_period.sec = 3600 * 24 * 7;

Line 1 (Figure 10.12 QoS for an Aperiodic, Bursty Writer on the previous page): This is the default setting
for a writer, shown here strictly for clarity.
Line 2 (Figure 10.12 QoS for an Aperiodic, Bursty Writer on the previous page): Since we do not want
any data lost, we want the History kind set to KEEP_ALL.
Line 3 (Figure 10.12 QoS for an Aperiodic, Bursty Writer on the previous page): The default Connext
DDS reliable writer will push, but we want the reader to pull instead.
Line 5-Line 11 (Figure 10.12 QoS for an Aperiodic, Bursty Writer on the previous page): We assume a
single instance, in which case the maximum DDS sample count will be the same as the maximum DDS
sample count per writer. In contrast to the one-at-a-time case discussed in Aperiodic Use Case: One-at-aTime (Section 10.3.7.2 on page 655), the writer’s queue is large; as big as the burst size in fact, but no
more because this model tries to resolve a burst within a reasonable period, to be computed shortly. Of
course, we could block the writing thread in the middle of the burst, but that might complicate the design
of the sending application.
Line 13-Line 14 (Figure 10.12 QoS for an Aperiodic, Bursty Writer on the previous page): By a ‘piggyback’ Heartbeat, we mean only a Heartbeat that is appended to data being pushed from the writer. Strictly
speaking, the writer will also append a Heartbeat with each reply to a reader’s lost DDS sample request,
but we call that a ‘framing’ Heartbeat. Since data is pulled, heartbeats_per_max_samples is ignored.
Line 16-Line 17 (Figure 10.12 QoS for an Aperiodic, Bursty Writer on the previous page): Similar to the
previous aperiodic writer, this writer spends most of its time idle. But as the name suggests, even a single
new DDS sample implies more DDS sample to follow in a burst. Putting the writer into a fast mode
quickly will allow readers to be notified soon. Only when all DDS samples have been delivered, the writer
can rest.

661

10.3.7.3 Aperiodic, Bursty

Line 19- Line 23 (Figure 10.12 QoS for an Aperiodic, Bursty Writer on page 660): Similar to the one-at-atime case, there is no reason to delay response with only one reader. In this case, we can estimate the time
to resolve a burst with only a few parameters. Let’s say that the reader figures it can safely receive and process 20 DDS samples at a time without being overwhelmed, and that the time it takes a writer to fetch
these 20 DDS samples and send a single packet containing these 20 DDS samples, plus the time it takes a
reader to receive and process these DDS samples, and send another request back to the writer for the next
20 DDS samples is 11 ms. Even on the same hardware, if the reader’s processing time can be reduced, this
time will decrease; other factors such as the traversal time through Connext DDS and the transport are typically in microseconds range (depending on machines of course).
For example, let’s also say that the worst case burst is 1000 DDS samples. The writing thread will of
course not block because it is merely copying each of the 1000 DDS samples to the Connext DDS queue
on the writer side; on a typical modern machine, the act of writing these 1000 DDS samples will probably
take no more than a few ms. But it would take at least 1000/20 = 50 resend packets for the reader to catch
up to the writer, or 50 times 11 ms = 550 ms. Since the burst model deals with one burst at a time, we
would expect that another burst would not come within this time, and that we are allowed to block for at
least this period. Including a safety margin, it would appear that we can comfortably handle a burst of
1000 every second or so.
But what if there are multiple readers? The writer would then take more time to feed multiple readers, but
with a fast transport, a few more readers may only increase the 11 ms to only 12 ms or so. Eventually, however, the number of readers will justify the use of multicast. Even in pull mode, Connext DDS supports
multicast by measuring how many multicast readers have requested DDS sample repair. If the writer does
not delay response to NACK, then repairs will be sent in unicast. But a suitable NACK delay allows the
writer to collect potentially NACKs from multiple readers, and feed a single multicast packet. But as discussed in Aperiodic Use Case: One-at-a-Time (Section 10.3.7.2 on page 655), by delaying reply to
coalesce response, we may end up waiting much longer than desired. On a Windows system with 10 ms
minimum sleep achievable, the delay would add at least 10 ms to the 11 ms delay, so that the time to push
1000 DDS samples now increases to 50 times 21 ms = 1.05 seconds. It would appear that we will not be
able to keep up with incoming burst if it came at roughly 1 second, although we put fewer packets on the
wire by taking advantage of multicast.
Line 25-Line 28 (Use Cases (Section 10.3.7 on page 654)): We now understand how the writer feeds the
reader in response to the NACKs. But how does the reader realize that it is behind? The writer notifies the
reader with a Heartbeat to kick-start the exchange. Therefore, the latency will be lower bound by the
writer’s fast heartbeat period. If the application is not particularly sensitive to latency, the minimum wait
time supported by the OS (10 ms on Windows systems, for example) might be a reasonable value.
Line 29 (Figure 10.12 QoS for an Aperiodic, Bursty Writer on page 660): With a fast heartbeat period of
50 ms, a writer will take 500 ms (50 ms times the default max_heartbeat_retries of 10) to write-off an
unresponsive reader. If a reader crashes while we are writing a lot of DDS samples per second, the writer
queue may completely fill up before the writer has a chance to drop the crashed reader. Lowering max_
heartbeat_retries will prevent that scenario.

662

10.3.7.3 Aperiodic, Bursty

Line 31-Line 32 (Figure 10.12 QoS for an Aperiodic, Bursty Writer on page 660): For an aperiodic writer,
turning off slow periodic Heartbeats will remove unwanted traffic from the network.
Figure 10.13 QoS for an Aperiodic, Bursty Reader below shows example code for a corresponding aperiodic, bursty reader.
Figure 10.13 QoS for an Aperiodic, Bursty Reader
1. qos->reliability.kind = DDS_RELIABLE_RELIABILITY_QOS;
2.

qos->history.kind = DDS_KEEP_ALL_HISTORY_QOS;

3.

qos->resource_limits.initial_samples =

4.

qos->resource_limits.max_samples =

5.

qos->reader_resource_limits.max_samples_per_remote_writer = 32;

6.
7.

//use these hard coded value until you use key

8.

qos->resource_limits.max_samples_per_instance =

9.

qos->resource_limits.max_samples;

10.

qos->resource_limits.initial_instances =

11.

qos->resource_limits.max_instances = 1;

12.
13.

// the writer probably has more for the reader; ask right away

14.

qos->protocol.rtps_reliable_reader.min_heartbeat_response_delay.sec = 0;

15.
0;

qos->protocol.rtps_reliable_reader.min_heartbeat_response_delay.nanosec =

16.

qos->protocol.rtps_reliable_reader.max_heartbeat_response_delay.sec = 0;

17.
0;

qos->protocol.rtps_reliable_reader.max_heartbeat_response_delay.nanosec =

Line 1-Line 2 (Figure 10.13 QoS for an Aperiodic, Bursty Reader above): Unlike a writer, the reader’s
default reliability setting is best-effort, so reliability must be turned on. Since we don’t want to drop anything, we choose KEEP_ALL for the History QoS kind.
Line 3-Line 5 (Figure 10.13 QoS for an Aperiodic, Bursty Reader above): Unlike the writer, the reader’s
queue can be kept small, since the reader is free to send ACKs for as much as it wants anyway. In general,
the larger the queue, the larger the packet needs to be, and the higher the throughput will be. When the
reader NACKs for lost DDS sample, it will only ask for this much.
Line 7-Line 11 (Figure 10.13 QoS for an Aperiodic, Bursty Reader above): We do not use keys in this
example.

663

10.3.7.4 Periodic

Line 13-Line 17 (Figure 10.13 QoS for an Aperiodic, Bursty Reader on the previous page): We respond
immediately to catch up as soon as possible. When there are many readers, this may cause a NACK storm,
as discussed in the reader code for one-at-a-time reliable reader.
10.3.7.4 Periodic
In a periodic reliable model, we can use the writer and the reader queue to keep the data flowing at a
smooth rate. The data flows from the sending application to the writer queue, then to the transport, then to
the reader queue, and finally to the receiving application. Unless the sending application or any one of the
receiving applications becomes unresponsive (including a crash) for a noticeable duration, this flow should
continue uninterrupted.
The latency will be low in most cases, but will be several times higher for the recovered and many subsequent DDS samples. In the event of a disruption (e.g., loss in transport, or one of the readers becoming
temporarily unresponsive), the writer’s queue level will rise, and may even block in the worst case. If the
writing thread must not block, the writer’s queue must be sized sufficiently large to deal with any fluctuation in the system. Figure 10.14 QoS for a Periodic Reliable Writer below shows an example, with lineby-line analysis below.
Figure 10.14 QoS for a Periodic Reliable Writer
1. qos->reliability.kind = DDS_RELIABLE_RELIABILITY_QOS;
2.

qos->history.kind = DDS_KEEP_ALL_HISTORY_QOS;

3.

qos->protocol.push_on_write = DDS_BOOLEAN_TRUE;

4.
5.

//use these hard coded value until you use key

6.

qos->resource_limits.initial_instances =

7.

qos->resource_limits.max_instances = 1;

8.
9.
10.

int unresolvedSamplePerRemoteWriterMax =
worstCaseApplicationDelayTimeInMs * dataRateInHz / 1000;

11.

qos->resource_limits.max_samples = unresolvedSamplePerRemoteWriterMax;

12.

qos->resource_limits.initial_samples = qos->resource_limits.max_samples/2;

13.

qos->resource_limits.max_samples_per_instance =

14.

qos->resource_limits.max_samples;

15.
16.

int piggybackEvery = 8;

17.

qos->protocol.rtps_reliable_writer.heartbeats_per_max_samples =

664

10.3.7.4 Periodic

18.

qos->resource_limits.max_samples / piggybackEvery;

19.
20.

qos->protocol.rtps_reliable_writer.high_watermark = piggybackEvery * 4;

21.

qos->protocol.rtps_reliable_writer.low_watermark = piggybackEvery * 2;

22.

qos->reliability.max_blocking_time = blockingTime;

23.
24.

qos->protocol.rtps_reliable_writer.min_nack_response_delay.sec = 0;

25.

qos->protocol.rtps_reliable_writer.min_nack_response_delay.nanosec = 0;

26.
27.

qos->protocol.rtps_reliable_writer.max_nack_response_delay.sec = 0;

28.

qos->protocol.rtps_reliable_writer.max_nack_response_delay.nanosec = 0;

29.
30.

qos->protocol.rtps_reliable_writer.fast_heartbeat_period.sec = 0;

31.

qos->protocol.rtps_reliable_writer.fast_heartbeat_period.nanosec =

32.

`

33.

qos->protocol.rtps_reliable_writer.max_heartbeat_retries = 7;

alertReaderWithinThisMs * 1000000;

34.
35.

// essentially turn off slow HB period

36.

qos->protocol.rtps_reliable_writer.heartbeat_period.sec = 3600 * 24 * 7;

Line 1 (Figure 10.14 QoS for a Periodic Reliable Writer on the previous page): This is the default setting
for a writer, shown here strictly for clarity.
Line 2 (Figure 10.14 QoS for a Periodic Reliable Writer on the previous page): Since we do not want any
data lost, we set the History kind to KEEP_ALL.
Line 3 (Figure 10.14 QoS for a Periodic Reliable Writer on the previous page): This is the default setting
for a writer, shown here strictly for clarity. Pushing will yield lower latency than pulling.
Line 5-Line 7 (Figure 10.14 QoS for a Periodic Reliable Writer on the previous page): We do not use keys
in this example, so there is only one instance.
Line 9-Line 11 (Figure 10.14 QoS for a Periodic Reliable Writer on the previous page): Though a
simplistic model of queue, this is consistent with the idea that the queue size should be proportional to the
data rate and the wort case jitter in communication.
Line 12 (Figure 10.14 QoS for a Periodic Reliable Writer on the previous page): Even though we have
sized the queue according to the worst case, there is a possibility for saving some memory in the normal
case. Here, we initially size the queue to be only half of the worst case, hoping that the worst case will not
occur. When it does, Connext DDS will keep increasing the queue size as necessary to accommodate new
DDS samples, until the maximum is reached. So when our optimistic initial queue size is breached, we

665

10.3.7.4 Periodic

will incur the penalty of dynamic memory allocation. Furthermore, you will wind up using more memory,
as the initially allocated memory will be orphaned (note: does not mean a memory leak or dangling
pointer); if the initial queue size is M_i and the maximal queue size is M_m, where M_m = M_i * 2^n, the
memory wasted in the worst case will be (M_m - 1) * sizeof(DDS sample) bytes. Note that the memory
allocation can be avoided by setting the initial queue size equal to its max value.
Line 13-Line 14 (Figure 10.14 QoS for a Periodic Reliable Writer on page 664): If there is only one
instance, maximum DDS samples per instance is the same as maximum DDS samples allowed.
Line 16-Line 18 (Figure 10.14 QoS for a Periodic Reliable Writer on page 664): Since we are pushing out
the data at a potentially rapid rate, the piggyback heartbeat will be useful in letting the reader know about
any missing DDS samples. The piggybackEvery can be increased if the writer is writing at a fast rate,
with the cost that more DDS samples will need to queue up for possible resend. That is, you can consider
the piggyback heartbeat to be taking over one of the roles of the periodic heartbeat in the case of a push.
So sending fewer DDS samples between piggyback heartbeats is akin to decreasing the fast heartbeat
period seen in previous sections. Please note that we cannot express piggybackEvery directly as its own
QoS, but indirectly through the maximum DDS samples.
Line 20-Line 22 (Figure 10.14 QoS for a Periodic Reliable Writer on page 664): If piggybackEvery was
exactly identical to the fast heartbeat, there would be no need for fast heartbeat or the high watermark. But
one of the important roles for the fast heartbeat period is to allow a writer to abandon inactive readers
before the queue fills. If the high watermark is set equal to the queue size, the writer would not doubt the
status of an unresponsive reader until the queue completely fills—blocking on the next write (up to blockingTime). By lowering the high watermark, you can control how vigilant a writer is about checking the
status of unresponsive readers. By scaling the high watermark to piggybackEvery, the writer is expressing
confidence that an alive reader will respond promptly within the time it would take a writer to send 4 times
piggybackEvery DDS samples. If the reader does not delay the response too long, this would be a good
assumption. Even if the writer estimated on the low side and does go into fast mode (suspecting that the
reader has crashed) when a reader is temporarily unresponsive (e.g., when it is performing heavy computation for a few milliseconds), a response from the reader in question will resolve any doubt, and data
delivery can continue uninterrupted. As the reader catches up to the writer and the queue level falls below
the low watermark, the writer will pop out to the normal, relaxed mode.
Line 24-Line 28 (Figure 10.14 QoS for a Periodic Reliable Writer on page 664): When a reader is behind
(including a reader whose Durability QoS is non-VOLATILE and therefore needs to catch up to the writer
as soon as it is created), how quickly the writer responds to the reader’s request will determine the catch-up
rate. While a multicast writer (that is, a writer with multicast readers) may consider delaying for some time
to take advantage of coalesced multicast packets. Keep in mind the OS delay resolution issue discussed in
the previous section.
Line 30-Line 33 (Figure 10.14 QoS for a Periodic Reliable Writer on page 664): The fast heartbeat mechanism allows a writer to detect a crashed reader and move along with the remaining readers when a reader
does not respond to any of the max_heartbeat_retries number of heartbeats sent at the fast_heartbeat_

666

10.3.7.4 Periodic

period rate. So if you want a more cautious writer, decrease either numbers; conversely, increasing either
number will result in a writer that is more reluctant to write-off an unresponsive reader.
Line 35-Line 36 (Figure 10.14 QoS for a Periodic Reliable Writer on page 664): Since this a periodic
model, a separate periodic heartbeat to notify the writer’s status would seem unwarranted; the piggyback
heartbeat sent with DDS samples takes over that role.
Figure 10.15 QoS for a Periodic Reliable Reader below shows how to set the QoS for a matching reader,
followed by a line-by-line explanation.
Figure 10.15 QoS for a Periodic Reliable Reader
1. qos->reliability.kind = DDS_RELIABLE_RELIABILITY_QOS;
2.

qos->history.kind = DDS_KEEP_ALL_HISTORY_QOS;

3.

qos->resource_limits.initial_samples =

4.

qos->resource_limits.max_samples =

5.

qos->reader_resource_limits.max_samples_per_remote_writer =

6.

((2*piggybackEvery - 1) + dataRateInHz * delayInMs / 1000);

7.
8.

//use these hard coded value until you use key

9.

qos->resource_limits.max_samples_per_instance =

10.
11.
12.

qos->resource_limits.max_samples;
qos->resource_limits.initial_instances =
qos->resource_limits.max_instances = 1;

13.
14.

qos->protocol.rtps_reliable_reader.min_heartbeat_response_delay.sec = 0;

15.
0;

qos->protocol.rtps_reliable_reader.min_heartbeat_response_delay.nanosec =

16.

qos->protocol.rtps_reliable_reader.max_heartbeat_response_delay.sec = 0;

17.
0;

qos->protocol.rtps_reliable_reader.max_heartbeat_response_delay.nanosec =

Line 1-Line 2 (Figure 10.15 QoS for a Periodic Reliable Reader above): Unlike a writer, the reader’s
default reliability setting is best-effort, so reliability must be turned on. Since we don’t want to drop anything, we choose KEEP_ALL for the History QoS.
Line 3-Line 6 (Figure 10.15 QoS for a Periodic Reliable Reader above) Unlike the writer, the reader
queue is sized not according to the jitter of the reader, but rather how many DDS samples you want to
cache speculatively in case of a gap in sequence of DDS samples that the reader must recover. Remember

667

10.4 Auto Throttling for DataWriter Performance—Experimental Feature

that a reader will stop giving a sequence of DDS samples as soon as an unintended gap appears, because
the definition of strict reliability includes in-order delivery. If the queue size were 1, the reader would have
no choice but to drop all subsequent DDS samples received until the one being sought is recovered. Connext DDS uses speculative caching, which minimizes the disruption caused by a few dropped DDS
samples. Even for the same duration of disruption, the demand on reader queue size is greater if the writer
will send more rapidly. In sizing the reader queue, we consider 2 factors that comprise the lost DDS
sample recovery time:
l

How long it takes a reader to request a resend to the writer.
The piggyback heartbeat tells a reader about the writer’s state. If only DDS samples between two
piggybacked DDS samples are dropped, the reader must cache piggybackEvery DDS samples
before asking the writer for resend. But if a piggybacked DDS sample is also lost, the reader will not
get around to asking the writer until the next piggybacked DDS sample is received. Note that in this
worst case calculation, we are ignoring stand-alone heartbeats (i.e., not piggybacked heartbeat from
the writer). Of course, the reader may drop any number of heartbeats, including the stand-alone
heartbeat; in this sense, there is no such thing as the absolute worst case—just reasonable worst case,
where the probability of consecutive drops is acceptably low. For the majority of applications, even
two consecutive drops is unlikely, in which case we need to cache at most (2*piggybackEvery - 1)
DDS samples before the reader will ask the writer to resend, assuming no delay (Line 14-Line 17,
Figure 10.15 QoS for a Periodic Reliable Reader on the previous page).

l

How long it takes for the writer to respond to the request.
Even ignoring the flight time of the resend request through the transport, the writer takes a finite time
to respond to the repair request--mostly if the writer delays reply for multicast readers. In case of
immediate response, the processing time on the writer end, as well as the flight time of the messages
to and from the writer do not matter unless very larger data rate; that is, it is the product term that matters. In case the delay for multicast is random (that is, the minimum and the maximum delay are not
equal), one would have to use the maximum delay to be conservative.

Line 8-Line 12 (Figure 10.15 QoS for a Periodic Reliable Reader on the previous page): Since we are not
using keys, there is just one instance.
Line 14-Line 17 (Figure 10.15 QoS for a Periodic Reliable Reader on the previous page): If we are not
using multicast, or the number of readers being fed by the writer, there is no reason to delay.

10.4 Auto Throttling for DataWriter Performance—Experimental Feature
Auto Throttling is an experimental feature that allows you to configure a DataWriter to automatically
adjust its writing rate and send window size to provide the best latency/throughput tradeoff as system conditions change.

668

10.4 Auto Throttling for DataWriter Performance—Experimental Feature

When DataWriters and DataReaders are configured to be reliable, lost DDS samples are repaired automatically by Connext DDS. However, the repair path consumes bandwidth and increases latency. A high
number of lost DDS samples can reduce the throughput and increase the communication latency. With
Auto Throttling, the number of repair (lost) DDS samples is reduced by using feedback provided by
DataReaders in terms of ACK and NACK messages to adjust the DataWriter's write rate and send window size.
To configure Auto Throttling, use the following properties:
dds.domain_participant.auto_throttle.enable: Configures the DomainParticipant to gather internal
measurements (during DomainParticipant creation) that are required for the Auto Throttle feature. This
allows DataWriters belonging to this DomainParticipant to use the Auto Throttle feature. Default: false.
dds.data_writer.auto_throttle.enable: Enables automatic throttling in the DataWriter so it can automatically adjust the writing rate and the send window size; this minimizes the need for repair DDS samples
and improves latency. Default: false.
Note: This property takes effect only in DataWriters that belong to a DomainParticipant that has set the
property dds.domain_participant.auto_throttle.enable (described above) to true.
When Auto throttling is enabled, the size of the send window size is adjusted within the interval [min_
send_window_size, max_send_window_size] configured in DATA_WRITER_PROTOCOL
QosPolicy (DDS Extension) (Section 6.5.3 on page 347)

669

Chapter 11 Collaborative DataWriters
The Collaborative DataWriters feature allows you to have multiple DataWriters publishing DDS
samples from a common logical data source. The DataReaders will combine the DDS samples
coming from these DataWriters in order to reconstruct the correct order in which they were produced at the source. This combination process for the DataReaders can be configured using the
AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on page 337). It requires the middleware to provide a way to uniquely identify every DDS sample published in a DDS domain independently of the actual DataWriter that published the DDS sample.
In Connext DDS, every modification (DDS sample) to the global dataspace made by a DataWriter
within a DDS domain is identified by a pair (virtual GUID, sequence number).
The virtual GUID (Global Unique Identifier) is a 16-byte character identifier associated with the
logical data source. DataWriters can be assigned a virtual GUID using virtual_guid in the
DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).
The virtual sequence number is a 64-bit integer that identifies changes within the logical data
source.
Several DataWriters can be configured with the same virtual GUID. If each of these DataWriters
publishes a DDS sample with sequence number '0', the DDS sample will only be received once by
the DataReaders subscribing to the content published by the DataWriters (see Figure 11.1 Global
Dataspace Changes on the next page).

670

11.1 Collaborative DataWriters Use Cases

Figure 11.1 Global Dataspace Changes

11.1 Collaborative DataWriters Use Cases
l

Ordered delivery of DDS samples in high availability scenarios
One example of this is RTI Persistence Service1. When a late-joining DataReader configured with
DURABILITY QosPolicy (Section 6.5.7 on page 368) set to PERSISTENT or TRANSIENT
joins a DDS domain, it will start receiving DDS samples from multiple DataWriters. For example, if
the original DataWriter is still alive, the newly created DataReader will receive DDS samples from
the original DataWriter and one or more RTI Persistence Service DataWriters (PRSTDataWriters).

l

Ordered delivery of DDS samples in load-balanced scenarios

l

Multiple instances of the same application can work together to process and deliver DDS samples.
When the DDS samples arrive through different data-paths out of order, the DataReader will be able
to reconstruct the order at the source. An example of this is when multiple instances of RTI Persistence Service are used to persist the data. Persisting data to a database on disk can impact performance. By dividing the workload (e.g., DDS samples larger than 10 are persisted by Persistence
Service 1, DDS samples smaller or equal to 10 are persisted by Persistence Service 2) across different instances of RTI Persistence Service using different databases the user can improve scalability
and performance.
Ordered delivery of DDS samples with Group Ordered Access
The Collaborative DataWriters feature can also be used to configure the DDS sample ordering process when the Subscriber is configured with PRESENTATION QosPolicy (Section 6.4.6 on page

1For more information on Persistence Service, see Part 6: RTI Persistence Service (Section on page 932).

671

11.2 DDS Sample Combination (Synchronization) Process in a DataReader

330) access_scope set to GROUP. In this case, the Subscriber must deliver in order the DDS
samples published by a group of DataWriters that belong to the same Publisher and have access_
scope set to GROUP.
Figure 11.2 Load-Balancing with Persistence Service

11.2 DDS Sample Combination (Synchronization) Process in a
DataReader
A DataReader will deliver a DDS sample (VGUIDn, VSNm) to the application only when if one of the
following conditions is satisfied:
l
l

l

(VGUIDn, VSNm-1) has already been delivered to the application.
All the known DataWriters publishing VGUIDn have announced that they do not have (VGUIDn,
VSNm-1).
None of the known DataWriters publishing VGUIDn have announced potential availability of
(VGUIDn, VSNm-1) and a configurable timeout (max_data_availability_waiting_time) expires.

For additional details on how the reconstruction process works see the AVAILABILITY QosPolicy
(DDS Extension) (Section 6.5.1 on page 337).

672

11.3 Configuring Collaborative DataWriters

11.3 Configuring Collaborative DataWriters
11.3.1 Assocating Virtual GUIDs with DDS Data Samples
There are two ways to associate a virtual GUID with the DDS samples published by a DataWriter.
l

l

Per DataWriter: Using virtual_guid in DATA_WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).
Per DDS Sample: By setting the writer_guid in the identity field of the WriteParams_t structure
provided to the write_w_params operation (see Writing Data (Section 6.3.8 on page 283)). Since
the writer_guid can be set per DDS sample, the same DataWriter can potentially write DDS
samples from independent logical data sources. One example of this is RTI Persistence Service
where a single persistence service DataWriter can write DDS samples on behalf of multiple original
DataWriters.

11.3.2 Assocating Virtual Sequence Numbers with DDS Data Samples
You can associate a virtual sequence number with a DDS sample published by a DataWriter by setting the
sequence_number in the identity field of the WriteParams_t structure provided to the write_w_params
operation (see Writing Data (Section 6.3.8 on page 283)). Virtual sequence numbers for a given virtual
GUID must be strictly monotonically increasing. If you try to write a DDS sample with a sequence number less than or equal to the last sequence number, the write operation will fail.

11.3.3 Specifying which DataWriters will Deliver DDS Samples to the
DataReader from a Logical Data Source
The required_matched_endpoint_groups field in the AVAILABILITY QosPolicy (DDS Extension)
(Section 6.5.1 on page 337) can be used to specify the set of DataWriter groups that are expected to
provide DDS samples for the same data source (virtual GUID). The quorum count in a group represents
the number of DataWriters that must be discovered for that group before the DataReader is allowed to
provide non-consecutive DDS samples to the application.
A DataWriter becomes a member of an endpoint group by configuring the role_name in ENTITY_
NAME QosPolicy (DDS Extension) (Section 6.5.9 on page 374).

11.3.4 Specifying How Long to Wait for a Missing DDS Sample
A DataReader’s AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on page 337) specifies
how long to wait for a missing DDS sample. For example, this is important when the first DDS sample is
received: how long do you wait to determine the lowest sequence number available in the system?
l

The max_data_availability_waiting_time defines how much time to wait before delivering a DDS
sample to the application without having received some of the previous DDS samples.

673

11.4 Collaborative DataWriters and Persistence Service

l

The max_endpoint_availability_waiting_time defines how much time to wait to discover
DataWriters providing DDS samples for the same data source (virtual GUID).

11.4 Collaborative DataWriters and Persistence Service
The DataWriters created by persistence service are automatically configured to do collaboration:
l
l

Every DDS sample published by the Persistence Service DataWriter keeps its original identity.
Persistence Service associates the role name PERSISTENCE_SERVICE with all the DataWriters
that it creates. You can overwrite that setting by changing the DataWriter QoS configuration in persistence service.

For more information, see Part 6: RTI Persistence Service (Section on page 932).

674

Chapter 12 Mechanisms for Achieving
Information Durability and
Persistence
12.1 Introduction
Connext DDS offers the following mechanisms for achieving durability and persistence:
l

l

l

Durable Writer History This feature allows a DataWriter to persist its historical cache, perhaps locally, so that it can survive shutdowns, crashes and restarts. When an application
restarts, each DataWriter that has been configured to have durable writer history automatically load all of the data in this cache from disk and can carry on sending data as if it had
never stopped executing. To the rest of the system, it will appear as if the DataWriter had
been temporarily disconnected from the network and then reappeared.
Durable Reader State This feature allows a DataReader to persist its state and remember
which data it has already received. When an application restarts, each DataReader that has
been configured to have durable reader state automatically loads its state from disk and can
carry on receiving data as if it had never stopped executing. Data that had already been
received by the DataReader before the restart will be suppressed so that it is not even sent
over the network.
Data Durability This feature is a full implementation of the OMG DDS Persistence Profile.
The DURABILITY QosPolicy (Section 6.5.7 on page 368) allows an application to configure a DataWriter so that the information written by the DataWriter survives beyond the
lifetime of the DataWriter. In this manner, a late-joining DataReader can subscribe to and
receive the information even after the DataWriter application is no longer executing. To use
this feature, you need Persistence Service, a separate application described in Introduction to
RTI Persistence Service (Section Chapter 26 on page 933).

675

12.1.1 Scenario 1. DataReader Joins after DataWriter Restarts (Durable Writer History)

These features can be configured separately or in combination. To use Durable Writer State and Durable
Reader State, you need a relational database, which is not included with Connext DDS. Supported databases are listed in the Release Notes. Persistence Service does not require a database when used in
TRANSIENT mode (see RTI Persistence Service (Section 12.5.1 on page 692)) or in PERSISTENT
mode with file-system storage (see RTI Persistence Service (Section 12.5.1 on page 692) and Configuring
Remote Administration (Section 27.5 on page 942)).
To understand how these features interact we will examine the behavior of the system using the following
scenarios:
l

l

l

Scenario 1. DataReader Joins after DataWriter Restarts (Durable Writer History) (Section 12.1.1
below)
Scenario 2: DataReader Restarts While DataWriter Stays Up (Durable Reader State) (Section 12.1.2
on the facing page)
Scenario 3. DataReader Joins after DataWriter Leaves Domain (Durable Data) (Section 12.1.3 on
page 679)

12.1.1 Scenario 1. DataReader Joins after DataWriter Restarts (Durable
Writer History)
In this scenario, a DomainParticipant joins the domain, creates a DataWriter and writes some data, then
the DataWriter shuts down (gracefully or due to a fault). The DataWriter restarts and a DataReader joins
the domain. Depending on whether the DataWriter is configured with durable history, the late-joining
DataReader may or may not receive the data published already by the DataWriter before it restarted. This
is illustrated in Figure 12.1 Durable Writer History on the facing page. For more information, see Durable
Writer History (Section 12.3 on page 681)

676

12.1.2 Scenario 2: DataReader Restarts While DataWriter Stays Up (Durable Reader State)

Figure 12.1 Durable Writer History

12.1.2 Scenario 2: DataReader Restarts While DataWriter Stays Up (Durable
Reader State)
In this scenario, two DomainParticipants join a domain; one creates a DataWriter and the other a
DataReader on the same Topic. The DataWriter publishes some data ("a" and "b") that is received by the

677

12.1.2 Scenario 2: DataReader Restarts While DataWriter Stays Up (Durable Reader State)

DataReader. After this, the DataReader shuts down (gracefully or due to a fault) and then restarts—all
while the DataWriter remains present in the domain.
Depending on whether the DataReader is configured with Durable Reader State, the DataReader may or
may not receive a duplicate copy of the data it received before it restarted. This is illustrated in Figure 12.2
Durable Reader State below. For more information, see Durable Reader State (Section 12.4 on page 686).
Figure 12.2 Durable Reader State

678

12.1.3 Scenario 3. DataReader Joins after DataWriter Leaves Domain (Durable Data)

12.1.3 Scenario 3. DataReader Joins after DataWriter Leaves Domain
(Durable Data)
In this scenario, a DomainParticipant joins a domain, creates a DataWriter, publishes some data on a
Topic and then shuts down (gracefully or due to a fault). Later, a DataReader joins the domain and subscribes to the data. Persistence Service is running.
Depending on whether Durable Data is enabled for the Topic, the DataReader may or may not receive the
data previous published by the DataWriter. This is illustrated in Figure 12.3 Durable Data below. For
more information, see Data Durability (Section 12.5 on page 692)
Figure 12.3 Durable Data

679

12.2 Durability and Persistence Based on Virtual GUIDs

This third scenario is similar to Scenario 1. DataReader Joins after DataWriter Restarts (Durable Writer
History) (Section 12.1.1 on page 676) except that in this case the DataWriter does not need to restart for
the DataReader to get the data previously written by the DataWriter. This is because Persistence Service
acts as an intermediary that stores the data so it can be given to late-joining DataReaders.

12.2 Durability and Persistence Based on Virtual GUIDs
Every modification to the global dataspace made by a DataWriter is identified by a pair (virtual GUID,
sequence number).
l

l

The virtual GUID (Global Unique Identifier) is a 16-byte character identifier associated with a
DataWriter or DataReader; it is used to uniquely identify this entity in the global data space.
The sequence number is a 64-bit identifier that identifies changes published by a specific
DataWriter.

Several DataWriters can be configured with the same virtual GUID. If each of these DataWriters publishes a sample with sequence number '0', the sample will only be received once by the DataReaders subscribing to the content published by the DataWriters (see Figure 12.4 Global Dataspace Changes below).
Figure 12.4 Global Dataspace Changes

Additionally, Connext DDS uses the virtual GUID to associate a persisted state (state in permanent storage) to the corresponding Entity.
For example, the history of a DataWriter will be persisted in a database table with a name generated from
the virtual GUID of the DataWriter. If the DataWriter is restarted, it must have associated the same virtual
GUID to restore its previous history.

680

12.3 Durable Writer History

Likewise, the state of a DataReader will be persisted in a database table whose name is generated from the
DataReader virtual GUID (see Figure 12.5 History/State Persistence Based on Virtual GUID below).
Figure 12.5 History/State Persistence Based on Virtual GUID

l

l

A DataWriter’s virtual GUID can be configured using the member virtual_guid in the DATA_
WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347).
A DataReader’s virtual GUID can be configured using the member virtual_guid in the DATA_
READER_PROTOCOL QosPolicy (DDS Extension) (Section 7.6.1 on page 511).

The DDS_PublicationBuiltinTopicData and DDS_SubscriptionBuiltinTopicData structures include the virtual GUID associated with the discovered publication or subscription (see Built-in DataReaders (Section
16.2 on page 773)).

12.3 Durable Writer History
The DURABILITY QosPolicy (Section 6.5.7 on page 368) controls whether or not, and how, published
samples are stored by the DataWriter application for DataReaders that are found after the samples were initially written. The samples stored by the DataWriter constitute the DataWriter’s history.
Connext DDS provides the capability to make the DataWriter history durable, by persisting its content in a
relational database. This makes it possible for the history to be restored when the DataWriter restarts. See
the RTI Connext DDS Core Libraries Release Notes for the list of supported relational databases.

681

12.3.1 Durable Writer History Use Case

The association between the history stored in the database and the DataWriter is done using the virtual
GUID.

12.3.1 Durable Writer History Use Case
The following use case describes the durable writer history functionality:
1. A DataReader receives two samples with sequence number 1 and 2 published by a DataWriter with
virtual GUID 1.

2. The process running the DataWriter is stopped and a new late-joining DataReader is created.
The new DataReader with virtual GUID 2 does not receive samples 1 and 2 because the original
DataWriter has been destroyed. If the samples must be available to late-joining DataReaders after
the DataWriter deletion, you can use Persistence Service, described in Introduction to RTI Persistence Service (Section Chapter 26 on page 933).

3. The DataWriter is restarted using the same virtual GUID.
After being restarted, the DataWriter restores its history. The late-joining DataReader will receive
samples 1 and 2 because they were not received previously. The DataReader with virtual GUID 1
will not receive samples 1 and 2 because it already received them

682

12.3.2 How To Configure Durable Writer History

4. The DataWriter publishes two new samples.

The two new samples with sequence numbers 3 and 4 will be received by both DataReaders.

12.3.2 How To Configure Durable Writer History
Connext DDS allows a DataWriter’s history to be stored in a relational database that provides an ODBC
driver.
For each DataWriter history that is configured to be durable, Connext DDS will create a maximum of two
tables:
l

l

The first table is used to store the samples associated with the writer history. The name of that table
is WS<32 uuencoding of the writer virtual GUID>.
The second table is only created for keyed-topic and it is used to store the instances associated with
the writer history. The name of the second table is WI<32 uuencoding of the writer virtual GUID>.

To configure durable writer history, use the PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on
page 394) associated with DataWriters and DomainParticipants.
A ‘durable writer history’ property defined in the DomainParticipant will be applicable to all the
DataWriters belonging to the DomainParticipant unless it is overwritten by the DataWriter. Table 12.1
Durable Writer History Properties lists the supported ‘durable writer history’ properties.

683

12.3.2 How To Configure Durable Writer History

Table 12.1 Durable Writer History Properties
Property

Description

Required.
dds.data_
writer.history.plugin_ Must be set to "dds.data_writer.history.odbc_plugin.builtin" to enable durable writer history in the
name
DataWriter.
dds.data_
writer.history.odbc_
plugin.
dsn
dds.data_
writer.history.odbc_
plugin.
driver
dds.data_
writer.history.odbc_
plugin.
username

Required.
The ODBC DSN (Data Source Name) associated with the database where the writer history must be persisted.

Tells Connext DDS which ODBC driver to load. If the property is not specified, Connext DDS will try to use the
standard ODBC driver manager library (UnixOdbc on UNIX/Linux systems, the Windows ODBC driver manager
on Windows systems).

Configures the username/password used to connect to the database.

dds.data_
writer.history.odbc_
plugin.
password

Default: No password or username

dds.data_
writer.history.odbc_
plugin.
shared

When set to 1, Connext DDS will create a single connection per DSN that will be shared across DataWriters within
the same Publisher.
A DataWriter can be configured to create its own database connection by setting this property to 0 (the default).

684

12.3.2 How To Configure Durable Writer History

Table 12.1 Durable Writer History Properties
Property

Description

dds.data_
writer.history.odbc_
These properties configure the resource limits associated with the ODBC writer history caches.
plugin.
instance_cache_max_ To minimize the number of accesses to the database, Connext DDS uses two caches, one for samples and one for
size
instances. The initial size and the maximum size of these caches are configured using these properties.
dds.data_
writer.history.odbc_
plugin.
instance_cache_init_
size

The resource limits, initial_instances, max_instances, initial_samples, max_samples, and max_samples_per_
instance defined in RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405) are used to configure the
maximum number of samples and instances that can be stored in the relational database.

dds.data_
writer.history.odbc_
plugin.
sample_cache_max_
size

instance_cache_init_size: initial_instances in RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)

dds.data_
writer.history.odbc_
plugin.
sample_cache_init_
size

Defaults:
instance_cache_max_size: max_instances in RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)

sample_cache_max_size: 32
sample_cache_init_size: 32

If in_memory_state (see below in this table) is 1, instance_cache_max_size is always equal
to max_instances in RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)—it
cannot be changed.

This property indicates whether or not the persisted writer history must be restored once the DataWriter is restarted.
dds.data_
writer.history.odbc_
plugin.
restore

If this property is 0, the content of the database associated with the DataWriter being restarted will be deleted.
If it is 1, the DataWriter will restore its previous state from the database content.
Default: 1
This property determines how much state will be kept in memory by the ODBC writer history in order to avoid
accessing the database.

dds.data_
writer.history.odbc_
plugin.
in_memory_state

If this property is 1, then the property instance_cache_max_size (see above in this table) is always equal to max_
instances in RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405)—it cannot be changed. In addition,
the ODBC writer history will keep in memory a fixed state overhead of 24 bytes per sample. This mode provides
the best ODBC writer history performance. However, the restore operation will be slower and the maximum
number of samples that the writer history can manage is limited by the available physical memory.
If it is 0, all the state will be kept in the underlying database. In this mode, the maximum number of samples in the
writer history is not limited by the physical memory available.
Default: 1

Durable Writer History is not supported for Multi-channel DataWriters (see Multi-channel
DataWriters (Section Chapter 18 on page 824)) or when Batching is enabled (see BATCH
QosPolicy (DDS Extension) (Section 6.5.2 on page 341)); an error is reported if this type of
DataWriter tries to configure Durable Writer History.

685

12.4 Durable Reader State

See also: Durable Reader State (Section 12.4 below).
Example C++ Code
/* Get default QoS */
...
retcode = DDSPropertyQosPolicyHelper::add_property (writerQos.property,
"dds.data_writer.history.plugin_name",
"dds.data_writer.history.odbc_plugin.builtin",
DDS_BOOLEAN_FALSE);
if (retcode != DDS_RETCODE_OK) {
/* Report error */
}
retcode = DDSPropertyQosPolicyHelper::add_property (writerQos.property,
"dds.data_writer.history.odbc_plugin.dsn",
"",
DDS_BOOLEAN_FALSE);
if (retcode != DDS_RETCODE_OK) {
/* Report error */
}
retcode = DDSPropertyQosPolicyHelper::add_property (writerQos.property,
"dds.data_writer.history.odbc_plugin.driver",
"",
DDS_BOOLEAN_FALSE);
if (retcode != DDS_RETCODE_OK) {
/* Report error */
}
retcode = DDSPropertyQosPolicyHelper::add_property (writerQos.property,
"dds.data_writer.history.odbc_plugin.shared",
"<0|1>",
DDS_BOOLEAN_FALSE);
if (retcode != DDS_RETCODE_OK) {
/* Report error */
}
/* Create Data Writer */
...

12.4 Durable Reader State
Durable reader state allows a DataReader to locally store its state in disk and remember the data that has
already been processed by the application1. When an application restarts, each DataReader configured to
have durable reader state automatically reads its state from disk. Data that has already been processed by
the application before the restart will not be provided to the application again.
Important: The DataReader does not persist the full contents of the data in its historical cache; it only persists an identification (e.g. sequence numbers) of the data the application has processed. This distinction is
not meaningful if your application always uses the ‘take’ methods to access your data, since these methods
remove the data from the cache at the same time they deliver it to your application. (See Read vs. Take

1The circumstances under which a data sample is considered “processed by the application” are described in the sections

that follow.

686

12.4.1 Durable Reader State With Protocol Acknowledgment

(Section 7.4.3.1 on page 494)) However, if your application uses the ‘read’ methods, leaving the data in
the DataReader's cache after you've accessed it for the first time, those previously viewed samples will not
be restored to the DataReader's cache in the event of a restart.
Connext DDS requires a relational database to persist the state of a DataReader. This database is accessed
using ODBC. See the RTI Connext DDS Core Libraries Release Notes for the list of supported relational
databases.

12.4.1 Durable Reader State With Protocol Acknowledgment
For each DataReader configured to have durable state, Connext DDS will create one database table with
the following naming convention: RS<32 uuencoding of the reader virtual GUID>. This table will
store the last sequence number processed from each virtual GUID. For DataReaders on keyed topics
requesting instance-ordering (see PRESENTATION QosPolicy (Section 6.4.6 on page 330)), this state
will be stored per instance per virtual GUID..
Criteria to consider a sample “processed by the application”
l

For the read/take methods that require calling return_loan(), a sample 's1' with sequence number
's1_seq_num' and virtual GUID ‘vg1’ is considered processed by the application when the
DataReader’s return_loan() operation is called for sample 's1' or any other sample with the same
virtual GUID and a sequence number greater than 's1_seq_num'. For example:
retcode = Foo_reader->take(data_seq, info_seq,
DDS_LENGTH_UNLIMITED, DDS_ANY_SAMPLE_STATE,
DDS_ANY_VIEW_STATE, DDS_ANY_INSTANCE_STATE);
if (retcode == DDS_RETCODE_NO_DATA) {
return;
} else if (retcode != DDS_RETCODE_OK) {
/* report error */
return;
}
for (i = 0; i < data_seq.length(); ++i) {
/* Operate with the data */
}
/* Return the loan */
retcode = Foo_reader->return_loan(data_seq, info_seq);
if (retcode != DDS_RETCODE_OK) {
/* Report and error */
}
/* At this point the samples contained in data_seq
will be considered as received. If the DataReader
restarts, the samples will not be received again */

l

For the read/take methods that do not require calling return_loan(), a sample 's1' with sequence
number 's1_seq_num' and virtual GUID ‘vg1’ will be considered processed after the application

687

12.4.1.1 Bandwidth Utilization

reads or takes the sample 's1' or any other sample with the same virtual GUID and with a sequence
number greater than 's1_seq_num'. For example:
retcode = Foo_reader->take_next_sample(data,info);
/* At this point the sample contained in data will be
considered as received. All the samples with a sequence
number smaller than the sequence number associated with
data will also be considered as received.
If the DataReader restarts, these sample will not
be received again */

If you access the samples in the DataReader cache out of order—for example via QueryCondition,
specifying an instance state, or reading by instance when the PRESENTATION QoS is not set to
INSTANCE_PRESENTATION_QOS—then the samples that have not yet been taken or read by
the application may still be considered as ”processed by the application”.
12.4.1.1 Bandwidth Utilization
To optimize network usage, if a DataReader configured with durable reader state is restarted and it discovers a DataWriter with a virtual GUID ‘vg’, the DataReader will ACK all the samples with a sequence
number smaller than ‘sn’, where ‘sn’ is the first sequence number that has not been being processed by the
application for ‘vg’.
Notice that the previous algorithm can significantly reduce the number of duplicates on the wire. However,
it does not suppress them completely in the case of keyed DataReaders where the durable state is kept per
(instance, virtual GUID). In this case, and assuming that the application has read samples out of order
(e.g., by reading different instances), the ACK is sent for the lowest sequence number processed across all
instances and may cause samples already processed to flow on the network again. These redundant
samples waste bandwidth, but they will be dropped by the DataReader and not be delivered to the application.

12.4.2 Durable Reader State with Application Acknowledgment
This section assumes you are familiar with the concept of Application Acknowledgment as described in
Application Acknowledgment (Section 6.3.12 on page 288).
For each DataReader configured to be durable and that uses application acknowledgement (see Application Acknowledgment (Section 6.3.12 on page 288)), Connext DDS will create one database table with
the following naming convention: RS<32 uuencoding of the reader virtual GUID>. This table will
store the list of sequence number intervals that have been acknowledged for each virtual GUID. The size
of the column that stores the sequence number intervals is limited to 32767 bytes. If this size is exceeded
for a given virtual GUID, the operation that persists the DataReader state into the database will fail.

688

12.4.2.1 Bandwidth Utilization

12.4.2.1 Bandwidth Utilization
To optimize network usage, if a DataReader configured with durable reader state is restarted and it discovers a DataWriter with a virtual GUID ‘vg’, the DataReader will send an APP_ACK message with all
the samples that were auto-acknowledged or explicitly acknowledged in previous executions.
Notice that this algorithm can significantly reduce the number of duplicates on the wire. However, it does
not suppress them completely since the DataReader may send a NACK and receive some samples from
the DataWriter before the DataWriter receives the APP_ACK message.

12.4.3 Durable Reader State Use Case
The following use case describes the durable reader state functionality:
1. A DataReader receives two samples with sequence number 1 and 2 published by a DataWriter with
virtual GUID 1. The application takes those samples.

2. After the application returns the loan on samples 1 and 2, the DataReader considers them as processed and it persists the state change.

3. The process running the DataReader is stopped.

689

12.4.4 How To Configure a DataReader for Durable Reader State

4. The DataReader is restarted.
Because all the samples with sequence number smaller or equal than 2 were considered received,
the reader will not ask for these samples from the DataWriter.

12.4.4 How To Configure a DataReader for Durable Reader State
To configure a DataReader with durable reader state, use the PROPERTY QosPolicy (DDS Extension)
(Section 6.5.17 on page 394) associated with DataReaders and DomainParticipants.
A property defined in the DomainParticipant will be applicable to all the DataReaders contained in the
participant unless it is overwritten by the DataReaders. Table 12.2 Durable Reader State Properties lists
the supported properties.
Table 12.2 Durable Reader State Properties
Property

Description
Required.

dds.data_
reader.state.odbc.dsn

The ODBC DSN (Data Source Name) associated with the database where the DataReader state must be
persisted.
To enable durable reader state, this property must be set to 1.

dds.data_reader.state.
filter_redundant_samples

When set to 0, the reader state is not maintained and Connext DDS does not filter duplicate samples that may
be coming from the same virtual writer.
Default: 1

dds.data_
reader.state.odbc.driver

This property indicates which ODBC driver to load. If the property is not specified, Connext DDS will try to
use the standard ODBC driver manager library (UnixOdbc on UNIX/Linux systems, the Windows ODBC
driver manager on Windows systems).

690

12.4.4 How To Configure a DataReader for Durable Reader State

Table 12.2 Durable Reader State Properties
Property

Description

dds.data_
reader.state.odbc.username These two properties configure the username and password used to connect to the database.
Default: No password or username
dds.data_
reader.state.odbc.password
This property indicates if the persisted DataReader state must be restored or not once the DataReader is
restarted.
dds.data_
reader.state.restore

If this property is 0, the previous state will be deleted from the database. If it is 1, the DataReader will restore
its previous state from the database content.
Default: 1
This property controls how often the reader state is stored into the database. A value of N means store the state
once every N samples.

dds.data_reader.state.
checkpoint_frequency

A high frequency will provide better performance. However, if the reader is restarted it may receive some
duplicate samples. These samples will be filtered by Connext DDS and they will not be propagated to the
application.
Default: 1

dds.data_
reader.state.persistence_
service.request_depth

This property indicates how many of the most recent historical samples the persisted DataReader wants to
receive upon start-up.
Default: 0

Example (C++ code):
/* Get default QoS */
...
retcode = DDSPropertyQosPolicyHelper::add_property(
readerQos.property,
"dds.data_reader.state.odbc.dsn",
"", DDS_BOOLEAN_FALSE);
if (retcode != DDS_RETCODE_OK) {
/* Report error */
}
retcode = DDSPropertyQosPolicyHelper::add_property(readerQos.property,
"dds.data_reader.state.odbc.driver",
"", DDS_BOOLEAN_FALSE);
if (retcode != DDS_RETCODE_OK) {
/* Report error */
}
retcode = DDSPropertyQosPolicyHelper::add_property(readerQos.property,
"dds.data_reader.state.restore", "<0|1>",
DDS_BOOLEAN_FALSE);
if (retcode != DDS_RETCODE_OK) {
/* Report error */

691

12.5 Data Durability

}
/* Create Data Reader */
...

12.5 Data Durability
The data durability feature is an implementation of the OMG DDS Persistence Profile. The
DURABILITY QosPolicy (Section 6.5.7 on page 368) allows an application to configure a DataWriter
so that the information written by the DataWriter survives beyond the lifetime of the DataWriter.
Connext DDS implements TRANSIENT and PERSISTENT durability using an external service called
Persistence Service, available for purchase as a separate RTI product.
Persistence Service receives information from DataWriters configured with TRANSIENT or
PERSISTENT durability and makes that information available to late-joining DataReaders—even if the
original DataWriter is not running.
The samples published by a DataWriter can be made durable by setting the kind field of the
DURABILITY QosPolicy (Section 6.5.7 on page 368) to one of the following values:
l

l

DDS_TRANSIENT_DURABILITY_QOS:Connext DDS will store previously published samples
in memory using Persistence Service, which will send the stored data to newly discovered
DataReaders.
DDS_PERSISTENT_DURABILITY_QOS: Connext DDS will store previously published
samples in permanent storage, like a disk, using Persistence Service, which will send the stored data
to newly discovered DataReaders.

A DataReader can request TRANSIENT or PERSISTENT data by setting the kind field of the corresponding DURABILITY QosPolicy (Section 6.5.7 on page 368). A DataReader requesting
PERSISTENT data will not receive data from DataWriters or Persistence Service applications that are configured with TRANSIENT durability.

12.5.1 RTI Persistence Service
Persistence Service is a Connext DDS application that is configured to persist topic data. Persistence Service is included with the Connext DDS Professional, Evaluation, and Basic package types. For each one
of the topics that must be persisted for a specific domain, the service will create a DataWriter (known as
PRSTDataWriter) and a DataReader (known as PRSTDataReader). The samples received by the
PRSTDataReaders will be published by the corresponding PRSTDataWriters to be available for late-joining DataReaders.
For more information on Persistence Service, please see:

692

12.5.1 RTI Persistence Service

l

Introduction to RTI Persistence Service (Section Chapter 26 on page 933)

l

Configuring Persistence Service (Section Chapter 27 on page 934)

l

Running RTI Persistence Service (Section Chapter 28 on page 962)

Persistence Service can be configured to operate in PERSISTENT or TRANSIENT mode:
l

l

TRANSIENT mode The PRSTDataReaders and PRSTDataWriters will be created with
TRANSIENT durability and Persistence Service will keep the received samples in memory.
Samples published by a TRANSIENT DataWriter will survive the DataWriter lifecycle but will not
survive the lifecycle of Persistence Service (unless you are running multiple copies).
PERSISTENT mode The PRSTDataWriters and PRSTDataReaders will be created with
PERSISTENT durability and Persistence Service will store the received samples in files or in an
external relational database. Samples published by a PERSISTENT DataWriter will survive the
DataWriter lifecycle as well as any restarts of Persistence Service.

Peer-to-Peer Communication:
By default, a PERSISTENT/TRANSIENT DataReader will receive samples directly from the original
DataWriter if it is still alive. In this scenario, the DataReader may also receive the same samples from Persistence Service. Duplicates will be discarded at the middleware level. This Peer-To-Peer communication
pattern is illustrated in Figure 12.6 Peer-to-Peer Communication below. To use this peer-to-peer communication pattern, set the direct_communication field in the DURABILITY QosPolicy (Section 6.5.7
on page 368) to TRUE. A PERSISTENT/TRANSIENT DataReader will receive information directly
from PERSISTENT/TRANSIENT DataWriters.
Figure 12.6 Peer-to-Peer Communication

693

12.5.1 RTI Persistence Service

Relay Communication
A PERSISTENT/TRANSIENT DataReader may also be configured to not receive samples from the original DataWriter. In this case the traffic is relayed by Persistence Service. This ‘relay communication’ pattern is illustrated in Figure 12.7 Relay Communication below. To use relay communication, set the direct_
communication field in the DURABILITY QosPolicy (Section 6.5.7 on page 368) to FALSE. A
PERSISTENT/TRANSIENT DataReader will receive all the information from Persistence Service.
Figure 12.7 Relay Communication

694

Chapter 13 Guaranteed Delivery of Data
13.1 Introduction
Some application scenarios need to ensure that the information produced by certain producers is
delivered to all the intended consumers. This chapter describes the mechanisms available in Connext DDS to guarantee the delivery of information from producers to consumers such that the delivery is robust to many kinds of failures in the infrastructure, deployment, and even the
producing/consuming applications themselves.
Guaranteed information delivery is not the same as protocol-level reliability (described in Reliable
Communications (Section Chapter 10 on page 629)) or information durability (described in Mechanisms for Achieving Information Durability and Persistence (Section Chapter 12 on page 675)).
Guaranteed information delivery is an end-to-end application-level QoS, whereas the others are
middleware-level QoS. There are significant differences between these two:
l

l

With protocol-level reliability alone, the producing application knows that the information is
received by the protocol layer on the consuming side. However the producing application
cannot be certain that the consuming application read that information or was able to successfully understand and process it. The information could arrive in the consumer’s protocol
stack and be placed in the DataReader cache but the consuming application could either
crash before it reads it from the cache, not read its cache, or read the cache using queries or
conditions that prevent that particular DDS data sample from being accessed. Furthermore,
the consuming application could access the DDS sample, but not be able to interpret its
meaning or process it in the intended way.
With information durability alone, there is no way to specify or characterize the intended consumers of the information. Therefore the infrastructure has no way to know when the information has been consumed by all the intended recipients. The information may be persisted
such that it is not lost and is available to future applications, but the infrastructure and producing applications have no way to know that all the intended consumers have joined the
system, received the information, and processed it successfully.

695

13.1 Introduction

The guaranteed data-delivery mechanism provided in Connext DDS overcomes the limitations described
above by providing the following features:
l

l

l

Required subscriptions. This feature provides a way to configure, identify and detect the applications that are intended to consume the information. See Required Subscriptions (Section 6.3.13 on
page 294).
Application-level acknowledgments. This feature provides the means ensure that the information
was successfully processed by the application-layer in a consumer application. See Application
Acknowledgment (Section 6.3.12 on page 288).
Durable subscriptions. This feature leverages the RTI Persistence Service to persist DDS DDS
samples intended for the required subscriptions such that they are delivered even if the originating
application is not available. See Configuring Durable Subscriptions in Persistence Service (Section
27.9 on page 955).

These features used in combination with the mechanisms provided for Information Durability and Persistence (see Mechanisms for Achieving Information Durability and Persistence (Section Chapter 12 on
page 675)) enable the creation of applications where the information delivery is guaranteed despite application and infrastructure failures. Scenarios (Section 13.2 on page 700) describes various guaranteed-delivery scenarios and how to configure the applications to achieve them.
When implementing an application that needs guaranteed data delivery, we have to consider three key
aspects:
Related Features and QoS

Key Aspects to Consider

Required subscriptions
Durable subscriptions
Identifying the required consumers of information
EntityName QoS policy
Availability QoS policy
Application-level acknowledgment
Ensuring the intended consumer applications process the data
successfully

Acknowledgment by a quorum of required and durable
subscriptions
Reliability QoS policy (acknowledgment mode)
Availability QoS policy
Persistence Service
Durable Subscriptions

Ensuring information is available to late joining applications
Durability QoS
Durable Writer History

696

13.1.1 Identifying the Required Consumers of Information

13.1.1 Identifying the Required Consumers of Information
The first step towards ensuring that information is processed by the intended consumers is the ability to specify and recognize those intended consumers. This is done using the required subscriptions feature
(Required Subscriptions (Section 6.3.13 on page 294)) configured via the ENTITY_NAME QosPolicy
(DDS Extension) (Section 6.5.9 on page 374) and AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on page 337)).
Connext DDS DataReader entities (as well as DataWriter and DomainParticipant entities) can have a
name and a role_name. These names are configured using the ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9 on page 374), which is propagated via DDS discovery and is available as part of the
builtin-topic data for the Entity (see Built-In Topics (Section Chapter 16 on page 772)).
The DDS DomainParticipant, DataReader and DataWriter entities created by RTI-provided applications
and services, specifically services such as RTI Persistence Service, automatically configure the ENTITY_
NAME QoS policy according to their function. For example the DataReaders created by RTI Persistence
Service have their role_name set to “PERSISTENCE_SERVICE”.
Unless explicitly set by the user, the DomainParticipant, DataReader and DataWriter entities created by
end-user applications have their name and role_name set to NULL. However applications may modify
this using the ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9 on page 374).
Connext DDS uses the role_name of DataReaders to identify the consumer’s logical function. For this
reason Connext DDS’s required subscriptions feature relies on the role_name to identify intended consumers of information. The use of the DataReader’s role_name instead of the name is intentional. From
the point of view of the information producer, the important thing is not the concrete DataReader (identified by its name, for example, “Logger123”) but rather its logical function in the system (identified by its
role_name, for example “LoggingService”).
A DataWriter that needs to ensure its information is delivered to all the intended consumers uses the
AVAILABILITY QosPolicy (DDS Extension) (Section 6.5.1 on page 337) to configure the role names
of the consumers that must receive the information.
The AVAILABILITY QoS Policy set on a DataWriter lets an application configure the required consumers of the data produced by the DataWriter. The required consumers are specified in the required_
matched_endpoint_groups attribute within the AVAILABILITY QoS Policy. This attribute is a
sequence of DDS EndpointGroup structures. Each EndpointGroup represents a required information consumer characterized by the consumer’s role_name and quorum_count. The role_name identifies a
logical consumer; the quorum_count specifies the minimum number of consumers with that role_name
that must acknowledge the DDS sample before the DataWriter can consider it delivered to that required
consumer.
For example, an application that wants to ensure data written by a DataWriter is delivered to at least two
Logging Services and one Display Service would configure the DataWriter’s AVAILABILITY QoS
Policy with a required_matched_endpoint_groups consisting of two elements. The first element would
specify a required consumer with the role_name “LoggingService” and a quorum_count of 2. The

697

13.1.2 Ensuring Consumer Applications Process the Data Successfully

second element would specify a required consumer with the role_name “DisplayService” and a quorum_
count of 1. Furthermore, the application would set the logging service DataReader ENTITY_NAME
policy to have a role_name of “LoggingService” and similarly the display service DataReader ENTITY_
NAME policy to have the role_name of “DisplayService.”
A DataWriter that has been configured with an AVAILABILITY QoS policy will not remove DDS
samples from the DataWriter cache until they have been “delivered” to both the already-discovered
DataReaders and the minimum number (quorum_count) of DataReaders specified for each role. In particular, DDS samples will be retained by the DataWriter if the quorum_count of matched DataReaders
with a particular role_name have not been discovered yet.
We used the word “delivered” in quotes above because the level of assurance a DataWriter has that a particular DDS sample has been delivered depends on the setting of the RELIABILITY QosPolicy (Section
6.5.19 on page 400). We discuss this next in Ensuring Consumer Applications Process the Data Successfully (Section 13.1.2 below).

13.1.2 Ensuring Consumer Applications Process the Data Successfully
Identifying the Required Consumers of Information (Section 13.1.1 on the previous page) described mechanisms by which an application could configure who the required consumers of information are. This section is about the criteria, mechanisms, and assurance provided by Connext DDS to ensure consumers have
the information delivered to them and process it in a successful manner.
RTI provides four levels of information delivery guarantee. You can set your desired level using the
RELIABILITY QosPolicy (Section 6.5.19 on page 400). The levels are:
l

l

l

Best-effort, relying only on the underlying transportThe DataWriter considers the DDS sample
delivered/acknowledged as soon as it is given to the transport to send to the DataReader’s destination. Therefore, the only guarantee is the one provided by the underlying transport itself. Note
that even if the underlying transport is reliable (e.g., shared memory or TCP) the reliability is limited
to the transport-level buffers. There is no guarantee that the DDS sample will arrive to the
DataReader cache because after the transport delivers to the DataReader’s transport buffers, it is
possible for the DDS sample to be dropped because it exceeds a resource limit, fails to deserialize
properly, the receiving application crashes, etc.
Reliable with protocol acknowledgmentThe DDS-RTPS reliability protocol used by Connext
DDS provides acknowledgment at the RTPS protocol level: a DataReader will acknowledge it has
deserialized the DDS sample correctly and stored it in the DataReader’s cache. However, there is
no guarantee the application actually processed the DDS sample. The application might crash before
processing the DDS sample, or it might simply fail to read it from the cache.
Reliable with Application Acknowledgment (Auto)Application Acknowledgment in Auto mode
causes Connext DDS to send an additional application-level acknowledgment (above and beyond
the RTPS protocol level acknowledgment) after the consuming application has read the DDS

698

13.1.3 Ensuring Information is Available to Late-Joining Applications

sample from the DataReader cache and the application has subsequently called the DataReader’s
return_loan() operation (see Loaning and Returning Data and SampleInfo Sequences (Section
7.4.2 on page 492)) for that DDS sample. This mode guarantees that the application has fully read
the DDS sample all the way until it indicates it is done with it. However it does not provide a guarantee that the application was able to successfully interpret or process the DDS sample. For
example, the DDS sample could be a command to execute a certain action and the application may
read the DDS sample and not understand the command or may not be able to execute the action.
l

Reliable with Application Acknowledgment (Explicit)Application Acknowledgment in Explicit
mode causes Connext DDS to send an application-level acknowledgment only after the consuming
application has read the DDS sample from the DataReader cache and subsequently called the
DataReader’s acknowledge_sample() operation (see Acknowledging DDS Samples (Section 7.4.4
on page 502)) for that DDS sample. This mode guarantees that the application has fully read the
DDS sample and completed operating on it as indicated by explicitly calling acknowledge_sample
(). In contrast with the Auto mode described above, the application can delay the acknowledgment
of the DDS sample beyond the time it holds onto the data buffers, allowing it to be process in a
more flexible manner. Similar to the Auto mode, it does not provide a guarantee that the application
was able to successfully interpret or process the DDS sample. For example, the DDS sample could
be a command to execute a certain action and the application may read the DDS sample and not
understand the command or may not be able to execute the action. Applications that need guarantees
that the data was successfully processed and interpreted should use a request-reply interaction,
which is available as part of the Connext DDS Professional, Evaluation, and Basic package types
(see Part 4: Request-Reply Communication Pattern (Section on page 873)).

13.1.3 Ensuring Information is Available to Late-Joining Applications
The third aspect of guaranteed data delivery addresses situations where the application needs to ensure that
the information produced by a particular DataWriter is available to DataReaders that join the system after
the data was produced. The need for data delivery may even extend beyond the lifetime of the producing
application; that is, it may be required that the information is delivered to applications that join the system
after the producing application has left the system.
Connext DDS provides four mechanisms to handle these scenarios:
l

The DDS Durability QoS Policy. The DURABILITY QosPolicy (Section 6.5.7 on page 368) specifies whether DDS samples should be available to late joiners. The policy is set on the DataWriter
and the DataReader and supports four kinds: VOLATILE, TRANSIENT_LOCAL,
TRANSIENT, or PERSISTENT. If the DataWriter’s Durability QoS policy is set to VOLATILE
kind, the DataWriter’s DDS samples will not be made available to any late joiners. If the
DataWriter’s policy kind is set to TRANSIENT_LOCAL, TRANSIENT, or PERSISTENT, the
DDS samples will be made available for late-joining DataReaders who also set their
DURABILITY QoS policy kind to something other than VOLATILE.

699

13.2 Scenarios

l

l

l

Durable Writer History. A DataWriter configured with a DURABILITY QoS policy kind other
than VOLATILE keeps its data in a local cache so that it is available when the late-joining application appears. The data is maintained in the DataWriter’s cache until it is considered to be no longer
needed. The precise criteria depends on the configuration of additional QoS policies such as
LIFESPAN QoS Policy (Section 6.5.12 on page 381), HISTORY QosPolicy (Section 6.5.10 on
page 376), RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405), etc. For the purposes
of guaranteeing information delivery it is important to note that the DataWriter’s cache can be configured to be a memory cache or a durable (disk-based) cache. A memory cache will not survive an
application restart. However, a durable (disk-based) cache can survive the restart of the producing
application. The use a durable writer history, including the use of an external ODBC database as a
cache is described in Durable Writer History (Section 12.3 on page 681).
RTI Persistence Service. This service allows the information produced by a DataWriter to survive
beyond the lifetime of the producing application. Persistence Service is an stand-alone application
that runs on many supported platforms. This service complies with the Persistent Profile of the
OMG DDS specification. The service uses DDS to subscribe to the DataWriters that specify a
DURABILITY QosPolicy (Section 6.5.7 on page 368) kind of TRANSIENT or PERSISTENT.
Persistence Service receives the data from those DataWriters, stores the data in its internal caches,
and makes the data available via DataWriters (which are automatically created by Persistence Service) to late-joining DataReaders that specify a Durability kind of TRANSIENT or PERSISTENT.
Persistence Service can operate as a relay for the information from the original writer, preserving the
source_timestamp of the data, as well as the original DDS sample virtual writer GUID (see RTI
Persistence Service (Section 12.5.1 on page 692)). In addition, you can configure Persistence Service itself to use a memory-based cache or a durable (disk-based or database-based) cache. See Configuring Persistent Storage (Section 27.6 on page 943). Configuration of redundant and loadbalanced persistence services is also supported.
Durable Subscriptions. This is a Persistence Service configuration setting that allows configuration
of the required subscriptions (Identifying the Required Consumers of Information (Section 13.1.1
on page 697)) for the data stored by Persistence Service (Managing Data Instances (Working with
Keyed Data Types) (Section 6.3.14 on page 296)). Configuring required subscriptions for Persistence Service ensures that the service will store the DDS samples until they have been delivered
to the configured number (quorum_count) of DataReaders that have each of the specified roles.

13.2 Scenarios
In each of the scenarios below, we assume both the DataWriter and DataReader are configured for strict
reliability (RELIABLE ReliabilityQosPolicyKind and KEEP_ALL HistoryQosPolicyKind, see Controlling Queue Depth with the History QosPolicy (Section 10.3.3 on page 644)). As a result, when the
DataWriter’s cache is full of unacknowledged DDS samples, the write() operation will block until DDS
samples are acknowledged by all the intended consumers.

700

13.2.1 Scenario 1: Guaranteed Delivery to a-priori Known Subscribers

13.2.1 Scenario 1: Guaranteed Delivery to a-priori Known Subscribers
A common use case is to guarantee delivery to a set of known subscribers. These subscribers may be
already running and have been discovered, they may be temporarily non-responsive, or it could be that
some of those subscribers are still not present in the system. See Figure 13.1 Guaranteed Delivery Scenario
1 on the next page.
To guarantee delivery, the list of required subscribers should be configured using the AVAILABILITY
QosPolicy (DDS Extension) (Section 6.5.1 on page 337) on the DataWriters to specify the role_name
and quorum_count for each required subscription. Similarly the ENTITY_NAME QosPolicy (DDS
Extension) (Section 6.5.9 on page 374) should be used on the DataReaders to specify their role_name.
In addition we use Application Acknowledgment (Section 6.3.12 on page 288) to guarantee the DDS
sample was delivered and processed by the DataReader.

701

13.2.1 Scenario 1: Guaranteed Delivery to a-priori Known Subscribers

Figure 13.1 Guaranteed Delivery Scenario 1

The DataWriter's and DataReader's RELIABILITY QoS Policy can be configured for either AUTO or
EXPLICIT application acknowledgment kind. As the DataWriter publishes the DDS sample, it will await
acknowledgment from the DataReader (through the protocol-level acknowledgment) and from the subscriber application (though the additional application-level acknowledgment). The DataWriter will only

702

13.2.2 Scenario 2: Surviving a Writer Restart when Delivering DDS Samples to a priori Known

consider the DDS sample acknowledged when it has been acknowledged by all discovered active
DataReaders and also by the quorum_count of each required subscription.
In this specific scenario, DataReader #1 is configured for EXPLICIT application acknowledgment. After
reading and processing the DDS sample, the subscribing application calls acknowledge_sample() or
acknowledge_all() (see Acknowledging DDS Samples (Section 7.4.4 on page 502)). As a result, Connext DDS will send an application-level acknowledgment to the DataWriter, which will in its turn confirm
the acknowledgment.
If the DDS sample was lost in transit, the reliability protocol will repair the DDS sample. Since it has not
been acknowledged, it remains available in the writer’s queue to be automatically resent by Connext DDS.
The DDS sample will remain available until acknowledged by the application. If the subscribing application crashes while processing the DDS sample and restarts, Connext DDS will repair the unacknowledged DDS sample. DDS samples which already been processed and acknowledged will not be
resent.
In this scenario, DataReader #2 may be a late joiner. When it starts up, because it is configured with
TRANSIENT_LOCAL Durability, the reliability protocol will re-send the DDS samples previously sent
by the writer. These DDS samples were considered unacknowledged by the DataWriter because they had
not been confirmed yet by the required subscription (identified by its role_name: ‘logger’).
DataReader #2 does not explicitly acknowledge the DDS samples it reads. It is configured to use AUTO
application acknowledgment, which will automatically acknowledge DDS samples that have been read or
taken after the application calls the DataReader return_loan operation.
This configuration works well for situations where the DataReader may not be immediately available or
may restart. However, this configuration does not provide any guarantee if the DataWriter restarts. When
the DataWriter restarts, DDS samples previously unacknowledged are lost and will no longer be available
to any late joining DataReaders.

13.2.2 Scenario 2: Surviving a Writer Restart when Delivering DDS Samples
to a priori Known Subscribers
Scenario 1 describes a use case where DDS samples are delivered to a list of a priori known subscribers.
In that scenario, Connext DDS will deliver DDS samples to the late-joining or restarting subscriber.
However, if the producer is re-started the DDS samples it had written will no longer be available to future
subscribers.
To handle a situation where the producing application is restarted, we will use the Durable Writer History
(Section 12.3 on page 681) feature. See Figure 13.2 Guaranteed Delivery Scenario 2 on the next page.
A DataWriter can be configured to maintain its data and state in durable storage. This configuration is
done using the PROPERTY QoS policy as described in How To Configure Durable Writer History (Section 12.3.2 on page 683).. With this configuration the DDS data samples written by the DataWriter and
any necessary internal state is persisted by the DataWriter into durable storage As a result, when the
DataWriter restarts, DDS samples which had not been acknowledged by the set of required subscriptions
703

13.2.3 Scenario 3: Delivery Guaranteed by Persistence Service (Store and Forward) to a priori Known

will be resent and late-joining DataReaders specifying DURABILITY kind different from VOLATILE
will receive the previously-written DDS samples.
Figure 13.2 Guaranteed Delivery Scenario 2

13.2.3 Scenario 3: Delivery Guaranteed by Persistence Service (Store and
Forward) to a priori Known Subscribers
Previous scenarios illustrated that using the DURABILITY, RELIABILITY, and AVAILABILITY QoS
policies we can ensure that as long as the DataWriter is present in the system, DDS samples written by a
DataWriter will be delivered to the intended consumers. The use of the durable writer history in the previous scenario extended this guarantee even in the presence of a restart of the application writing the data.
This scenario addresses the situation where the originating application that produced the data is no longer
available. For example, the network could have become partitioned, the application could have been terminated, it could have crashed and not have been restarted, etc.

704

13.2.3 Scenario 3: Delivery Guaranteed by Persistence Service (Store and Forward) to a priori Known

In order to deliver data to applications that appear after the producing application is no longer available on
the network it is necessary to have another service that stores those DDS samples and delivers them. This
is the purpose of the RTI Persistence Service.
The RTI Persistence Service can be configured to automatically discover DataWriters that specify a
DURABILITY QoS with kind TRANSIENT or PERSISTENT and automatically create pairs
(DataReader, DataWriter) that receive and store that information (see Introduction to RTI Persistence Service (Section Chapter 26 on page 933)). All the DataReaders created by the RTI Persistence Service have
the ENTITY_QOS policy set with the role_name of “PERSISTENCE_SERVICE”. This allows an
application to specify Persistence Service as one of the required subscriptions for its DataWriters.
In this third scenario, we take advantage of this capability to configure the DataWriter to have the RTI Persistence Service as a required subscription. See Figure 13.3 Guaranteed Delivery Scenario 3 below.
Figure 13.3 Guaranteed Delivery Scenario 3

The RTI Persistence Service can also have its DataWriters configured with required subscriptions. This
feature is known as Persistence Service “durable subscriptions”. DataReader #1 is pre configured in Persistence Service as a Durable Subscription. (Alternatively, DataReader #1 could have registered itself
dynamically as Durable Subscription using the DomainParticipant register_durable_subscription() operation).

705

13.2.3.1 Variation: Using Redundant Persistence Services

We also configure the RELIBILITY QoS policy setting of the AcknowledgmentKind to
APPLICATION_AUTO_ACKNOWLEDGMENT_MODE in order to ensure DDS samples are stored
in the Persistence Service and properly processed on the consuming application prior to them being
removed from the DataWriter cache.
With this configuration in place the DataWriter will deliver DDS samples to the DataReader and to the
Persistence Service reliably and wait for the Application Acknowledgment from both. Delivery of DDS
samples to DataReader #1 and the Persistence Service occurs concurrently. The Persistence Service in
turn takes responsibility to deliver the DDS samples to the configured “logger” durable subscription. If the
original publisher is no longer available, DDS samples can still be delivered by the Persistence Service. to
DataReader #1 and any other late-joining DataReaders.
When DataReader #1 acknowledges the DDS sample through an application-acknowledgment message,
both the original DataWriter and Persistence Service will receive the application-acknowledgment. Connext DDS takes advantage of this to reduce or eliminate delivery if duplicate DDS samples, that is, the Persistence Service can notice that DataReader #1 has acknowledged a DDS sample and refrain from
separately sending the same DDS sample to DataReader #1.
13.2.3.1 Variation: Using Redundant Persistence Services
Using a single Persistence Service to guarantee delivery can still raise concerns about having the Persistence Service as a single point of failure. To provide a level of added redundancy, the publisher may be
configured to await acknowledgment from a quorum of multiple persistence services (role_name remains
PERSISTENCE). Using this configuration we can achieve higher levels of redundancy

706

13.2.3.2 Variation: Using Load-Balanced Persistent Services

Figure 13.4 Guaranteed Delivery Scenario 3 with Redundant Persistence Service

The RTI Persistence Services will automatically share information to keep each other synchronized. This
includes both the data and also the information on the durable subscriptions. That is, when a Persistence
Service discovers a durable subscription, information about durable subscriptions is automatically replicated and synchronized among persistence services (CITE: New section to be written in Persistence Service Chapter).
13.2.3.2 Variation: Using Load-Balanced Persistent Services
The Persistence Service will store DDS samples on behalf of many DataWriters and, depending on the
configuration, it might write those DDS samples to a database or to disk. For this reason the Persistence
Service may become a bottleneck in systems with high durable DDS sample throughput.
It is possible to run multiple instances of the Persistence Service in a manner where each is only responsible for the guaranteed delivery of certain subset of the durable data being published. These Persistence
Service can also be run different computers and in this manner achieve much higher throughput. For
example, depending on the hardware, using typical hard-drives a single a Persistence Service may be able
to store only 30000 DDS samples per second. By running 10 persistence services in 10 different computers we would be able to handle storing 10 times that system-wide, that is, 300000 DDS samples per
second.
The data to be persisted can be partitioned among the persistence services by specifying different Topics to
be persisted by each Persistence Service. If a single Topic has more data that can be handled y a single Persistence Service it is also possible to specify a content-filter so that only the data within that Topic that

707

13.2.3.2 Variation: Using Load-Balanced Persistent Services

matches the filter will be stored by the Persistence Service. For example assume the Topic being persisted
has an member named “x” of type float. It is possible to configure two Persistence Services one with the filter “x>10”, and the other “x <=10”, such that each only stores a subject of the data published on the Topic.
See also: Configuring Durable Subscriptions in Persistence Service (Section 27.9 on page 955).

708

Chapter 14 Discovery
This section discusses how Connext DDS objects on different nodes find out about each other
using the default Simple Discovery Protocol (SDP). It describes the sequence of messages that are
passed between Connext DDS on the sending and receiving sides.
This section includes:
l

What is Discovery? (Section 14.1 on the next page)

l

Configuring the Peers List Used in Discovery (Section 14.2 on page 711)

l

Discovery Implementation (Section 14.3 on page 717)

l

Debugging Discovery (Section 14.4 on page 735)

l

Ports Used for Discovery (Section 14.5 on page 738)

The discovery process occurs automatically, so you do not have to implement any special code.
We recommend that all users read What is Discovery? (Section 14.1 on the next page) and Configuring the Peers List Used in Discovery (Section 14.2 on page 711). The remaining sections contain advanced material for those who have a particular need to understand what is happening
‘under the hood.’ This information can help you debug a system in which objects are not communicating.
You may also be interested in reading Transport Plugins (Section Chapter 15 on page 743) , as
well as learning about these QosPolicies:
l

TRANSPORT_SELECTION QosPolicy (DDS Extension) (Section 6.5.23 on page 411)

l

TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section 8.5.7 on page 606)

l

TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412)

l

TRANSPORT_MULTICAST QosPolicy (DDS Extension) (Section 7.6.5 on page 529)

709

14.1 What is Discovery?

14.1 What is Discovery?
Discovery is the behind-the-scenes way in which Connext DDS objects (DomainParticipants,
DataWriters, and DataReaders) on different nodes find out about each other. Each DomainParticipant
maintains a database of information about all the active DataReaders and DataWriters that are in the same
DDS domain. This database is what makes it possible for DataWriters and DataReaders to communicate.
To create and refresh the database, each application follows a common discovery process.
This chapter describes the default discovery mechanism known as the Simple Discovery Protocol, which
includes two phases: Simple Participant Discovery (Section 14.1.1 below) and Simple Endpoint Discovery (Section 14.1.2 on the facing page). (Discovery can also be performed using the Enterprise Discovery Protocol—this requires a separately purchased package, RTI Enterprise Discovery Service.)
The goal of these two phases is to build, for each DomainParticipant, a complete picture of all the entities
that belong to the remote participants that are in its peers list. The peers list is the list of nodes with which a
participant may communicate. It starts out the same as the initial_peers list that you configure in the
DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580). If the accept_unknown_peers
flag in that same QosPolicy is TRUE, then other nodes may also be added as they are discovered; if it is
FALSE, then the peers list will match the initial_peers list, plus any peers added using the DomainParticipant’s add_peer() operation.

14.1.1 Simple Participant Discovery
This phase of the Simple Discovery Protocol is performed by the Simple Participant Discovery Protocol
(SPDP).
During the Participant Discovery phase, DomainParticipants learn about each other. The DomainParticipant’s details are communicated to all other DomainParticipants in the same DDS domain by sending
participant declaration messages, also known as participant DATA submessages. The details include the
DomainParticipant’s unique identifying key (GUID or Globally Unique ID described below), transport
locators (addresses and port numbers), and QoS. These messages are sent on a periodic basis using besteffort communication.
Participant DATAs are sent periodically to maintain the liveliness of the DomainParticipant. They are also
used to communicate changes in the DomainParticipant’s QoS. Only changes to QosPolicies that are part
of the DomainParticipant’s built-in data (namely, the USER_DATA QosPolicy (Section 6.5.26 on page
417)) need to be propagated.
When a DomainParticipant is deleted, a participant DATA (delete) submessage with the DomainParticipant's identifying GUID is sent.
The GUID is a unique reference to an entity. It is composed of a GUID prefix and an Entity ID. By
default, the GUID prefix is calculated from the IP address and the process ID. (For more on how the
GUID is calculated, see Controlling How the GUID is Set (rtps_auto_id_kind) (Section 8.5.9.4 on page
614).) The IP address and process ID are stored in the DomainParticipant’s WIRE_PROTOCOL

710

14.1.2 Simple Endpoint Discovery

QosPolicy (DDS Extension) (Section 8.5.9 on page 610). The entityID is set by Connext DDS (you may
be able to change it in a future version).
Once a pair of remote participants have discovered each other, they can move on to the Endpoint Discovery phase, which is how DataWriters and DataReaders find each other.

14.1.2 Simple Endpoint Discovery
This phase of the Simple Discovery Protocol is performed by the Simple Endpoint Discovery Protocol
(SEDP).
During the Endpoint Discovery phase, Connext DDS matches DataWriters and DataReaders. Information
(GUID, QoS, etc.) about your application’s DataReaders and DataWriters is exchanged by sending publication/subscription declarations in DATA messages that we will refer to as publication DATAs and subscription DATAs. The Endpoint Discovery phase uses reliable communication.
As described in Discovery Implementation (Section 14.3 on page 717), these declaration or DATA messages are exchanged until each DomainParticipant has a complete database of information about the participants in its peers list and their entities. Then the discovery process is complete and the system switches
to a steady state. During steady state, participant DATAs are still sent periodically to maintain the liveliness
status of participants. They may also be sent to communicate QoS changes or the deletion of a DomainParticipant.
When a remote DataWriter/DataReader is discovered, Connext DDS determines if the local application
has a matching DataReader/DataWriter. A ‘match’ between the local and remote entities occurs only if
the DataReader and DataWriter have the same Topic, same data type, and compatible QosPolicies (which
includes having the same partition name string, see PARTITION QosPolicy (Section 6.4.5 on page 323)).
Furthermore, if the DomainParticipant has been set up to ignore certain DataWriters/DataReaders, those
entities will not be considered during the matching process. See Ignoring Publications and Subscriptions
(Section 16.4.2 on page 786) for more on ignoring specific publications and subscriptions.
This ‘matching’ process occurs as soon as a remote entity is discovered, even if the entire database is not
yet complete: that is, the application may still be discovering other remote entities.
A DataReader and DataWriter can only communicate with each other if each one’s application has
hooked up its local entity with the matching remote entity. That is, both sides must agree to the connection.
Discovery Implementation (Section 14.3 on page 717) describes the details about the discovery process.

14.2 Configuring the Peers List Used in Discovery
As part of the participant phase of the discovery process, Connext DDS will announce itself within the
DDS domain. Connext DDS will try to contact all possible participants in the ‘initial peers list,’ specified
in the DomainParticipant’s DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580).
Note, however, it is not known if there are actually Connext DDS applications running on the hosts in the
inital peers list. The initial peers list may include both unicast and multicast peer locators.

711

14.2 Configuring the Peers List Used in Discovery

After startup, you can add to the ‘peers list’ with the add_peer() operation (see Adding and Removing
Peers List Entries (Section 8.5.2.3 on page 581)). The ‘peers list’ may also grow as peers are automatically discovered (if accept_unknown_peers is TRUE, see Controlling Acceptance of Unknown Peers
(Section 8.5.2.6 on page 583)).
When you call get_default_participant_qos() for a DomainParticipantFactory, the values used for the
DiscoveryQosPolicy’s initial_peers and multicast_receive_addresses may come from the following:
l

l

l

A file named NDDS_DISCOVERY_PEERS, which is formatted as described in NDDS_
DISCOVERY_PEERS File Format (Section 14.2.3 on page 717). The file must be in the same directory as your application’s executable.
An environment variable named NDDS_DISCOVERY_PEERS, defined as a comma-separated list
of peer descriptors (see NDDS_DISCOVERY_PEERS Environment Variable Format (Section
14.2.2 on page 716)).
The value specified in the default XML QoS profile (see Configuring QoS with XML (Section 17.4
on page 803)).

If NDDS_DISCOVERY_PEERS (file or environment variable) does not contain a multicast address,
then multicast_receive_addresses is cleared and the RTI discovery process will not listen for discovery
messages via multicast.
If NDDS_DISCOVERY_PEERS (file or environment variable) contains one or more multicast
addresses, the addresses are stored in multicast_receive_addresses, starting at element 0. They will be
stored in the order in which they appear in NDDS_DISCOVERY_PEERS.
Note: Setting initial_peers in the default XML QoS Profile does not modify the value of multicast_
receive_address.
If both the file and environment variable are found, the file takes precedence and the environment variable
will be ignored.1 The settings in the default XML QoS Profile take precedence over the file and environment variable. In the absence of a file, environment variable, or default XML QoS profile values, Connext DDS will use a default value. See the API Reference HTML documentation for details (in the section
on the DISCOVERY QosPolicy).
If initial peers are specified in both the currently loaded QoS XML profile and in the NDDS_
DISCOVERY_PEERS file, the values in the profile take precedence.
The file, environment variable, and default XML QoS Profile make it easy to reconfigure which nodes
will take part in the discovery process—without recompiling your application.

1This is true even if the file is empty.

712

14.2.1 Peer Descriptor Format

The file, environment variable, and default XML QoS Profile are the possible sources for the default initial
peers list. You can, of course, explicitly set the initial list by changing the values in the QoS provided to
the DomainParticipantFactory's create_participant() operation, or by adding to the list after startup with
the DomainParticipant’s add_peer() operation (see Adding and Removing Peers List Entries (Section
8.5.2.3 on page 581)).
If you set NDDS_DISCOVERY_PEERS and You Want to Communicate over Shared Memory:
Suppose you want to communicate with other Connext DDS applications on the same host and you are
explicitly setting NDDS_DISCOVERY_PEERS (generally in order to use unicast discovery with applications on other hosts).
If the local host platform does not support the shared memory transport, then you can include the name of
the local host in the NDDS_DISCOVERY_PEERS list. (To check if your platform supports shared
memory, see the RTI Connext DDS Core Libraries Platform Notes.)
If the local host platform supports the shared memory transport, then you must do one of the following:
l

Include "shmem://" in the NDDS_DISCOVERY_PEERS list. This will cause shared memory to
be used for discovery and data traffic for applications on the same host.

or:
l

Include the name of the local host in the NDDS_DISCOVERY_PEERS list, and disable the
shared memory transport in the TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section
8.5.7 on page 606) of the DomainParticipant. This will cause UDP loopback to be used for discovery and data traffic for applications on the same host.

14.2.1 Peer Descriptor Format
A peer descriptor string specifies a range of participants at a given locator. Peer descriptor strings are used
in the DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580) initial_peers field (see Setting the ‘Initial Peers’ List (Section 8.5.2.2 on page 581)) and the DomainParticipant’s add_peer() and
remove_peer() operations (see Adding and Removing Peers List Entries (Section 8.5.2.3 on page 581)).
The anatomy of a peer descriptor is illustrated in Example Peer Descriptor Address Strings (Section Figure
14.1 on the next page) using a special "StarFabric" transport example.

713

14.2.1.1 Locator Format

Figure 14.1 Example Peer Descriptor Address Strings

A peer descriptor consists of:
l

l

[optional] A participant ID limit. If a simple integer is specified, it indicates the maximum participant ID to be contacted by the Connext DDS discovery mechanism at the given locator. If that
integer is enclosed in square brackets (e.g., [2]), then only that Participant ID will be used. You can
also specify a range in the form of [a-b]: in this case only the Participant IDs in that specific range
are contacted. If omitted, a default value of 4 is implied and participant IDs 0, 1, 2, 3, and 4 will be
contacted.
A locator, as described in Locator Format (Section 14.2.1.1 below).

These are separated by the '@' character. The separator may be omitted if a participant ID limit is not explicitly specified.
The "participant ID limit" only applies to unicast locators; it is ignored for multicast locators (and therefore
should be omitted for multicast peer descriptors).
14.2.1.1 Locator Format
A locator string specifies a transport and an address in string format. Locators are used to form peer
descriptors. A locator is equivalent to a peer descriptor with the default participant ID limit (4).
A locator consists of:
714

14.2.1.2 Address Format

l

l

[optional] Transport name (alias or class). This identifies the set of transport plug-ins (transport aliases) that may be used to parse the address portion of the locator. Note that a transport class name is
an implicit alias used to refer to all the transport plug-in instances of that class.
[optional] An address, as described in Address Format (Section 14.2.1.2 below).

These are separated by the "://" string. The separator is specified if and only if a transport name is specified.
If a transport name is specified, the address may be omitted; in that case all the unicast addresses (across all
transport plug-in instances) associated with the transport class are implied. Thus, a locator string may specify several addresses.
If an address is specified, the transport name and the separator string may be omitted; in that case all the
available transport plug-ins for the Entity may be used to parse the address string.
The transport names for the built-in transport plug-ins are:
l

shmem - Shared Memory Transport

l

udpv4 - UDPv4 Transport

l

udpv6 - UDPv6 Transport

14.2.1.2 Address Format
An address string specifies a transport-independent network address that qualifies a transport-dependent
address string. Addresses are used to form locators. Addresses are also used in the DISCOVERY
QosPolicy (DDS Extension) (Section 8.5.2 on page 580) multicast_receive_addresses and the DDS_
TransportMulticastSettings_t::receive_address fields. An address is equivalent to a locator in which the
transport name and separator are omitted.
An address consists of:
l

l

[optional] A network address in IPv4 or IPv6 string notation. If omitted, the network address of the
transport is implied.
[optional] A transport address, which is a string that is passed to the transport for processing. The
transport maps this string into NDDS_Transport_Property_t::address_bit_count bits. If omitted,
the network address is used as the fully qualified address.

The network and transport addressed are separated by the '#' character. If a separator is specified, it must
be followed by a non-empty string that is passed to the transport plug-in. If the separator is omitted, it is
treated as a transport address with an implicit network address (of the transport plugin). The implicit network address is the address used when registering the transport: e.g., the UDPv4 implicit network address
is 0.0.0.0.0.0.0.0.0.0.0.0.

715

14.2.2 NDDS_DISCOVERY_PEERS Environment Variable Format

The bits resulting from the transport address string are prepended with the network address. The least significant NDDS_Transport_Property_t::address_bit_count bits of the network address are ignored.

14.2.2 NDDS_DISCOVERY_PEERS Environment Variable Format
You can set the default value for the initial peers list in an environment variable named NDDS_
DISCOVERY_PEERS. Multiple peer descriptor entries must be separated by commas. Table 14.1
NDDS_DISCOVERY_PEERS Environment Variable Examples shows some examples. The examples
use an implied maximum participant ID of 4 unless otherwise noted. (If you need instructions on how to
set environment variables, see the RTI Connext DDS Core Libraries Getting Started Guide).
Table 14.1 NDDS_DISCOVERY_PEERS Environment Variable Examples
NDDS_DISCOVERY_
PEERS

Description of Host(s)

239.255.0.1

multicast

localhost

localhost

192.168.1.1

10.10.30.232 (IPv4)

FAA0::1

FAA0::0 (IPv6)

himalaya,gangotri

himalaya and gangotri

1@himalaya,1@gangotri

himalaya and gangotri (with a maximum participant ID of 1 on each host)

FAA0::0#localhost

FAA0::0#localhost (could be a UDPv4 transport plug-in registered at network address of FAA0::0)
(IPv6)

udpv4://himalaya

himalaya accessed using the "udpv4" transport plug-in (IPv4)

udpv4://FAA0::0#localhost

localhost using the "udpv4" transport plug-in registered at network address FAA0::0

0/0/R
0/0/R (StarFabric)
#0/0/R
starfabric://0/0/R
0/0/R (StarFabric) using the "starfabric" (StarFabric) transport plug-ins
starfabric://#0/0/R
starfabric://FBB0::0#0/0/R

0/0/R (StarFabric) using the "starfabric" (StarFabric) transport plug-ins registered at network address
FAA0::0

starfabric://

all unicast addresses accessed via the "starfabric" (StarFabric) transport plug-ins

shmem://FCC0::0

all unicast addresses accessed via the "shmem" (shared memory) transport plug-ins registered at network
address FCC0::0

716

14.2.3 NDDS_DISCOVERY_PEERS File Format

14.2.3 NDDS_DISCOVERY_PEERS File Format
You can set the default value for the initial peers list in a file named NDDS_DISCOVERY_PEERS. The
file must be in the your application’s current working directory.
The file is optional. If it is found, it supersedes the values in any environment variable of the same name.
Entries in the file must contain a sequence of peer descriptors separated by whitespace or the comma (',')
character. The file may also contain comments starting with a semicolon (';') character until the end of the
line.
Example file contents:
;; NDDS_DISCOVERY_PEERS - Discovery Configuration File
;; Multicast builtin.udpv4://239.255.0.1 ; default discovery multicast addr
;; Unicast
localhost,192.168.1.1
FAA0::1 FAA0::0#localhost
1@himalaya
1@gangotri
;; UDPv4
udpv4://himalaya
udpv4://FAA0::0#localhost
;; Shared Memory
shmem://
builtin.shmem://
shmem://FCC0::0
;; StarFabric
0/0/R
starfabric://0/0/R
starfabric://FBB0::0#0/0/R
starfabric://

; A comma can be used a separator
; Whitespace can be used as a separator
; Max participant ID of 1 on 'himalaya'

; 'himalaya' via 'udpv4' transport plugin(s)
; 'localhost' via 'updv4' transport plugin
; registered at network address FAA0::0
;
;
;
;

All 'shmem' transport plugin(s)
The builtin builtin 'shmem' transport plugin
Shared memory transport plugin registered
at network address FCC0::0

;
;
;
;
;
;

StarFabric node 0/0/R
0/0/R accessed via 'starfabric'
transport plugin(s)
StarFabric transport plugin registered
at network address FBB0::0
All 'starfabric' transport plugin(s)

14.3 Discovery Implementation
Note: this section contains advanced material not required by most users.
Discovery is implemented using built-in DataWriters and DataReaders. These are the same class of entities your application uses to send/receive data. That is, they are also of type
DDSDataWriter/DDSDataReader. For each DomainParticipant, three built-in DataWriters and three
built-in DataReaders are automatically created for discovery purposes. Figure 14.2 Built-in Writers and
Readers for Discovery on the next page shows how these objects are used. (For more on built-in
DataReaders and DataWriters, see Built-In Topics (Section Chapter 16 on page 772)).

717

14.3.1 Participant Discovery

Figure 14.2 Built-in Writers and Readers for Discovery

For each DomainParticipant, there are six objects automatically created for discovery purposes. The top two objects
are used to send/receive participant DATA messages, which are used in the Participant Discovery phase to find remote
DomainParticipants. This phase uses best-effort communications. Once the participants are aware of each other, they
move on to the Endpoint Discovery Phase to learn about each other’s DataWriters and DataReaders. This phase uses
reliable communications.

The implementation is split into two separate protocols:
Simple Participant Discovery Protocol (SPDP)
+ Simple Endpoint Discovery Protocol (SEDP)
= Simple Discovery Protocol (SDP)

14.3.1 Participant Discovery
When a DomainParticipant is created, a DataWriter and a DataReader are automatically created to
exchange participant DATA messages in the network. These DataWriters and DataReaders are "special"
because the DataWriter can send to a given list of destinations, regardless of whether there is a Connext
DDS application at the destination, and the DataReader can receive data from any source, whether the
source is previously known or not. In other words, these special readers and writers do not need to discover the remote entity and perform a match before they can communicate with each other.

718

14.3.1 Participant Discovery

When a DomainParticipant joins or leaves the network, it needs to notify its peer participants. The list of
remote participants to use during discovery comes from the peer list described in the DISCOVERY
QosPolicy (DDS Extension) (Section 8.5.2 on page 580). The remote participants are notified via participant DATA messages. In addition, if a participant’s QoS is modified in such a way that other participants need to know about the change (that is, changes to the USER_DATA QosPolicy (Section 6.5.26
on page 417)), a new participant DATA will be sent immediately.
Participant DATAs are also used to maintain a participant’s liveliness status. These are sent at the rate set
in the participant_liveliness_assert_period in the DISCOVERY_CONFIG QosPolicy (DDS Extension)
(Section 8.5.3 on page 585).
Let’s examine what happens when a new remote participant is discovered. If the new remote participant is
in the local participant's peer list, the local participant will add that remote participant into its database. If
the new remote participant is not in the local application's peer list, it may still be added, if the accept_
unknown_peers field in the DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580) is set
to TRUE.
Once a remote participant has been added to the Connext DDS database, Connext DDS keeps track of
that remote participant’s participant_liveliness_lease_duration. If a participant DATA for that participant
(identified by the GUID) is not received at least once within the participant_liveliness_lease_duration, the
remote participant is considered stale, and the remote participant, together with all its entities, will be
removed from the database of the local participant.
To keep from being purged by other participants, each participant needs to periodically send a participant
DATA to refresh its liveliness. The rate at which the participant DATA is sent is controlled by the participant_liveliness_assert_period in the participant’s DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585). This exchange, which keeps Participant A from appearing ‘stale,’ is
illustrated in Figure 14.3 Periodic ‘participant DATAs’ on the next page. Figure 14.4 Ungraceful Termination of a Participant on page 721 shows what happens when Participant A terminates ungracefully
and therefore needs to be seen as ‘stale.’

719

14.3.1 Participant Discovery

Figure 14.3 Periodic ‘participant DATAs’

The DomainParticipant on Node A sends a ‘participant DATA’ to Node B, which is in Node A’s peers list. This occurs
regardless of whether or not there is a Connext DDS application on Node B.

 The green short dashed lines are periodic participant DATAs. The time between these messages is controlled by the
participant_liveliness_assert_period in the DiscoveryConfig QosPolicy.
k In addition to the periodic participant DATAs, ‘initial repeat messages’ (shown in blue, with longer dashes) are sent
from A to B. These messages are sent at a random time between min_initial_participant_announcement_period and
max_initial_participant_announcement_period (in A’s DiscoveryConfig QosPolicy). The number of these initial
repeat messages is set in initial_participant_announcements.

720

14.3.1 Participant Discovery

Figure 14.4 Ungraceful Termination of a Participant

Participant A is removed from participant B’s database if it is not refreshed within the liveliness lease duration.
Dashed lines are periodic participant DATA messages.
(Periodic resends of ‘participant B DATA’ from B to A are omitted from this diagram for simplicity. Initial repeat messages from A to B are also omitted from this diagram—these messages are sent at a random time between min_initial_
participant_announcement_period and max_initial_participant_announcement_period, see Figure 14.3 Periodic
‘participant DATAs’ on the previous page.)

721

14.3.1.1 Refresh Mechanism

14.3.1.1 Refresh Mechanism
To ensure that a late-joining participant does not need to wait until the next refresh of the remote participant DATA to discover the remote participant, there is a resend mechanism. If the received participant
DATA is from a never-before-seen remote participant, and it is in the local participant's peers list, the application will resend its own participant DATA to all its peers. This resend can potentially be done multiple
times, with a random sleep time in between. Figure 14.5 Resending ‘participant DATA’ to a Late-Joiner
on the facing page illustrates this scenario.
The number of retries and the random amount of sleep between them are controlled by each participant’s
DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585) (see Figure 14.5
Resending ‘participant DATA’ to a Late-Joiner on the facing page).

722

14.3.1.1 Refresh Mechanism

Figure 14.5 Resending ‘participant DATA’ to a Late-Joiner

Participant A has Participant B in its peers list. Participant B does not have Participant A in its peers list, but [DiscoveryQosPolicy.accept_unknown_peers] is set to DDS_BOOLEAN_TRUE. Participant A joins the system after B has
sent its initial announcement. After B discovers A, it waits for time Á, then resends its participant DATA.

723

14.3.1.2 Maintaining DataWriter Liveliness for kinds AUTOMATIC and MANUAL_BY_

(Initial repeat messages are omitted from this diagram for simplicity, see Figure 14.3 Periodic ‘participant DATAs’
on page 720.)

Figure 14.6 Participant Discovery Summary below provides a summary of the messages sent during the
participant discovery phase.
Figure 14.6 Participant Discovery Summary

Participants A and B both have each other in their peers lists. Participant A is created first.

14.3.1.2 Maintaining DataWriter Liveliness for kinds AUTOMATIC and MANUAL_BY_
PARTICIPANT
To maintain the liveliness of DataWriters that have a LIVELINESS QosPolicy (Section 6.5.13 on page
382) kind field set to AUTOMATIC or MANUAL_BY_PARTICIPANT, Connext DDS uses a builtin DataWriter and DataReader pair, referred to as the inter-participant reader and inter-participant writer.
If the DomainParticipant has any DataWriters with Liveliness QosPolicy kind set to AUTOMATIC, the
inter-participant writer will reliably broadcast an AUTOMATIC liveliness message at a period equal to
724

14.3.1.2 Maintaining DataWriter Liveliness for kinds AUTOMATIC and MANUAL_BY_

the shortest lease_duration of these DataWriters. (The lease_duration is a field in the LIVELINESS
QosPolicy (Section 6.5.13 on page 382).) Figure 14.7 DataWriter with AUTOMATIC Liveliness below
illustrates this scenario.
Figure 14.7 DataWriter with AUTOMATIC Liveliness

A liveliness message is sent automatically when a DataWriter with AUTOMATIC Liveliness kind is created, and then
periodically, every DDS_DataWriterQos.liveliness.lease_duration.

If the DomainParticipant has any DataWriters with Liveliness QosPolicy kind set to MANUAL_BY_
PARTICIPANT, Connext DDS will periodically check to see if any of them have called write(),
assert_liveliness(), dispose() or unregister(). The rate of this check is every X seconds, where X is the
smallest lease_duration among all the DomainParticipant's MANUAL_BY_PARTICIPANT
DataWriters. (The lease_duration is a field in the LIVELINESS QosPolicy (Section 6.5.13 on page

725

14.3.1.2 Maintaining DataWriter Liveliness for kinds AUTOMATIC and MANUAL_BY_

382).) If any of the MANUAL_BY_PARTICIPANT DataWriters have called any of those operations,
the inter-participant writer will reliably broadcast a MANUAL liveliness message.
If a DomainParticipant's assert_liveliness() operation is called, and that DomainParticipant has any
MANUAL_BY_PARTICIPANT DataWriters, the inter-participant writer will reliably broadcast a
MANUAL liveliness message within the above-defined X time period. These MANUAL liveliness messages are used to update the liveliness of all the DomainParticipant's MANUAL_BY_PARTICIPANT
DataWriters, as well as the liveliness of the DomainParticipant itself. Figure 14.8 DataWriter with
MANUAL_BY_PARTICIPANT Liveliness on the facing page shows an example sequence.

726

14.3.1.2 Maintaining DataWriter Liveliness for kinds AUTOMATIC and MANUAL_BY_

Figure 14.8 DataWriter with MANUAL_BY_PARTICIPANT Liveliness

Once a MANUAL_BY_PARTICIPANT DataWriter is created, subsequent calls to assert_liveliness, write, dispose, or
unregister_instance will trigger Liveliness messages, which update the liveliness status of all the participant’s
DataWriters, and the participant itself.

The inter-participant reader receives data from remote inter-participant writers and asserts the liveliness of
remote DomainParticipants endpoints accordingly.
If the DomainParticipant has no DataWriters with LIVELINESS QosPolicy (Section 6.5.13 on page
382) kind set to AUTOMATIC or MANUAL_BY_PARTICIPANT, then no liveliness messages are
ever sent from the inter-participant writer.

727

14.3.2 Endpoint Discovery

14.3.2 Endpoint Discovery
As we saw in Built-in Writers and Readers for Discovery (Section Figure 14.2 on page 718), reliable
DataReaders and Datawriters are automatically created to exchange publication/subscription information
for each DomainParticipant. We will refer to these as ‘discovery endpoint readers and writers.’ However,
nothing is sent through the network using these entities until they have been ‘matched’ with their remote
counterparts. This ‘matching’ is triggered by the Participant Discovery phase. The goal of the Endpoint
Discovery phase is to add the remote endpoint to the local database, so that user-created endpoints (your
application’s DataWriters/DataReaders) can communicate with each other.
When a new remote DomainParticipant is discovered and added to a participant’s database, Connext
DDS assumes that the remote DomainParticipant is implemented in the same way and therefore is creating the appropriate counterpart entities. Therefore, Connext DDS will automatically add two remote discovery endpoint readers and two remote discovery endpoint writers for that remote DomainParticipant
into the local database. Once that is done, there is now a match with the local discovery endpoint writers
and readers, and publication DATAs and subscription DATAs can then be sent between the discovery endpoint readers/writers of the two DomainParticipant.
When you create a DataWriter/DataReader for your user data, a publication/subscription DATA describing the newly created object is sent from the local discovery endpoint writer to the remote discovery endpoint readers of the remote DomainParticipants that are currently in the local database.
If your application changes any of the following QosPolicies for a local user-data DataWriter/DataReader,
a modified subscription/publication DATA is sent to propagate the QoS change to other DomainParticipants:
l

TOPIC_DATA QosPolicy (Section 5.2.1 on page 209)

l

GROUP_DATA QosPolicy (Section 6.4.4 on page 320)

l

USER_DATA QosPolicy (Section 6.5.26 on page 417)

l

OWNERSHIP_STRENGTH QosPolicy (Section 6.5.16 on page 393)

l

PARTITION QosPolicy (Section 6.4.5 on page 323)

l

TIME_BASED_FILTER QosPolicy (Section 7.6.4 on page 526)

l

LIFESPAN QoS Policy (Section 6.5.12 on page 381)

What the above QosPolicies have in common is that they are all changeable and part of the built-in data
(see Built-In Topics (Section Chapter 16 on page 772)).
Similarly, if the application deletes any user-data writers/readers, the discovery endpoint writer/readers
send delete publication/subscription DATAs. In addition to sending publication/subscription DATAs, the
discovery endpoint writer will check periodically to see if the remote discovery endpoint reader is up-todate. (The rate for this check is the publication_writer.heartbeat_period or subscription_writer.heartbeat_
period in the DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585). If the

728

14.3.2 Endpoint Discovery

discovery endpoint writer has not been acknowledged by the remote discovery endpoint reader regarding
receipt of the latest DATA, the discovery endpoint writer will send a special Heartbeat (HB) message with
the Final bit set to 0 (F=0) to request acknowledgement from the remote discovery endpoint reader, as seen
in Figure 14.9 Endpoint Discovery Summary on the next page.

729

14.3.2 Endpoint Discovery

Figure 14.9 Endpoint Discovery Summary

730

14.3.2 Endpoint Discovery

Assume participants A and B have been discovered on both sides. A’s DiscoveryConfigQosPolicy.publication_writer.heartbeats_per_max_samples = 0, so no HB is piggybacked with the publication DATA. A HB with F=0 is a request for
an ACK/NACK. The periodic and initial repeat participant DATAs are omitted from the diagram.

Discovery endpoint writers and readers have their HISTORY QosPolicy (Section 6.5.10 on page 376) set
to KEEP_LAST, and their DURABILITY QosPolicy (Section 6.5.7 on page 368) set to TRANSIENT_
LOCAL. Therefore, even if the remote DomainParticipant has not yet been discovered at the time the
local user’s DataWriter/DataReader is created, the remote DomainParticipant will still be informed about
the previously created DataWriter/DataReader. This is achieved by the HB and ACK/NACK that are
immediately sent by the built-in endpoint writer and built-in endpoint reader respectively when a new
remote participant is discovered. Figure 14.10 DataWriter Discovered by Late-Joiner, Triggered by HB
below and Figure 14.11 DataWriter Discovered by Late-Joiner, Triggered by ACKNACK on the next
page illustrate this sequence for HB and ACK/NACK triggers, respectively.
Figure 14.10 DataWriter Discovered by Late-Joiner, Triggered by HB

Writer C is created on Participant A before Participant A discovers Participant B. Assuming DiscoveryConfigQosPolicy.publication_writer.heartbeats_per_max_samples = 0, no HB is piggybacked with the publication
DATA. Participant B has A in its peer list, but not vice versa. Accept_unknown_locators is true. On A, in response to
receiving the new participant B DATA message, a participant A DATA message is sent to B. The discovery endpoint

731

14.3.2 Endpoint Discovery

reader on A will also send an ACK/NACK to the discovery endpoint writer on B. (Initial repeat participant messages
and periodic participant messages are omitted from this diagram for simplicity, see Figure 14.3 Periodic ‘participant
DATAs’ on page 720 in Participant Discovery (Section 14.3.1 on page 718).)

Figure 14.11 DataWriter Discovered by Late-Joiner, Triggered by ACKNACK

Writer C is created on Participant A before Participant A discovers Participant B. Assuming DiscoveryConfigQosPolicy.publication_writer.heartbeats_per_max_samples = 0, no HB is piggybacked with the publication
DATA message. Participant A has B in its peer list, but not vice versa. Accept_unknown_locators is true. In response to
receiving the new Participant A DATA message on node B, a participant B DATA message will be sent to A. The discovery endpoint writer on Node B will also send a HB to the discovery endpoint reader on Node A. These are omitted
in the diagram for simplicity. (Initial repeat participant messages and periodic participant messages are omitted from
this diagram, see Figure 14.3 Periodic ‘participant DATAs’ on page 720 in Participant Discovery (Section 14.3.1
on page 718).)

Endpoint discovery latency is determined by the following members of the DomainParticipant’s
DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585):

732

14.3.3 Discovery Traffic Summary

l

publication_writer

l

subscription_writer

l

publication_reader

l

subscription_reader

When a remote entity record is added, removed, or changed in the database, matching is performed with
all the local entities. Only after there is a successful match on both ends can an application’s user-created
DataReaders and DataWriters communicate with each other.
For more information about reliable communication, see Reliable Communications (Section Chapter 10 on
page 629).

14.3.3 Discovery Traffic Summary

This diagram shows both phases of the discovery process. Participant A is created first, followed by Participant B.
Each has the other in its peers list. After they have discovered each other, a DataWriter is created on Participant A.
Periodic participant DATAs, HBs and ACK/NACKs are omitted from this diagram.

733

14.3.4 Discovery-Related QoS

14.3.4 Discovery-Related QoS
Each DomainParticipant needs to be uniquely identified in the DDS domain and specify which other
DomainParticipants it is interested in communicating with. The WIRE_PROTOCOL QosPolicy (DDS
Extension) (Section 8.5.9 on page 610) uniquely identifies a DomainParticipant in the DDS domain. The
DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580) specified the peer participants it
is interested in communicating with.
There is a trade-off between the amount of traffic on the network for the purposes of discovery and the
delay in reaching steady state when the DomainParticipant is first created.
For example, if the DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580)’s participant_liveliness_assert_period and participant_liveliness_lease_duration fields are set to small values, the
discovery of stale remote DomainParticipants will occur faster, but more discovery traffic will be sent
over the network. Setting the participant’s heartbeat_period1 to a small value can cause late-joining
DomainParticipants to discover remote user-data DataWriters and DataReaders at a faster rate, but Connext DDS might send HBs to other nodes more often. This timing can be controlled by the following
DomainParticipant QosPolicies:
l

l

l

l

DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580) — specifies how other
DomainParticipants in the network can communicate with this DomainParticipant, and which
other DomainParticipants in the network this DomainParticipant is interested in communicating
with. See also: Ports Used for Discovery (Section 14.5 on page 738).
DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585) — specifies the
QoS of the discovery readers and writers (parameters that control the HB and ACK rates of discovery endpoint readers/writers, and periodic refreshing of participant DATA from discovery participant readers/writers). It also allow you to configure asynchronous writers in order to send data
with a larger size than the transport message size.
DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4
on page 593) — specifies the number of local and remote entities expected in the system.
WIRE_PROTOCOL QosPolicy (DDS Extension) (Section 8.5.9 on page 610) — specifies the
rtps_app_id and rtps_host_id that uniquely identify the participant in the DDS domain.

The other important parameter is the domain ID: DomainParticipants can only discover each other if they
belong to the same DDS domain. The domain ID is a parameter passed to the create_participant() operation (see Creating a DomainParticipant (Section 8.3.1 on page 556)).

1heartbeat_period is part of the DDS_RtpsReliableWriterProtocol_t structure used in the DISCOVERY QosPolicy (DDS

Extension) (Section 8.5.2 on page 580)’s publication_writer and subscription_writer fields.

734

14.4 Debugging Discovery

14.4 Debugging Discovery
To understand the flow of messages during discovery, you can increase the verbosity of the messages
logged by Connext DDS so that you will see whenever a new entity is discovered, and whenever there is
a match between a local entity and a remote entity.
This can be achieved with the logging API:
NDDSConfigLogger::get_instance()->set_verbosity_by_category (NDDS_CONFIG_LOG_CATEGORY_
ENTITIES, NDDS_CONFIG_LOG_VERBOSITY_STATUS_REMOTE);

Using the scenario in the summary diagram in Discovery Traffic Summary (Section 14.3.3 on page 733),
these are the messages as seen on DomainParticipant A:
[D0049|ENABLE]DISCPluginManager_onAfterLocalParticipantEnabled:announcing new local
participant: 0XA0A01A1,0X5522,0X1,0X1C1
[D0049|ENABLE]DISCPluginManager_onAfterLocalParticipantEnabled:at {46c614d9,0C43B2DC}

(The above messages mean: First participant A DATA sent out when participant A is enabled.)
DISCSimpleParticipantDiscoveryPluginReaderListener_onDataAvailable:discovered new
participant: host=0x0A0A01A1, app=0x0000552B, instance=0x00000001
DISCSimpleParticipantDiscoveryPluginReaderListener_onDataAvailable:at {46c614dd,8FA13C1F}
DISCParticipantDiscoveryPlugin_assertRemoteParticipant:plugin discovered/updated remote
participant: 0XA0A01A1,0X552B,0X1,0X1C1
DISCParticipantDiscoveryPlugin_assertRemoteParticipant:at {46c614dd,8FACE677}
DISCParticipantDiscoveryPlugin_assertRemoteParticipant:plugin accepted new remote
participant: 0XA0A01A1,0X552B,0X1,0X1C1
DISCParticipantDiscoveryPlugin_assertRemoteParticipant:at {46c614dd,8FACE677}

(The above messages mean: Received participant B DATA.)
DISCSimpleParticipantDiscoveryPlugin_remoteParticipantDiscovered:re-announcing participant
self: 0XA0A01A1,0X5522,0X1,0X1C1
DISCSimpleParticipantDiscoveryPlugin_remoteParticipantDiscovered:at {46c614dd,8FC02AF7}

(The above messages mean: Resending participant A DATA to the newly discovered remote participant.)
PRESPsService_linkToLocalReader:assert remote 0XA0A01A1,0X552B,0X1,0X200C2, local 0x000200C7
in reliable reader service
PRESPsService_linkToLocalWriter:assert remote 0XA0A01A1,0X552B,0X1,0X200C7, local 0x000200C2
in reliable writer service

735

14.4 Debugging Discovery

PRESPsService_linkToLocalWriter:assert remote 0XA0A01A1,0X552B,0X1,0X4C7, local 0x000004C2 in
reliable writer service
PRESPsService_linkToLocalWriter:assert remote 0XA0A01A1,0X552B,0X1,0X3C7, local 0x000003C2 in
reliable writer service
PRESPsService_linkToLocalReader:assert remote 0XA0A01A1,0X552B,0X1,0X4C2, local 0x000004C7 in
reliable reader service
PRESPsService_linkToLocalReader:assert remote 0XA0A01A1,0X552B,0X1,0X3C2, local 0x000003C7 in
reliable reader service
PRESPsService_linkToLocalReader:assert remote 0XA0A01A1,0X552B,0X1,0X100C2, local 0x000100C7
in best effort reader service

(The above messages mean: Automatic matching of the discovery readers and writers. A built-in remote
endpoint's object ID always ends with Cx.)
DISCSimpleParticipantDiscoveryPluginReaderListener_onDataAvailable:discovered modified
participant: host=0x0A0A01A1, app=0x0000552B, instance=0x00000001
DISCParticipantDiscoveryPlugin_assertRemoteParticipant:plugin discovered/updated remote
participant: 0XA0A01A1,0X552B,0X1,0X1C1
DISCParticipantDiscoveryPlugin_assertRemoteParticipant:at {46c614dd,904D876C}

(The above messages mean: Received participant B DATA.)
DISCPluginManager_onAfterLocalEndpointEnabled:announcing new local publication:
0XA0A01A1,0X5522,0X1,0X80000003
DISCPluginManager_onAfterLocalEndpointEnabled:at {46c614d9,1013B9F0}
DISCSimpleEndpointDiscoveryPluginPDFListener_onAfterLocalWriterEnabled:announcing new
publication: 0XA0A01A1,0X5522,0X1,0X80000003
DISCSimpleEndpointDiscoveryPluginPDFListener_onAfterLocalWriterEnabled:at {46c614d9,101615EB}

(The above messages mean: Publication C DATA has been sent.)
DISCSimpleEndpointDiscoveryPlugin_subscriptionReaderListenerOnDataAvailable:discovered
subscription: 0XA0A01A1,0X552B,0X1,0X80000004
DISCSimpleEndpointDiscoveryPlugin_subscriptionReaderListenerOnDataAvailable:at
{46c614dd,94FAEFEF}
DISCEndpointDiscoveryPlugin_assertRemoteEndpoint:plugin discovered/updated remote endpoint:
0XA0A01A1,0X552B,0X1,0X80000004
DISCEndpointDiscoveryPlugin_assertRemoteEndpoint:at {46c614dd,950203DF}

736

14.4 Debugging Discovery

(The above messages mean: Receiving subscription D DATA from Node B.)
PRESPsService_linkToLocalWriter:assert remote 0XA0A01A1,0X552B,0X1,0X80000004, local
0x80000003 in best effort writer service

(The above message means: User-created DataWriter C and DataReader D are matched.)
[D0049|DELETE_CONTAINED]DISCPluginManager_onAfterLocalEndpointDeleted:announcing disposed
local publication: 0XA0A01A1,0X5522,0X1,0X80000003
[D0049|DELETE_CONTAINED]DISCPluginManager_onAfterLocalEndpointDeleted:at {46c61501,288051C8}
[D0049|DELETE_CONTAINED]DISCSimpleEndpointDiscoveryPluginPDFListener_
onAfterLocalWriterDeleted:announcing disposed publication: 0XA0A01A1,0X5522,0X1,0X80000003
[D0049|DELETE_CONTAINED]DISCSimpleEndpointDiscoveryPluginPDFListener_
onAfterLocalWriterDeleted:at {46c61501,28840E15}

(The above messages mean: Publication C DATA(delete) has been sent.)
DISCPluginManager_onBeforeLocalParticipantDeleted:announcing before disposed local
participant: 0XA0A01A1,0X5522,0X1,0X1C1
DISCPluginManager_onBeforeLocalParticipantDeleted:at {46c61501,28A11663}

(The above messages mean: Participant A DATA(delete) has been sent.)
DISCParticipantDiscoveryPlugin_removeRemoteParticipantsByCookie:plugin removing 3 remote
entities by cookie
DISCParticipantDiscoveryPlugin_removeRemoteParticipantsByCookie:at {46c61501,28E38A7C}
DISCParticipantDiscoveryPlugin_removeRemoteParticipantI:plugin discovered disposed remote
participant: 0XA0A01A1,0X552B,0X1,0X1C1
DISCParticipantDiscoveryPlugin_removeRemoteParticipantI:at {46c61501,28E68E3D}
DISCParticipantDiscoveryPlugin_removeRemoteParticipantI:remote entity removed from database:
0XA0A01A1,0X552B,0X1,0X1C1
DISCParticipantDiscoveryPlugin_removeRemoteParticipantI:at {46c61501,28E68E3D}

(The above messages mean: Removing discovered entities from local database, before shutting down.)
As you can see, the messages are encoded, since they are primarily used by RTI support personnel.
For more information on the message logging API, see Controlling Messages from Connext DDS (Section
21.2 on page 865).
If you notice that a remote entity is not being discovered, check the QoS related to discovery (see Discovery-Related QoS (Section 14.3.4 on page 734)).

737

14.5 Ports Used for Discovery

If a remote entity is discovered, but does not match with a local entity as expected, check the QoS of both
the remote and local entity.

14.5 Ports Used for Discovery
There are two kinds of traffic in a Connext DDS application: discovery (meta) traffic, and user traffic.
Meta-traffic is for data (declarations) that is sent between the automatically-created discovery writers and
readers; user traffic is for data that is sent between user-created DataWriters and DataReaders. To keep
the two kinds of traffic separate, Connext DDS uses different ports, as described below.
Note: The ports described in this section are used for incoming data. Connext DDS uses ephemeral ports
for outbound data.
Connext DDS uses the RTPS wire protocol. The discovery protocols defined by RTPS rely on wellknown ports to initiate discovery. These well-known ports define the multicast and unicast ports on which
a Participant will listen for meta-traffic from other Participants. The meta-traffic contains the information
required by Connext DDS to establish the presence of remote Entities in the network.
The well-known incoming ports are defined by RTPS in terms of port mapping expressions with several
tunable parameters. This allows you to customize what network ports are used for receiving data by Connext DDS. These parameters are shown in Table 14.2 WireProtocol QosPolicy’s rtps_well_known_ports
(DDS_RtpsWellKnownPorts_t). (For defaults and valid ranges, please see the API Reference HTML documentation.)
Table 14.2 WireProtocol QosPolicy’s rtps_well_known_ports (DDS_RtpsWellKnownPorts_t)
Type

Field Name
port_base
domain_id_gain
participant_id_gain

DDS_
Long

builtin_multicast_
port_offset
builtin_unicast_
port_offset

Description
The base port offset. All mapped well-known ports are offset by this value. Resulting ports must be within
the range imposed by the underlying transport.
Tunable gain parameters. See Tuning domain_id_gain and participant_id_gain (Section 14.5.4 on page
740).

Additional offset for meta-traffic port. See Inbound Ports for Meta-Traffic (Section 14.5.1 on the facing
page).

user_multicast_
port_offset
Additional offset for user traffic port. See Inbound Ports for User Traffic (Section 14.5.2 on page 740).
user_unicast_port_
offset

738

14.5.1 Inbound Ports for Meta-Traffic

In order for all Participants in a system to correctly discover each other, it is important that they all use the
same port mapping expressions.
In addition to the parameters listed in Table 14.2 WireProtocol QosPolicy’s rtps_well_known_ports
(DDS_RtpsWellKnownPorts_t), the port formulas described below depend on:
l

l

The domain ID specified when the DomainParticipant is created (see Creating a DomainParticipant
(Section 8.3.1 on page 556)). The domain ID ensures no port conflicts exist between Participants
belonging to different domains. This also means that discovery traffic in one DDS domain is not visible to DomainParticipants in other DDS domains.
The participant_id is a field in the WIRE_PROTOCOL QosPolicy (DDS Extension) (Section
8.5.9 on page 610), see Choosing Participant IDs (Section 8.5.9.1 on page 611). The participant_
id ensures that unique unicast port numbers are assigned to DomainParticipants belonging to the
same DDS domain on a given host.

Backwards Compatibility: Connext DDS supports the standard DDS Interoperability Wire Protocol
based on the Real-time Publish-Subscribe (RTPS) protocol. This protocol is not compatible with the one
used by earlier releases (4.2c or lower). Therefore, applications built with 4.2d or higher will not interoperate with applications built with 4.2c or lower. The default port mapping from domainID and participant index has also been changed according to the new interoperability specification. The message
types and formats used by RTPS have also changed.
Port Aliasing: When modifying the port mapping parameters, avoid port aliasing. This would result in
undefined discovery behavior. The chosen parameter values will also determine the maximum possible
number of DDS domains in the system and the maximum number of participants per DDS domain. Additionally, any resulting mapped port number must be within the range imposed by the underlying transport.
For example, for UDPv4, this range typically equals [1024 - 65535].

14.5.1 Inbound Ports for Meta-Traffic
The Wire Protocol QosPolicy’s rtps_well_known_ports.metatraffic_unicast_port determines the port
used for receiving meta-traffic using unicast:
metatraffic_unicast_port = port_base +
(domain_id_gain * Domain ID) +
(participant_id_gain * participant_id) +
builtin_unicast_port_offset

Similarly, rtps_well_known_ports.metatraffic_multicast_port determines the port used for receiving
meta-traffic using multicast. The corresponding multicast group addresses are specified via multicast_
receive_addresses (see Configuring Multicast Receive Addresses (Section 8.5.2.4 on page 582)).
metatraffic_multicast_port = port_base +
(domain_id_gain * Domain ID) +
builtin_multicast_port_offset

739

14.5.2 Inbound Ports for User Traffic

Note: Multicast is only used for meta-traffic if a multicast address is specified in the NDDS_
DISCOVERY_PEERS environment variable or file or if the multicast_receive_addresses field of the
DISCOVERY_CONFIG QosPolicy (DDS Extension) (Section 8.5.3 on page 585) is set.

14.5.2 Inbound Ports for User Traffic
RTPS also defines the default multicast and unicast ports on which DataReaders and DataWriters receive
user traffic. These default ports can be overridden using the DataReader’s TRANSPORT_MULTICAST
QosPolicy (DDS Extension) (Section 7.6.5 on page 529) and TRANSPORT_UNICAST QosPolicy
(DDS Extension) (Section 6.5.24 on page 412), or the DataWriter’s TRANSPORT_UNICAST
QosPolicy (DDS Extension) (Section 6.5.24 on page 412).
The WireProtocol QosPolicy’s rtps_well_known_ports.usertraffic_unicast_port determines the port
used for receiving user data using unicast:
usertraffic_unicast_port =
port_base +
(domain_id_gain * Domain ID) +
(participant_id_gain * participant_id)+
user_unicast_port_offset

Similarly, rtps_well_known_ports.usertraffic_multicast_port determines the port used for receiving
user data using multicast. The corresponding multicast group addresses can be configured using the
TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412).
usertraffic_multicast_port =
port_base +
(domain_id_gain * Domain ID) +
user_multicast_port_offset

14.5.3 Automatic Selection of participant_id and Port Reservation
The WIRE_PROTOCOL QosPolicy (DDS Extension) (Section 8.5.9 on page 610)rtps_reserved_
ports_mask field determines what type of ports are reserved when the DomainParticipant is enabled. See
Choosing Participant IDs (Section 8.5.9.1 on page 611).

14.5.4 Tuning domain_id_gain and participant_id_gain
The domain_id_gain is used as a multiplier of the domain ID. Together with participant_id_gain (Tuning domain_id_gain and participant_id_gain (Section 14.5.4 above)), these values determine the highest
domain ID and participant_id allowed on this network.
In general, there are two ways to set up the domain_id_gain and participant_id_gain parameters.
l

If domain_id_gain > participant_id_gain, it results in a port mapping layout where all
DomainParticipants in a DDS domain occupy a consecutive range of domain_id_gain ports.

740

14.5.4 Tuning domain_id_gain and participant_id_gain

Precisely, all ports occupied by the DDS domain fall within:
(port_base + (domain_id_gain * Domain ID))

and:
(port_base + (domain_id_gain * (Domain ID + 1)) - 1)

In this case, the highest domain ID is limited only by the underlying transport's maximum port. The
highest participant_id, however, must satisfy:
max_participant_id < (domain_id_gain / participant_id_gain)
l

Or if domain_id_gain <= participant_id_gain, it results in a port mapping layout where a given
DDS domain's DomainParticipant instances occupy ports spanned across the entire valid port range
allowed by the underlying transport. For instance, it results in the following potential mapping:
Mapped Port

Domain ID

Participant ID

1
2
0
higher port number

1
1
0
1
0

lower port number

0

In this case, the highest participant_id is limited only by the underlying transport's maximum port. The
highest domain_id, however, must satisfy:
max_domain_id < (participant_id_gain / domain_id_gain)

The domain_id_gain also determines the range of the port-specific offsets:
domain_id_gain >
abs(builtin_multicast_port_offset - user_multicast_port_offset)

and
domain_id_gain >
abs(builtin_unicast_port_offset - user_unicast_port_offset)

Violating this may result in port aliasing and undefined discovery behavior.

741

14.5.4 Tuning domain_id_gain and participant_id_gain

The participant_id_gain also determines the range of builtin_unicast_port_offset and user_unicast_
port_offset.
participant_id_gain >
abs(builtin_unicast_port_offset - user_unicast_port_offset)

In all cases, the resulting ports must be within the range imposed by the underlying transport.

742

Chapter 15 Transport Plugins
Connext DDS has a pluggable-transports architecture. The core of Connext DDS is transport
agnostic—it does not make any assumptions about the actual transports used to send and receive
messages. Instead, Connext DDS uses an abstract "transport API" to interact with the transport plugins that implement that API. A transport plugin implements the abstract transport API, and performs the actual work of sending and receiving messages over a physical transport.
There are essentially three categories of transport plugins:
l

l

l

Builtin Transport Plugins Connext DDS comes with a set of commonly used transport plugins. These ‘builtin’ plugins include UDPv4, UDPv6, and shared memory. So that Connext
DDS applications can work out-of-the-box, some of these are enabled by default (see
TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section 8.5.7 on page 606)).
Extension Transport Plugins RTI offers extension transports, including RTI Secure WAN
Transport (see Part 5: RTI Secure WAN Transport (Section on page 900) and RTI TCP
Transport (see Part 8: RTI TCP Transport (Section on page 987)).
Custom-developed Transport Plugins RTI supports the use of custom transport plugins.
This is a powerful capability that distinguishes Connext DDS from competing middleware
approaches. If you are interested in developing a custom transport plugin for Connext DDS,
please contact your local RTI representative or email sales@rti.com.

15.1 Builtin Transport Plugins
There are two ways in which the builtin transport plugins may be registered:
l

Default builtin Transport Instances: Builtin transports that are turned "on" in the
TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section 8.5.7 on page 606) are
implicitly registered when (a) the DomainParticipant is enabled, (b) the first DataWriter/DataReader is created, or (c) you look up a builtin DataReader (by calling lookup_
datareader() on a Subscriber), whichever happens first. The builtin transport plugins have

743

15.2 Extension Transport Plugins

default properties. If you want to change these properties, do so before1 the transports are registered.
l

Other Transport Instances: There are two ways to install non-default builtin transport instances:
l Transport plugins may be explicitly registered by first creating an instance of the transport plugin (by calling NDDS_Transport_UDPv4_new(), NDDS_Transport_UDPv6_new() or
NDDS_Transport_Shmem_new(), see Explicitly Creating Builtin Transport Plugin
Instances (Section 15.4 on page 746)), then calling register_transport() (Installing Additional
Builtin Transport Plugins with register_transport() (Section 15.7 on page 765)). (For example,
suppose you want an extra instance of a transport.) (Not available for the Java or .NET API.)
l

Additional builtin transport instances can also be installed through the PROPERTY
QosPolicy (DDS Extension) (Section 6.5.17 on page 394).

To configure the properties of the builtin transports:
l

Set properties by calling set_builtin_transport_property() (see Setting Builtin Transport Properties
of Default Transport Instance—get/set_builtin_transport_properties() (Section 15.5 on page 746))

or
l

Specify predefined property strings in the DomainParticipant’s PropertyQosPolicy, as described in
Setting Builtin Transport Properties with the PropertyQosPolicy (Section 15.6 on page 748).

For other builtin transport instances:
l

l

If the builtin transport plugin is created with NDDS_Transport_UDPv4_new(), NDDS_Transport_UDPv6_new() or NDDS_Transport_Shmem_new(), properties can be specified during creation time. See Explicitly Creating Builtin Transport Plugin Instances (Section 15.4 on page 746).
If the additional builtin transport instances are installed through the PROPERTY QosPolicy (DDS
Extension) (Section 6.5.17 on page 394), the properties of the builtin transport plugins can also be
specified through that same QosPolicy.

15.2 Extension Transport Plugins
If you want to change the properties for an extension transport plugin, do so before the plugin is registered.
Any transport property changes made after the plugin is registered will have no effect.
There are two ways to install an extension transport plugin:

1Any transport property changes made after the plugin is registered will have no effect.

744

15.3 The NDDSTransportSupport Class

l

Implicit Registration: Transports can be installed through the predefined strings in the DomainParticipant’s PropertyQosPolicy. Once the transport’s properties are specified in the PropertyQosPolicy, the transport will be implicitly registered when (a) the DomainParticipant is enabled,
(b) the first DataWriter/DataReader is created, or (c) you look up a builtin DataReader (by calling
lookup_datareader() on a Subscriber), whichever happens first.
QosPolicies can also be configured from XML resources (files, strings)—with this approach, you
can change the QoS without recompiling the application. The QoS settings are automatically loaded
by the DomainParticipantFactory when the first DomainParticipant is created. For more information, see Configuring QoS with XML (Section Chapter 17 on page 791).

l

Explicit Registration: Transports may be explicitly registered by first creating an instance of the
transport plugin (see Explicitly Creating Builtin Transport Plugin Instances (Section 15.4 on the next
page)) and then calling register_transport() (see Installing Additional Builtin Transport Plugins with register_transport() (Section 15.7 on page 765)).

15.3 The NDDSTransportSupport Class
The register_transport() and set_builtin_transport_property() operations are part of the NDDSTransportSupport class, which includes the operations listed in Table 15.1 Transport Support Operations.
Table 15.1 Transport Support Operations
Operation

Description

Reference

get_transport_
plugin

Retrieves a previously registered
transport plugin.

register_transport

Registers a transport plugin for use
with a DomainParticipant.

get_builtin_
transport_
property

Gets the properties used to create a
builtin transport plugin.

set_builtin_
transport_
property

Sets the properties used to create a
builtin transport plugin.

add_send_route

Adds a route for outgoing messages. Adding a Send Route (Section 15.9.1 on page 769)

add_receive_
route

Adds a route for incoming
messages.

lookup_transport

Looks up a transport plugin within a
Looking Up a Transport Plugin (Section 15.9.3 on page 771)
DomainParticipant.

Installing Additional Builtin Transport Plugins with register_transport() (Section
15.7 on page 765)

Setting Builtin Transport Properties of Default Transport Instance—get/set_
builtin_transport_properties() (Section 15.5 on the next page)

Adding a Receive Route (Section 15.9.2 on page 770)

745

15.4 Explicitly Creating Builtin Transport Plugin Instances

15.4 Explicitly Creating Builtin Transport Plugin Instances
The builtin transports (UDPv4, UDPv6, and Shared Memory) are implicitly created by default (if they are
enabled via the TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section 8.5.7 on page 606)).
Therefore, you only need to explicitly create a new instance if you want an extra instance (suppose you
want two UDPv4 transports, one with special settings).
Transport plugins may be explicitly registered by first creating an instance of the transport plugin and then
calling register_transport() (Installing Additional Builtin Transport Plugins with register_transport() (Section 15.7 on page 765)). (For example, suppose you want an extra instance of a transport.) (Not available
for the Java API.)
To create an instance of a builtin transport plugin, use one of the following functions:
NDDS_Transport_Plugin* NDDS_Transport_UDPv4_new (
const struct NDDS_Transport_UDPv4_Property_t * property_in)
NDDS_Transport_Plugin* NDDS_Transport_UDPv4_new (
const struct NDDS_Transport_UDPv4_Property_t * property_in)
NDDS_Transport_Plugin* NDDS_Transport_Shmem_new (
const struct NDDS_Transport_Shmem_Property_t * property_in)

Where:
property_in

Desired behavior of this transport. May be NULL for default properties.

For details on using these functions, please see the API Reference HTML documentation.
Your application may create and register multiple instances of these transport plugins with Connext DDS.
This may be done to partition the network interfaces across multiple DDS domains. However, note that the
underlying transport, the operating system's IP layer, is still a "singleton." For example, if a unicast transport has already bound to a port, and another unicast transport tries to bind to the same port, the second
attempt will fail.

15.5 Setting Builtin Transport Properties of Default Transport Instance—
get/set_builtin_transport_properties()
Perhaps you want to use one of the builtin transports, but need to modify the properties. (For default values, please see the API Reference HTML documentation.) Used together, the two operations below allow
you to customize properties of the builtin transport when it is implicitly registered (see Builtin Transport
Plugins (Section 15.1 on page 743)).
Note: Another way to change the properties is with the Property QosPolicy, see Setting Builtin Transport
Properties with the PropertyQosPolicy (Section 15.6 on page 748). Changing properties with the Property
QosPolicy will overwrite the properties set by calling set_builtin_transport_property().
DDS_ReturnCode_t
NDDSTransportSupport::get_builtin_transport_property(

746

15.5 Setting Builtin Transport Properties of Default Transport Instance—get/set_builtin_transport_

DDSDomainParticipant * participant_in,
DDS_TransportBuiltinKind builtin_transport_kind_in,
struct NDDS_Transport_Property_t
&builtin_transport_property_inout)
DDS_ReturnCode_t
NDDSTransportSupport::set_builtin_transport_property(
DDSDomainParticipant * participant_in,
DDS_TransportBuiltinKind builtin_transport_kind_in,
const struct NDDS_Transport_Property_t
&builtin_transport_property_in)

Where:
participant_in

A valid non-NULL DomainParticipant that has not been enabled. If the
DomainParticipant if already enabled when this operation is called, your transport
property changes will not be reflected in the transport used by the
DomainParticipant's DataWriters and DataReaders.

builtin_transport_kind_ The builtin transport kind for which to specify the properties.
in
builtin_transport_
property_inout

(Used by the “get” operation only.) The storage area where the retrieved property will
be output. The specific type required by the builtin_transport_kind_in must be
used.

builtin_transport_
property_in

(Used by the “set” operation only.) The new transport property that will be used to
the create the builtin transport plugin. The specific type required by the builtin_
transport_kind_in must be used.

In this example, we want to use the builtin UDPv4 transport, but with modified properties.
/* Before this point, create a disabled DomainParticipant */
struct NDDS_Transport_UDPv4_Property_t property =
NDDS_TRANSPORT_UDPV4_PROPERTY_DEFAULT;
if (NDDSTransportSupport::get_builtin_transport_property(
participant, DDS_TRANSPORTBUILTIN_UDPv4,
(struct NDDS_Transport_Property_t&)property) !=
DDS_RETCODE_OK) {
printf("**Error: get builtin transport property\n");
}
/* Make your desired changes here */
/* For example, to increase the UDPv4 max msg size to 64K: */
property.parent.message_size_max = 65535;
property.recv_socket_buffer_size = 65535;
property.send_socket_buffer_size = 65535;
if (NDDSTransportSupport::set_builtin_transport_property(
participant, DDS_TRANSPORTBUILTIN_UDPv4,
(struct NDDS_Transport_Property_t&)property)
!= DDS_RETCODE_OK) {
printf("***Error: set builtin transport property\n");
}
/* Enable the participant to turn on communications with
other participants in the DDS domain using the new

747

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

properties for the automatically registered builtin
transport plugins */
if (entity->enable() != DDS_RETCODE_OK) {
printf("***Error: failed to enable entity\n");
}

Note:Builtin transport property changes will have no effect after the builtin transport has been registered.
The builtin transports are implicitly registered when (a) the DomainParticipant is enabled, (b) the first
DataWriter/DataReader is created, or (c) you lookup a builtin DataReader, whichever happens first.

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy
The PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394) allows you to set name/value pairs of data and attach them to an entity, such as a DomainParticipant.
To assign properties, use the add_property() operation:
DDS_ReturnCode_t DDSPropertyQosPolicyHelper::add_property
(DDS_PropertyQosPolicy policy,
const char * name,
const char * value,
DDS_Boolean propagate)

For more information on add_property() and the other operations in the DDSPropertyQosPolicyHelper
class, please see Table 6.57 PropertyQoSPolicyHelper Operations, as well as the API Reference HTML
documentation.
The ‘name’ part of the name/value pairs is a predefined string. The property names for the builtin transports are described in these tables:
l

Table 15.2 Properties for the Builtin UDPv4 Transport

l

Table 15.3 Properties for Builtin UDPv6 Transport

l

Table 15.4 Properties for Builtin Shared-Memory Transport

See also:
l

l

l

Setting the Maximum Gather-Send Buffer Count for UDPv4 and UDPv6 (Section 15.6.1 on page
763)
Setting the Maximum Gather-Send Buffer Count for UDPv4 and UDPv6 (Section 15.6.1 on page
763)
Formatting Rules for IPv6 ‘Allow’ and ‘Deny’ Address Lists (Section 15.6.2 on page 765)

Note:

748

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Changing properties with the PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394)
will overwrite any properties set by calling set_builtin_transport_property().
Table 15.2 Properties for the Builtin UDPv4 Transport
Property Name
(prefix with
‘dds.transport.UDPv4.builtin.’)

Property Value Description

Number of bits in a 16-byte address that are used by the transport. Should be between 0 and
128.
parent.address_bit_count
For example, for an address range of 0-255, the address_bit_count should be set to 8. For the
range of addresses used by IPv4 (4 bytes), it should be set to 32.
A bitmap that defines various properties of the transport to the Connext DDS core.
parent.properties_bitmap

Currently, the only property supported is whether or not the transport plugin will always loan a
buffer when Connext DDS tries to receive a message using the plugin. This is in support of a
zero-copy interface.
Specifies the maximum number of buffers that Connext DDS can pass to the send() method of a
transport plugin.
The transport plugin send() API supports a gather-send concept, where the send() call can take
several discontiguous buffers, assemble and send them in a single message. This enables
Connext DDS to send a message from parts obtained from different sources without first having
to copy the parts into a single contiguous buffer.

parent.
gather_send_buffer_count_max

However, most transports that support a gather-send concept have an upper limit on the number
of buffers that can be gathered and sent. Setting this value will prevent Connext DDS from
trying to gather too many buffers into a send call for the transport plugin.
Connext DDS requires all transport-plugin implementations to support a gather-send of least a
minimum number of buffers. This minimum number is NDDS_TRANSPORT_PROPERTY_
GATHER_SEND_BUFFER_COUNT_MIN.
See Setting the Maximum Gather-Send Buffer Count for UDPv4 and UDPv6 (Section 15.6.1
on page 763).
The maximum size of a message in bytes that can be sent or received by the transport plugin.

parent.message_size_max

This value must be set before the transport plugin is registered, so that Connext DDS can
properly use the plugin.

749

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.2 Properties for the Builtin UDPv4 Transport
Property Name
Property Value Description

(prefix with
‘dds.transport.UDPv4.builtin.’)

A list of strings, each identifying a range of interface addresses or an interface name. Interfaces
must be specified as comma-separated strings, with each comma delimiting an interface.
For example, the following are acceptable strings:
192.168.1.1
192.168.1.*
192.168.*
192.*
parent.allow_interfaces_list

ether0
If the list is non-empty, this "white" list is applied before the parent.deny_interfaces_list (Section
below) list. The DomainParticipant will use the resulting list of interfaces to inform its remote
participant(s) about which unicast addresses may be used to contact the DomainParticipant.
The resulting list restricts reception to a particular set of interfaces for unicast UDP. Multicast
output will still be sent and may be received over the interfaces in the list (if multicast is
supported on the platform).
You must manage the memory of the list. The memory may be freed after the
DomainParticipant is deleted.
A list of strings, each identifying a range of interface addresses or an interface name. If the list is
non-empty, deny the use of these interfaces.
Interfaces must be specified as comma-separated strings, with each comma delimiting an
interface.
For example, the following are acceptable strings:
192.168.1.1
192.168.1.*
192.168.*

parent.deny_interfaces_list

192.*
ether0
This "black" list is applied after the parent.allow_interfaces_list (Section above) list and filters
out the interfaces that should not be used for receiving data.
The resulting list restricts reception to a particular set of interfaces for unicast UDP. Multicast
output will still be sent and may be received over the interfaces in the list (if multicast is
supported on the platform).
You must manage the memory of the list. The memory may be freed after the
DomainParticipant is deleted.

750

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.2 Properties for the Builtin UDPv4 Transport
Property Name
(prefix with
‘dds.transport.UDPv4.builtin.’)

Property Value Description

A list of strings, each identifying a range of interface addresses or an interface name. If the list is
non-empty, allow the use of multicast only on these interfaces. If the list is empty, allow the use
of all the allowed interfaces.
Interfaces must be specified as comma-separated strings, with each comma delimiting an
interface.
parent.
allow_multicast_interfaces_list

This list sub-selects from the allowed interfaces that are obtained after applying the parent.allow_
interfaces_list (Section on the previous page) "white" list and the parent.deny_interfaces_list
(Section on the previous page) "black" list. From that resulting list, parent.deny_multicast_
interfaces_list (Section below) is applied. Multicast output will be sent and may be received over
the interfaces in the resulting list (if multicast is supported on the platform).
If this list is empty, all the allowed interfaces may potentially be used for multicast.
You must manage the memory of the list. The memory may be freed after the
DomainParticipant is deleted.
A list of strings, each identifying a range of interface addresses or an interface name. If the list is
non-empty, deny the use of those interfaces for multicast.
Interfaces should be specified as comma-separated strings, with each comma delimiting an
interface.

parent.
deny_multicast_interfaces_list

This "black" list is applied after the parent.allow_multicast_interfaces_list (Section above) list
and filters out the interfaces that should not be used for multicast. The final resulting list will be
those interfaces that—if multicast is available—will be used for multicast sends.
You must manage the memory of the list. The memory may be freed after the
DomainParticipant is deleted.
Size in bytes of the send buffer of a socket used for sending. On most operating systems,
setsockopt() will be called to set the SENDBUF to the value of this parameter.

send_socket_buffer_size

This value must be greater than or equal to the property,
parent.message_size_max (Section on page 749).
The maximum value is operating system-dependent.
If NDDS_TRANSPORT_UDPV4_SOCKET_BUFFER_SIZE_OS_DEFAULT, then
setsockopt() (or equivalent) will not be called to size the send buffer of the socket.

751

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.2 Properties for the Builtin UDPv4 Transport
Property Name
(prefix with
‘dds.transport.UDPv4.builtin.’)

Property Value Description

Size in bytes of the receive buffer of a socket used for receiving.
On most operating systems, setsockopt() will be called to set the RECVBUF to the value of this
parameter.
recv_socket_buffer_size

This value must be greater than or equal to the property, parent.message_size_max (Section on
page 749). The maximum value is operating system-dependent.
Default: NDDS_TRANSPORT_UDPV4_MESSAGE_SIZE_MAX_DEFAULT.
If NDDS_TRANSPORT_UDPV4_SOCKET_BUFFER_SIZE_OS_DEFAULT, then
setsockopt() (or equivalent) will not be called to size the receive buffer of the socket.

unicast_enabled

Allows the transport plugin to use unicast UDP for sending and receiving. By default, it will be
turned on. Also by default, it will use all the allowed network interfaces that it finds up and
running when the plugin is instanced.
Can be 1 (enabled) or 0 (disabled).

multicast_enabled

Allows the transport plugin to use multicast for sending and receiving. You can turn multicast
on or off for this plugin. The default is that multicast is on and the plugin will use the all network
interfaces allowed for multicast that it finds up and running when the plugin is instanced.
Can be 1 (enabled) or 0 (disabled).

multicast_ttl

Value for the time-to-live parameter for all multicast sends using this plugin. This is used to set
the TTL of multicast packets sent by this transport plugin.
Prevents the transport plugin from putting multicast packets onto the loopback interface.
If disabled, then when sending multicast packets, do not put a copy on the loopback interface.
This will prevent other applications on the same node (including itself) from receiving those
packets.

multicast_loopback_disabled
This is set to 0 by default. So multicast loopback is enabled. Turning off multicast loopback (set
to 1) may result in minor performance gains when using multicast.
Note: Windows CE does not support multicast loopback. This field is ignored for Windows CE
targets.

752

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.2 Properties for the Builtin UDPv4 Transport
Property Name
Property Value Description

(prefix with
‘dds.transport.UDPv4.builtin.’)

Prevents the transport plugin from using the IP loopback interface. Three values are allowed:
l

0: Forces local traffic to be sent over loopback, even if a more efficient transport (such
as shared memory) is installed (in which case traffic will be sent over both transports).
l

1: Disables local traffic via this plugin. The IP loopback interface will not be used, even
if no NICs are discovered. This is useful when you want applications running on the
same node to use a more efficient transport (such as shared memory) instead of the IP
loopback.

ignore_loopback_interface

l

-1: Automatic. Enables local traffic via this plugin. To avoid redundant traffic, Connext
DDS will selectively ignore the loopback destinations that are also reachable through
shared memory.
This property is only supported on Windows platforms with statically configured IP addresses.
It allows/disallows the use of interfaces that are not reported as UP (by the operating system) in
the UDPv4 transport. Two values are allowed:
l

0: Allow interfaces that are reported as DOWN.
Setting this value to 0 supports communication scenarios in which interfaces are enabled
after the participant is created. Once the interfaces are enabled, discovery will not occur
until the participant sends the next periodic announcement (controlled by the parameter
participant_qos.discovery_config.participant_liveliness_
assert_period).

ignore_nonup_interfaces

To reduce discovery time, you may want to decrease the value of participant_
liveliness_assert_period.
For the above scenario, there is one caveat: non-UP interfaces must have a static IP
assigned.
l

1 (default): Do not allow interfaces that are reported as DOWN.

interface_poll_period

If ignore_nonup_interfaces is 0, the UDPv4 transport creates a new thread to query the status
of the interfaces. The interface_poll_period specifies the polling period in milliseconds for
performing this query.
This property’s value is ignored if ignore_nonup_interfaces is 1.

753

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.2 Properties for the Builtin UDPv4 Transport
Property Name
Property Value Description

(prefix with
‘dds.transport.UDPv4.builtin.’)

Prevents the transport plugin from using a network interface that is not reported as RUNNING
by the operating system.

ignore_nonrunning_interfaces

The transport checks the flags reported by the operating system for each network interface upon
initialization. An interface which is not reported as UP will not be used. This property allows the
same check to be extended to the IFF_RUNNING flag implemented by some operating systems.
The RUNNING flag is defined to mean that "all resources are allocated", and may be off if there
is no link detected, e.g., the network cable is unplugged. Two values are allowed:
l

0: Do not check the RUNNING flag when enumerating interfaces, just make sure the
interface is UP.
l

1: Check the flag when enumerating interfaces, and ignore those that are not reported as
RUNNING. This can be used on some operating systems to cause the transport to
ignore interfaces that are enabled but not connected to the network.
Prevents the transport plugin from doing a zero copy.
By default, this plugin will use the zero copy on OSs that offer it. While this is good for
performance, it may sometime tax the OS resources in a manner that cannot be overcome by the
application.
no_zero_copy

The best example is if the hardware/device driver lends the buffer to the application itself. If the
application does not return the loaned buffers soon enough, the node may error or malfunction.
In case you cannot reconfigure the hardware, device driver, or the OS to allow the zero-copy
feature to work for your application, you may have no choice but to turn off zero-copy.
By default this is set to 0, so Connext DDS will use the zero-copy API if offered by the OS.
Controls the blocking behavior of send sockets. CHANGING THIS FROM THE DEFAULT
CAN CAUSE SIGNIFICANT PERFORMANCE PROBLEMS. Currently two values are
defined:

send_blocking

NDDS_TRANSPORT_UDPV4_BLOCKING_ALWAYS: Sockets are blocking (default
socket options for operating system).
NDDS_TRANSPORT_UDPV4_BLOCKING_NEVER: Sockets are modified to make them
non-blocking. This is not a supported configuration and may cause significant
performance problems.

754

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.2 Properties for the Builtin UDPv4 Transport
Property Name
(prefix with
‘dds.transport.UDPv4.builtin.’)

transport_priority_mask

Property Value Description

Mask for the transport priority field. This is used in conjunction with transport_priority_
mapping_low (Section below) and transport_priority_mapping_high (Section below) to define
the mapping from the TRANSPORT_PRIORITY QosPolicy (Section 6.5.22 on page 409) to
the IPv4 TOS field. Defines a contiguous region of bits in the 32-bit transport priority value that
is used to generate values for the IPv4 TOS field on an outgoing socket.
For example, the value 0x0000ff00 causes bits 9-16 (8 bits) to be used in the mapping. The
value will be scaled from the mask range (0x0000 - 0xff00 in this case) to the range specified by
low and high.
If the mask is set to zero, then the transport will not set IPv4 TOS for send sockets.
Sets the low and high values of the output range to IPv4 TOS.

transport_priority_mapping_low
These values are used in conjunction with transport_priority_mask (Section above) to define the
mapping from the TRANSPORT_PRIORITY QosPolicy (Section 6.5.22 on page 409) to the
IPv4 TOS field. Defines the low and high values of the output range for scaling.
transport_priority_mapping_high
Note that IPv4 TOS is generally an 8-bit value.

reuse_multicast_receive_resource

Controls whether or not to reuse receive resources. Setting this to 0 (FALSE) prevents multicast
crosstalk by uniquely configuring a port and creating a receive thread for each multicast group
address.
Affects Linux systems only; ignored for non-Linux systems.
Maximum size in bytes of protocol overhead, including headers.

protocol_overhead_max

This value is the maximum size, in bytes, of protocol-related overhead. Normally, the overhead
accounts for UDP and IP headers. The default value is set to accommodate the most common
UDP/IP header size.
Note that when parent.message_size_max (Section on page 749) plus this overhead is larger
than the UDPv4 maximum message size (65535 bytes), the middleware will automatically
reduce the effective message_size_max to 65535 minus this overhead.

755

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.2 Properties for the Builtin UDPv4 Transport
Property Name
Property Value Description

(prefix with
‘dds.transport.UDPv4.builtin.’)

Public IP address associated with the transport instantiation.
Setting the public IP address is only necessary to support communication over WAN that
involves Network Address Translation (NAT).
Typically, the address is the public address of the IP NAT router that provides access to the
WAN.
By default, the DomainParticipant creating the transport will announce the IP addresses
obtained from the NICs to other DomainParticipants in the system.
When this property is set, the DomainParticipant will announce the IP address corresponding to
the property value instead of the LAN IP addresses associated with the NICs.
Notes:

public_address

Setting this property is necessary, but is not a sufficient condition for sending and receiving data
over the WAN. You must also configure the IP NAT router to allow UDP traffic and to map the
public IP address specified by this property to the DomainParticipant's private LAN IP address.
This is typically done with one of these mechanisms:
Port Forwarding: You must map the private ports used to receive discovery and user data
traffic to the corresponding public ports (see Table 8.20 DDS_RtpsWellKnownPorts_t). Public
and private ports must be the same since the transport does not allow you to change the
mapping.
1:1 NAT: You must add a 1:1 NAT entry that maps the public IP address specified in this
property to the private LAN IP address of the DomainParticipant.
By setting this property, the DomainParticipant only announces its public IP address to other
DomainParticipants. Therefore, communication with DomainParticipants within the LAN that
are running on different nodes will not work unless the NAT router is configured to enable
NAT reflection (hairpin NAT).
There is another way to achieve simultaneous communication with DomainParticipants running
in the LAN and WAN, that does not require hairpin NAT. This way uses a gateway application
such as RTI Routing Service to provide access to the WAN.

756

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.3 Properties for Builtin UDPv6 Transport
Property Name (prefix with
‘dds.transport.UDPv6.builtin.’)

Description
Number of bits in a 16-byte address that are used by the transport. Should be between 0 and
128.

parent.address_bit_count
For example, for an address range of 0-255, this address_bit_count should be set to 8. For the
range of addresses used by IPv4 (4 bytes), it should be set to 32.
A bitmap that defines various properties of the transport to the Connext DDS core.
parent.properties_bitmap

Currently, the only property supported is whether or not the transport plugin will always loan a
buffer when Connext DDS tries to receive a message using the plugin. This is in support of a
zero-copy interface.
Specifies the maximum number of buffers that Connext DDS can pass to the send() method of a
transport plugin.

parent.gather_send_buffer_
count_max

The transport plugin send() API supports a gather-send concept, where the send() call can take
several discontiguous buffers, assemble and send them in a single message. This enables
Connext DDS to send a message from parts obtained from different sources without first having
to copy the parts into a single contiguous buffer.
However, most transports that support a gather-send concept have an upper limit on the number
of buffers that can be gathered and sent. Setting this value will prevent Connext DDS from
trying to gather too many buffers into a send call for the transport plugin.
Connext DDS requires all transport-plugin implementations to support a gather-send of least a
minimum number of buffers. This minimum number is NDDS_TRANSPORT_PROPERTY_
GATHER_SEND_BUFFER_COUNT_MIN.
The maximum size of a message in bytes that can be sent or received by the transport plugin.

parent.message_size_max

This value must be set before the transport plugin is registered, so that Connext DDS can
properly use the plugin.
A list of strings, each identifying a range of interface addresses or an interface name.
Interfaces must be specified as comma-separated strings, with each comma delimiting an
interface. See Formatting Rules for IPv6 ‘Allow’ and ‘Deny’ Address Lists (Section 15.6.2 on
page 765).

parent.allow_interfaces_list

If the list is non-empty, this "white" list is applied before the parent.deny_interfaces_list (Section
on the next page) list. The DomainParticipant will use the resulting list of interfaces to inform
its remote participant(s) about which unicast addresses may be used to contact the
DomainParticipant.
The resulting list restricts reception to a particular set of interfaces for unicast UDP. Multicast
output will still be sent and may be received over the interfaces in the list (if multicast is
supported on the platform).
You must manage the memory of the list. The memory may be freed after the
DomainParticipant is deleted.

757

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.3 Properties for Builtin UDPv6 Transport
Property Name (prefix with
‘dds.transport.UDPv6.builtin.’)

Description
A list of strings, each identifying a range of interface addresses or an interface name. If the list is
non-empty, deny the use of these interfaces.
Interfaces must be specified as comma-separated strings, with each comma delimiting an
interface. See Formatting Rules for IPv6 ‘Allow’ and ‘Deny’ Address Lists (Section 15.6.2 on
page 765).

parent.deny_interfaces_list

This "black" list is applied after the parent.allow_interfaces_list (Section on the previous page)
list and filters out the interfaces that should not be used.
The resulting list restricts reception to a particular set of interfaces for unicast UDP. Multicast
output will still be sent and may be received over the interfaces in the list (if multicast is
supported on the platform).
You must manage the memory of the list. The memory may be freed after the
DomainParticipant is deleted.
A list of strings, each identifying a range of interface addresses or an interface name. If the list is
non-empty, allow the use of multicast only these interfaces; otherwise allow the use of all the
allowed interfaces.
Interfaces must be specified as comma-separated strings, with each comma delimiting an
interface. See Formatting Rules for IPv6 ‘Allow’ and ‘Deny’ Address Lists (Section 15.6.2 on
page 765).

parent.
allow_multicast_interfaces_list

This list sub-selects from the allowed interfaces that are obtained after applying the parent.allow_
interfaces_list (Section on the previous page) "white" list and the parent.deny_interfaces_list
(Section above) "black" list. Finally, the parent.deny_multicast_interfaces_list (Section below) is
applied. Multicast output will be sent and may be received over the interfaces in the resulting list
(if multicast is supported on the platform).
If this list is empty, all the allowed interfaces may potentially be used for multicast.
You must manage the memory of the list. The memory may be freed after the
DomainParticipant is deleted.
A list of strings, each identifying a range of interface addresses or an interface name. If the list is
non-empty, deny the use of those interfaces for multicast.
Interfaces must be specified as comma-separated strings, with each comma delimiting an
interface. See Formatting Rules for IPv6 ‘Allow’ and ‘Deny’ Address Lists (Section 15.6.2 on
page 765).

parent.
deny_multicast_interfaces_list

This "black" list is applied after the parent.allow_multicast_interfaces_list (Section above) list
and filters out the interfaces that should not be used for multicast. Multicast output will be sent
and may be received over the interfaces in the resulting list (if multicast is supported on the
platform).
You must manage the memory of the list. The memory may be freed after the
DomainParticipant is deleted.

758

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.3 Properties for Builtin UDPv6 Transport
Property Name (prefix with
‘dds.transport.UDPv6.builtin.’)

Description
Size in bytes of the send buffer of a socket used for sending.
On most operating systems, setsockopt() will be called to set the SENDBUF to the value of this
parameter.

send_socket_buffer_size

This value must be greater than or equal to parent.message_size_max. The maximum value is
operating system-dependent.
If NDDS_TRANSPORT_UDPV6_SOCKET_BUFFER_SIZE_OS_DEFAULT, then
setsockopt() (or equivalent) will not be called to size the send buffer of the socket.
Size in bytes of the receive buffer of a socket used for receiving.
On most operating systems, setsockopt() will be called to set the RECVBUF to the value of this
parameter.

recv_socket_buffer_size

This value must be greater than or equal to parent.message_size_max. The maximum value is
operating system-dependent.
If NDDS_TRANSPORT_UDPV6_SOCKET_BUFFER_SIZE_OS_DEFAULT, then
setsockopt() (or equivalent) will not be called to size the receive buffer of the socket.

unicast_enabled

Allows the transport plugin to use unicast UDP for sending and receiving. By default, it will be
turned on (1). Also by default, it will use all the allowed network interfaces that it finds up and
running when the plugin is instanced.
Can be 1 (enabled) or 0 (disabled).
Allows the transport plugin to use multicast for sending and receiving.

multicast_enabled

You can turn multicast UDP on or off for this plugin. By default, it will be turned on (1). Also
by default, it will use the all network interfaces allowed for multicast that it finds up and running
when the plugin is instanced.
Can be 1 (enabled) or 0 (disabled).
Value for the time-to-live parameter for all multicast sends using this plugin.

multicast_ttl
This is used to set the TTL of multicast packets sent by this transport plugin
Prevents the transport plugin from putting multicast packets onto the loopback interface.

multicast_loopback_disabled

If disabled, then when sending multicast packets, Connext DDS will not put a copy on the
loopback interface. This will prevent applications on the same node (including itself) from
receiving those packets.
This is set to 0 by default, meaning multicast loopback is enabled. Disabling multicast loopback
off (setting this value to 1) may result in minor performance gains when using multicast.

759

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.3 Properties for Builtin UDPv6 Transport
Property Name (prefix with
‘dds.transport.UDPv6.builtin.’)

Description
Prevents the transport plugin from using the IP loopback interface. Three values are allowed:
l

0: Enable local traffic via this plugin. This plugin will only use and report the IP
loopback interface if there are no other network interfaces (NICs) up on the system.
l

1: Disable local traffic via this plugin. Do not use the IP loopback interface even if no
NICs are discovered. This is useful when you want applications running on the same
node to use a more efficient plugin like Shared Memory instead of the IP loopback.

ignore_loopback_interface

l

-1: Automatic. Enables local traffic via this plugin. To avoid redundant traffic, Connext
DDS will selectively ignore the loopback destinations that are also reachable through
shared memory.
Prevents the transport plugin from using a network interface that is not reported as RUNNING
by the operating system.

ignore_nonrunning_interfaces

The transport checks the flags reported by the operating system for each network interface upon
initialization. An interface which is not reported as UP will not be used. This property allows the
same check to be extended to the IFF_RUNNING flag implemented by some operating systems.
The RUNNING flag is defined to mean that "all resources are allocated", and may be off if there
is no link detected, e.g., the network cable is unplugged. Two values are allowed:
l

0: Do not check the RUNNING flag when enumerating interfaces, just make sure the
interface is UP.
l

1: Check the flag when enumerating interfaces, and ignore those that are not reported as
RUNNING. This can be used on some operating systems to cause the transport to
ignore interfaces that are enabled but not connected to the network.
Prevents the transport plugin from doing a zero copy.
By default, this plugin will use the zero copy on OSs that offer it. While this is good for
performance, it may sometime tax the OS resources in a manner that cannot be overcome by the
application.
no_zero_copy

The best example is if the hardware/device driver lends the buffer to the application itself. If the
application does not return the loaned buffers soon enough, the node may error or malfunction.
In case you cannot reconfigure the H/W, device driver, or the OS to allow the zero-copy feature
to work for your application, you may have no choice but to turn off zero-copy.
By default this is set to 0, so Connext DDS will use the zero-copy API if offered by the OS.

760

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.3 Properties for Builtin UDPv6 Transport
Property Name (prefix with
‘dds.transport.UDPv6.builtin.’)

Description
Controls the blocking behavior of send sockets. CHANGING THIS FROM THE DEFAULT
CAN CAUSE SIGNIFICANT PERFORMANCE PROBLEMS. Currently two values are
defined:
l

NDDS_TRANSPORT_UDPV4_BLOCKING_ALWAYS: Sockets are blocking
(default socket options for Operating System).

send_blocking
l

NDDS_TRANSPORT_UDPV4_BLOCKING_NEVER: Sockets are modified to
make them non-blocking. This is not a supported configuration and may cause
significant performance problems.
Specifies whether the UDPv6 transport will process IPv4 addresses.
enable_v4mapped

Set this to 1 to turn on processing of IPv4 addresses. Note that this may make it incompatible
with use of the UDPv4 transport within the same DomainParticipant.
Sets a mask for use of transport priority field.
If transport priority mapping is supported on the platform1, this mask is used in conjunction
with transport_priority_mapping_low (Section below) and transport_priority_mapping_high
(Section below) to define the mapping from the DDS transport priority TRANSPORT_
PRIORITY QosPolicy (Section 6.5.22 on page 409) to the IPv6 TCLASS field.

transport_priority_mask

Defines a contiguous region of bits in the 32-bit transport priority value that is used to generate
values for the IPv6 TCLASS field on an outgoing socket.
For example, the value 0x0000ff00 causes bits 9-16 (8 bits) to be used in the mapping. The
value will be scaled from the mask range (0x0000 - 0xff00 in this case) to the range specified by
low and high.
If the mask is set to zero, then the transport will not set IPv6 TCLASS for send sockets.
Sets the low and high values of the output range to IPv6 TCLASS.

transport_priority_mapping_low
These values are used in conjunction with transport_priority_mask (Section above) to define the
mapping from DDS transport priority to the IPv6 TCLASS field. Defines the low and high
values of the output range for scaling.
transport_priority_mapping_high

Note that IPv6 TCLASS is generally an 8-bit value.

1See the Platform Notes to find out if the transport priority is supported on a specific platform.

761

15.6 Setting Builtin Transport Properties with the PropertyQosPolicy

Table 15.4 Properties for Builtin Shared-Memory Transport
Property Name
(prefix with
‘dds.transport.shmem.builtin.’)

Property Value Description

Number of bits in a 16-byte address that are used by the transport. Should be between 0 and
128.
parent.address_bit_count
For example, for an address range of 0-255, this address_bit_count should be set to 8. For the
range of addresses used by IPv4 (4 bytes), it should be set to 32.
A bitmap that defines various properties of the transport to the Connext DDS core.
parent.properties_bitmap

Currently, the only property supported is whether or not the transport plugin will always loan a
buffer when Connext DDS tries to receive a message using the plugin. This is in support of a
zero-copy interface.
Specifies the maximum number of buffers that Connext DDS can pass to the send() method
of a transport plugin.

parent.gather_send_buffer_
count_max

The transport plugin send() API supports a gather-send concept, where the send() call can take
several discontiguous buffers, assemble and send them in a single message. This enables
Connext DDS to send a message from parts obtained from different sources without first having
to copy the parts into a single contiguous buffer.
However, most transports that support a gather-send concept have an upper limit on the number
of buffers that can be gathered and sent. Setting this value will prevent Connext DDS from
trying to gather too many buffers into a send call for the transport plugin.
Connext DDS requires all transport-plugin implementations to support a gather-send of least a
minimum number of buffers. This minimum is NDDS_TRANSPORT_PROPERTY_
GATHER_SEND_BUFFER_COUNT_MIN.
The maximum size of a message in bytes that can be sent or received by the transport plugin.

parent.message_size_max

This value must be set before the transport plugin is registered, so that Connext DDS can
properly use the plugin.

parent.allow_interfaces_list
parent.deny_interfaces_list
parent.
allow_multicast_interfaces_list

Not applicable to the Shared-Memory Transport

parent.
deny_multicast_interfaces_list

762

15.6.1 Setting the Maximum Gather-Send Buffer Count for UDPv4 and UDPv6

Table 15.4 Properties for Builtin Shared-Memory Transport
Property Name
(prefix with
‘dds.transport.shmem.builtin.’)

received_message_count_max

Property Value Description

Number of messages that can be buffered in the receive queue. This is the maximum number of
messages that can be buffered in a RecvResource of the Transport Plugin. This does not
guarantee that the Transport-Plugin will actually be able to buffer received_message_count_
max messages of the maximum size set in parent.message_size_max (Section on the previous
page).
The total number of bytes that can be buffered for a RecvResource is actually controlled by
receive_buffer_size (Section below).
The total number of bytes that can be buffered in the receive queue.
This number controls how much memory is allocated by the plugin for the receive queue (on a
per RecvResource basis). The actual number of bytes allocated is:
size = receive_buffer_size + message_size_max +
received_message_count_max * fixedOverhead
where fixedOverhead is some small number of bytes used by the queue data structure.
If receive_buffer_size <
(message_size_max * received_message_count_max), the transport plugin will not be able to
store received_message_count_max messages of size message_size_max.

receive_buffer_size

If receive_buffer_size >
(message_size_max * received_message_count_max), then there will be memory allocated
that cannot be used by the plugin and thus wasted.
To optimize memory usage, specify a receive queue size less than that required to hold the
maximum number of messages which are all of the maximum size.
In most situations, the average message size may be far less than the maximum message size. So
for example, if the maximum message size is 64K bytes, and you configure the plugin to buffer
at least 10 messages, then 640K bytes of memory would be needed if all messages were 64K
bytes. Should this be desired, then receive_buffer_size should be set to 640K bytes.
However, if the average message size is only 10K bytes, then you could set the receive_buffer_
size to 100K bytes. This allows you to optimize the memory usage of the plugin for the average
case and yet allow the plugin to handle the extreme case.
The queue will always be able to hold 1 message of message_size_max bytes, regardless of the
value of receive_buffer_size.

15.6.1 Setting the Maximum Gather-Send Buffer Count for UDPv4 and
UDPv6
To minimize memory copies, Connext DDS uses the "gather send" API that may be available on the transport.
Some operating systems limit the number of gather buffers that can be given to the gather-send function.
This limits Connext DDS's ability to concatenate multiple DDS samples into a single network message.

763

15.6.1 Setting the Maximum Gather-Send Buffer Count for UDPv4 and UDPv6

An example is the UDP transport's sendmsg() call, which on some OSs (such as Solaris) can only take 16
gather buffers, limiting the number of DDS samples that can be concatenated to five or six.
To match this limitation, Connext DDS sets the UDPv4 and UDPv6 transport plug-ins' gather_send_buffer_count_max to 16 by default for all operating systems. This field is part of the NDDS_Transport_
Property_t structure.
l
l

l

On VxWorks 5.5 operating systems, gather_send_buffer_count_max can be set as high as 63.
On Windows and INTEGRITY operating systems, gather_send_buffer_count_max can be set as
high as 128.
On most other operating systems, gather_send_buffer_count_max can be set as high as 16.

If you are using an OS that allows more than 16 gather buffers for a sendmsg() call, you may increase the
UDPv4 or UDPv6 transport plug-in's gather_send_buffer_count_max from the default up to your OS's
limit (but no higher than 128).
For example, if your OS imposes a limit of 64 gather buffers, you may increase the gather_send_buffer_
count_max up to 64. However, if your OS's gather-buffer limit is 1024, you may only increase the
gather_send_buffer_count_max up to 128.
By changing gather_send_buffer_count_max, you can increase performance in the following situations:
l

l

l

When a DataWriter is sending multiple packets to a DataReader either because the DataReader is a
late-joiner and needs to catch up, or because several packets were dropped and need to be resent.
Changing the setting will help when the DataWriter needs to send or resend more than five or six
packets at a time.
If your application has more than five or six DataWriters or DataReaders in a participant. (In this
case, the change will make the discovery process more efficient.)
When using an asynchronous DataWriter, DDS samples are sent asynchronously by a separate
thread. DDS samples may not be sent immediately, but may be queued instead, depending on the settings of the associated FlowController. If multiple DDS samples in the queue must be sent to the
same destination, they will be coalesced into as few network packets as possible. The number of
DDS samples that can be put in a single message is directly proportional to gather_send_buffer_
count_max. Therefore, by maximizing gather_send_buffer_count_max, you can minimize the
number of packets on the wire.

764

15.6.2 Formatting Rules for IPv6 ‘Allow’ and ‘Deny’ Address Lists

15.6.2 Formatting Rules for IPv6 ‘Allow’ and ‘Deny’ Address Lists
This section describes how to format the strings in the properties that create “allow” and “deny” lists:
l

dds.transport.UDPv6.builtin.parent.allow_interfaces_list (Section on page 750)

l

dds.transport.UDPv6.builtin.parent.deny_interfaces_list (Section on page 750)

l

dds.transport.UDPv6.builtin.parent.allow_multicast_interfaces_list (Section on page 751)

l

dds.transport.UDPv6.builtin.parent.deny_multicast_interfaces_list (Section on page 751)

These properties may contain a list of strings, each identifying a range of interface addresses or an interface
name. Interfaces should be specified as comma-separated strings, with each comma delimiting an interface.
The strings can be addresses and patterns in IPv6 notation. They are case-insensitive.
They may contain a wildcard '*' and can expand up to 4 digits in a block. The wildcard must be either leading or trailing (cannot be in the middle of the string). Multiple wildcards can be specified in a single filter,
but only one wildcard can be specified per block (between colons). Table 15.5 Examples of IPv6 Address
Filters shows some examples.
Table 15.5 Examples of IPv6 Address Filters
Example Filter

Equivalent Filters

Matches

*:*:*:*:*:*:*:*
FE80::*:*

fe80::*:*,
Fe80:0:0::*:*
Fe80:0:0:0:0:0:*:*

FE80:aBC::202:2*:*:*2

FE80:0000:0000:0000:0000:0000:xxxx:xxxx
FE80:0ABC:0000:0000:0202:2xxx:xxxx:xxx2

15.7 Installing Additional Builtin Transport Plugins with register_
transport()
After you create an instance of a transport plugin (see Explicitly Creating Builtin Transport Plugin
Instances (Section 15.4 on page 746)) , you have to register it.
The builtin transports (UDPv4, UDPv6, and Shared Memory) are implicitly registered by default (if they
are enabled via the TRANSPORT_BUILTIN QosPolicy (DDS Extension) (Section 8.5.7 on page 606)).
Therefore, you only need to explicitly register a builtin transport if you want an extra instance of it (suppose you want two UDPv4 transports, one with special settings).

765

15.7.1 Transport Lifecycles

The register_transport() operation registers a transport plugin for use with a DomainParticipant and
assigns it a network address. (Note: this operation is only available in the APIs other than Java or .NET. If
you are using Java or .NET, use the Property QosPolicy to install additional transport plugins.)
NDDS_Transport_Handle_t NDDSTransportSupport::register_transport(
DDSDomainParticipant * participant_in,
NDDS_Transport_Plugin * transport_in,
const DDS_StringSeq & aliases_in,
const NDDS_Transport_Address_t & network_address_in)

Where:
participant_in A non-NULL, disabled DomainParticipant.
transport_in

A non-NULL transport plugin that is currently not registered with another DomainParticipant.

aliases_in

A non-NULL sequence of strings used as aliases to refer to the transport plugin symbolically.
The transport plugin will be "available for use" to an Entity contained in the DomainParticipant,
if the transport alias list associated with the Entity contains one of these transport aliases. An
empty alias list represents a WILDCARD and matches ALL aliases. See Transport Aliases
(Section 15.7.2 on the facing page).

network_
address_in

The network address at which to register this transport plugin. The least significant transport_
in.property.address_bit_count will be truncated. The remaining bits are the network address of
the transport plugin. See Transport Network Addresses (Section 15.7.3 on page 768).

Note: You must ensure that the transport plugin instance is only used by one DomainParticipant at a time.
See Transport Lifecycles (Section 15.7.1 below).
Upon success, a valid non-NIL transport handle is returned, representing the association between the
DomainParticipant and the transport plugin. If the transport cannot be registered, NDDS_TRANSPORT_
HANDLE_NIL is returned.
Note that a transport plugin's class name is automatically registered as an implicit alias for the plugin. Thus,
a class name can be used to refer to all the transport plugin instances of that class.
The C and C++ APIs also have a operation to retrieve a registered transport plugin, get_transport_plugin
().
NDDS_Transport_Plugin* get_transport_plugin(
DDSDomainParticipant* participant_in,
const char* alias_in);

15.7.1 Transport Lifecycles
If you create and register a transport plugin with a DomainParticipant, you are responsible for deleting it
by calling its destructor. Builtin transport plugins are automatically managed by Connext DDS if they are
implicitly registered through the TransportBuiltinQosPolicy.

766

15.7.2 Transport Aliases

User-created transport plugins must not be deleted while they are is still in use by a DomainParticipant.
This generally means that a user-created transport plugin instance can only be deleted after the DomainParticipant with which it was registered is deleted. Note that a transport plugin cannot be "unregistered" from
a DomainParticipant.
A transport plugin instance cannot be registered with more than one DomainParticipant at a time. This
requirement is necessary to guarantee the multi-threaded safety of the transport API.
Thus, if the same physical transport resources are to be used with multiple DomainParticipants in the same
address space, the transport plugin should be written in such a way so that it can be instantiated multiple
times—once for each DomainParticipant in the address space. Note that it is always possible to write the
transport plugin so that multiple transport plugin instances share the same underlying resources; however
the burden (if any) of guaranteeing multi-threaded safety to access shared resource shifts to the transport
plugin developer.

15.7.2 Transport Aliases
In order to use a transport plugin instance in a Connext DDS application, it must be registered with a
DomainParticipant using the register_transport() operation (Installing Additional Builtin Transport Plugins with register_transport() (Section 15.7 on page 765)). register_transport() takes a pointer to the transport plugin instance, and in addition allows you to specify a sequence of "alias" strings to symbolically
refer to the transport plugin. The same alias strings can be used to register more than one transport plugin.
Multiple transport plugins can be registered with a DomainParticipant. An alias symbolically refers to one
or more transport plugins registered with the DomainParticipant. Pre-configured builtin transport plugin
instances can be referred to using preconfigured aliases.
A transport plugin's class name is automatically used as an implicit alias. It can be used to refer to all the
transport plugin instance of that class.
You can use aliases to refer to transport plugins in order to specify:
l

l

l

l

Transport plugins to use for discovery (see enabled_transports in DISCOVERY QosPolicy (DDS
Extension) (Section 8.5.2 on page 580)), and for DataWriters and DataReaders (see
TRANSPORT_SELECTION QosPolicy (DDS Extension) (Section 6.5.23 on page 411)).
Multicast addresses on which to receive discovery messages (see multicast_receive_addresses in
DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580)), and the multicast
addresses and ports on which to receive user data (DDS_DataReaderQos::multicast).
Unicast ports used for user data (see TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412)) on both DataWriters and DataReaders.
Transport plugins used to parse an address string in a locator.

767

15.7.3 Transport Network Addresses

A DomainParticipant (and its contained entities) will start using a transport plugin after the DomainParticipant is enabled (see Enabling DDS Entities (Section 4.1.2 on page 154)). An entity will use all the
transport plugins that match the specified transport QoS policy. All transport plugins are treated uniformly,
regardless of how they were created or registered; there is no notion of some transports being more "special" that others.

15.7.3 Transport Network Addresses
The address bits not used by the transport plugin for its internal addressing constitute its network address
bits.
In order for Connext DDS to properly route the messages, each unicast interface in the DDS domain must
have a unique address.
You specify the network address when installing a transport plugin via the register_transport() operation
(Installing Additional Builtin Transport Plugins with register_transport() (Section 15.7 on page 765)).
Choose the network address for a transport plugin so that the resulting fully qualified 128-bit address will
be unique in the DDS domain.
If two instances of a transport plugin are registered with a DomainParticipant, they need different network
addresses so that their unicast interfaces will have unique, fully qualified 128-bit addresses.
While it is possible to create multiple transports with the same network address (this can be useful for certain situations), this requires special entity configuration for most transports to avoid clashes in resource use
(e.g., sockets for UDPv4 transport).

15.8 Installing Additional Builtin Transport Plugins with
PropertyQosPolicy
Similar to default builtin transport instances, additional builtin transport instances can also be configured
through PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394).
To install additional instances of builtin transport, the Properties listed in Table 15.6 Properties for Dynamically Loading and Registering Additional Builtin Transport Plugins are required.
Table 15.6 Properties for Dynamically Loading and Registering Additional Builtin Transport
Plugins
Property
Name

Description

dds.transport.load_
Comma-separated list of . Up to 8 entries may be specified.
plugins

768

15.9 Other Transport Support Operations

Table 15.6 Properties for Dynamically Loading and Registering Additional Builtin Transport
Plugins
Property
Name

Description
Indicates the additional builtin transport instances to be installed, and must be in one of the following form, where
 can be any string other than “builtin”:
dds.transport.shmem.



dds.transport.UDPv4.
dds.transport.UDPv6.
In the following examples in this table,  is used to indicate one element of this string that is
used as a prefix in the property names for all the settings that are related to the plugin.
Optional.

.
aliases

Aliases used to register the transport to the DomainParticipant. Refer to the aliases_in parameter in register_
transport() (see Installing Additional Builtin Transport Plugins with register_transport() (Section 15.7 on page 765)).
Aliases should be specified as a comma separated string, with each comma delimiting an alias. If it is not specified,
 is used as the default alias for the plugin.
Optional.

.
network_address

Network address used to register the transport to the DomainParticipant. Refer to network_address_in parameter in
register_transport() (see Installing Additional Builtin Transport Plugins with register_transport() (Section 15.7 on
page 765)). If it is not specified, the network_address_out output parameter from NDDS_Transport_create_
plugin is used. The default value is a zeroed out network address.
Optional.

.


Property for creating the transport plugin. More than one . can be
specified. See Table 15.2 Properties for the Builtin UDPv4 Transport through Table 15.4 Properties for Builtin
Shared-Memory Transport for the property names that can be used to configure the additional builtin transport
instances. The only difference is that the property name will be prefixed by dds.transport.., where  is configured through the dds.transport.load_plugins
property instead of dds.transport..builtin.

15.9 Other Transport Support Operations
15.9.1 Adding a Send Route
By default, a transport plugin will send outgoing messages using the network address range at which the
plugin was registered.
The add_send_route() operation allows you to control the routing of outgoing messages, so that a transport plugin will only send messages to certain ranges of destination addresses.
Before using this operation, the DomainParticipant to which the transport is registered must be disabled.
DDS_ReturnCode_t NDDSTransportSupport::add_send_route(
const NDDS_Transport_Handle_t & transport_handle_in,

769

15.9.2 Adding a Receive Route

const NDDS_Transport_Address_t & address_range_in,
DDS_Long address_range_bit_count_in)

Where:
transport_handle_in

A valid non-NIL transport handle as a result of a call to register_transport()
(Installing Additional Builtin Transport Plugins with register_transport()
(Section 15.7 on page 765)).

address_range_in

The outgoing address range for which to use this transport plugin.

address_range_bit_count_in The number of most significant bits used to specify the address range.

It returns one of the standard return codes or DDS_RETCODE_PRECONDITION_NOT_MET.
The method can be called multiple times for a transport plugin, with different address ranges. You can set
up a routing table to restrict the use of a transport plugin to send messages to selected addresses ranges.
Outgoing Address Range 1

->

Transport Plugin

...

->

...

Outgoing Address Range K

->

Transport Plugin

15.9.2 Adding a Receive Route
By default, a transport plugin will receive incoming messages using the network address range at which
the plugin was registered.
The add_receive_route() operation allows you to configure a transport plugin so that it will only receive
messages on certain ranges of addresses.
Before using this operation, the DomainParticipant to which the transport is registered must be disabled.
DDS_ReturnCode_t NDDSTransportSupport::add_receive_route(
const NDDS_Transport_Handle_t & transport_handle_in,
const NDDS_Transport_Address_t &
DDS_Long

address_range_in,

address_range_bit_count_in)

Where:
transport_handle_in

A valid non-NIL transport handle as a result of a call to register_transport()
(Installing Additional Builtin Transport Plugins with register_transport()
(Section 15.7 on page 765)).

address_range_in

The incoming address range for which to use this transport plugin.

address_range_bit_count_in The number of most significant bits used to specify the address range.

770

15.9.3 Looking Up a Transport Plugin

It returns one of the standard return codes or DDS_RETCODE_PRECONDITION_NOT_MET.
The method can be called multiple times for a transport plugin, with different address ranges.
Transport Plugin

<-

Incoming Address Range 1

...

<-

...

Transport Plugin

<-

Incoming Address Range M

You can set up a routing table to restrict the use of a transport plugin to receive messages from selected
ranges. For example, you may restrict a transport plugin to:
Receive messages from a certain multicast address range.
Receive messages only on certain unicast interfaces (when multiple unicast interfaces are available on the
transport plugin).

15.9.3 Looking Up a Transport Plugin
If you need to get the handle associated with a transport plugin that is registered with a
DomainParticipant, use the lookup_transport() operation.
NDDS_Transport_Handle_t NDDSTransportSupport::lookup_transport(
DDSDomainParticipant * participant_in,
DDS_StringSeq & aliases_out,
NDDS_Transport_Address_t & network_address_out,
NDDS_Transport_Plugin * transport_in )

Where:
participant_in A non-NULL DomainParticipant.
aliases_out

A sequence of strings where the aliases used to refer to the transport plugin symbolically will be
returned. NULL if not interested.

network_
address_out

The network address at which to register the transport plugin will be returned here. NULL if not
interested.

transport_in

A non-NULL transport plugin that is already registered with the DomainParticipant.

If successful, this operation returns a valid non-NIL transport handle, representing the association between
the DomainParticipant and the transport plugin; otherwise it returns a NDDS_TRANSPORT_
HANDLE_NIL upon failure.

771

Chapter 16 Built-In Topics
This chapter discusses how to use Built-in Topics.
Connext DDS must discover and keep track of remote entities, such as new participants in the
DDS domain. This information may also be important to the application itself, which may want to
react to this discovery or access it on demand. To support these needs, Connext DDS provides
built-in Topics (“DCPSParticipant”, “DCPSPublication”, “DCPSSubscription” in Built-in Writers
and Readers for Discovery (Section Figure 14.2 on page 718)) and the corresponding built-in
DataReaders that you can use to access this discovery information.
The discovery information is accessed just as if it is normal application data. This allows the application to know (either via listeners or by polling) when there are any changes in those values. Note
that only entities that belong to a different DomainParticipant are being discovered and can be
accessed through the built-in readers. Entities that are created within the local DomainParticipant
are not included as part of the data that can be accessed by the built-in readers.
Built-in topics contain information about the remote entities, including their QoS policies. These
QoS policies appear as normal fields inside the topic’s data, which can be read by means of the
built-in Topic. Additional information is provided to identify the entity and facilitate the application
logic.

16.1 Listeners for Built-in Entities
Built-in entities have default listener settings:
l

l

The built-in Subscriber and its built-in topics have 'nil' listeners—all status bits are set in the
listener masks, but the listener is NULL. This effectively creates a NO-OP listener that does
not reset communication status.
Built-in DataReaders have null listeners with no status bits set in their masks.

This approach prevents callbacks to the built-in DataReader listeners from invoking your
DomainParticipant’s listeners, and at the same time ensures that the status changed flag is not

772

16.2 Built-in DataReaders

reset. For more information, see Table 4.4 Effect of Different Combinations of Listeners and Status Bit
Masks and Hierarchical Processing of Listeners (Section 4.4.4 on page 180).

16.2 Built-in DataReaders
Built-in DataReaders belong to a built-in Subscriber, which can be retrieved by using the DomainParticipant’s get_builtin_subscriber() operation. You can retrieve the built-in DataReaders by using the Subscriber’s lookup_datareader() operation, which takes the Topic name as a parameter. The built-in
DataReader is created when lookup_datareader() is called on a built-in topic for the first time.
To conserve memory, built-in Subscribers and DataReaders are created only if and when you look them
up. Therefore, if you do not want to miss any built-in data, you should look up the built-in readers before
the DomainParticipant is enabled.
The following tables describe the built-in topics and their data types. The USER_DATA QosPolicy (Section 6.5.26 on page 417), TOPIC_DATA QosPolicy (Section 5.2.1 on page 209) and GROUP_DATA
QosPolicy (Section 6.4.4 on page 320) are included as part of the built-in data type and are not used by
Connext DDS. Therefore, you can use them to send application-specific information.
Built-in topics can be used in conjunction with the ignore_*() operations to ignore certain entities (see
Restricting Communication—Ignoring Entities (Section 16.4 on page 784)).
Table 16.1 Participant Built-in Topic’s Data Type (DDS_ParticipantBuiltinTopicData)
Type

Field

Description

DDS_
BuiltinTopicKey

key

Key to distinguish the discovered DomainParticipant

DDS_
UserDataQosPolicy

user_data

Data that can be set when the related DomainParticipant is created (via the USER_DATA QosPolicy
(Section 6.5.26 on page 417)) and that the application may use as it wishes (e.g., to perform some
security checking).

DDS_
PropertyQosPolicy

property

Pairs of names/values to be stored with the DomainParticipant. See PROPERTY QosPolicy (DDS
Extension) (Section 6.5.17 on page 394). The usage is strictly application-dependent.

DDS_
ProtocolVersion_t

rtps_
protocol_
version

Version number of the RTPS wire protocol used.

DDS_VendorId_t

rtps_
vendor_id

ID of vendor implementing the RTPS wire protocol.

DDS_UnsignedLong

dds_
builtin_
endpoints

Bitmap set by the discovery plugins.
Each bit in this field indicates a built-in endpoint present for discovery.

773

16.2 Built-in DataReaders

Table 16.1 Participant Built-in Topic’s Data Type (DDS_ParticipantBuiltinTopicData)
Type

DDS_LocatorSeq

Field

default_
unicast_
locators

Description
If the TransportUnicastQosPolicy is not specified when a DataWriter/DataReader is created, the
unicast_locators in the corresponding Publication/Subscription built-in topic data will be empty. When
the unicast_locators in the Publication/SubscriptionBuiltinTopicData is empty, the default_unicast_
locators in the corresponding Participant Builtin Topic Data is assumed.
If default_unicast_locators is empty, it defaults to DomainParticipantQos.default_unicast.

DDS_
ProductVersion_t

product_
version

Vendor-specific parameter. The current version of Connext DDS.

DDS_
participant_ Name and role_name assigned to the DomainParticipant. See ENTITY_NAME QosPolicy (DDS
EntityNameQosPolicy name
Extension) (Section 6.5.9 on page 374).
DDS_DomainId_t

domain_id

Domain ID associated with the discovered participant.
A sequence of DDS_TransportInfo_t containing information about each of the installed transports of
the discovered DomainParticipant.
A DDS_TransportInfo_t structure contains the class_id and message_size_max for a single transport.

DDS_
TransportInfoSeq

transport_
info

The maximum length of this sequence is controlled by the DOMAIN_PARTICIPANT_
RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593) transport_info_list_
max_length (see Table 8.12 DDS_DomainParticipantResourceLimitsQosPolicy ).
Connext DDS uses the transport information propagated via discovery to detect potential
misconfigurations in a Connext DDS distributed system. If two DomainParticipants that discover
each other have one common transport with different values for message_size_max, Connext DDS
prints a warning message about that condition.

Table 16.2 Publication Built-in Topic’s Data Type (DDS_PublicationBuiltinTopicData)
Type

Field

Description

DDS_
BuiltinTopicKey_t

key

DDS_
BuiltinTopicKey_t

participant_
Key to distinguish the participant to which the discovered DataWriter belongs
key

DDS_String

topic_name

Topic name of the discovered DataWriter

DDS_String

type_name

Type name attached to the topic of the discovered DataWriter

Key to distinguish the discovered DataWriter

774

16.2 Built-in DataReaders

Table 16.2 Publication Built-in Topic’s Data Type (DDS_PublicationBuiltinTopicData)
Type

Field

DDS_
DurabilityQosPolicy

durability

DDS_
DurabilityServiceQosPolicy

durability_
service

DDS_
DeadlineQosPolicy

deadline

DDS_
DestinationOrderQosPolicy

destination_
order

DDS_LatencyBudgetQosPolicy

latency_
budget

DDS_
LivelinessQosPolicy

liveliness

DDS_
ReliabilityQosPolicy

reliability

DDS_
LifespanQosPolicy

lifespan

DDS_
UserDataQosPolicy

user_data

DDS_
OwnershipQosPolicy

ownership

DDS_
OwnershipStrengthQosPolicy

ownership_
strength

Description

QosPolicies of the discovered DataWriter

Data that can be set when the DataWriter is created (via the USER_DATA QosPolicy (Section
6.5.26 on page 417)) and that the application may use as it wishes.

QosPolicies of the discovered DataWriter
DDS_
DestinationOrderQosPolicy

destination_
order

DDS_
presentation
PresentationQosPolicy
DDS_
PartitionQosPolicy

partition

Name of the partition, set in the PARTITION QosPolicy (Section 6.4.5 on page 323) for the
publisher to which the discovered DataWriter belongs

775

16.2 Built-in DataReaders

Table 16.2 Publication Built-in Topic’s Data Type (DDS_PublicationBuiltinTopicData)
Type

Field

Description

DDS_
TopicDataQosPolicy

topic_data

Data that can be set when the Topic (with which the discovered DataWriter is associated) is created
(via the TOPIC_DATA QosPolicy (Section 5.2.1 on page 209)) and that the application may use as
it wishes.

DDS_
GroupDataQosPolicy

group_data

Data that can be set when the Publisher to which the discovered DataWriter belongs is created (via
the GROUP_DATA QosPolicy (Section 6.4.4 on page 320)) and that the application may use as it
wishes.

DDS_TypeObject *

type

Describes the type of the remote DataReader.
See the API Reference HTML documentation.
DDS_TypeCode *

type_code

Type code information about this Topic. See Using Generated Types without Connext DDS
(Standalone) (Section 3.7 on page 139).

DDS_
BuiltinTopicKey_t

publisher_
key

The key of the Publisher to which the DataWriter belongs.

DDS_
PropertyQosPolicy

property

Properties (pairs of names/values) assigned to the corresponding DataWriter. Usage is strictly
application-dependent. See PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page
394).

DDS_LocatorSeq

unicast_
locators

If the TransportUnicastQosPolicy is not specified when a DataWriter/DataReader is created, the
unicast_locators in the corresponding Publication/Subscription built-in topic data will be empty.
When the unicast_locators in the Publication/SubscriptionBuiltinTopicData is empty, the default_
unicast_locators in the corresponding Participant Builtin Topic Data is assumed.

DDS_GUID_t

virtual_guid

Virtual GUID for the corresponding DataWriter. For more information, see Durability and
Persistence Based on Virtual GUIDs (Section 12.2 on page 680).

DDS_
ServiceQosPolicy

service

Service associated with the discovered DataWriter.

DDS_
ProtocolVersion_t

rtps_
protocol_
version

Version number of the RTPS wire protocol in use.

DDS_VendorId_t

rtps_
vendor_id

ID of the vendor implementing the RTPS wire protocol.

DDS_Product_
Version_t

product_
version

Vendor-specific value. For RTI, this is the current version of Connext DDS.

776

16.2 Built-in DataReaders

Table 16.2 Publication Built-in Topic’s Data Type (DDS_PublicationBuiltinTopicData)
Type

Field

DDS_
locator_
LocatorFilterQosPolicy filter

Description
When the MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386) is used
on the discovered DataWriter, the locator_filter contains the sequence of LocatorFilters in that
policy.
There is one LocatorFilter per DataWriter channel. A channel is defined by a filter expression and a
sequence of multicast locators.
See LOCATOR_FILTER QoS Policy (DDS Extension) (Section 16.2.1 on page 782).

DDS_Boolean

disable_
positive_
acks

DDS_
EntityNameQosPolicy

publication_ Name and role_name assigned to the DataWriter. See ENTITY_NAME QosPolicy (DDS
name
Extension) (Section 6.5.9 on page 374).

Vendor specific parameter. Determines whether matching DataReaders send positive
acknowledgements for reliability.

Table 16.3 Subscription Built-in Topic’s Data Type (DDS_SubscriptionBuiltinTopicData)
Type

Field

Description

DDS_BuiltinTopicKey_t

key

Key to distinguish the discovered DataReader.

DDS_BuiltinTopicKey_t

participant_
key

Key to distinguish the participant to which the discovered DataReader belongs.

char *

topic_name

Topic name of the discovered DataReader.

char *

type_name

Type name attached to the Topic of the discovered DataReader.

DDS_DurabilityQosPolicy

durability

DDS_DeadlineQosPolicy

deadline

DDS_LatencyBudget-QosPolicy

latency_
budget

DDS_LivelinessQosPolicy

liveliness

DDS_ReliabilityQosPolicy

reliability

DDS_OwnershipQosPolicy

ownership

DDS_
DestinationOrderQosPolicy

destination_
order

QosPolicies of the discovered DataReader

777

16.2 Built-in DataReaders

Table 16.3 Subscription Built-in Topic’s Data Type (DDS_SubscriptionBuiltinTopicData)
Type

Field

Description
Data that can be set when the DataReader is created (via the USER_DATA
QosPolicy (Section 6.5.26 on page 417)) and that the application may use as it
wishes.

DDS_UserDataQosPolicy

user_data

DDS_
TimeBasedFilterQosPolicy

time_based_
filter

DDS_PresentationQosPolicy

presentation

DDS_PartitionQosPolicy

partition

Name of the partition, set in the PARTITION QosPolicy (Section 6.4.5 on page
323) for the Subscriber to which the discovered DataReader belongs.

DDS_TopicDataQosPolicy

topic_data

Data that can be set when the Topic to which the discovered DataReader belongs
is created (via the TOPIC_DATA QosPolicy (Section 5.2.1 on page 209)) and
that the application may use as it wishes.

DDS_GroupDataQosPolicy

group_data

Data that can be set when the Publisher to which the discovered DataReader
belongs is created (via the GROUP_DATA QosPolicy (Section 6.4.4 on page
320)) and that the application may use as it wishes.

DDS_TypeObject *

type

QosPolicies of the discovered DataReader

Describes the type of the remote DataReader.
See the API Reference HTML documentation.

DDS_
type_
TypeConsistencyEnforcementQosPolicy consistency

Indicates the type-consistency requirements of the remote DataReader. See
TYPE_CONSISTENCY_ENFORCEMENT QosPolicy (Section 7.6.6 on page
532) and the RTI Connext DDS Core Libraries Getting Started Guide Addendum
for Extensible Types.

DDS_TypeCode *

type_code

Type code information about this Topic. See Using Generated Types without
Connext DDS (Standalone) (Section 3.7 on page 139).

DDS_BuiltinTopicKey_t

subscriber_
key

Key of the Subscriber to which the DataReader belongs.

DDS_PropertyQosPolicy

property

Properties (pairs of names/values) assigned to the corresponding DataReader.
Usage is strictly application-dependent. See PROPERTY QosPolicy (DDS
Extension) (Section 6.5.17 on page 394).

DDS_LocatorSeq

unicast_
locators

If the TransportUnicastQosPolicy is not specified when a
DataWriter/DataReader is created, the unicast_locators in the corresponding
Publication/Subscription builtin topic data will be empty. When the unicast_
locators in the Publication/SubscriptionBuiltinTopicData is empty, the default_
unicast_locators in the corresponding Participant Builtin Topic Data is assumed.

DDS_LocatorSeq

multicast_
locators

Custom multicast locators that the endpoint can specify.

778

16.2 Built-in DataReaders

Table 16.3 Subscription Built-in Topic’s Data Type (DDS_SubscriptionBuiltinTopicData)
Type

Field

Description

DDS_ContentFilter-Property_t

content_
filter_
property

Provides all the required information to enable content filtering on the writer side.

DDS_GUID_t

virtual_guid

Virtual GUID for the corresponding DataReader. For more information, see
Durability and Persistence Based on Virtual GUIDs (Section 12.2 on page 680).

DDS_ServiceQosPolicy

service

Service associated with the discovered DataReader.

DDS_ProtocolVersion_t

rtps_
protocol_
version

Version number of the RTPS wire protocol in use.

DDS_VendorId_t

rtps_vendor_
ID of the vendor implementing the RTPS wire protocol.
id

DDS_Product_Version_t

product_
version

Vendor-specific value. For RTI, this is the current version of Connext DDS.

DDS_Boolean

disable_
positive_
acks

Vendor specific parameter. Determines whether matching DataReaders send
positive acknowledgements for reliability.

DDS_EntityNameQosPolicy

subscription_ Name and role_name assigned to the DataReader. See ENTITY_NAME
name
QosPolicy (DDS Extension) (Section 6.5.9 on page 374).

Table 16.4 Topic Built-in Topic’s Data Type (DDS_TopicBuiltinTopicData)
Type

Field

Description

DDS_BuiltinTopicKey_t

key

Key to distinguish the discovered Topic

DDS_String

name

Topic name

DDS_String

type_name

type name attached to the Topic

779

16.2 Built-in DataReaders

Table 16.4 Topic Built-in Topic’s Data Type (DDS_TopicBuiltinTopicData)
Type

Field

DDS_DurabilityQosPolicy

Description

durability

DDS_
durability_
DurabilityServiceQosPolicy service
DDS_DeadlineQosPolicy

deadline

DDS_
LatencyBudgetQosPolicy

latency_
budget

DDS_LivelinessQosPolicy

liveliness

DDS_ReliabilityQosPolicy

reliability

DDS_
transport_
TransportPriorityQosPolicy priority
DDS_LifespanQosPolicy

QosPolicy of the discovered Topic

lifespan

DDS_
destination_
DestinationOrderQosPolicy order
DDS_HistoryQosPolicy

history

DDS_
ResourceLimitsQosPolicy

resource_
limits

DDS_
OwnershipQosPolicy

ownership

DDS_TopicDataQosPolicy topic_data

Data that can be set when the Topic to which the discovered DataReader
belongs is created (via the TOPIC_DATA QosPolicy (Section 5.2.1 on
page 209)) and that the application may use as it wishes.

Table 16.5 QoS of Built-in Subscriber and DataReader lists the QoS of the built-in Subscriber and
DataReader created for accessing discovery data. These are provided for your reference only; they cannot
be changed.
Table 16.5 QoS of Built-in Subscriber and DataReader
QosPolicy

Value

Deadline

period = infinite

DestinationOrder

kind = BY_RECEPTION_TIMESTAMP_DESTINATIONORDER_QOS

780

16.2 Built-in DataReaders

Table 16.5 QoS of Built-in Subscriber and DataReader
QosPolicy

Value

Durability

kind = TRANSIENT_LOCAL_DURABILITY_QOS

EntityFactory

autoenable_created_entities = TRUE

GroupData

value = empty sequence
kind = KEEP_LAST_HISTORY_QOS

History
depth = 1
LatencyBudget

duration = 0
kind = AUTOMATIC_LIVELINESS_QOS

Liveliness
lease_duration = infinite
Ownership

kind = SHARED_OWNERSHIP_QOS

Ownership Strength value = 0
access_scope = TOPIC_PRESENTATION_QOS
Presentation

coherent_access = FALSE
ordered_access = FALSE

Partition

name = empty sequence

ReaderDataLifecycle autopurge_nowriter_samples_delay = infinite
kind = RELIABLE_RELIABILITY_QOS
Reliability
max_blocking_time is irrelevant for the DataReader
Depends on setting of DomainParticipantResourceLimitsQosPolicy and
DiscoveryConfigQosPolicy in DomainParticipantQos:

ResourceLimits

max_samples = domainParticipantQos.discovery_config.
[participant/publication/subscription]_reader_resource_limits.max_samples
max_instances = domainParticipantQos.resource_limits.
[remote_writer/reader/participant]_allocation.max_count
max_samples_per_instance = 1

TimeBasedFilter

minimum_separation = 0

TopicData

value = empty sequence

UserData

value = empty sequence

781

16.2.1 LOCATOR_FILTER QoS Policy (DDS Extension)

Note:
The DDS_TopicBuiltinTopicData built-in topic (described in Table 16.4 Topic Built-in Topic’s Data
Type (DDS_TopicBuiltinTopicData) ) is meant to convey information about discovered Topics. However,
this topic's data is not sent separately and therefore a DataReader for DDS_TopicBuiltinTopicData will
not receive any data. Instead, DDS_TopicBuiltinTopicData data is included in the information carried by
the built-in topics for Publications and Subscriptions (DDS_PublicationBuiltinTopicData and DDS_SubscriptionBuiltinTopicData) and can be accessed with their built-in DataReaders.

16.2.1 LOCATOR_FILTER QoS Policy (DDS Extension)
The LocatorFilter QoS Policy is only applicable to the built-in topic for a Publication (see Table 16.2
Publication Built-in Topic’s Data Type (DDS_PublicationBuiltinTopicData)).
Table 16.6 DDS_LocatorFilterQosPolicy
Type

Field
Name

Description

DDS_
locator_ A sequence of locator filters, described in Table 16.7 DDS_LocatorFilter_t. There is one locator filter per
LocatorFilterSeq filters
DataWriter channel. If the length of the sequence is zero, the DataWriter is not using multi-channel.
Name of the filter class used to describe the locator filter expressions. The following two values are supported:
char *

filter_
name

l

DDS_SQLFILTER_NAME
l

DDS_STRINGMATCHFILTER_NAME

Table 16.7 DDS_LocatorFilter_t
Type

Field
Name

DDS_
locators
LocatorSeq

char *

Description

A sequence of multicast address locators for the locator filter. See Table 16.8 DDS_Locator_t.

A logical expression used to determine if the data will be published in the channel associated with this locator
filter_
filter. See SQL Filter Expression Notation (Section 5.4.6 on page 222) and STRINGMATCH Filter Expression
expression
Notation (Section 5.4.7 on page 231) for information about the expression syntax.

782

16.3 Accessing the Built-in Subscriber

Table 16.8 DDS_Locator_t
Type

Field
Name

Description
If the locator kind is DDS_LOCATOR_KIND_UDPv4, the address contains an IPv4 address. The leading 12
octets of the address must be zero. The last 4 octets store the IPv4 address.

DDS_Long

kind

If the locator kind is DDS_LOCATOR_KIND_UDPv6, the address contains an IPv6 address. IPv6 addresses
typically use a shorthand hexadecimal notation that maps one-to-one to the 16 octets of the address.
In C#, the locator kinds for UDPv4 and UDPv6 addresses are Locator_t.LOCATOR_KIND_UDPv4 and
Locator_t.LOCATOR_KIND_UDPv6.

DDS_Octet
[16]

address The locator address.

DDS_
port
UnsignedLong

The locator port number.

16.3 Accessing the Built-in Subscriber
Getting the built-in subscriber allows you to retrieve the built-in readers of the built-in topics through the
Subscriber’s lookup_datareader() operation. By accessing the built-in reader, you can access discovery
information about remote entities.
// Lookup built-in reader
DDSDataReader *builtin_reader =
builtin_subscriber->lookup_datareader(DDS_PUBLICATION_TOPIC_NAME);
if (builtin_reader == NULL) {
// ... error
}
// Register listener to built-in reader
MyPublicationBuiltinTopicDataListener builtin_reader_listener =
new MyPublicationBuiltinTopicDataListener();
if (builtin_reader->set_listener(builtin_reader_listener,
DDS_DATA_AVAILABLE_STATUS) != DDS_RETCODE_OK) {
// ... error

783

16.4 Restricting Communication—Ignoring Entities

}
// enable DomainParticipant
if (participant->enable() != DDS_RETCODE_OK) {
// ... error
}

For example, you can call the DomainParticipant’s get_builtin_subscriber() operation, which will provide
you with a built-in Subscriber. Then you can use that built-in Subscriber to call the Subscriber’s lookup_
datareader() operation; this will retrieve the built-in reader. Another option is to register a Listener on the
built-in subscriber instead, or poll for the status of the built-in subscriber to see if any of the built-in data
readers have received data.

16.4 Restricting Communication—Ignoring Entities
The ignore_participant() operation allows an application to ignore all communication from a specific
DomainParticipant. Or for even finer control you can use the ignore_publication(), ignore_subscription(),
and ignore_topic() operations. These operations are described below.
DDS_ReturnCode_t

ignore_participant (const DDS_InstanceHandle_t &handle)

DDS_ReturnCode_t

ignore_publication (const DDS_InstanceHandle_t &handle)

DDS_ReturnCode_t

ignore_subscription (const DDS_InstanceHandle_t &handle)

DDS_ReturnCode_t

ignore_topic (const DDS_InstanceHandle_t &handle)

The entity to ignore is identified by the handle argument. It may be a local or remote entity. For ignore_
publication(), the handle will be that of a local DataWriter or a discovered remote DataWriter. For
ignore_subscription(), that handle will be that of a local DataReader or a discovered remote DataReader.
The safest approach for ignoring an entity is to call the ignore operation within the Listener callback of the
built-in reader, or before any local entities are enabled. This will guarantee that the local entities (entities
that are created by the local DomainParticipant) will never have a chance to establish communication with the
remote entities (entities that are created by another DomainParticipant) that are going to be ignored.
If the above is not possible and a remote entity is to be ignored after the communication channel has been
established, the remote entity will still be removed from the database of the local application as if it never
existed. However, since the remote application is not aware that the entity is being ignored, it may potentially be expecting to receive messages or continuing to send messages. Depending on the QoS of the
remote entity, this may affect the behavior of the remote application and may potentially stop the remote
application from communicating with other entities.

784

16.4.1 Ignoring Specific Remote DomainParticipants

You can use this operation in conjunction with the ParticipantBuiltinTopicData to implement access control. You can pass application data associated with a DomainParticipant in the USER_DATA QosPolicy
(Section 6.5.26 on page 417). This application data is propagated as a field in the built-in topic. Your
application can use the data to implement an access control policy.
Ignore operations, in conjunction with the Built-in Topic Data, can be used to implement access control.
You can pass data associated with an entity in the USER_DATA QosPolicy (Section 6.5.26 on page
417), GROUP_DATA QosPolicy (Section 6.4.4 on page 320) or TOPIC_DATA QosPolicy (Section
5.2.1 on page 209). This data is propagated as a field in the built-in topic. When data for a built-in topic is
received, the application can check the user_data, group_data or topic_data field of the remote entity,
determine if it meets the security requirement, and ignore the remote entity if necessary.
See also: Discovery (Section Chapter 14 on page 709).

16.4.1 Ignoring Specific Remote DomainParticipants
The ignore_participant() operation is used to instruct Connext DDS to locally ignore a remote
DomainParticipant. It causes Connext DDS to locally behave as if the remote DomainParticipant does
not exist.
DDS_ReturnCode_t ignore_participant

(const DDS_InstanceHandle_t &

handle)

After invoking this operation, Connext DDS will locally ignore any Topic, publication, or subscription
that originates on that DomainParticipant. (If you only want to ignore specific publications or subscriptions, see Ignoring Publications and Subscriptions (Section 16.4.2 on the facing page) instead.) Ignoring Participants (Section Figure 16.1 below) provides an example.
By default, the maximum number of participants that can be ignored is limited by ignored_entity_allocation.max_count in the DOMAIN_PARTICIPANT_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4 on page 593). However, that behavior can be changed by using ignore_entity_
replacement_kind in the same QoS policy.
See also: Resource Limits Considerations for Ignored Entities (Section 16.4.4 on page 788).
Caution: There is no way to reverse this operation. You can add to the peer list, however—see Adding
and Removing Peers List Entries (Section 8.5.2.3 on page 581).
Figure 16.1 Ignoring Participants
class MyParticipantBuiltinTopicDataListener :
public DDSDataReaderListener {
public:
virtual void on_data_available(DDSDataReader *reader);
// ......
};
void MyParticipantBuiltinTopicdataListener::on_data_available(

785

16.4.2 Ignoring Publications and Subscriptions

DDSDataReader *reader) {
DDSParticipantBuiltinTopicDataDataReader
*builtinTopicDataReader =
DDSParticipantBuiltinTopicDataDataReader *) reader;
DDS_ParticipantBuiltinTopicDataSeq data_seq;
DDS_SampleInfoSeq info_seq;
int = 0;
if (builtinTopicDataReader->take(data_seq, info_seq,
DDS_LENGTH_UNLIMITED, DDS_ANY_SAMPLE_STATE,
DDS_ANY_VIEW_STATE, DDS_ANY_INSTANCE_STATE) !=
DDS_RETCODE_OK){
// ... error
}
for (i = 0; i < data_seq.length(); ++i) {
if (info_seq[i].valid_data) {
// check user_data for access control
if (data_seq[i].user_data[0] != 0x9) {
if (builtinTopicDataReader->get_subscriber()
->get_participant()
->ignore_participant(
info_seq[i].instance_handle)
!= DDS_RETCODE_OK) {
// ... error
}
}
}
}
if (builtinTopicDataReader->return_loan(
data_seq, info_seq) != DDS_RETCODE_OK) {
// ... error
}
}

16.4.2 Ignoring Publications and Subscriptions
You can instruct Connext DDS to locally ignore a publication or subscription. A publication/subscription
is defined by the association of a Topic name, user data and partition set on the Publisher/Subscriber. After
this call, any data written related to associated DataWriter/DataReader will be ignored.
The entity to ignore is identified by the handle argument. For ignore_publication(), the handle will be that
of a DataWriter. For ignore_subscription(), that handle will be that of a DataReader.
This operation can be used to ignore local and remote entities:
l

l

For local entities, you can obtain the handle argument by calling the get_instance_handle() operation for that particular entity.
For remote entities, you can obtain the handle argument from the DDS_SampleInfo structure
retrieved when reading DDS data samples available for the entity’s built-in DataReader.

786

16.4.2 Ignoring Publications and Subscriptions

DDS_ReturnCode_t ignore_publication (const DDS_InstanceHandle_t &
DDS_ReturnCode_t ignore_subscription (const DDS_InstanceHandle_t &

handle)
handle)

Caution: There is no way to reverse these operations.
Figure 16.2 Ignoring Publications below provides an example.
Figure 16.2 Ignoring Publications
class MyPublicationBuiltinTopicDataListener : public DDSDataReaderListener
{
public:
virtual void on_data_available(DDSDataReader *reader);
// ......
};
void MyPublicationBuiltinTopicdataListener::on_data_available(
DDSDataReader *reader) {
DDSPublicationBuiltinTopicDataReader *builtinTopicDataReader =
(DDS_PublicationBuiltinTopicDataReader *)reader;
DDS_PublicationBuiltinTopicDataSeq data_seq;
DDS_SampleInfoSeq info_seq;
int = 0;
if (builtinTopicDataReader->take(data_seq, info_seq,
DDS_LENGTH_UNLIMITED, DDS_ANY_SAMPLE_STATE,
DDS_ANY_VIEW_STATE, DDS_ANY_INSTANCE_STATE)
!= DDS_RETCODE_OK)
{
// ... error
}
for (i = 0; i < data_seq.length(); ++i) {
if (info_seq[i].valid_data) {
// check user_data for access control
if (data_seq[i].user_data[0] != 0x9) {
if (builtinTopicDataReader->get_subscriber()
->get_participant()
->ignore_publication(
info_seq[i].instance_handle)
!= DDS_RETCODE_OK) {
// ... error
}
}
}
}
if (builtinTopicDataReader->return_loan(data_seq, info_seq) !=
DDS_RETCODE_OK) {
...

787

16.4.3 Ignoring Topics

16.4.3 Ignoring Topics
The ignore_topic() operation instructs Connext DDS to locally ignore a Topic. This means it will locally
ignore any publication or subscription to the Topic.
DDS_ReturnCode_t ignore_topic (const DDS_InstanceHandle_t & handle)

Caution: There is no way to reverse this operation.
If you know that your application will never publish or subscribe to data under certain topics, you can use
this operation to save local resources.
The Topic to ignore is identified by the handle argument. This handle is the one that appears in the DDS_
SampleInfo retrieved when reading the DDS data samples from the built-in DataReader to the Topic.

16.4.4 Resource Limits Considerations for Ignored Entities
When an entity is ignored, Connext DDS adds it to an internal ‘ignore’ table whose resource limits are configured using the ignored_entity_allocation.max_count in the DOMAIN_PARTICIPANT_
RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 8.5.4). The behavior of Connext DDS when
this limit is exceeded can be modified by using the ignored_entity_replacement_kind in the same QoS
policy.
The default value for ignored_entity_replacement_kind is DDS_NO_REPLACEMENT_
IGNORED_ENTITY_REPLACEMENT, meaning that a call to the DomainParticipant’s ignore_
participant(), ignore_publication(), or ignore_subscription() will fail if the DomainParticipant has
ignored more entities than the limit set in ignored_entity_allocation.max_count entities.
When ignored_entity_replacement_kind is set to DDS_NOT_ALIVE_FIRST_IGNORED_
ENTITY_REPLACEMENT, a call to ignore_participant() will not fail when ignored_entity_allocation.max_count is exceeded, as long as there is one DomainParticipant already ignored. Instead, the call
will replace one of the existing DomainParticipants in the internal table. The remote DomainParticipant
that will be replaced is the one for which the local DomainParticipant had not received any message for
the longest time.
When a remote DomainParticipant is replaced in the ‘ignore’ table, it becomes un-ignored. Thus, the local
DomainParticipant would have to call ignore_participant() again to re-ignore the replaced entity.
Note: In this release, ignored publications and subscriptions are never replaced in the ‘ignore’ table. Since
this table also contains the ignored DomainParticipants, a call to ignore_participant() will fail if
ignored_entity_allocation.max_count is reached and none of the ignored entities is a DomainParticipant.

16.4.5 Supervising Endpoint Discovery
It is possible to control for which DomainParticipants endpoint discovery may occur. You can configure
this behavior with the enable_endpoint_discovery field in the DISCOVERY QosPolicy (DDS

788

16.4.5 Supervising Endpoint Discovery

Extension) (Section 8.5.2 on page 580):
l

l

When set to TRUE (the default value), endpoint discovery will automatically occur for every discovered DomainParticipant. This is the normal operation of the discovery process.
When set to FALSE, endpoint discovery will be disabled for every discovered DomainParticipant.
Then applications will have to manually enable endpoint discovery (described below) for the
DomainParticipants they are interested in communicating with. By disabling endpoint discovery,
the DomainParticipant will not store any state about remote endpoints and will not send local endpoint information to remote DomainParticipants.

When enable_endpoint_discovery is set to FALSE, you have two options after a remote DomainParticipant is discovered:
l

Call the DomainParticipant’s resume_endpoint_discovery() operation to enable endpoint discovery. After invoking this operation, the DomainParticipant will start to exchange endpoint information so that matching and communication can occur with the remote DomainParticipant.
DDS_ReturnCode_t resume_endpoint_discovery(
const DDS_InstanceHandle_t & remote_participant_handle)

Or
l

Call the DomainParticipant’s ignore_participant() operation to permanently ignore endpoint discovery with the remote DomainParticipant.

Setting enable_endpoint_discovery to FALSE enables application-level authentication use cases, in
which a DomainParticipant will resume endpoint discovery with a remote DomainParticipant after successful authentication at the application level. The following example shows how to provide access control
using this feature:
class MyParticipantBuiltinTopicDataListener :
public DDSDataReaderListener {
public:
virtual void on_data_available(DDSDataReader *reader);
// ...
};
void MyParticipantBuiltinTopicdataListener::on_data_available(
DDSDataReader *reader) {
DDSParticipantBuiltinTopicDataDataReader
*builtinTopicDataReader =
DDSParticipantBuiltinTopicDataDataReader *) reader;
DDS_ParticipantBuiltinTopicDataSeq data_seq;
DDS_SampleInfoSeq info_seq;
int = 0;
if (builtinTopicDataReader->take(

789

16.4.5 Supervising Endpoint Discovery

data_seq, info_seq,
DDS_LENGTH_UNLIMITED,
DDS_ANY_SAMPLE_STATE,
DDS_ANY_VIEW_STATE,
DDS_ANY_INSTANCE_STATE)!= DDS_RETCODE_OK){
// ... error
}
for (i = 0; i < data_seq.length(); ++i) {
if (info_seq[i].valid_data) {
DDSDomainParticipant * localParticipant =
builtinTopicDataReader->
get_subscriber()->get_participant();
DDS_ReturnCode_t retCode;
// check user_data for access control
if (data_seq[i].user_data[0] != 0x9) {
retCode = localParticipant->
ignore_participant(
info_seq[i].instance_handle);
}else {
retCode = localParticipant->
resume_endpoint_discovery(
info_seq[i].instance_handle)
}
}
}
if (builtinTopicDataReader->return_loan(
data_seq, info_seq)
!= DDS_RETCODE_OK) {
// ... error }
}

790

Chapter 17 Configuring QoS with XML
Connext DDS entities are configured by means of Quality of Service (QoS) policies, which may
be set programmatically in one of the following ways:
l

l
l

Directly when the entity is created as an additional argument to the create_() operation (or the Entity's constructor in the Modern C++ API).
Directly via the set_qos() operation on the entity.
Indirectly as a default QoS on the factory for the entity (set_default__qos() operations on Publisher, Subscriber, DomainParticipant, DomainParticipantFactory)

Entities can also be configured from an XML file or XML string. With this feature, you can
change QoS configurations simply by changing the XML file or string—you do not have to recompile the application. This chapter describes how to configure Connext DDS entities using XML:

17.1 Example XML File
The QoS configuration of a Entity can be loaded from an XML file or string.
The file contents must follow an important hierarchy: the file contains one or more libraries; each
library contains one or more profiles; each profile contains QoS settings.
Let's look at a very basic configuration file, just to get an idea of its contents. You will learn the
meaning of each line as you read the rest of this chapter:









791

17.2 QoS Libraries

KEEP_ALL_HISTORY_QOS


RELIABLE_RELIABILITY_QOS




KEEP_ALL_HISTORY_QOS


RELIABLE_RELIABILITY_QOS






KEEP_ALL_HISTORY_QOS






KEEP_ALL_HISTORY_QOS


RELIABLE_RELIABILITY_QOS


792

17.3 QoS Profiles




KEEP_ALL_HISTORY_QOS


RELIABLE_RELIABILITY_QOS






A QoS library can be reopened within the same configuration file or across different configuration files.
For example:


...

...

...



17.3 QoS Profiles
A QoS profile groups a set of related QoS, usually one per entity, identified by a name. For example:



KEEP_ALL_HISTORY_QOS


RELIABLE_RELIABILITY_QOS




KEEP_ALL_HISTORY_QOS


RELIABLE_RELIABILITY_QOS




Duplicate QoS profiles are not allowed. To overwrite a QoS profile, use QoS Profile Inheritance (Section
17.3.3 on page 797).

793

17.3.1 Built-in QoS Profiles

There are functions that allow you to create Entities using profiles, such as create_participant_with_profile() (Creating a DomainParticipant (Section 8.3.1 on page 556)), create_topic_with_profile() (Creating
Topics (Section 5.1.1 on page 202)), etc.
If you create an entity using a profile without a QoS definition or an inherited QoS definition (see QoS Profile Inheritance (Section 17.3.3 on page 797)) for that class of entity, Connext DDS uses the default QoS.
Example 1:



true




The DataReader QoS value in the profile BatchStrictReliableCommunicationProfile is inherited from
the profile StrictReliableCommunicationProfile.
Example 2:



true




The DataReader QoS value in the profile BatchProfile is the default Connext DDS QoS.

17.3.1 Built-in QoS Profiles
Several QoS profiles are built into the Connext DDS core libraries and can be used as starting points when
configuring QoS for your Connext DDS applications. There are two provided libraries, BuiltinQosLib
and BuiltinQosLibExp, and 34 different profiles. You can use any of these profiles as base profiles when
creating your own XML configurations or simply use these profiles directly in the DDS_*_create_*_
with_profile() APIs.
There are three types of built-in profiles:
l

Baseline.X.X.X profiles represent the QoS defaults for Connext DDS version X.X.X. The defaults
for the latest Connext DDS version can be accessed using the BuiltinQosLib::Baseline profile.

794

17.3.1 Built-in QoS Profiles

l

l

Generic.X profiles allow you to easily configure different features and communication use-cases
with Connext DDS. For example, there is a Generic.StrictReliable profile for use when your
application has a requirement for no data loss, regardless of the application domain.
Pattern.X profiles inherit from Generic.X profiles and allow you to configure various domain-specific communication use cases. For example, there is a Pattern.Alarm profile that can be used to
manage the generation and consumption of alarm events.

The USER_QOS_PROFILES.xml file generated by RTI Code Generator contains a profile that inherits
from the BuiltinQosLibExp::Generic.StrictReliable profile as an example of how to use these profiles in
your own application.
Example use-cases for these profiles:
l

l

l

To quickly enable RTI Monitoring Library by inheriting from the BuiltinQosLib::Generic.Monitoring.Common profile. (See note below.)
To easily revert to the default QoS values from a previous Connext DDS version by inheriting from
the correct BuiltinQosLib::Baseline.X.X.X profile.
To set up common use-case configurations and patterns such as strict reliability or large data communication by inheriting from one of the BuiltinQosLibExp::Generic.X or Pattern.X profiles.

To see the contents of the built-in QoS profiles:
In /resource/xml, you will find:
l

l

BaselineRoot.documentationONLY.xml—This file contains the root baseline QoS profile corresponding to the default values of Connext DDS 5.0.0.
BuiltinProfiles.documentationONLY.xml—This file contains the rest of the built-in QoS profiles.

Notes:
l

The built-in QoS profiles that enable RTI Monitoring Library set the property rti.monitor.create_
function. Consequently, they only work in Connext DDS applications in which the monitoring library can be loaded dynamically. Specifically, the built-in monitoring profiles will not work in these
situations:
l When the Connext DDS application links the monitoring libraries statically
l

When using a VxWorks 6.7 or 6.8 platform with Java1.

1VxWorks 6.7 and 6.8 Java platforms require custom supported libraries.

795

17.3.2 Overwriting Default QoS Values

For more information, see Part 9: RTI Monitoring Library (Section on page 1022)).
l

Some of the built-in profiles are experimental. All the experimental profiles are contained within the
library BuiltinQosLibExp.

17.3.2 Overwriting Default QoS Values
There are two ways to overwrite the default QoS used for new entities with values from a profile: programmatically and with an XML attribute.
l

l

l

You can overwrite the default QoS programmatically with set_default__qos_with_profile
() (where  is participant, topic, publisher, subscriber, datawriter, or datareader)
You can overwrite the default QoS using the XML attribute is_default_qos with the 
tag
Only for the DomainParticipantFactory: You can overwrite the default QoS using the XML attribute
is_default_participant_factory_profile. This attribute has precedence over is_default_qos if both
are set.

In the following example, the DataWriter and DataReader default QoS will be overwritten with the values
specified in a profile named ‘StrictReliableCommunicationProfile’:



KEEP_ALL_HISTORY_QOS


RELIABLE_RELIABILITY_QOS




KEEP_ALL_HISTORY_QOS


RELIABLE_RELIABILITY_QOS




If multiple profiles are configured to overwrite the default QoS, only the last one parsed applies.
Example:
In this example, the profile used to configure the default QoSs will be StrictReliableCommunicationProfile.

796

17.3.3 QoS Profile Inheritance


...


...


17.3.3 QoS Profile Inheritance
An individual QoS or profile can inherit values from other QoSs or profiles described in the XML file by
using the attribute, base_name.
Inheriting from other XML Files:
A QoS or QoS Profile may inherit values from other QoSs or QoS Profiles described in different XML
files. A QoS or profile can only inherit from other QoS policies or profiles that have already been loaded.
The order in which XML resources are loaded is described in How to Load XML-Specified QoS Settings
(Section 17.5 on page 810).
The following examples show how to inherit from other profiles:
Example 1:



...


...




...


...




The writer_qos and reader_qos in DerivedProfile inherit their values from the corresponding QoS in
BaseProfile.
Example 2:



797

17.3.3 QoS Profile Inheritance

...




The datareader_qos in DerivedProfile inherits its values from the datareader_qos of BaseProfile. In this
example, the datareader_qos definition is a shortcut for a profile definition with a single QoS.
Example 3:



...


...




...


...




The datawriter_qos in Profile2 inherits its values from the datawriter_qos in Profile1. The datareader_
qos in Profile2 will not inherit the values from the corresponding QoS in Profile1.
Example 4:



...


...




...


...


798

17.3.4 Topic Filters




...


...




The datawriter_qos in Profile3 inherits its values from the datawriter_qos in Profile2. The datareader_
qos in Profile3 inherits its values from the datareader_qos in Profile1.
Example 5:


...



...




The datareader_qos in DerivedProfile inherits its values from the datareader_qos in BaseProfile.

17.3.4 Topic Filters
A QoS profile may contain several writer, reader and topic QoSs. Connext DDS will select a QoS based
on the evaluation of a filter expression on the topic name. The filter expression is specified as an attribute
in the XML QoS definition. For example:



KEEP_ALL_HISTORY_QOS


RELIABLE_RELIABILITY_QOS




KEEP_ALL_HISTORY_QOS


RELIABLE_RELIABILITY_QOS



799

17.3.4 Topic Filters

128
128

128
1
1


...


If topic_filter is not specified in a QoS, Connext DDS will assume the filter '*'. The QoSs with an explicit
topic_filter attribute definition will be evaluated in order; they have precedence over a QoS without a
topic_filter expression.
The topic_filter attribute is only used with the following APIs:
DomainParticipantFactory:
l

get__qos_from_profile_w_topic_name() (where  may be topic, datareader, or
datareader; see Getting QoS Values from a QoS Profile (Section 8.2.5 on page 547))

DomainParticipant:
l

create_datawriter_with_profile() (see Creating DataWriters (Section 6.3.1 on page 266))

l

create_datareader_with_profile() (see Creating DataReaders (Section 7.3.1 on page 463)

l

create_topic_with_profile() (see Creating Topics (Section 5.1.1 on page 202))

Publisher:
l

create_datawriter_with_profile() (see Creating DataWriters (Section 6.3.1 on page 266))

Subscriber:
l

create_datareader_with_profile() (see Creating DataReaders (Section 7.3.1 on page 463))

Topic:
l

set_qos_with_profile() (see Setting Topic QosPolicies (Section 5.1.3 on page 204))

DataWriter:

800

17.3.4 Topic Filters

l

set_qos_with_profile() (see Changing QoS Settings After the Publisher Has Been Created (Section
6.2.4.3 on page 254))

DataReader:
l

set_qos_with_profile() (see Setting DataReader QosPolicies (Section 7.3.8 on page 482))

Other APIs will ignore QoSs with a topic_filter value different than "*". A QoS Profile with QoSs using
topic_filter can also inherits from other QoS Profiles. In this case, inheritance will consider the value of
the topic_filter expression.
Example 1:



...


...


...




...


...


...




The datawriter_qos with topic_filter T11 in DerivedProfile will inherit its values from the datawriter_
qos with topic_filter T1* in BaseProfile. The datawriter_qos with topic_filter T21 in DerivedProfile
will inherit its values from the datawriter_qos with topic_filter T2* in BaseProfile. The datawriter_qos
with topic_filter T31 in DerivedProfile will inherit its values from the datawriter_qos without topic_filter in BaseProfile.
Example 2:



...


801

17.3.5 QoS Profiles with a Single QoS


...




...


...




Although the topic_filter expressions do not match, the datawriter_qos with topic_filter T11 in
DerivedProfile will inherit its values from the datawriter_qos with topic_filter T2* in BaseProfile. topic_
filter is not used with inheritance from QoS to QoS. The datawriter_qos with topic_filter T21 in
DerivedProfile will inherit its values from the datawriter_qos with topic_filter T2* in BaseProfile.
Example 3:


...


...



In the case of a single QoS profile, although the topic_filter expressions do not match, the datawriter_qos
named DerivedQos with topic_filter T2 will inherit its values from the datawriter_qos named BaseQos
with topic_filter T1.

17.3.5 QoS Profiles with a Single QoS
The definition of an individual QoS outside a profile is a shortcut for defining a QoS profile with a single
QoS. For example:


KEEP_ALL_HISTORY_QOS



is equivalent to:




802

17.4 Configuring QoS with XML

KEEP_ALL_HISTORY_QOS




17.4 Configuring QoS with XML
To configure the QoS for an Entity using XML, use the following tags:
l


Note: The only QoS policies that can be configured for the DomainParticipantFactory are  and .

l



l



l



l



l

 or  (writer_qos is valid only with DTD validation)

l

 or  (reader_qos is valid only with DTD validation)

Each QoS can be identified by a name. The QoS can inherit its values from other QoSs described in the
XML file. For example:


KEEP_ALL_HISTORY_QOS



In the above example, the datawriter_qos named 'DerivedWriterQos' inherits the values from 'BaseWriterQos' in the library 'Lib'. The HistoryQosPolicy kind is set to KEEP_ALL_HISTORY_QOS.
Each XML tag with an associated name can be uniquely identified by its fully qualified name in C++
style.
The writer, reader and topic QoSs can also contain an attribute called topic_filter that will be used to associate a set of topics to a specific QoS when that QoS is part of a QoS profile. See Topic Filters (Section
17.3.4 on page 799) and URL Groups (Section 17.8 on page 814).

17.4.1 QosPolicies
The fields in a QosPolicy are described in XML using a 1-to-1 mapping with the equivalent C representation. For example, the Reliability QosPolicy is represented with the following C structures:

803

17.4.2 Sequences

struct DDS_Duration_t {
DDS_Long sec;
DDS_UnsignedLong nanosec;
}
struct DDS_ReliabilityQosPolicy {
DDS_ReliabilityQosPolicyKind kind;
DDS_Duration_t max_blocking_time;
}

The equivalent representation in XML is as follows:








17.4.2 Sequences
In general, sequences in QosPolicies are described with the following XML format:

...
...
...


Each element of the sequence is enclosed in an  tag. For example:



my name
my value


my name2
my value2




A sequence without elements represents a sequence of length 0. For example:





For sequences that may have a default initialization that is not empty (such as the initial_peers field in the
DISCOVERY QosPolicy (DDS Extension) (Section 8.5.2 on page 580)), using the above construct

804

17.4.2 Sequences

would result in an empty list and not the default value. So to simply show a sequence for the sake of completeness, but not change its default value, comment it out, as follows:





As a general rule, sequences defined in a derived1 QoS will replace the corresponding sequences in the
base QoS. For example, consider the following:




192.168.1.1
192.168.1.2








192.168.1.3





The initial peers sequence defined above in the participant QoS of MyDerivedProfile will contain a single
element with a value 192.168.1.3. The elements 192.168.1.1 and 192.168.1.2 will not be inherited.
However, there is one exception to this behavior. The  tag provides an attribute called inherit
that allows you to choose the inheritance behavior for the sequence defined within the tag.
The  tag provides an attribute called inherit that allows you to choose the inheritance behavior
for the sequence defined within the tag.
By default, the value of the attribute inherit is true. Therefore, the  tag defined within a derived
QoS profile will inherit its elements from the  tag defined within a base QoS profile.
In the following example, the property sequence defined in the participant QoS of MyDerivedProfile will
contain two properties:
l

dds.transport.UDPv4.builtin.send_socket_buffer_size will be inherited from the base profile and
have the value 524288.

1The concepts of derived and base QoS are described in QoS Profile Inheritance (Section 17.3.3 on page 797).

805

17.4.2 Sequences

l

dds.transport.UDPv4.builtin.recv_socket_buffer_size will overwrite the value defined in the
base QoS profile with 1048576.







dds.transport.UDPv4.builtin.send_socket_buffer_size

524288



dds.transport.UDPv4.builtin.recv_socket_buffer_size

2097152











dds.transport.UDPv4.builtin.recv_socket_buffer_size

1048576






To discard all the properties defined in the base QoS profile, set inherit to false.
In the following example, the property sequence defined in the participant QoS of MyDerivedProfile will
contain a single property named dds.transport.UDPv4.builtin.recv_socket_buffer_size, with a value of
1048576. The property dds.transport.UDPv4.builtin.send_socket_buffer_size will not be inherited.






dds.transport.UDPv4.builtin.send_socket_buffer_size

524288


806

17.4.3 Arrays



dds.transport.UDPv4.builtin.recv_socket_buffer_size

2097152










dds.transport.UDPv4.builtin.recv_socket_buffer_size

1048576






17.4.3 Arrays
In general, the arrays contained in the QosPolicies are described with the following XML format:

...
...
...


Each element of the array is enclosed in an  tag.
As a special case, arrays of octets are represented with a single XML tag enclosing an array of decimal/hexadecimal values between 0..255 separated with commas.
For example:

...



1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16





807

17.4.4 Enumeration Values

17.4.4 Enumeration Values
Enumeration values are represented using their C or Java string representation. For example:

DDS_KEEP_ALL_HISTORY_QOS


or

KEEP_ALL_HISTORY_QOS


When the XSD document is used for validation during editing (see XML File Validation During Editing
(Section 17.9.2 on page 816)), only the Java representation is valid.

17.4.5 Time Values (Durations)
You can use the following special values for fields that require seconds or nanoseconds:
l

DURATION_INFINITE_SEC or DDS_DURATION_INFINITE_SEC,

l

DURATION_ZERO_SEC or DDS_DURATION_ZERO_SEC,

l

DURATION_INFINITE_NSEC or DDS_DURATION_INFINITE_NSEC

l

DURATION_ZERO_NSEC or DDS_DURATION_ZERO_NSEC

For example:


DURATION_INFINITE_SEC
DURATION_INFINITE_NSEC



When the XSD document is used for validation during editing (see XML File Validation During Editing
(Section 17.9.2 on page 816)), only the values without the DDS prefix are considered valid.

17.4.6 Transport Properties
You can configure transport plugins using the DomainParticipant’s PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394).
l

Properties for the builtin transports are described in Setting Builtin Transport Properties with the
PropertyQosPolicy (Section 15.6 on page 748).

808

17.4.7 Thread Settings

l

Properties for other transport plugins such as RTI TCP Transport1are described in their respective
chapters in this manual.

For example:





dds.transport.UDPv4.builtin.parent.message_size_max

65507



dds.transport.UDPv4.builtin.send_socket_buffer_size

131072



dds.transport.UDPv4.builtin.recv_socket_buffer_size

131072





17.4.7 Thread Settings
See Table 19.1 XML Tags for ThreadSettings_t.

17.4.8 Entity Names
The name and role_name fields in the ENTITY_NAME QosPolicy (DDS Extension) (Section 6.5.9 on
page 374) have three distinct possible values: NULL, an empty string, and a non-empty string. Each of
these three states are specified in XML in a different way.
To specify that the name or role_name of an entity is NULL, use the xsi:nil attribute. The xsi:nil attribute
can be set to either "true" or "false". For example, to set the participant name to NULL:




To specify the empty string, leave the XML element empty:

1RTI TCP Transport is included with Connext DDS, but is not enabled by default.

809

17.5 How to Load XML-Specified QoS Settings





To specify a non-empty string:

"My Participant's Name"


17.5 How to Load XML-Specified QoS Settings
There are several ways to load XML QoS profiles into your application. In C, Traditional C++, Java and
.NET, it's the singleton DomainParticipantFactory that loads these profiles. Applications using the Modern
C++ API can create any number of instances of dds::core::QosProvider with different parameters to load
different QoS profiles or, they can use the singleton QosProvider::Default(). The profiles configured in the
default QosProvider are used when creating an Entity without a explicit QoS parameter.
Here are the various approaches, listed in load order:
l

l

l

l

l

$NDDSHOME/resource/xml/NDDS_QOS_PROFILES.xml
This file is loaded automatically if it exists (not the default) and ignore_resource_profile in the
PROFILE QosPolicy (DDS Extension) (Section 8.4.2 on page 573) is FALSE (the default).
NDDS_QOS_PROFILES.xml does not exist by default. However, NDDS_QOS_PROFILES.example.xml is shipped with the host bundle of the product; you can copy it to NDDS_QOS_
PROFILES.xml and modify it for your own use. The file contains the default QoS values that will
be used for all entity kinds. (First to be loaded)
URL Groups in NDDS_QOS_PROFILES
URL groups (see URL Groups (Section 17.8 on page 814)) separated by semicolons referenced by
the environment variable NDDS_QOS_PROFILES are loaded automatically if they exist and
ignore_environment_profile in PROFILE QosPolicy (DDS Extension) (Section 8.4.2 on page
573) is FALSE (the default).
/USER_QOS_PROFILES.xml
This file is loaded automatically if it exists and ignore_user_profile in PROFILE QosPolicy (DDS
Extension) (Section 8.4.2 on page 573) is FALSE (the default).
URL groups in url_profile
URL groups (see URL Groups (Section 17.8 on page 814)) referenced by url_profile (in
PROFILE QosPolicy (DDS Extension) (Section 8.4.2 on page 573)) will be loaded automatically if
specified.
XML strings in string_profile
The sequence of XML strings referenced by string_profile (in PROFILE QosPolicy (DDS Extension) (Section 8.4.2 on page 573)) will be loaded automatically if specified. (Last to be loaded)

810

17.5.1 Loading, Reloading and Unloading Profiles

You may use a combination of the above approaches.
The location of the XML documents (only files and strings are supported) is specified using URL (Uniform Resource Locator) format. For example:
l

File Specification: file:///usr/local/default_dds.xml

l

String Specification: str://"…"

If you omit the URL schema name, Connext DDS will assume a file name. For example:
l

File Specification: /usr/local/default_dds.xml

Duplicate QoS profiles are not allowed. Connext DDS will report an error message in these scenarios. To
overwrite a QoS profile, use QoS Profile Inheritance (Section 17.3.3 on page 797).
Several QoS profiles are built into the Connext DDS core libraries and can be used as starting points when
configuring QoS for your Connext DDS applications. For details, see Configuring QoS with XML (Section 17.4 on page 803).

17.5.1 Loading, Reloading and Unloading Profiles
You do not have to explicitly call load_profiles(). QoS profiles are loaded when any of these DomainParticipantFactory operations are called:
l

create_participant() (see Creating a DomainParticipant (Section 8.3.1 on page 556))

l

create_participant_with_profile() (see Creating a DomainParticipant (Section 8.3.1 on page 556))

l

l

l

l

get__qos_from_profile() (where  is participant, topic, publisher, subscriber,
datawriter, or datareader) (see Getting QoS Values from a QoS Profile (Section 8.2.5 on page
547))
get__qos_from_profile_w_topic_name() (where  is topic, datawriter, or
datareader) (see Getting QoS Values from a QoS Profile (Section 8.2.5 on page 547))
get_default_participant_qos() (see Getting and Setting Default QoS for DomainParticipants (Section 8.2.2 on page 545))
get_qos_profile_libraries() (See Retrieving a List of Available Libraries (Section 17.10.1 on page
823))

l

get_qos_profiles() (See Configuring QoS with XML (Section 17.4 on page 803))

l

load_profiles()

l

set_default_participant_qos_with_profile() (see Getting and Setting Default QoS for DomainParticipants (Section 8.2.2 on page 545))

811

17.6 XML File Syntax

l

l

set_default_library() (see Getting and Setting the Publisher’s Default QoS Profile and Library (Section 6.2.4.4 on page 255))
set_default_profile() (see Getting and Setting the Publisher’s Default QoS Profile and Library (Section 6.2.4.4 on page 255))

In the Modern C++ API, the previous operations cause the default QosProvider (QosProvider::Default())
to load the QoS profiles. Any other QosProvider that an application instantiates will load the QoS Profiles
it is configured to load in its constructor.
QoS profiles are reloaded when either of these DomainParticipantFactory operations are called:
l

reload_profiles()

l

set_qos() (see Getting, Setting, and Comparing QosPolicies (Section 4.1.7 on page 158))

It is important to distinguish between loading and reloading:
l

l

Loading only happens when there are no previously loaded profiles. This could be when the profiles
are loaded the first time or after a call to unload_profiles().
Reloading replaces all previously loaded profiles. Reloading a profile does not change the QoS of
entities that have already been created with previously loaded profiles.

The DomainParticipantFactory also has an unload_profiles() operation that frees the resources associated
with the XML QoS profiles.
DDS_ReturnCode_t unload_profiles()

17.6 XML File Syntax
The contents of the XML configuration file must follow an important hierarchy: the file contains one or
more libraries; each library contains one or more profiles; each profile contains QoS settings.
In addition, the file must follow these syntax rules:
l

The syntax is XML and the character encoding is UTF-8.

l

Opening tags are enclosed in <>; closing tags are enclosed in .

l

A tag value is a UTF-8 encoded string. Legal values are alphanumeric characters. The middleware’s
parser will remove all leading and trailing spacesafrom the string before it is processed.

aLeading and trailing spaces in enumeration fields will not be considered valid if you use the distributed XSD document to

do validation at run-time with a code editor (see URL Groups (Section 17.8 on page 814)).

812

17.6.1 Using Environment Variables in XML

l

For example,  value  is the same as value.

l

All values are case-sensitive unless otherwise stated.

l

Comments are enclosed as follows: .

l

The root tag of the configuration file must be  and end with .

l

The primitive types for tag values are specified in Table 17.1 Supported Tag Values.

Table 17.1 Supported Tag Values
Type

Format

Notes

yesa, 1, true, BOOLEAN_TRUE or DDS_BOOLEAN_TRUE: these all
mean TRUE
DDS_Boolean

Not case-sensitive
no, 0, false, BOOLEAN_FALSE or DDS_BOOLEAN_FALSE: these all
mean FALSE

DDS_Enum

A string. Legal values are those listed in the API Reference HTML
documentation for the C or Java API.

Must be specified as a string. (Do not use
numeric values.)

-2147483648 to 2147483647
or 0x80000000 to 0x7fffffff
DDS_Long

A 32-bit signed integer
or LENGTH_UNLIMITED
or DDS_LENGTH_UNLIMITED

0 to 4294967296
DDS_
or
UnsignedLong
0 to 0xffffffff

A 32-bit unsigned integer

String

All leading and trailing spaces are ignored
between two tags

UTF-8 character string

17.6.1 Using Environment Variables in XML
The text within an XML tag and attribute can refer to environment variable. To do so, use the following
notation:
$(MY_VARIABLE)

For example:

aThese values will not be considered valid if you use the distributed XSD document to do validation at run-time with a code

editor (see URL Groups (Section 17.8 on the next page)).

813

17.7 XML String Syntax


The name is $(MY_NAME)
The value is $(MY_VALUE)


When the Connext DDS XML parser parses the above tags, it will replace the references to environment
variables with their actual values.

17.7 XML String Syntax
XML profiles can be described using strings. This configuration is useful for architectures without a file
system.
There are two different ways to configure Entities via XML strings:
l

String URLs are prefixed by the URI schema str:// and enclosed in double quotes. For example:
str://"..."

l

The string URLs can be specified in the environment variable NDDS_QOS_PROFILES as well as
in the field url_profile in PROFILE QosPolicy (DDS Extension) (Section 8.4.2 on page 573).
Each string URL must contain a whole XML document.
The string_profile field in the PROFILE QosPolicy (DDS Extension) (Section 8.4.2 on page 573)
allows you to split an XML document into multiple strings. For example:
const char * MyXML[4] =
{
"",
"",
"",
""
};
factoryQos.profile.string_profile.from_array(MyXML,4);

Only one XML document can be specified with the string_profile field.

17.8 URL Groups
To provide redundancy and fault tolerance, you can specify multiple locations for a single XML document
via URL groups. The syntax of a URL group is:
[URL1 | URL2 | URL2 | ... | URLn]

For example:
[file:///usr/local/default_dds.xml | file:///usr/local/alternative_default_dds.xml]

Only one of the elements in the group will be loaded by Connext DDS, starting from the left.
814

17.9 How the XML is Validated

Brackets are not required for groups with a single URL.
The NDDS_QOS_PROFILES environment variable contains a set of URL groups separated by semicolons. For example, on Linux and Solaris systems (note: this should be entered in a single command line):
setenv NDDS_QOS_PROFILES
[file:///usr/local/default_dds.xml|file:///usr/local/alternative_default_dds.xml];
[str://""]

The url_profile field in the PROFILE QosPolicy (DDS Extension) (Section 8.4.2 on page 573) will contain a sequence of URL groups.

17.9 How the XML is Validated
17.9.1 Validation at Run-Time
Connext DDS validates the input XML files using a builtin Document Type Definition (DTD).
You can find a copy of the builtin DTD in /resource/schema/rti_dds_qos_profiles.dtd.
(This is only a copy of what the Connext DDS core uses. Changing this file has no effect unless you specify its path with the  tag, described below.)
You can overwrite the builtin DTD by using the XML tag, . For example, the following
indicates that Connext DDS must use a DTD file from a user’s directory to perform validation:


l

The DTD path can be absolute, or relative to the application's current working directory.

l

If the specified file does not exist, you will see the following error:
RTIXMLDtdParser_parse:!open DTD file

l
l

If you do not specify the DOCTYPE tag in the XML file, the builtin DTD is used.
The XML files used by Connext DDS can be versioned using the attribute version in the  tag.
For example:

...


815

17.9.2 XML File Validation During Editing

Although the attribute version is not required during the validation process, it helps to detect DTD
incompatibility scenarios by providing better error messages.
For example, if an application using Connext DDS 5.x.y tries to load an XML file from Connext
DDS 4.5z and there is some incompatibility in the XML content, the following parsing error will be
printed:
ATTENTION: The version declared in this file (4.5z) is different from the
version of Connext DDS (5.x.y). If these versions are not compatible, that
incompatibility could be the cause of this error.

17.9.2 XML File Validation During Editing
Connext DDS provides DTD and XSD files that describe the format of the XML content. We recommend
including a reference to one of these documents in the XML file that contains the QoS profiles—this
provides helpful features in code editors such as Visual Studio and Eclipse, including validation and autocompletion while you are editing the XML file.
The DTD and XSD definitions of the XML elements are in
/resource/schema/rti_dds_qos_profiles.dtd and /resource/schema/rti_dds_qos_profiles.xsd, respectively. ( is described in Paths Mentioned in Documentation (Section on page xxxviii).)
To include a reference to the XSD document in your XML file, use the attribute
xsi:noNamespaceSchemaLocation in the  tag. For example:


...


To include a reference to the DTD document in your XML file use the  tag. For example:

/resource/schema/rti_dds_qos_profiles.dtd">

...


We recommend including a reference to the XSD file in the XML documents because it provides stricter
validation and better auto-completion than the corresponding DTD file.

816

17.10 Using QoS Profiles in Your Connext DDS Application

17.10 Using QoS Profiles in Your Connext DDS Application
You can use the operations listed in Table 17.2 Operations for Working with QoS Profiles to refer to and
use QoS profiles (see URL Groups (Section 17.8 on page 814)) described in XML files and XML strings.
Table 17.2 Operations for Working with QoS Profiles
Working With ...

Profile-Related
Operations

Reference

DataReaders

set_qos_with_profile

Changing QoS Settings After DataReader Has Been Created (Section 7.3.8.3
on page 487)

DataWriters

set_qos_with_profile

Changing QoS Settings After the DataWriter Has Been Created (Section
6.3.15.3 on page 305)

817

17.10 Using QoS Profiles in Your Connext DDS Application

Table 17.2 Operations for Working with QoS Profiles
Working With ...

Profile-Related
Operations

Reference

create_datareader_with_profile

Creating DataReaders (Section 7.3.1 on page 463)

create_datawriter_with_profile

Creating DataWriters (Section 6.3.1 on page 266)

create_publisher_with_profile

Creating Publishers (Section 6.2.2 on page 249)

create_subscriber_with_profile

Creating Subscribers (Section 7.2.2 on page 445)

create_topic_with_profile

Creating Topics (Section 5.1.1 on page 202)

get_default_library
get_default_profile

Getting and Setting DomainParticipant’s Default QoS Profile and Library
(Section 8.3.6.4 on page 567)

get_default_profile_library
set_default_datareader_qos_
with_profile
DomainParticipants
set_default_datawriter_qos_
with_profile
set_default_library
set_default_profile

Getting and Setting Default QoS for Child Entities (Section 8.3.6.5 on page
568)

Getting and Setting DomainParticipant’s Default QoS Profile and Library
(Section 8.3.6.4 on page 567)

set_default_publisher_qos_
with_profile
set_default_subscriber_qos_
with_profile

Getting and Setting Default QoS for Child Entities (Section 8.3.6.5 on page
568)

set_default_topic_qos_with_
profile
set_qos_with_profile

Changing QoS Settings After DomainParticipant Has Been Created (Section
8.3.6.3 on page 566)

818

17.10 Using QoS Profiles in Your Connext DDS Application

Table 17.2 Operations for Working with QoS Profiles
Working With ...

Profile-Related
Operations

DomainParticipantFactory create_participant_with_profile

Reference
Creating a DomainParticipant (Section 8.3.1 on page 556)

819

17.10 Using QoS Profiles in Your Connext DDS Application

Table 17.2 Operations for Working with QoS Profiles
Working With ...

Profile-Related
Operations
get_datareader_qos_from_
profile

Reference

Getting QoS Values from a QoS Profile (Section 8.2.5 on page 547)

820

17.10 Using QoS Profiles in Your Connext DDS Application

Table 17.2 Operations for Working with QoS Profiles
Working With ...

Profile-Related
Operations

Reference

get_datawriter_qos_from_
profile
get_datawriter_qos_from_
profile_w_topic_name
get_datareader_qos_from_
profile_w_topic_name
get_default_library
get_default_profile

Getting and Setting the DomainParticipantFactory’s Default QoS Profile and
Library (Section 8.2.1.1 on page 544)

get_default_profile_library
get_participant_qos_from_
profile
get_publisher_qos_from_profile
get_subscriber_qos_from_
profile

Getting QoS Values from a QoS Profile (Section 8.2.5 on page 547)

get_topic_qos_from_profile
get_topic_qos_from_profile_w_
topic_name
get_qos_profiles

Retrieving a List of Available QoS Profiles (Section 17.10.2 on page 823)

get_qos_profile_libraries

Retrieving a List of Available Libraries (Section 17.10.1 on page 823)

load_profiles
Loading, Reloading and Unloading Profiles (Section 17.5.1 on page 811)
reload_profiles
set_default_participant_qos_
with_profile
set_default_library
set_default_profile
unload_profiles

Getting and Setting Default QoS for DomainParticipants (Section 8.2.2 on
page 545)
Getting and Setting the DomainParticipantFactory’s Default QoS Profile and
Library (Section 8.2.1.1 on page 544)
Loading, Reloading and Unloading Profiles (Section 17.5.1 on page 811)

821

17.10 Using QoS Profiles in Your Connext DDS Application

Table 17.2 Operations for Working with QoS Profiles
Working With ...

Profile-Related
Operations
create_datawriter_with_profile

Reference
Creating Publishers (Section 6.2.2 on page 249)

get_default_library
get_default_profile

Getting and Setting the Publisher’s Default QoS Profile and Library (Section
6.2.4.4 on page 255)

get_default_profile_library
Publishers

set_default_datawriter_qos_
with_profile
set_default_library
set_default_profile

Getting and Setting Default QoS for DataWriters (Section 6.2.4.5 on page
256)
Getting and Setting the Publisher’s Default QoS Profile and Library (Section
6.2.4.4 on page 255)

set_qos_with_profile

Changing QoS Settings After the Publisher Has Been Created (Section
6.2.4.3 on page 254)

create_datareader_with_profile

Creating DataReaders (Section 7.3.1 on page 463)

get_default_library
get_default_profile

Getting and Settings Subscriber’s Default QoS Profile and Library (Section
7.2.4.4 on page 451)

get_default_profile_library
Subscribers

set_default_datareader_qos_
with_profile
set_default_library
set_default_profile

Topics

Getting and Setting Default QoS for DataReaders (Section 7.2.4.5 on page
452)
Getting and Settings Subscriber’s Default QoS Profile and Library (Section
7.2.4.4 on page 451)

set_qos_with_profile

Changing QoS Settings After Subscriber Has Been Created (Section 7.2.4.3
on page 450)

set_qos_with_profile

Setting Topic QosPolicies (Section 5.1.3 on page 204)

Note: For the Modern C++ API, please refer to the RTI Connext DDS API Reference
HTML documentation, Configuring QoS Profiles with XML.

822

17.10.1 Retrieving a List of Available Libraries

17.10.1 Retrieving a List of Available Libraries
To get a list of available QoS libraries, call the DomainParticipantFactory’s get_qos_profile_libraries()
operation, which returns the names of all QoS libraries that have been loaded by Connext DDS.
DDS_ReturnCode_t get_qos_profile_libraries (struct DDS_StringSeq *profile_names)

17.10.2 Retrieving a List of Available QoS Profiles
To get a list of available QoS profiles, call the DomainParticipantFactory’s get_qos_profiles() operation,
which returns the names of all profiles within a specified QoS library. Either the input QoS library name
must be specified or the default profile library must have been set prior to calling this function.
DDS_ReturnCode_t get_qos_profiles (struct DDS_StringSeq *profile_names,
const char *library_name)

17.11 Configuring Logging Via XML
Logging can be configured via XML using the DomainParticipantFactory’s LoggingQosPolicy. See Configuring Logging via XML (Section 21.2.2 on page 871) for additional details.

823

Chapter 18 Multi-channel DataWriters
In Connext DDS, producers publish data to a Topic, identified by a topic name; consumers subscribe to a Topic and optionally to specific content by means of a content-filter expression.
A Market Data Example:
A producer can publish data on the Topic "MarketData" which can be defined as a structured
record containing fields that identify the exchange (e.g., "NYSE" or "NASDAQ"), the stock symbol (e.g., "APPL" or "JPM"), volume, bid and ask prices, etc.
Similarly, a consumer may want to subscribe to data on the "MarketData" Topic, but only if the
exchange is "NYSE" or the symbol starts with the letter "M." Or the consumer may want all the
data from the "NYSE" whose volume exceeds a certain threshold, or may want MarketData for a
specific stock symbol, regardless of the exchange, and so on.
The middleware’s efficient implementation of content-filtering is critical for scenarios such as the
above "Market Data" example, where there are large numbers of consumers, large volumes of
data, or Topics that transmit information about many data-objects or subjects (e.g., individual
stocks).
Traditionally, middleware products use four approaches to implement content filtering: Producerbased, Consumer-based, Server-based, and Network Switch-based.
l

Producer-based approaches push the burden of filtering to the producer side. The producer knows what each consumer wants and delivers to the consumer only the data that
matches the consumer's filter. This approach is suitable when using point-to-point protocols
such as TCP—it saves bandwidth and lowers the load on the consumer—but it does not
work if data is distributed via multicast. Also, this approach does not scale to large numbers
of consumers, because the producer would be overburdened by the need to filter for each
individual consumer.

824

18.1 What is a Multi-channel DataWriter?

l

l

l

Consumer-based approaches push the burden of filtering to the consumer side. The producer
sends all the data to every consumer and the middleware on the consumer side decides whether the
application wants it or not, automatically filtering the unwanted data. This approach is simple and
fits well in systems that use multicast protocols as a transport. But the approach is not efficient for
consumers that want small subsets of the data, since the consumers have to spend a lot of time filtering unwanted data. This approach is also unsuitable for systems with large volumes of data, such
as the above Market Data system.
Server-based approaches push the burden of filtering to a third component: a server or broker.
This approach has some scalability advantages—the server can be run on a more powerful computer
and can be federated to handle a large number of consumers. Some providers also provide hardware-assisted filtering in the server. However, the server-based approach significantly increases
latency and jitter. It is also far more expensive to deploy and manage.
Network Switch-based approaches leverage the network hardware, specifically advanced (IGMP
snooping) network switches, to offload most of the burden of filtering from the producers and consumers without introducing additional hardware, servers or proxies. This approach preserves the low
latency and ease of deployment of the brokerless approaches while still providing most of the offloading and scalability benefits of the broker.

RTI supports the producer-based, consumer-based and network-switch approaches to content filtering:
l

l

RTI automatically uses the producer-based and consumer-based approaches as soon as it detects a
consumer that specifies a content filter. The producer-based approach is used if the consumer is
receiving data over a point-to-point protocol (i.e., not multicast) and the number of consumers that
specify filters is reasonably low (below 32). Otherwise, RTI uses a subscriber-based approach.
To use the more scalable network-switched based approach, an application must configure the
DataWriter as a Multi-channel DataWriter. This concept is described in the following section.

18.1 What is a Multi-channel DataWriter?
A Multi-channel DataWriter is a DataWriter that is configured to send data over multiple multicast
addresses, according to some filtering criteria applied to the data.
To determine which multicast addresses will be used to send the data, the middleware evaluates a set of filters that are configured for the DataWriter. Each filter "guards" a channel—a set of multicast addresses.
Each time a multi-channel DataWriter writes data, the filters are applied. If a filter evaluates to true, the
data is sent over that filter’s associated channel (set of multicast addresses). We refer to this type of filter as
a Channel Guard filter.

825

18.1 What is a Multi-channel DataWriter?

Figure 18.1 Multi-channel Data Flow

826

18.1 What is a Multi-channel DataWriter?

Figure 18.2 Multi-Channel Evaluation

Multi-channel DataWriters can be used to trade off network bandwidth with the unnecessary processing of
unwanted data for situations where there are multiple DataReaders who are interested in different subsets
of data that come from the same data stream (Topic). For example, in Financial applications, the data
stream may be quotes for different stocks at an exchange. Applications usually only want to receive data
(quotes) for only a subset of the stocks being traded. In tracking applications, a data stream may carry
information on hundreds or thousands of objects being tracked, but again, applications may only be interested in a subset.
The problem is that the most efficient way to deliver data to multiple applications is to use multicast so that
a data value is only sent once on the network for any number of subscribers to the data. However, using
multicast, an application will receive all of the data sent and not just the data in which it is interested, thus
extra CPU time is wasted to throw away unwanted data. With this QoS, you can analyze the data-usage
patterns of your applications and optimize network vs. CPU usage by partitioning the data into multiple
multicast streams. While network bandwidth is still being conserved by sending data only once using multicast, most applications will only need to listen to a subset of the multicast addresses and receive a reduced
amount of unwanted data.
Note: Your system can gain more of the benefits of using multiple multicast groups if your network uses
Layer 2 Ethernet switches. Layer 2 switches can be configured to only route multicast packets to those
ports that have added membership to specific multicast groups. Using those switches will ensure that only
the multicast packets used by applications on a node are routed to the node; all others are filtered-out by
the switch.

827

18.2 How to Configure a Multi-channel DataWriter

18.2 How to Configure a Multi-channel DataWriter
To configure a multi-channel DataWriter, simply define a list of all its channels in the DataWriter’s
MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page 386).
Each channel consists of filter criterion to apply to the data and a set of multicast destinations (transport,
address, port) that will be used for sending data that matches the filter. You can think of this sequence of
channels as a table like the one shown below:
If the Data Matches this Filter...

Send the Data to these Multicast Destinations

Symbol MATCH '[A-K]*

UDPv4:225.0.0.1:9000

Symbol MATCH '[L-Q]*

UDPv4:225.0.0.2:9001

Symbol MATCH '[P-Z]*

UDPv4:225.0.0.3:9002; 225.0.0.4:9003;

The example C++ code in Using the MULTI_CHANNEL QosPolicy (Section Figure 18.3 on the next
page) shows how to configure the channels.

828

18.2.1 Limitations

Figure 18.3 Using the MULTI_CHANNEL QosPolicy

// initialize writer_qos with default values
publisher->get_default_datawriter_qos(writer_qos);
// Initialize MULTI_CHANNEL Qos Policy
// Assign the filter name
// Possible options: DDS_STRINGMATCHFILTER_NAME, DDS_SQLFILTER_NAME
writer_qos.multi_channel.filter_name =
(char*) DDS_STRINGMATCHFILTER_NAME;
// Create two channels
writer_qos.multi_channel.channels.ensure_length(2,2);
// First channel
writer_qos.multi_channel.channels[0].filter_expression =
DDS_String_dup("Symbol MATCH '[A-M]*'");
writer_qos.multi_channel.channels[0].
multicast_settings.ensure_length(1,1);
writer_qos.multi_channel.channels[0].
multicast_settings[0].receive_port = 8700;
writer_qos.multi_channel.channels[0].
multicast_settings[0].receive_address =
DDS_String_dup("239.255.1.1");
// Second channel
writer_qos.multi_channel.channels[1].
multicast_settings.ensure_length(1,1);
writer_qos.multi_channel.channels[1].
multicast_settings[0].receive_port = 8800;
writer_qos.multi_channel.channels[1].
multicast_settings[0].receive_address =
DDS_String_dup("239.255.1.2");
writer_qos.multi_channel.channels[1].filter_expression =
DDS_String_dup("Symbol MATCH '[N-Z]*'");
// Create writer
writer = publisher->create_datawriter(
topic, writer_qos, NULL, DDS_STATUS_MASK_NONE);

The MULTI_CHANNEL QosPolicy is propagated along with discovery traffic. The value of this policy
is available in the builtin topic for the publication (see the locator_filter field in Table 16.2 Publication
Built-in Topic’s Data Type (DDS_PublicationBuiltinTopicData)).

18.2.1 Limitations
When considering use of a multi-channel DataWriter, please be aware of the following limitations:
l

A DataWriter that uses the MULTI_CHANNEL QosPolicy will ignore multicast and unicast
addresses specified on the reader side through the TRANSPORT_MULTICAST QosPolicy (DDS
Extension) (Section 7.6.5 on page 529) and TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24 on page 412). The DataWriter will not publish DDS samples on these

829

18.3 Multi-Channel Configuration on the Reader Side

locators.
l

l
l

l

l

Multi-channel DataWriters cannot be configured to use the Durable Writer History feature
(described in Durable Writer History (Section 12.3 on page 681)).
Multi-channel DataWriters do not support fragmentation of large data.
Multi-channel DataWriters cannot be configured for asynchronous publishing (described in
ASYNCHRONOUS_PUBLISHER QosPolicy (DDS Extension) (Section 6.4.1 on page 313)).
Multi-channel DataWriters rely on the rtps_object_id in the DATA_WRITER_PROTOCOL
QosPolicy (DDS Extension) (Section 6.5.3 on page 347) to be DDS_RTPS_AUTO_ID (which
causes automatic assignment of object IDs to channels).
To guarantee reliable delivery, a DataReader's PRESENTATION QosPolicy (Section 6.4.6 on
page 330) must be set to per-instance ordering (DDS_INSTANCE_PRESENTATION_QOS, the
default value), instead of per-topic ordering (DDS_TOPIC_PRESENTATION_QOS), and the
matching DataWriter's MULTI_CHANNEL QosPolicy (DDS Extension) (Section 6.5.14 on page
386) must use expressions that only refer to key fields.

18.3 Multi-Channel Configuration on the Reader Side
No special changes are required in a subscribing application to get data from a multi-channel DataWriter.
If you want the DataReader to subscribe to only a subset of the channels, use a ContentFilteredTopic, as
described in ContentFilteredTopics (Section 5.4 on page 212). For example:
// Create a content filtered topic
contentFilter =
participant->create_contentfilteredtopic_with_filter(
"FilteredTopic",
topic,
"symbol MATCH 'NYE/BAC,NASDAQ/MSFT,NASDAQ/GOOG",
parameters,
DDS_STRINGMATCHFILTER_NAME);
// Create a DataReader that uses the content filtered topic
reader = subscriber->create_datareader(contentFilter,
DDS_DATAREADER_QOS_DEFAULT,
NULL,0);

From there, Connext DDS takes care of all the necessary steps:
l

l

l

The DataReader automatically discovers all the DataWriters—including multi-channel
DataWriters—for the Topic it subscribes to.
When the DataReader discovers a multi-channel DataWriter, it also discovers the list of channels
used by that DataWriter.
When the multi-channel DataWriter discovers a DataReader, it also discovers the content filters specified by that DataReader, if any.

830

18.3 Multi-Channel Configuration on the Reader Side

With all this information, Connext DDS automatically determines which channels are of "interest" to the
DataReader.
A DataReader is interested in a channel if and only if the set of data values for which the channel guard filter evaluates to TRUE intersects the set of data values for which the DataReader's content filter evaluates
to TRUE. If a DataReader does not use a content filter, then it is interested in all the channels.
Figure 18.4 Filter Intersection

In this scenario, the DataReader is interested in Channel1 and Channel2, but not Channel3.

Market Data Example, continued:
If the channel guard filter for Channel 1 is 'Symbol MATCH '[A-K]*' then the channel will only transfer
data for stocks whose symbol starts with a letter in the A to K range.
That is, it will transfer data on 'APPL', "GOOG', and 'IBM', but not on 'MSFT', 'ORCL', or 'YHOO'.
Channel 1 will be of interest to DataReaders whose content filter includes at least one stock whose symbol
starts with a letter in the A to K range.

831

18.4 Where Does the Filtering Occur?

A DataReader that specifies a content filter such as "Symbol MATCH 'IBM, YHOO' " will be interested
in Channel1.
A DataReader that specifies a content filter such as "Symbol MATCH '[G-M]*'" will also be interested in
Channel1.
A DataReader that specifies a content filter such as "Symbol MATCH '[M-T]*' " will not be interested in
Channel1.

18.4 Where Does the Filtering Occur?
If multi-channel DataWriters are used, the filtering can occur in three places:
l

Filtering at the DataWriter (Section 18.4.1 below)

l

Filtering at the DataReader (Section 18.4.2 below)

l

Filtering on the Network Hardware (Section 18.4.3 on the next page)

18.4.1 Filtering at the DataWriter
Each time data is written, the DataWriter evaluates each of the channel guard filters to determine which
channels will transmit the data. This filtering occurs on the DataWriter.
Filtering on the DataWriter side is scalable because the number of filter evaluations depends only on the
number of channels, not on the number of DataReaders. Usually, the number of channels is smaller than
the number of possible DataReaders.
As explained in Performance Considerations (Section 18.7 on page 835), if the channel guard filters are
configured to only look at the "key" fields in the data, the channel filtering becomes a very efficient lookup
operation.

18.4.2 Filtering at the DataReader
The DataReader will listen on the multicast addresses that correspond to the channels of interest (see
Using the MULTI_CHANNEL QosPolicy (Section Figure 18.3 on page 829)). When a channel is 'of
interest', it means that it is possible for the channel to transmit data that meets the content filter of the
DataReader, however the channel may also transmit data that does not pass the DataReader's content filter. Therefore, the DataReader has to filter all incoming data on that channel to determine if it passes its
content filter.
Market Data Example, continued:
Channel 1, identified by guard filter "Symbol MATCH '[A-M]*'", will be of interest to DataReaders
whose content filter includes at least one stock whose symbol starts with a letter in the A to K range.

832

18.4.3 Filtering on the Network Hardware

A DataReader with content filter "Symbol MATCH 'GOOG'" will listen on Channel1.
In addition to 'GOOG', the DataReader will also receive DDS samples corresponding to stock symbols
such as 'MSFT' and 'APPL'. The DataReader must filter these DDS samples out.
As explained in Performance Considerations (Section 18.7 on page 835), if the DataReader’s content filters are configured to only look at the "key" fields in the data, the DataReader filtering becomes a very efficient lookup operation.

18.4.3 Filtering on the Network Hardware
DataReaders will only listen to multicast addresses that correspond to the channels of interest. The multicast traffic generated in other channels will be filtered out by the network hardware (routers, switches).
Layer 3 routers will only forward multicast traffic to the actual destination ports. However, by default,
layer 2 switches treat multicast traffic as broadcast traffic. To take advantage of network filtering with layer
2 devices, they must be configured with IGMP snooping enabled (see Network-Switch Filtering (Section
18.7.1 on page 835)).

18.5 Fault Tolerance and Redundancy
To achieve fault tolerance and redundancy, configure the DataWriter’s MULTI_CHANNEL QosPolicy
(DDS Extension) (Section 6.5.14 on page 386) to publish a DDS sample over multiple channels or over
different multicast addresses within a single channel. Figure 18.5 Using the MULTI_CHANNEL
QosPolicy with Overlapping Channels below shows how to use overlapping channels.
If a DDS sample is published to multiple multicast addresses, a DataReader may receive multiple copies
of the DDS sample. By default, duplicates are discarded by the DataReader and not provided to the application. To change this default behavior, use the Durable Reader State property, dds.data_reader.state.filter_redundant_samples (see How To Configure a DataReader for Durable Reader State
(Section 12.4.4 on page 690)).
Figure 18.5 Using the MULTI_CHANNEL QosPolicy with Overlapping Channels
// initialize writer_qos with default values
publisher->get_default_datawriter_qos(writer_qos);
// Initialize MULTI_CHANNEL Qos Policy
// Assign the filter name
// Possible options: DDS_STRINGMATCHFILTER_NAME and DDS_SQLFILTER_NAME
writer_qos.multi_channel.filter_name = (char*) DDS_STRINGMATCHFILTER_NAME;
// Create two channels
writer_qos.multi_channel.channels.ensure_length(2,2);
// First channel
writer_qos.multi_channel.channels[0].filter_expression =
DDS_String_dup("Symbol MATCH '[A-M]*'");

833

18.6 Reliability with Multi-Channel DataWriters

writer_qos.multi_channel.channels[0].multicast_settings.ensure_length(2,2);
writer_qos.multi_channel.channels[0].multicast_settings[0].receive_port = 8700;
writer_qos.multi_channel.channels[0].multicast_settings[0].receive_address =
DDS_String_dup("239.255.1.1");
// Second channel
writer_qos.multi_channel.channels[1].multicast_settings.ensure_length(1,1);
writer_qos.multi_channel.channels[1].multicast_settings[0].receive_port = 8800;
writer_qos.multi_channel.channels[1].multicast_settings[0].receive_address =
DDS_String_dup("239.255.1.2");
writer_qos.multi_channel.channels[1].filter_expression =
DDS_String_dup("Symbol MATCH '[C-Z]*'");
// Symbols starting with [C-M] will be published in two different channels
// Create writer
writer = publisher->create_datawriter(
topic, writer_qos, NULL, DDS_STATUS_MASK_NONE);

18.6 Reliability with Multi-Channel DataWriters
18.6.1 Reliable Delivery
Reliable delivery is only guaranteed when the access_scope in the Subscriber's PRESENTATION
QosPolicy (Section 6.4.6 on page 330) is set to DDS_INSTANCE_PRESENTATION_QOS (default
value) and the filters in the DataWriter's MULTI_CHANNEL QosPolicy (DDS Extension) (Section
6.5.14 on page 386)) are keyed-only based.
Market Data Example, continued:
Given the following IDL description for our MarketData topic type:
Struct MarketData {
string<255> Symbol; //@key
double Price;
}

A guard filter "Symbol MATCH 'APPL'" is keyed-only based.
A guard filter "Symbol MATCH 'APPL' and Price < 100" is not keyed-only based.
If any of the guard filters are based on non-key fields, Connext DDS only guarantees reception of the
most recent data from the multi-channel DataWriter.

18.6.2 Reliable Protocol Considerations
Reliability is maintained on a per-channel basis. Each channel has its own reliability channel send window:
l

low_watermark and high_watermark: The low and high watermarks control the send-window
levels (when not using batching, this is a number of DDS samples; when using batching, this is a

834

18.7 Performance Considerations

number of batches) that determine when to switch between regular and fast heartbeat rates (see High
and Low Watermarks (Section 6.5.3.1 on page 352)). With multi-channel DataWriters, high_
watermark and low_watermark are computed from the channel with the smaller send-window
size and they apply to all the channels. Therefore, because the watermark is determined by the channel with the smallest send-window, periodic heartbeating cannot be controlled on a per-channel
basis.
l

heartbeats_per_max_samples: This field defines the number of piggyback heartbeats per current
send-window. For multi-channel DataWriters, piggyback heartbeats are sent per channel. The sendwindow size that is used to calculate the piggyback heartbeat rate is the smallest across all channels..

18.7 Performance Considerations
18.7.1 Network-Switch Filtering
By default, multicast traffic is treated as broadcast traffic by layer 2 switches. To avoid flooding the network with broadcast traffic and take full advantage of network filtering, the layer 2 switches should be configured to use IGMP snooping. Refer to your switch’s manual for specific instructions.
When IGMP snooping is enabled, a switch can route a multicast packet to just those ports that subscribe to
it, as seen in IGMP Snooping (Section Figure 18.6 below).
Figure 18.6 IGMP Snooping

18.7.2 DataWriter and DataReader Filtering
Where Does the Filtering Occur? (Section 18.4 on page 832) describes the three places where filtering can
occur with Multi-channel DataWriters. To improve performance when filtering occurs on the reader
and/or writer sides, use filter expressions that are only based on keys (see DDS Samples, Instances, and

835

18.7.2 DataWriter and DataReader Filtering

Keys (Section 2.3.1 on page 14)). Then the results of the filter are cached in a hash table on a per-key
basis.

Market Data Example, continued:
The filter expressions in the Market Data example are based on the value of the field, Symbol. To make
filter operations on this field more efficient, declare Symbol as a key. For example:
struct {
string Symbol; //@key
}

You can also improve performance by increasing the number of buckets associated with the hash table. To
do so, use the instance_hash_buckets field in the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on
page 405) on both the writer and reader sides. A higher number of buckets will provide better performance, but requires more resources.

836

Chapter 19 Connext DDS Threading Model
This chapter describes the internal threads that Connext DDS uses for sending and receiving data,
maintaining internal state, and calling user code when events occur such as the arrival of new DDS
data samples. It may be important for you to understand how these threads may interact with your
application.
A DomainParticipant uses three types of threads. The actual number of threads depends on the
configuration of various QosPolicies as well as the implementation of the transports used by the
DomainParticipant to send and receive data.
Through various QosPolicies, the user application can configure the priorities and other properties
of the threads created by Connext DDS. In real-time systems, the user often needs to set the priorities of all threads in an application relative to each other for the proper operation of the system.
This chapter includes:

19.1 Database Thread
Connext DDS uses internal data structures to store information about locally-created and remotelydiscovered Entities. In addition, it will store various objects and data used by Connext DDS for
maintaining proper communications between applications. This “database” is created for each
DomainParticipant.
As Entities and objects are created and deleted during the normal operation of the user application,
different entries in the database may be created and deleted as well. Because multiple threads may
access objects stored in the database simultaneously, the deletion and removal of an object from the
database happens in two phases to support thread safety.
When an entry/object in the database is deleted either through the actions of user code or as a result
of a change in system state, it is only marked for deletion. It cannot be actually deleted and
removed from the database until Connext DDS can be sure that no threads are still accessing the

837

19.2 Event Thread

object. Instead, the actual removal of the object is delegated to an internal thread that Connext DDS
spawns to periodically wake up and purge the database of deleted objects.
This thread is known as the Database thread (also referred to as the database cleanup thread).
l

Only one Database thread is created for each DomainParticipant.

The DATABASE QosPolicy (DDS Extension) (Section 8.5.1 on page 577) of the DomainParticipant
configures both the resources used by the database as well as the properties of the cleanup thread. Specifically, the user may want to use this QosPolicy to set the priority, stack size and thread options of the
cleanup thread. You must set these options before the DomainParticipant is created, because once the
cleanup thread is started as a part of participant creation, these properties cannot be changed.
The period at which the database-cleanup thread wakes up to purge deleted objects is also set in the
DATABASE QosPolicy. Typically, this period is set to a long time (on the order of a minute) since there
is no need to waste CPU cycles to wake up a thread only to find nothing to do.
However, when a DomainParticipant is destroyed, all of the objects created by the DomainParticipant
will be destroyed as well. Many of these objects are stored in the database, and thus must be destroyed by
the cleanup thread. The DomainParticipant cannot be destroyed until the database is empty and is destroyed itself. Thus, there is a different parameter in the DATABASE QosPolicy, shutdown_cleanup_
period, that is used by the database cleanup thread when the DomainParticipant is being destroyed. Typically set to be on the order of a second, this parameter reduces the additional time needed to destroy a
DomainParticipant simply due to waiting for the cleanup thread to wake up and purge the database.

19.2 Event Thread
During operation, Connext DDS must wake up at different intervals to check the condition of many different time-triggered or periodic events. These events are usually to determine if something happened or
did not happen within a specified time. Often the condition must be checked periodically as long as the
Entity for which the condition applies still exists. Also, the DomainParticipant may need to do something
periodically to maintain connections with remote Entities.
For example, the DEADLINE QosPolicy (Section 6.5.5 on page 363) is used to ensure that DataWriters
have published data or DataReaders have received data within a specified time period. Similarly, the
LIVELINESS QosPolicy (Section 6.5.13 on page 382) configures Connext DDS both to check periodically to see if a DataWriter has sent a liveliness message and to send liveliness messages periodically
on the behalf of a DataWriter. As a last example, for reliable connections, heartbeats must be sent periodically from the DataWriter to the DataReader so that the DataReader can acknowledge the data that it
has received, see Reliable Communications (Section Chapter 10 on page 629).
Connext DDS uses an internal thread, known as the Event thread, to do the following:

838

19.3 Receive Threads

l

Check whether or not deadlines have been missed

l

Invoke user-installed Listener callbacks to notify the application of missed deadlines

l

Send heartbeats to maintain reliable connections

Note: Only one Event thread is created per DomainParticipant.
The EVENT QosPolicy (DDS Extension) (Section 8.5.5 on page 602) of the DomainParticipant configures both the properties and resources of the Event thread. Specifically, the user may want to use this
QosPolicy to set the priority, stack size and thread options of the Event thread. You must set these options
before the DomainParticipant is created, because once the Event thread is started as a part of participant
creation, these properties cannot be changed.
The EVENT QosPolicy also configures the maximum number of events that can be handled by the Event
thread. While the Event thread can only service a single event at a time, it must maintain a queue to hold
events that are pending. The initial_count and max_count parameters of the QosPolicy set the initial and
maximum size of the queue.
The priority of the Event thread should be carefully set with respect to the priorities of the other threads in
a system. While many events can tolerate some amount of latency between the time that the event expires
and the time that the Event thread services the event, there may be application-specific events that must be
handled as soon as possible.
For example, if an application uses the liveliness of a remote DataWriter to infer the correct operation of a
remote application, it may be critical for the user code in the DataReader Listener callback, on_liveliness_
changed(), to be called by the Event thread as soon as it can be determined that the remote application has
died. The operating system uses the priority of the Event thread to schedule this action.

19.3 Receive Threads
Connext DDS uses internal threads, known as Receive threads, to process the data packets received via
underlying network transports. These data packets may contain meta-traffic exchanged by DomainParticipants for discovery, or user data (and meta-data to support reliable connections) destined for local
DataReaders.
As a result of processing packets received by a transport, a Receive thread may respond by sending packets on the network. Discovery packets may be sent to other DomainParticipants in response to ones
received. ACK/NACK packets are sent in response to heartbeats to support a reliable connection.
When a DDS sample arrives, the Receive thread is responsible for deserializing and storing the data in the
receive queue of a DataReader as well as invoking the on_data_available() DataReaderListener callback
(see Setting Up DataReaderListeners (Section 7.3.4 on page 466)).
The number of Receive threads that Connext DDS will create for a DomainParticipant depends on how
you have configured the QosPolicies of DomainParticipants, DataWriters and DataReaders as well as on
the implementation of a particular transport. The behavior of the builtin transports is well specified.

839

19.3 Receive Threads

However, if a custom transport is installed for a DomainParticipant, you will have to understand how the
custom transport works to predict how many Receive threads will be created.
The following discussion applies on a per-transport basis. A single Receive thread will only service a
single transport.
Connext DDS will try to create receive resources1 for every port of every transport on which it is configured to receive messages. The TRANSPORT_UNICAST QosPolicy (DDS Extension) (Section 6.5.24
on page 412) for DomainParticipant, DataWriters, and DataReaders, the TRANSPORT_MULTICAST
QosPolicy (DDS Extension) (Section 7.6.5 on page 529) for DataReaders and the DISCOVERY
QosPolicy (DDS Extension) (Section 8.5.2 on page 580) for DomainParticipants all configure the number of ports and the number of transports that Connext DDS will try to use for receiving messages.
Generally, transports will require Connext DDS to create a new receive resource for every unique port
number. However, this is both dependent on how the underlying physical transport works and the implementation of the transport plug-in used by Connext DDS. Sometimes Connext DDS only needs to create a
single receive resource for any number of ports.
When Connext DDS finds that it is configured to receive data on a port for a transport for which it has not
already created a receive resource, it will ask the transport if any of the existing receive resources created
for the transport can be shared. If so, then Connext DDS will not have to create a new receive resource. If
not, then Connext DDS will.
The TRANSPORT_UNICAST, TRANSPORT_MULTICAST, and DISCOVERY QosPolicies allow
you customize ports for receiving user data (on a per-DataReader basis) and meta-traffic (DataWriters and
DomainParticipants); ports can be also set differently for unicast and multicast.
How do receive resources relate to Receive threads? Connext DDS will create a Receive thread to service
every receive resource that is created. If you use a socket analogy, then for every socket created, Connext
DDS will use a separate thread to process the data received on that socket.
So how many thread will Connext DDS create by default–using only the builtin UDPv4 and shared
memory transports and without modifying any QosPolicies?
Three Receive threads are created for meta-traffic2:

1If UDPv4 was the only transport that Connext DDS supports, we would call these receive resources ‘sockets.’
2Meta-traffic refers to traffic internal to Connext DDS related to dynamic discovery (see Discovery (Section Chapter 14

on page 709)).

840

19.4 Exclusive Areas, Connext DDS Threads and User Listeners

l

2 for unicast (one for UDPv4, one for shared memory)

l

1 for multicast (for UDPv4)1

Two Receive threads created for user data:
l

2 for unicast (UDPv4, shared memory)

l

0 for multicast (because user data is not sent via multicast by default)

Therefore, by default, you will have a total of five Receive threads per DomainParticipant. By using only
a single transport and disabling multicast, a DomainParticipant can have as few as 2 Receive threads.
Similar to the Database and Event threads, a Receive thread is configured by the RECEIVER_POOL
QosPolicy (DDS Extension) (Section 8.5.6 on page 604). However, note that the thread properties in the
RECEIVER_POOL QosPolicy apply to all Receive threads created for the DomainParticipant.

19.4 Exclusive Areas, Connext DDS Threads and User Listeners
Connext DDS Event and Receive threads may invoke user code through the Listener callbacks installed
on different Entities while executing internal Connext DDS code. In turn, user code inside the callbacks
may invoke Connext DDS APIs that reenter the internal code space of Connext DDS. For thread safety,
Connext DDS allocates and uses mutual exclusion semaphores (mutexes).
As discussed in Exclusive Areas (EAs) (Section 4.5 on page 182), when multiple threads and multiple
mutexes are mixed together, deadlock may result. To prevent deadlock from occurring, Connext DDS is
designed using careful analysis and following rules that force mutexes to be taken in a certain order when
a thread must take multiple mutexes simultaneously.
However, because the Event and Receive threads already hold mutexes when invoking user callbacks, and
because the Connext DDS APIs that the user code can invoke may try to take other mutexes, deadlock
may still result. Thus, to prevent user code to cause internal Connext DDS threads to deadlock, we have
created a concept called Exclusive Areas (EA) that follow rules that prevent deadlock. The more EAs that
exist in a system, the more concurrency is allowed through Connext DDS code. However, the more EAs
that exist, the more restrictions on the Connext DDS APIs that are allowed to be invoked in Entity Listener
callbacks.
The EXCLUSIVE_AREA QosPolicy (DDS Extension) (Section 6.4.3 on page 318) control how many
EAs will be created by Connext DDS. For a more detailed discussion on EAs and the restrictions on the
use of Connext DDS APIs within Entity Listener methods, please see Exclusive Areas (EAs) (Section 4.5
on page 182).

1Multicast is not supported by shared memory transports.

841

19.5 Controlling CPU Core Affinity for RTI Threads

19.5 Controlling CPU Core Affinity for RTI Threads
Two fields in the DDS_ThreadSettings_t structure (see Thread Settings (Section 17.4.7 on page 809)) are
related to CPU core affinity: cpu_list and cpu_rotation.
Note: Although DDS_ThreadSettings_t is used in the Event, Database, ReceiverPool, and AsynchronousPublisher QoS policies, cpu_list and cpu_rotation are only relevant in the RECEIVER_POOL
QosPolicy (DDS Extension) (Section 8.5.6 on page 604).
While most thread-related QoS settings apply to a single thread, the ReceiverPool QoS policy’s thread-settings control every receive thread created. In this case, there are several schemes to map M threads to N processors; cpu_rotation controls which scheme is used.
The cpu_rotation determines how cpu_list affects processor affinity for thread-related QoS policies that
apply to multiple threads. If cpu_list is empty, cpu_rotation is irrelevant since no affinity adjustment will
occur. Suppose instead that cpu_list ={0,1} and that the middleware creates three receive threads: {A, B,
C}. If cpu_rotation is set to CPU_NO_ROTATION, threads A, B and C will have the same processor
affinities (0-1), and the OS will control thread scheduling within this bound.
CPU affinities are commonly denoted with a bitmask, where set bits represent allowed processors to run
on. This mask is printed in hex, so a CPU affinity of 0-1 can be represented by the mask 0x3.
If cpu_rotation is CPU_RR_ROTATION, each thread will be assigned in round-robin fashion to one of
the processors in cpu_list; perhaps thread A to 0, B to 1, and C to 0. Note that the order in which internal
middleware threads spawn is unspecified.
The RTI Connext DDS Core Libraries Platform Notes describe which architectures support this feature.

19.6 Configuring Thread Settings with XML
Table 19.1 XML Tags for ThreadSettings_t describes the XML tags that you can use to configure thread
settings. For more information on thread settings, see:
l

Thread Settings (Section 17.4.7 on page 809)

l

The RTI Connext DDS Core Libraries Platform Notes

l

The API Reference HTML documentation (select Modules, RTI Connext DDS API
Reference,Infrastructure Module, QoS Policies, Extended QoS Support, Thread Settings)

842

19.6 Configuring Thread Settings with XML

Table 19.1 XML Tags for ThreadSettings_t
Tags
within


Description

Number of
Tags
Allowed

Each  specifies a processor on which the thread may run.




value


0 or 1

Only applies to platforms that support controlling CPU core affinity (see Controlling CPU Core Affinity
for RTI Threads (Section 19.5 on the previous page) and the RTI Connext DDS Core Libraries Platform
Notes).
Determines how the CPUs in  will be used by the thread. The value can be either:
THREAD_SETTINGS_CPU_NO_ROTATION
The thread can run on any listed processor, as determined by OS scheduling.


THREAD_SETTINGS_CPU_RR_ROTATION

0 or 1

The thread will be assigned a CPU from the list in round-robin order.
Only applies to platforms that support controlling CPU core affinity (see the RTI Connext DDS Core
Libraries Platform Notes).
A collection of flags used to configure threads of execution. Not all of these options may be relevant for all
operating systems. May include these bits:



l

STDIO

l

FLOATING_POINT

l

REALTIME_PRIORITY

l

PRIORITY_ENFORCE

0 or 1

It can also be set to a combination of the above bits by using the “or” symbol (|), such as
STDIO|FLOATING_POINT.
Default: MASK_DEFAULT

843

19.7 User-Managed Threads

Table 19.1 XML Tags for ThreadSettings_t
Tags
within


Description

Number of
Tags
Allowed

Thread priority. The value can be specified as an unsigned integer or one of the following strings.
l

THREAD_PRIORITY_DEFAULT

l

THREAD_PRIORITY_HIGH

l

THREAD_PRIORITY_ABOVE_NORMAL

l

THREAD_PRIORITY_NORMAL

l

THREAD_PRIORITY_BELOW_NORMAL

l

THREAD_PRIORITY_LOW



0 or 1
When using an unsigned integer, the allowed range is platform-dependent.
When thread priorities are configured using XML, the values are considered native priorities.
Example:

STDIO|FLOATING_POINT
10
THREAD_STACK_SIZE_DEFAULT


When the XML file is loaded using the Java API, the priority is a native priority, not a Java thread priority.


Thread stack size, specified as an unsigned integer or set to the string THREAD_STACK_SIZE_
DEFAULT. The allowed range is platform-dependent.

0 or 1

19.7 User-Managed Threads
In certain scenarios, you may want full control over the internal threads created by your Connext DDS
applications. For instance, in memory-constrained systems, applications may want to manage the resources
required by internal Connext DDS threads. Also, you may want to use a different thread technology than
the one Connext DDS incorporates by default (i.e., pthread on POSIX platforms).
Connext DDS can create the internal threads from the application layer via the abstract factory pattern.
You can provide a Connext DDS application with a ThreadFactory implementation that DomainParticipants will use to create and delete all the threads.
The ThreadFactory interface exposes operations for creating and deleting threads. These operations are
called on demand as DomainParticipants require new threads or need to delete existing ones.
The same ThreadFactory instance can be used by multiple DomainParticipants. To select which
ThreadFactory to use, use the set_thread_factory() operation in the DomainParticipantFactory:

844

19.7 User-Managed Threads

MyThreadFactory myThreadFactory; // Implements DDSThreadFactory
retcode = DDSTheParticipantFactory->set_thread_factory(&myThreadFactory);

Then you can create DomainParticipants using any of the available APIs (i.e. create_participant(), create_participant_from_config(), etc). A DomainParticipant will use the ThreadFactory object that is set
in the DomainParticipantFactory at the time it is created and throughout its entire lifecycle. If a new
ThreadFactory is set, existing DomainParticipants will not be affected; they will still use the same
ThreadFactory with which they were created.
This feature is only available for the C/C++ APIs. For further information, please see the API Reference
HTML documentation.

845

Chapter 20 DDS Sample-Data and
Instance-Data Memory
Management
This chapter describes how Connext DDS manages the memory for the DDS data samples that are
sent by DataWriters and received by DataReaders.

20.1 DDS Sample-Data Memory Management for DataWriters
To configure DDS sample-data memory management on the writer side, use the PROPERTY
QosPolicy (DDS Extension) (Section 6.5.17 on page 394). Table 20.1 DDS Sample-Data
Memory Management Properties for DataWriters lists the supported memory-management properties for DataWriters.
Table 20.1 DDS Sample-Data Memory Management Properties for DataWriters
Property

Description
If the serialized size of the DDS sample is <= pool_buffer_max_size:
The buffer is obtained from a pre-allocated pool and released when the DataWriter is deleted.
If the serialized size of the DDS sample is > pool_buffer_max_size:
The buffer is dynamically allocated from the heap and returned to the heap when the DDS sample is removed
from the DataWriter’s queue.

dds.data_writer.
history.memory_
manager.
Default: -1 (UNLIMITED). All DDS sample buffers are obtained from the pre-allocated pool; the buffer size
fast_pool.pool_ is the maximum serialized size of the DDS samples, as returned by the type plugin get_serialized_sample_
buffer_max_size max_size() operation.
See Memory Management without Batching (Section 20.1.1 on the next page).
Note: This property also controls the memory allocation for the serialized key buffer that is stored with every
instance. See Instance-Data Memory Management for DataWriters (Section 20.3 on page 861).

846

20.1.1 Memory Management without Batching

Table 20.1 DDS Sample-Data Memory Management Properties for DataWriters
Property

Description
Only supported when using the Java API.

Defines the minimum size of the buffer that will be used to serialize DDS samples.
dds.data_writer.
history.memory_ When a DataWriter is created, the Java layer will allocate a buffer of this size and associate it with the
manager.
DataWriter.
java_
Default: -1 (UNLIMITED). This is a sentinel that refers to the maximum serialized size of a DDS sample, as
stream.min_size
returned by the type plugin get_serialized_sample_max_size() operation
See Writer-Side Memory Management when Using Java (Section 20.1.3 on page 851).
Only supported when using the Java API.
dds.data_writer.
history.memory_
manager.
java_
stream.trim_to_
size

A boolean value that controls the growth of the serialization buffer.
If set to 0 (default): The buffer will not be reallocated unless the serialized size of a new DDS sample is greater
than the current buffer size.
If set to 1: The buffer will be reallocated with each new DDS sample to a smaller size in order to just fit the
DDS sample serialized size. The new size cannot be smaller than min_size.
See Writer-Side Memory Management when Using Java (Section 20.1.3 on page 851).

20.1.1 Memory Management without Batching
When the write() operation is called on a DataWriter that does not have batching enabled, the DataWriter
serializes (marshals) the input DDS sample and stores it in the DataWriter’s queue (see Figure 20.1
DataWriter Actions when Batching is Disabled on the facing page). The size of this queue is limited by initial_samples/max_samples in the RESOURCE_LIMITS QosPolicy (Section 6.5.20 on page 405).

847

20.1.1 Memory Management without Batching

Figure 20.1 DataWriter Actions when Batching is Disabled

848

20.1.2 Memory Management with Batching

Each DDS sample in the queue has an associated serialization buffer in which the DataWriter will serialize
the DDS sample. This buffer is either obtained from a pre-allocated pool (if the serialized size of the DDS
sample is <= dds.data_writer.history.memory_manager.fast_pool.pool_buffer_max_size) or the buffer is dynamically allocated from the heap (if the serialized size of the DDS sample is > dds.data_writer.history.memory_manager.fast_pool.pool_buffer_max_size). See Table 20.1 DDS Sample-Data
Memory Management Properties for DataWriters,
The default value of pool_buffer_max_size is -1 (UNLIMITED). In this case, all the DDS samples come
from the pre-allocated pool and the size of the buffers is the maximum serialized size of the DDS samples
as returned by the type plugin get_serialized_sample_max_size() operation. The default value is optimum
for real-time applications where determinism and predictability is a must. The trade-off is higher memory
usage, especially in cases where the maximum serialized size of a DDS sample is large.

20.1.2 Memory Management with Batching
When the write() operation is called on a DataWriter for which batching is enabled (see BATCH
QosPolicy (DDS Extension) (Section 6.5.2 on page 341)), the DataWriter serializes (marshals) the input
DDS sample into the current batch buffer (see Figure 20.2 DataWriter Actions when Batching is Enabled
on the facing page). When the batch is flushed, it is stored in the DataWriter’s queue along with its DDS
samples. The DataWriter queue can be sized based on:
l

l

l

The number of DDS samples, using initial_samples/max_samples (both set in the RESOURCE_
LIMITS QosPolicy (Section 6.5.20 on page 405))
The number of batches, using initial_batches/max_batches (both set in the DATA_WRITER_
RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 6.5.4 on page 359))
Or a combination of max_samples and max_batches

849

20.1.2 Memory Management with Batching

Figure 20.2 DataWriter Actions when Batching is Enabled

850

20.1.3 Writer-Side Memory Management when Using Java

When batching is enabled, the memory associated with the batch buffers always comes from a pre-allocated pool. The size of the buffers is determined by the QoS values max_samples and max_data_bytes
(both set in the BATCH QosPolicy (DDS Extension) (Section 6.5.2 on page 341)) as follows:
l

l

If max_data_bytes is a finite value, the size of the buffer is the minimum of this value and the maximum serialized size of a DDS sample (max_sample_serialized_size) as returned by the type-plugin get_serialized_sample_max_size(), since that batch must contain at least one DDS sample).
Otherwise, the size of the buffer is calculated by
(batch.max_samples * max_sample_serialized_size).

Notice that for variable-size DDS samples (for example, DDS samples containing sequences) it is good
practice to size the buffer based on max_data_bytes, since this leads to more efficient memory usage.
Note: The value of the property dds.data_writer.history.memory_manager.fast_pool.pool_buffer_
max_size is ignored by DataWriters with batching enabled.

20.1.3 Writer-Side Memory Management when Using Java
When the Java API is used, Connext DDS allocates a Java buffer per DataWriter; this buffer is used to
serialize the Java DDS samples published by the DataWriters. After a DDS sample is serialized into a Java
buffer, the result is copied into the underlying native buffer described in Memory Management without
Batching (Section 20.1.1 on page 847) and Memory Management with Batching (Section 20.1.2 on page
849).
You can use the following two DataWriter properties to control memory allocation for the Java buffers
that are used for serialization (see Table 20.1 DDS Sample-Data Memory Management Properties for
DataWriters):
l

dds.data_writer.history.memory_manager.java_stream.min_size

l

dds.data_writer.history.memory_manager.java_stream.trim_to_size

20.1.4 Writer-Side Memory Management when Working with Large Data
Large DDS samples are DDS samples with a large maximum size relative to the memory available to the
application. Notice the use of the word maximum, as opposed to actual size.
As described in Memory Management without Batching (Section 20.1.1 on page 847), by default, the middleware preallocates the DDS samples in the DataWriter queue to their maximum serialized size. This may
lead to high memory-usage in DataWriters where the maximum serialized size of a DDS sample is large.
For example, let’s consider a video conferencing application:
struct VideoFrame {
boolean keyFrame;

851

20.1.4 Writer-Side Memory Management when Working with Large Data

sequence data;
};

The above IDL definition can be used to work with video streams.
Each frame is transmitted as a sequence of octets with a maximum size of 1 MB. In this example, the video
stream has two types of frames: I-Frames (also called key frames) and P-Frames (also called delta frames).
I-Frames represent full images and do not require information about the preceding frames in order to be
decoded. P-frames require information about the preceding frames in order to be decoded.
A video stream consists of a sequence of frames in which I-Frames are followed by multiple P-frames. The
number of P-frames between I-Frames affects the video quality since, in a non-reliable configuration, losing a P-frame will degrade the image quality until the next I-frame is received.
For our use case, let’s assume that I-frames may require 1 MB, while P-Frames require less than 32 KB.
Also, there are 20 times more P-Frames than I-Frames.
Although the actual size of the frames sent by the Connext DDS application is usually significantly smaller
than 1 MB since they are P-Frames, the default memory management will use 1 MB per frame in the
DataWriter queue. If resource_limits.max_samples is 256, the DataWriter may end up allocating 256
MB.
Using some domain-specific knowledge, such as the fact that most of the P-Frames have a size smaller
than 32 KB, we can optimize memory usage in the DataWriter’s queue while still maintaining determinism and predictability for the majority of the frames sent on the wire.
The following XML file shows how to optimize the memory usage for the previous example (rather than
focusing on efficient usage of the available network bandwidth).










32

1






dds.data_writer.history.memory_manager.fast_pool.pool_buffer_max_size

33792




dds.data_writer.history.memory_manager.java_stream.min_size

33792



dds.data_writer.history.memory_manager.java_stream.trim_to_size

1








Working with large data DDS samples will likely require throttling the network traffic generated by single
DDS samples. For additional information on shaping network traffic, see FlowControllers (DDS Extension) (Section 6.6 on page 422).

20.2 DDS Sample-Data Memory Management for DataReaders
The DDS data samples received by a DataReader are deserialized (demarshaled) and stored in the
DataReader’s queue (see Adding DDS Samples to DataReader’s Queue (Section Figure 20.3 on page

853

20.2.1 Memory Management for DataReaders Using Generated Type-Plugins

855)). The size of this queue is limited by initial_samples/max_samples in the RESOURCE_LIMITS
QosPolicy (Section 6.5.20 on page 405).

20.2.1 Memory Management for DataReaders Using Generated TypePlugins
Figure 20.3 Adding DDS Samples to DataReader’s Queue on the next page shows how DDS samples are
processed and added to the DataReader’s queue.

854

20.2.1 Memory Management for DataReaders Using Generated Type-Plugins

Figure 20.3 Adding DDS Samples to DataReader’s Queue

The RTPS DATA DDS samples received by a DataReader can be either batch DDS samples or individual DDS samples. The DataReader queue does not store batches. Therefore, each one of the DDS
samples within a batch will be deserialized and processed individually.
When the DataReader processes a new sample, it will deserialize it into a sample obtained from a pre-allocated pool. By default, to provide predictability and determinism, the sample obtained from the pool is
855

20.2.2 Reader-Side Memory Management when Using Java

allocated to its maximum size. For example, with the following IDL type, each sample in the DataReader
queue will consume 1 MB, even if the actual size is smaller.
struct VideoFrame {
boolean keyFrame;
sequence data;
};

In the above example, it is possible to reduce the memory consumption in C, C++, and .NET by declaring
the data sequence as unbounded and by generating code for the type with the command-line option unboundedSupport. In this case, the middleware will not preallocate 1 MB for the data member. Instead,
the generated code will deserialize incoming samples by dynamically allocating and deallocating memory
to accommodate the actual size of the data sequence.

20.2.2 Reader-Side Memory Management when Using Java
When the Java API is used with DataReaders using generated type-plugins, Connext DDS allocates a
Java buffer per DataReader; this buffer is used to copy the native serialized data, so that the received DDS
samples can be deserialized into the Java objects obtained from the DDS sample pool in Adding DDS
Samples to DataReader’s Queue (Section Figure 20.3 on the previous page).
You can use the DataReader properties in Table 20.2 DDS Sample-Data Memory Management Properties
for DataReaders when Using Java API to control memory allocation for the Java buffer used for deserialization:
Table 20.2 DDS Sample-Data Memory Management Properties for DataReaders when Using
Java API
Property

Description

Only supported when using the Java API.
dds.data_reader.
history.memory_ Defines the minimum size of the buffer used for the serialized data.
manager.
When a DataReader is created, the Java layer will allocate a buffer of this size and associate it with the DataReader.
java_
stream.min_size Default: -1 (UNLIMITED) This is a sentinel to refer to the maximum serialized size of a DDS sample, as returned by
the type plugin method get_serialized_sample_max_size().
Only supported when using the Java API.
dds.data_reader.
history.memory_
manager.
java_
stream.trim_to_
size

A Boolean value that controls the growth of the deserialization buffer.
If set to 0 (the default), the buffer will not be re-allocated unless the serialized size of a new DDS sample is greater than
the current buffer size.
If set to 1, the buffer will be re-allocated with each new DDS sample in order to just fit the DDS sample serialized size.
The new size cannot be smaller than min_size.

856

20.2.3 Memory Management for DynamicData DataReaders

20.2.3 Memory Management for DynamicData DataReaders
Unlike DataReaders that use generated type-plugin code, DynamicData DataReaders provide configuration mechanisms to optimize the memory usage for use cases involving large data DDS samples.
A DDS DynamicData sample stored in the DataReader’s queue has an associated underlying buffer that
contains the serialized representation of the DDS sample. The buffer is allocated according to the configuration provided in the serialization member of the DynamicDataProperty_t used to create the
DynamicDataTypeSupport (see Interacting Dynamically with User Data Types (Section 3.8 on page
141)).
struct DDS_DynamicDataProperty_t {
...
DDS_DynamicDataTypeSerializationProperty_t serialization;
}
struct DDS_DynamicDataTypeSerializationProperty_t {
...
DDS_UnsignedLong max_size_serialized;
DDS_UnsignedLong min_size_serialized;
DDS_Boolean trim_to_size;
}

Table 20.3 struct DDS_DynamicDataTypeSerializationProperty_t describes the members of DDS_
DynamicDataTypeSerializationProperty_t.
Table 20.3 struct DDS_DynamicDataTypeSerializationProperty_t
Name
max_
size_
serialized

Description
Defines the maximum size of the buffer that will contain the serialized DDS sample.
Default: 0xFFFFFFFF, indicates that Connext DDS must use the maximum serialized size of a DDS sample according to the
type information. Except in very specific scenarios, the value max_size_serialized should always be the default.

min_size_ Defines the minimum size of the buffer used to hold the serialized data in a DynamicData object.
serialized Default: 0xFFFFFFFF, a sentinel that indicates that this value must be equal to the value specified in max_size_serialized.
Controls the growth of the serialization buffer in a DynamicData object.
trim_to_
size

If set to 0 (default): The buffer will not be reallocated unless the serialized size of the incoming DDS sample is greater than the
current buffer size.
If set to 1: The buffer of a DynamicData object obtained from the DDS sample pool will be re-allocated to just fit the size of the
serialized data of the incoming sample. The new size cannot be smaller than min_size_serialized.

Figure 20.4 Allocation of DDS Samples in DataReader Queue for DynamicData DataReaders on the
facing page shows how DDS samples are allocated in the DataReader queue for DynamicData DataReaders.

857

20.2.3 Memory Management for DynamicData DataReaders

Figure 20.4 Allocation of DDS Samples in DataReader Queue for DynamicData DataReaders

858

20.2.4 Memory Management for Fragmented DDS Samples

20.2.4 Memory Management for Fragmented DDS Samples
When a DataWriter writes DDS samples with a serialized size greater than the minimum of the largest
transport message sizes across all transports installed with the DataWriter, the DDS samples are fragmented into multiple RTPS fragment messages.
The different fragments associated with a DDS sample are assembled in the DataReader side into a single
buffer that will contain the DDS sample serialized data after the last fragment is received.
By default, the DataReader keeps a pool of pre-allocated serialization buffers that will be used to reconstruct the serialized data of a DDS sample from the different fragments. Each buffer hold one individual
DDS sample and it has a size equal to the maximum serialized size of a DDS sample. The pool size can be
configured using the QoS values initial_fragmented_samples and max_fragmented_samples in
DATA_READER_RESOURCE_LIMITS QosPolicy (DDS Extension) (Section 7.6.2 on page 517).
The main disadvantage in pre-allocating the serialization buffers is an increase in memory usage, especially
when the maximum serialized of a DDS sample is quite large. Connext DDS offers a setting that allows
memory for a DDS sample to be allocated from the heap the first time a fragment is received. The amount
of memory allocated equals the amount of memory needed to store all fragments in the DDS sample.

20.2.5 Reader-Side Memory Management when Working with Large Data
This section describes how to configure the DataReader side of the videoconferencing application introduced in Writer-Side Memory Management when Working with Large Data (Section 20.1.4 on page 851)
to optimize memory usage.
The following XML file can be used to optimize the memory usage in the previous example:










KEEP_ALL_HISTORY_QOS


32

1



859

20.2.5 Reader-Side Memory Management when Working with Large Data



1







dds.data_reader.history.memory_manager.java_stream.min_size

33792



dds.data_reader.history.memory_manager.java_stream.trim_to_size

1







To avoid preallocation of the samples in the DataReader's queue to their maximum size for Type-Plugin
generated code in C, C++, and .NET, replace the bounded sequence in VideoFrame with an unbounded
sequence and generate code using the -unboundedSupport command-line option:
struct VideoFrame {
boolean keyFrame;
sequence data;
};

See Memory Management for DataReaders Using Generated Type-Plugins (Section 20.2.1 on page 854)
for more details.
To avoid preallocation of the samples in the DataReader's queue to their maximum size for DynamicData,
set the min_size_serialized property to avoid the allocation of 1MB buffers for the DataReader queue
samples (See Memory Management for DynamicData DataReaders (Section 20.2.3 on page 857).

860

20.3 Instance-Data Memory Management for DataWriters

20.3 Instance-Data Memory Management for DataWriters
When an instance is registered with a DataWriter, the DataWriter serializes the key value and stores it
with the instance.
Each instance maintained by the DataWriter has an associated buffer in which the DataWriter serializes
the key. This buffer is either:
l

l

Obtained from a pre-allocated pool (if the key’s serialized size is <= dds.data_writer.history.memory_manager.fast_pool.pool_buffer_max_size)
Dynamically allocated from the heap (if the key’s serialized size is > dds.data_writer.history.memory_manager.fast_pool.pool_buffer_max_size).

See Table 20.4 Instance-Data Memory Management Properties for DataWriters.
Table 20.4 Instance-Data Memory Management Properties for DataWriters
Property

Description
Controls the memory allocation for the serialized key buffer that is stored with every instance.

dds.data_writer.
history.memory_
manager.
fast_pool.pool_
buffer_max_size

Default: -1 (UNLIMITED). All DDS sample buffers are obtained from the pre-allocated pool. The buffer size is the
maximum serialized size of the DDS samples, as returned by the type plugin get_serialized_sample_max_size()
operation.
Note: This property also controls DDS sample-data memory management. See DDS Sample-Data Memory Management
for DataWriters (Section 20.1 on page 846).

20.4 Instance-Data Memory Management for DataReaders
When an instance is received and registered by a DataReader, the DataReader serializes the key value
and stores it with the instance.
Each instance maintained by the DataReader has an associated buffer in which the DataReader serializes
the key. This buffer is either:
Obtained from a pre-allocated pool (if the key’s serialized size is <= dds.data_reader.history.memory_
manager.fast_pool.pool_buffer_max_size)
Dynamically allocated from the heap (if the key’s serialized size is > dds.data_reader.history.memory_
manager.fast_pool.pool_buffer_max_size)
See Table 20.5 Instance-Data Memory Management Properties for DataReaders .

861

20.4 Instance-Data Memory Management for DataReaders

Table 20.5 Instance-Data Memory Management Properties for DataReaders
Property

Description

dds.data_
reader.
history.memory_
manager.
fast_pool.pool_
buffer_max_size

Controls the memory allocation for the serialized key buffer that is stored with every instance in the
DataReader’s queue.
Default: -1 (UNLIMITED). All buffers come from the pre-allocated pool. The size of the buffers is
the maximum serialized size of the key as returned by the type plugin get_serialized_key_max_
size() operation.

862

Chapter 21 Troubleshooting
This chapter contains tips on troubleshooting Connext DDS applications. For an up-to-date list of
frequently asked questions, see the RTI Support Portal, accessible from https://support.rti.com—
select the Find Solution link to see example code, general information on Connext DDS, performance information, troubleshooting tips, and technical details.
This chapter contains the following sections:

21.1 What Version am I Running?
There are two ways to obtain version information:
l

l

By looking at the revision files, as described in Finding Version Information in Revision
Files (Section 21.1.1 below).
Programmatically at run time, as described in Finding Version Information Programmatically
(Section 21.1.2 on the next page).

21.1.1 Finding Version Information in Revision Files
In the top-level directory of your Connext DDS installation (${NDDSHOME}), you will find text
files that include revision information. The files are named rev__rtidds.. For
example, you might see files called rev_host_rtidds.5.x.y and rev_persistence_rtidds5.x.y
(where x and y stand for the version numbers of the current release). Each file contains more
details, such as a patch level and if the product is license managed.
For example:
Host Build 5.x.y rev 04 (0x04050200)

The revision files for Connext DDS target libraries are in the same directory as the libraries
(${NDDSHOME}/lib/).

863

21.1.2 Finding Version Information Programmatically

21.1.2 Finding Version Information Programmatically
The methods in the NDDSConfigVersion class can be used to retrieve version information for the Connext DDS product, the core library, and the C, C++ or Java libraries.
The version information includes four fields:
l

A major version number

l

A minor version number

l

A release number

l

A build number

Table 21.4 NDDSConfigLogger Operations lists the available operations (they will vary somewhat
depending on the programming language you are using; consult the API Reference HTML documentation
for more information).
Table 21.1 NDDSConfigVersion Operations
Purpose

Operation

Description

get_product_
Gets version information for the Connext DDS product.
version

To retrieve version information in
a structured format

get_core_
version

Gets version information for the Connext DDS core library.

get_c_api_
version

Gets version information for the Connext DDS C library.

get_cpp_api_
Gets version information for the Connext DDS C++ library.
version
To retrieve version information in
string format

to_string

Converts the version information for each library into a string. The strings for each
library are put in a single hyphen-delimited list.

The get_product_version() operation returns a reference to a structure of type DDS_ProductVersion_t:
struct NDDS_Config_ProductVersion_t {
DDS_Char major;
DDS_Char minor;
DDS_Char release;
DDS_Char revision;
};

The other get_*_version() operations return a reference to a structure of type NDDS_Config_LibraryVersion_t:

864

21.2 Controlling Messages from Connext DDS

struct NDDS_Config_LibraryVersion_t {
DDS_Long major;
DDS_Long minor;
char release;
DDS_Long build;
};

The to_string() operation returns version information for the Connext DDS core, followed by the C and
C++ API libraries, separated by hyphens. For example:

21.2 Controlling Messages from Connext DDS
Connext DDS provides several types of messages to help you debug your system and alert you to errors
during run time. You can control how much information is reported and where it is logged.
How much information is logged is known as the verbosity setting. Table 21.2 Message Logging Verbosity Levels describes the increasing verbosity levels.
Table 21.2 Message Logging Verbosity Levels
Verbosity
(NDDS_
CONFIG_
LOG_
VERBOSITY_
*)
SILENT

Description

No messages will be logged. (lowest verbosity)

865

21.2 Controlling Messages from Connext DDS

Table 21.2 Message Logging Verbosity Levels
Verbosity
(NDDS_
CONFIG_
LOG_
VERBOSITY_
*)

Description

ERROR (default
level for all
categories)

Log only high-priority error messages. An error indicates something is wrong with how Connext DDS is
functioning. The most common cause of this type of error is an incorrect configuration.

WARNING

Additionally log warning messages. A warning indicates that Connext DDS is taking an action that may or may not
be what you intended. Some configuration information is also logged at this verbosity to aid in debugging.

STATUS_LOCAL

Additionally log verbose information about the lifecycles of local Connext DDS objects.

STATUS_
REMOTE

Additionally log verbose information about the lifecycles of remote Connext DDS objects.

STATUS_ALL

Additionally log verbose information about periodic activities and Connext DDS threads. (highest verbosity)

Note that the verbosities are cumulative: logging at a high verbosity means also logging all lower verbosity
messages. If you change nothing, the default verbosity will be set to NDDS_CONFIG_LOG_
VERBOSITY_ERROR.
Logging at high verbosities can be detrimental to your application's performance. You should
generally not set the verbosity above NDDS_CONFIG_LOG_VERBOSITY_WARNING, unless
you are debugging a specific problem.
You will typically change the verbosity of all of Connext DDS at once. However, in the event that such a
strategy produces too much output, you can further discriminate among the messages you would like to
see. The types of messages logged by Connext DDS fall into the categories listed in Table 21.3 Message
Logging Categories; each category can be set to a different verbosity level.
Table 21.3 Message Logging Categories
Category (NDDS_CONFIG_
LOG_CATEGORY_*)

Description

PLATFORM

Messages about the underlying platform (hardware and OS).

COMMUNICATION

Messages about data serialization and deserialization and network traffic.

DATABASE

Messages about the internal database of Connext DDS objects.

866

21.2 Controlling Messages from Connext DDS

Table 21.3 Message Logging Categories
Category (NDDS_CONFIG_
LOG_CATEGORY_*)

Description

ENTITIES

Messages about local and remote entities and the discovery process.

API

Messages about Connext DDS’s API layer (such as method argument validation).

The methods in the NDDSConfigLogger class can be used to change verbosity settings, as well as the destination for logged messages. Table 21.4 NDDSConfigLogger Operations lists the available operations;
consult the API Reference HTML documentation for more information.
Table 21.4 NDDSConfigLogger Operations
Purpose

Operation

Description
Gets the current verbosity.

Change Verbosity for
all Categories

Change Verbosity for a
Specific Category

get_verbosity

If per-category verbosities are used, returns the highest verbosity of any
category.

set_verbosity

Sets the verbosity of all categories.

get_verbosity_
by_category
Gets/Sets the verbosity for a specific category.
set_verbosity_
by_category
get_output_file

Returns the file to which messages are being logged, or NULL for the default destination
(standard output on most platforms).

set_output_file Redirects future logged messages to the specified file (or NULL to return to the default)
Change Destination of
Logged Messages

Change Message
Format

get_output_
device

Returns the logging device installed with the logger.

set_output_
device

Registers a specified logging device with the logger. See Customizing the Handling of
Generated Log Messages (Section 21.2.3 on page 872)

get_print_
format
set_print_
format

Gets/Sets the current message format that Connext DDS is using to log diagnostic information.
See Format of Logged Messages (Section 21.2.1 on the next page).

867

21.2.1 Format of Logged Messages

21.2.1 Format of Logged Messages
You can control the amount of information in each message with the set_print_format() operation. The
format options are listed in Table 21.5 Message Formats.
Table 21.5 Message Formats
Message Format
(NDDS_CONFIG_LOG_
PRINT_FORMAT_*)

Description

DEFAULT

Message, method name, and activity context.

TIMESTAMPED

Message, method name, activity context, and timestamp.

VERBOSE

Message with all available context information (includes thread identifier, activity context).

VERBOSE_TIMESTAMPED

Message with all available context information and timestamp.

DEBUG

Information for internal debugging by RTI personnel.

MINIMAL

Message number, method name.

MAXIMAL

All available fields.

Of course, you are not likely to recognize all of the method names; many of the operations that perform logging are deep within the implementation of Connext DDS. However, in case of errors, logging will typically take place at several points within the call stack; the output thus implies the stack trace at the time the
error occurred. You may only recognize the name of the operation that was the last to log its message (i.e.,
the function that called all the others); however, the entire stack trace is extremely useful to RTI support
personnel in the event that you require assistance.
You may notice that many of the logged messages begin with an exclamation point character. This convention indicates an error and is intended to be reminiscent of the negation operator in many programming
languages. For example, the message “!create socket”in the second line of the above stack trace means
“cannot create socket.”
21.2.1.1 Timestamps
Reported times are in seconds from a system-dependent starting time; these are equivalent to the output
format from Connext DDS. The timestamp is in the form "ssssss.mmmmmm" where  is a number
of seconds, and  is a fraction of a second expressed in microseconds. Enabling timestamps will
result in some additional overhead for clock access for every message that is logged.
Logging of timestamps is not enabled by default. To enable it, use NDDS_Config_Logger method set_
print_format().

868

21.2.1.2 Thread identification

21.2.1.2 Thread identification
Thread identification strings uniquely identify for active thread when a message is output to the console. A
thread may be a user (application) thread or one of several types of internal threads. The possible thread
types are:
user thread: U
receive thread: rR, where thread index is an integer identifying this
receive thread
event thread: revt
asynchronous publisher thread: rDsp
Logging of thread IDs are not enabled by default. To enable it, use NDDS_Config_Logger method set_
print_format().
21.2.1.3 Hierarchical Context
Many middleware APIs now store information in thread-specific storage about the current operation, as
well as information about which DDS domain (and participant ID) was active, and which entities were
being operated on. In the case of objects that are associated with topics, the topic name is also stored.
The context field is output by default.
21.2.1.4 Explanation of Context Strings
l

DDS domain context
Dxxyy

In this case, xx = participant ID, yy = domain #. For example, D0149 means “domain 49, participant 01.”
l

Entity context
Operation on an entity will specify the object and a numeric ID, such as Writer(001A1). The name
will be one of the following:
String

Object Type

Participant

DDS_DomainParticipant

Pub

DDS_Publisher

Sub

DDS_Subscriber

869

21.2.1.4 Explanation of Context Strings

String

l

Object Type

Topic

DDS_Topic

Writer

DDS_<*>DataWriter

Reader

DDS_<*>DataReader

Topic Context
T=Hello refers to topic "Hello."
The operations which report context include:
String

Operation

Entity operations:
ENABLE

Entity::enable

GET_QOS

Entity::get_qos

SET_QOS

Entity::set_qos

GET_LISTENER

Entity::get_listener

SET_LISTENER

Entity::set_listener

Factory operations (DP Factory, Participant, Pub/Sub):
CREATE 

Factory::create_

DELETE 

Factory::delete_

GET_DEFAULT_QOS 

Factory::get_default__qos

SET_DEFAULT_QOS 

Factory::set_default__qos

Participant-specific operations:
GET_PUBS

Participant::get_publishers

GET_SUBS

Participant::get_subscribers

LOOKUP Topic()

Participant::lookup_topicdescription

LOOKUP FlowController()

Participant::lookup_flowcontroller

IGNORE ()

Participant::ignore_

870

21.2.2 Configuring Logging via XML

21.2.2 Configuring Logging via XML
Logging can also be configured using the DomainParticipantFactory’s LOGGING QosPolicy (DDS
Extension) (Section 8.4.1 on page 572) with the tags, . The fields in
the LoggingQosPolicy are described in XML using a 1-to-1 mapping with the equivalent C representation
shown below:
struct DDS_LoggingQosPolicy {
NDDS_Config_LogVerbosity verbosity;
NDDS_Config_LogCategory category;
NDDS_Config_LogPrintFormat print_format;
char * output_file;
};

The equivalent representation in XML:









The attribute  can be set to true for the  tag to
indicate from which profile to use . If multiple QoS profiles have  set to true, the last profile with  set to true will be used.
If none of the profiles have set  to true, the profile with  set to true will be used.
In the following example, DefaultProfile2 will be used:






ALL
ENTITIES
MAXIMAL
LoggerOutput1.txt





871

21.2.3 Customizing the Handling of Generated Log Messages



WARNING
API
VERBOSE_TIMESTAMPED
LoggerOutput2.txt






ERROR
DATABASE
VERBOSE
LoggerOutput3.txt






Note: The LoggingQosPolicy is currently the only QoS policy that can be configured using the  tag.

21.2.3 Customizing the Handling of Generated Log Messages
By default, the log messages generated by Connext DDS are sent to the standard output. You can redirect
the log messages to a file by using the set_output_file() operation,
To further customize the management of the generated log messages, you can use the Logger’s set_output_device() operation to install a user-defined logging device. The logging device must implement an
interface with two operations: write() and close().
Connext DDS will call the write() operation to write a new log message to the input device. The log message provides the text and the verbosity corresponding to the message.
Connext DDS will call the close() operation when the logging device is uninstalled.
Note: It is not safe to make any calls to the Connext DDS core library including calls to DDS_
DomainParticipant_get_current_time() from any of the logging device operations.
For additional details on user-defined logging devices, see the API Reference HTML documentation
(under Modules, RTI Connext DDS API Reference, Configuration Utilities).

872

Part 4: Request-Reply Communication
Pattern
The Request-Reply communication pattern is only available with the Connext DDS
Professional, Evaluation, and Basic package types.
As real-time and embedded applications become more complex, and require integration with enterprise applications, you may need additional communication patterns besides publish-subscribe. Perhaps your application needs certain information only occasionally—such as changes in temperature
over the past hour, or even just once, such as application configuration data that is required only at
start up. To get information only when needed, Connext DDS supports a request-reply communication pattern, which is described in the following sections:
l

Introduction to the Request-Reply Communication Pattern (Section Chapter 22 on page 874)

l

Using the Request-Reply Communication Pattern (Section Chapter 23 on page 880)

873

Chapter 22 Introduction to the RequestReply Communication Pattern
This chapter describes the Request-Reply communication pattern, which is available with
the Connext DDS Professional, Evaluation, and Basic package types.
The fundamental communication pattern provided by Connext DDS is known as DDS data-centric
publish-subscribe. The data-centric publish-subscribe pattern is particularly well-suited in situations
where the same data must flow from one producer to many consumers, or when data is streaming
continuously from producers to consumers. For example, the values produced by a temperature
sensor may be observed by multiple applications, such as control applications, UI applications,
supervisory applications, historians, etc.

874

22.1 The Request-Reply Pattern

Figure 22.1 Publish-Subscribe Overview

Sending temperature updates using the publish-subscribe pattern

The publish-subscribe pattern supports multicast, which allows efficient distribution from a single source to
multiple applications, devices, or subscribers simultaneously. But even with a single subscriber, the publish-subscribe pattern is still advantageous, because the publisher can push new updates to a subscriber as
soon as they happen. That way the subscriber always has access to the latest data, with minimum delays,
and without incurring the overhead of periodically polling what may be stale data. This efficient, lowlatency access to the most current information is important for real-time applications.

22.1 The Request-Reply Pattern
As applications become more complex, it often becomes necessary to use other communication patterns in
addition to publish-subscribe. Sometimes an application needs to get a one-time snapshot of information;
for example, to make a query into a database or retrieve configuration parameters that never change. Other
times an application needs to ask a remote application to perform an action on its behalf; for example, to
invoke a remote procedure call or a service.
To support these scenarios, Connext DDS includes support for the request-reply communication pattern. It
is available with the Connext DDS Professional, Evaluation, and Basic package types.

875

22.1 The Request-Reply Pattern

Figure 22.2 Request-Reply Overview

Request-Reply communication pattern using a Requester and a Replier

The request-reply pattern has two roles: The requester (service consumer or client) sends a request message and waits for a reply message. The replier (service provider) receives the request message and
responds with a reply message.
Using the request-reply pattern with a Replier is straightforward. Connext DDS provides two Entities: the
Requester and the Replier manage all the interactions on behalf of the application. The Requester and
Replier automatically discover each other based on an application-specified service name. When the application invokes a request, the Requester sends a message (on an automatically-created request Topic) to the
Replier, which notifies the receiving application. The application, in turn, uses the Replier to receive the
request and send the reply message. The reply message is sent by Connext DDS back to the original
Requester (using a different automatically created reply Topic).
Connext DDS supports both blocking and non-blocking request-reply interactions:
l

l

In a blocking (a.k.a. synchronous) interaction, the requesting application blocks while waiting for
the reply. This is typical of applications desiring remote-procedure-call or remote-method-invocation
interactions.
In a non-blocking (a.k.a. asynchronous) interaction, the requesting application can proceed with
other work and gets notified when a reply is available.

Repliers (Section 23.2 on page 890) explains how an application can use the methods provided by the
Requester and the Replier to perform both blocking and non-blocking request-reply interactions.
The implementation of request-reply in Connext DDS is highly scalable. A Replier can receive requests
from thousands of Requesters at the same time. Connext DDS will efficiently deliver each reply only to the
original Requester, allowing the number of Requesters to grow without significantly impacting each other.

876

22.1.1 Request-Reply Correlation

22.1.1 Request-Reply Correlation
An application might have multiple outstanding requests, all originating from the same Requester. This can
be as a result of using a non-blocking request-reply interaction, or as a result of having multiple application
threads using the same Requester. Because of this, Connext DDS provides a way for the application to correlate a reply with the request it is associated with. This meta-data is provided as part of a SampleInfo structure that accompanies the reply.
When using a blocking request operation, Connext DDS provides an easy-to-use API that automatically
does the correlation for you.

22.2 Single-Request, Multiple-Replies
Connext DDS also supports the single-request multiple-reply pattern. This pattern is an extension of the
basic request-reply pattern in which multiple reply messages can flow back as a result of a single request.
The single-request multiple-reply pattern is very useful when getting large amounts of data as a reply, such
as when querying a system for all data that matches a certain criteria. Another common use-case is invoking a service that goes through multiple stages and provides updates on each: service commencement, progress reports, and final completion.
Figure 22.3 Single Request, Multiple Replies

Request/Reply communication pattern with multiple replies resulting from a single request

For example, a mobile asset management system may need to locate a particular asset (truck, locomotive,
etc.). The system sends out the request. The first reply that comes back will read “locating.” The service
has not yet determined the position, but it notifies the requester that the search operation has started. The

877

22.3 Multiple Repliers

second reply might provide a status update on the search, perhaps including a rough area of location. The
third and final reply will have the exact location of the asset.

22.3 Multiple Repliers
Connext DDS directly supports applications that obtain results from multiple providers in parallel instead
of in sequence, basically implementing functional parallelism.
To illustrate, consider a system managing a fleet of drones, like unmanned aerial vehicles (UAVs). Using
the single request-multiple reply pattern, the application can use a Requester to send a single ‘DroneInfo’
request to all the drones to query for their current mission and status. Each drone replies with the information on its own status and the Requester aggregates all the responses for the application.
As another example, consider a system that would like to locate the best printer to perform a particular job.
The application can use a Requester to query all the printers that are on-line for their characteristics and
load. The Requester receives the replies and accumulates them until an application-specified number of
replies is received (or a timeout elapses). The application can then use the Requester to access all the
replies, examine their contents, and select the best printer for the job.
Figure 22.4 Multiple Repliers

Request/Reply communication pattern with a single Requester and multiple Repliers

878

22.4 Combining Request-Reply and Publish-Subscribe

22.4 Combining Request-Reply and Publish-Subscribe
Under the hood, Connext DDS implements request-reply using the DDS data-centric publish-subscribe pattern. This has a key benefit in that the two patterns can be combined, and mapped without interference.
Figure 22.5 Combining Patterns

Combining Request-Reply and Publish-Subscribe patterns

For example, a pair of applications may be involved in a two-way conversation using request-reply. For
debugging purposes or regulatory compliance, you want to inspect those request-reply messages, but
without disrupting the conversation.
Since Connext DDS implements requests and replies using DDS data-centric publish subscribe, others can
simply subscribe to the request and reply messages. You can introduce a subscriber to the reply Topic,
without interfering with the two-way conversation between the Requester and the Replier. This pattern is
also known as a Wire Tap. For example, you can use RTI Recording Service to non-intrusively capture
request-reply traffic.

879

Chapter 23 Using the Request-Reply
Communication Pattern
This section explains how to use and configure the Request-Reply communication pattern,
which is only available with the Connext DDS Professional, Evaluation, and Basic
package types.
There are two basic Connext DDS entities used by the Request-Reply communication pattern:
Requester and Replier.
l

l

A Requester publishes a request Topic and subscribes to a reply Topic. See Requesters (Section 23.1 on the next page).
A Replier subscribes to the request Topic and publishes the reply Topic. See Repliers (Section 23.2 on page 890).
There is an alternate type of replier known as a SimpleReplier:
l A SimpleReplier is useful for cases where there is a single reply to each request and
the reply can be generated quickly, such as looking up some data from memory.
l

l

A SimpleReplier is used in combination with a user-provided SimpleReplierListener.
Requests are passed to a callback in the SimpleReplierListener, which returns the
reply.
The SimpleReplier is not suitable if the replier needs to generate more than one reply
for a single request or if generating the reply can take significant time or needs to
occur asynchronously. For more information, see SimpleRepliers (Section 23.3 on
page 896).

Additional resources. In addition to the information in this section, you can find more information
and example code here:

880

23.1 Requesters

l

l

The Connext DDS API Reference HTML documentation1contains example code that will show
you how to use API: From the Modules tab, navigate to Programming How-To’s, Request-Reply
Examples.
The Connext DDS API Reference HTML documentation also contains the full API documentation
for the Requester, Replier, and SimpleReplier. Under the Modules tab, navigate to RTI Connext
DDS API Reference, RTI Connext Request-Reply API Reference.

Typecodes are required when using the Request-Reply communication pattern. To use this pattern,
do not use RTI Code Generator's -noTypeCode flag. If typecodes are missing, the Requester will
log an exception.

23.1 Requesters
A Requester is an entity with two associated DDS Entities: a DDS DataWriter bound to a request Topic
and a DDS DataReader bound to a reply Topic. A Requester sends requests by publishing samples of the
request Topic, and receives replies for those requests by subscribing to the reply Topic.
Valid types for request and reply Topics can be:
l

l

For the C API:
l DDS types generated by RTI Code Generator
For all other APIs:
l DDS types generated by RTI Code Generator
l

Built-in DDS types, such as, String, KeyedString, Octets, and KeyedOctets

l

DDS DynamicData Types

To communicate, a Requester and Replier must use the same request Topic name, the same reply Topic
name, and be associated with the same DDS domain_id.
A Requester has an associated DomainParticipant, which can be shared with other requesters or Connext
DDS entities. All the other entities required for request-reply interaction, including the request and reply
Topics, the DataWriter for writing requests, and a DataReader for reading replies, are automatically created when the Requester is constructed.
Connext DDS guarantees that a Requester will only receive replies associated with the requests it sends.
The Requester uses the underlying DataReader not only to receive the replies, but also as a cache that can
hold replies to multiple outstanding requests or even multiple replies to a single request. Depending on the

1The API Reference HTML documentation is available for all supported programming languages. Open

/README.html.

881

23.1.1 Creating a Requester

HistoryQoSPolicy configuration of the DataReader, the Requester may allow replies to replace previous
replies based on the reply data having the same value for the Key fields (see DDS Samples, Instances, and
Keys (Section 2.3.1 on page 14)). The default configuration of the Requester does not allow replacing.
You can configure the QoS for the underlying DataWriter and DataReader in a QoS profile. By default,
the DataWriter and DataReader are created with default values (DDS_DATAWRITER_QOS_
DEFAULT and DDS_DATAREADER_QOS_DEFAULT, respectively) except for the following:
l

RELIABILITY QosPolicy (Section 6.5.19 on page 400): kind is set to RELIABLE.

l

HISTORY QosPolicy (Section 6.5.10 on page 376): kind is set to KEEP_ALL.

l

Several other protocol-related settings for Requesters (see the API Reference HTML documentation: select Modules, Programming How-To’s, Request-Reply Examples; then scroll
down to the section on Configuring Request-Reply QoS profiles).

23.1.1 Creating a Requester
Before you can create a Requester, you need a DomainParticipant and a service name.
Note: The example code snippets in this section use the C++ API. You can find more complete examples
in all the supported programming languages (C, C++, Java, C#) in the Connext DDS API Reference
HTML documentation and in the “example” directory found in your Connext DDS installation.
To create a Requester with the minimum set of parameters, you can use the basic constructor that receives
only an existing DDS DomainParticipant and the name of the service:
Requester  *requester =
new Requester  (
participant, “ServiceName”);

To create a Requester with specific parameters, you may use a different constructor that receives a
RequesterParams structure (described in Setting Requester Parameters (Section 23.1.3 on the next page)):
Requester (const RequesterParams ¶ms)

The ServiceName parameter is used to generate the names of the request and reply Topics that the
Requester and Replier will use to communicate. For example, if the service name is “MyService”, the
topic names for the Requester and Replier will be “MyServiceRequest” and “MyServiceReply”, respectively. Therefore, for communication to occur, you must use the same service name when creating the
Requester and the Replier entities.
If you want to use topic names different from the ones that would be derived from the ServiceName, you
can override the default names by setting the actual request and reply Topic names using the request_

882

23.1.2 Destroying a Requester

topic_name() and reply_topic_name() accessors to the RequesterParams structure prior to creating the
Requester.
Example: To create a Requester with default QoS and topic names derived from the service name, you
may use the following code:
Requester * requester =
new Requester(
participant,"MyService");

Example: To create a Requester with a specific QoS profile with library name “MyLibrary” and profile
“MyProfile” defined inside USER_QOS_PROFILES.xml in the current working directory, you may use
the following code:
Requester * requester = new Requester(
RequesterParams(participant).
service_name("MyService").qos_profile(
"MyLibrary", "MyProfile"));

Once you have created a Requester, you can use it to perform the operations in Table 23.2 Requester Operations.

23.1.2 Destroying a Requester
To destroy a Requester and free its underlying entities you may use the destructor:
virtual ~Requester ()

23.1.3 Setting Requester Parameters
To change the RequesterParams that can be used when creating a Requester, you can use the operations listed in Table 23.1 Operations to Set Requester Parameters.
Table 23.1 Operations to Set Requester Parameters
Operation

Description

datareader_
qos

Sets the QoS of the reply DataReader.

datawriter_
qos

Sets the QoS of the request DataWriter.

publisher

Sets a specific Publisher.

qos_profile

Sets a QoS profile for the DDS entities in this requester.

883

23.1.4 Summary of Requester Operations

Table 23.1 Operations to Set Requester Parameters
Operation

Description

request_
topic_name

Sets the name of the Topic used for the request. If this parameter is set, then you must also set the reply_topic_name
parameter and you should not set the service_name parameter.

reply_topic_
name

Sets the name of the Topic used for the reply. If this parameter is set, then you must also set the request_topic_name
parameter and you should not set the service_name parameter.

reply_type_
support

Sets the type support for the reply type.

request_
Sets the type support for the request type.
type_support
service_name

Sets the service name. This will automatically set the name of the request Topic and the reply Topic. If this parameter is set
you should not set the request_topic_name or the reply_topic_name.

subscriber

Sets a specific Subscriber.

23.1.4 Summary of Requester Operations
There are several kinds of operations an application can perform using the Requester:
l

Sending requests (i.e., publishing request samples on the request Topic)

l

Waiting for replies to be received.

l

l

l

Taking the reply data. This gets the reply data from the Requester and removes from the Requester
cache.
Reading the reply data. This gets the reply data from the Requester but leaves it in the Requester
cache so it remain accessible to future operations on the Requester.
Receiving replies (a convenience operation that is a combination of ‘waiting’ and ‘taking’ the data in
a single operation)

These operations are summarized in Table 23.2 Requester Operations
Table 23.2 Requester Operations
Operation
Sending
Requests

send_
request

Description
Sends a request.

Reference
Sending Requests (Section
23.1.5 on the next page)

884

23.1.5 Sending Requests

Table 23.2 Requester Operations
Operation

Description

Waiting for wait_for_
Waits for replies to any request or to a specific request.
Replies
replies

take_
reply
Taking
Reply Data
take_
replies

Reference
Waiting for Replies (Section
23.1.6.1 on the facing page)

Copies a single reply into a Sample container. There are variants that allow
getting the next reply available or the next reply to a specific request.
This operation removes the reply from the Requester cache. So subsequent calls
to take or read replies will not get the same reply again.
Returns a LoanedSamples container with the collection of replies received by
the Requester. There are variants that allow accessing all the replies available or
only the replies to a specific request.

Repliers (Section 23.2 on page
890)

This operation removes the returned replies from the Requester cache. So
subsequent calls to take or read replies will not get the same replies again.

read_
reply
Reading
Reply Data
read_
replies

Copies a single reply into a Sample container. There are variants that allow
getting the next reply available or the next reply to a specific request.
This operation leaves the reply on the Requester cache. So subsequent calls to
take or read replies can get the same reply again.
Returns a LoanedSamples container with the collection of replies received by
the Requester. There are variants that allow accessing all the replies available or
only the replies to a specific request.

Repliers (Section 23.2 on page
890)

This operation leaves the returned replies in the Requester cache. So subsequent
calls to take or read replies can get the same replies again.

Receiving
Replies

receive_
reply

Convenience function that combines a call to wait_for_replies with a call to
take_reply.

receive_
replies

Convenience function that combines a call to wait_for_replies with a call to
take_replies.

get_
request_ Retrieves the underlying DataWriter that writes requests.
datawriter

Getting
Underlying
Entities
get_
reply_
Retrieves the underlying DataReader that reads replies.
datareader

Receiving Replies (Section
23.1.6.3 on page 889)

Accessing Underlying
DataWriters and DataReaders
(Section 23.4 on page 898)

23.1.5 Sending Requests
To send a request, use the send_request() operation on the Requester. There are three variants of this operation, depending on the parameters that are passed in:

885

23.1.6 Processing Incoming Replies with a Requester

1. send_request (const TRequest &request)
2. send_request (WriteSample &request)
3. send_request (WriteSampleRef &request)
The first variant simply sends a request.
The second variant sends a request and gets back information about the request in a WriteSample container. This information can be used to correlate the request with future replies.
The third variant is just like the second, but puts the information in a WriteSampleRef, which holds references to the data and parameters. Both WriteSample and WriteSampleRef provide information about the
request that can be used to correlate the request with future replies.

23.1.6 Processing Incoming Replies with a Requester
The Requester provides several operations that can be used to wait for and access replies:
l
l

l

wait_for_replies(), see Waiting for Replies (Section 23.1.6.1 below)
take_reply(), take_replies(), read_reply() and read_replies(), see Getting Replies (Section
23.1.6.2 on the next page)
receive_reply() and receive_replies(), see Receiving Replies (Section 23.1.6.3 on page 889)

The wait_for_replies operations are used to wait until the replies arrive.
The take_reply, take_replies, read_reply, and read_replies() operations access the replies once they
have arrived.
The receive_reply and receive_replies are convenience functions that combine waiting and accessing the
replies and are equivalent to calling the ‘wait’ operation followed by the corresponding take_reply or
take_replies operations.
Each of these operations has several variants, depending on the parameters that are passed in.
23.1.6.1 Waiting for Replies
Use the wait_for_replies() operation on the Requester to wait for the replies to previously sent requests.
There are three variants of this operation, depending on the parameters that are passed in. All these variants
block the calling thread until either there are replies or a timeout occurs.
1. wait_for_replies (const DDS_Duration_t &max_wait)
2.

wait_for_replies (int min_count, const DDS_Duration_t &max_wait)

3.

wait_for_replies (int min_count,
const DDS_Duration_t &max_wait,
const SampleIdentity_t &related_request_id)

886

23.1.6.2 Getting Replies

The first variant (only passing in max_wait) blocks until a reply is available or until max_wait time has
elapsed, whichever comes first. The reply can be to any of the requests made by the Requester.
The second variant (passing in min_count and max_wait) blocks until at least min_count replies are available or until max_wait time has elapsed, whichever comes first. These replies may all be to the same
request or to different requests made by the Requester.
The third variant (passing in min_count, max_wait, and related_request_id) blocks until at least min_
count replies to the request identified by the related_request_id are available, or until max_wait time has
passed, whichever comes first. Note that unlike the previous variants, the replies must all be to the same
single request (identified by the related_request_id) made by the Requester.
Typically after waiting for replies, you will call take_reply, take_replies, read_reply, or read_replies(),
see Repliers (Section 23.2 on page 890).
If you call wait_for_replies() several times without ‘taking’ the replies (using the take_reply or take_
replies operation), future calls to wait_for_replies() will return immediately and will not wait for new
replies.
23.1.6.2 Getting Replies
You can use the following operations to access replies: take_reply, take_replies, read_reply, and read_
replies().
As mentioned in Summary of Requester Operations (Section 23.1.4 on page 884), the difference between
the ‘take’ operations (take_reply, take_replies) and the ‘read’ operations (read_reply, read_replies) is
that ‘take’ operations remove the replies from the Requester cache. This means that future calls to take_
reply, read_reply, read_reply, and read_reply will not get the same reply again.
The take_reply and read_reply operations access a single reply, whereas the take_replies and read_
replies can access a collection of replies.
There are four variants of the take_reply and read_reply operations, depending on the parameters that are
passed in:
1. take_reply (Sample &reply)
read_reply (Sample &reply)
2. take_reply (SampleRef reply)
read_reply (SampleRef reply)
3. take_reply (Sample &reply,
const SampleIdentity_t &related_request_id)
read_reply (Sample &reply,
const SampleIdentity_t &related_request_id)

887

23.1.6.2 Getting Replies

4. take_reply (SampleRef reply,
const SampleIdentity_t &related_request_id)
read_reply (SampleRef reply,
const SampleIdentity_t &related_request_id)

The first two variants provide access to the next reply in the Requester cache. This is the earliest reply to
any previous requests sent by the Requester that has not been ‘taken’ from the Requester cache. The
remaining two variants provide access to the earliest non-previously ‘taken’ reply to the request specified
by the related_request_id.
Notice that some of these variants use a Sample, while other use a SampleRef. A SampleRef can be used
much like a Sample, but it holds references to the reply data and DDS SampleInfo, so there is no additional
copy. In contrast using the Sample obtains a copy of both the data and DDS SampleInfo.
The take_replies and read_replies operations access a collection of (one or more) replies to previously
sent requests. These operations are convenient when you expect multiple replies to a single request, or
when issuing multiple requests concurrently without waiting for intervening replies.
The take_replies and read_replies operations return a LoanedSamples container that holds the replies. To
increase performance, the LoanedSamples does not copy the reply data. Instead it ‘loans’ the necessary
resources from the Requester. The resources loaned by the LoanedSamples container must be eventually
returned, either explicitly calling the return_loan() operation on the LoanedSamples or through the
destructor of the LoanedSamples.
There are three variants of the take_replies and read_replies operations, depending on the parameters that
are passed in:
1. take_replies (int max_count=DDS_LENGTH_UNLIMITED)
read_replies (int max_count=DDS_LENGTH_UNLIMITED)
2. take_replies (int max_count,
const SampleIdentity_t &related_request_id)
read_replies (int max_count,
const SampleIdentity_t &related_request_id)
3. take_replies (const SampleIdentity_t &related_request_id)
read_replies (const SampleIdentity_t &related_request_id)

The first variant (only passing in max_count) returns a container holding up to max_count replies.
The second variant (passing in max_count and related_request_id) returns a LoanedSamples container
holding up to max_count replies that correspond to the request identified by the related_request_id.

888

23.1.6.3 Receiving Replies

The third variant (only passing in related_request_id) returns a LoanedSamples container holding an
unbounded number of replies that correspond to the request identified by the related_request_id. This is
equivalent to the second variant with max_count = DDS_LENGTH_UNLIMITED.
The resources for the LoanedSamples container must be eventually be returned, either by calling the
return_loan() operation on the LoanedSamples or through the LoanedSamples destructor.
For multi-reply scenarios, in which a Requester receives multiple replies from a Replier for a given
request, the Requester can check if a reply is the last reply in a sequence of replies. To do so, see if the bit
INTERMEDIATE_REPLY_SEQUENCE_SAMPLE is set in DDS_SampleInfo’s flag field (see Table
7.18 DDS_SampleInfo Structure) after receiving each reply. This bit indicates it is NOT the last reply.
23.1.6.3 Receiving Replies
The receive_reply() operation is a shortcut that combines calls to wait_for_replies() and to take_reply().
Similarly the receive_replies() operation combines wait_for_replies() and take_replies().
There is only one variant of the receive_reply() operation:
1. receive_reply (Sample &reply, const DDS_Duration_t &timeout)

This operation blocks until either a reply is received or a timeout occurs. The contents of the reply are
copied into the provided sample (reply).
There are two variants of the receive_replies() operation, depending on the parameters that are passed in:
1. receive_replies (const DDS_Duration_t &max_wait)
2. receive_replies (int min_count, int max_count,
const DDS_Duration_t &max_wait)

These two variants block until multiple replies are available or a timeout occurs.
The first variant (only passing in max_wait) blocks until at least one reply is available or until max_wait
time has passed, whichever comes first. The operation returns a LoanedSamples container holding the
replies. Note that there could be more than one reply. This can occur if, for example, there were already
replies available in the Requester from previous requests that were not processed. This operation does not
limit the number of replies that can be returned on the LoanedSamples container.
The second variant (passing in min_count, max_count, and max_wait) will block until min_count
replies are available or until max_wait time has passed, whichever comes first. Up to max_count replies
will be stored into the LoanedSamples container which is returned to the caller.
The resources held in the LoanedSamples container must eventually be returned, either with an explicit
call to return_loan() on the LoanedSamples or through the LoanedSamples destructor.

889

23.2 Repliers

23.2 Repliers
A Replier is an entity with two associated DDS Entities: a DDS DataReader bound to a request Topic and
a DDS DataWriter bound to a reply Topic. The Replier receives requests by subscribing to the request
Topic and sends replies to those requests by publishing on the reply Topic.
Valid data types for these topics are the same as specified for the Requester, see Requesters (Section 23.1
on page 881).
For multi-reply scenarios in which a Replier generates more than one reply for a request, the Replier
should mark all intermediate replies (all but the last reply) with the INTERMEDIATE_REPLY_
SEQUENCE_SAMPLE bit-flag in the WriteParams_t flag field (see Table 6.16 DDS_WriteParams_t).
Much like a Requester, a Replier has an associated DDS DomainParticipant which can be shared with
other Connext DDS entities. All the other entities required for the request-reply interaction, including a
DataWriter for writing replies and a DataReader for reading requests, are automatically created when the
Replier is constructed.
You can configure the QoS for the underlying DataWriter and DataReader in a QoS profile. By default,
the DataWriter and DataReader are created with default QoS values (using DDS_DATAWRITER_
QOS_DEFAULT and DDS_DATAREADER_QOS_DEFAULT, respectively) except for the following:
l

RELIABILITY QosPolicy (Section 6.5.19 on page 400): kind is set to RELIABLE

l

HISTORY QosPolicy (Section 6.5.10 on page 376): kind is set to KEEP_ALL

The Replier API supports several ways in which the application can be notified of, and process, requests:
l

l

l

Blocking: The application thread blocks waiting for requests, processes them, and dispatches the
reply. In this situation, if the computation necessary to process the request and produce the reply is
small, you may consider using the SimpleReplier, which offers a simplified API.
Polling: The application thread checks (polls) for requests periodically but does not block to wait for
them. To check for data without blocking, call take_requests() or read_requests().
Asynchronous notification: The application installs a ReplierListener to receive notifications
whenever a request is received.

23.2.1 Creating a Replier
To create a Replier with the minimum set of parameters you can use the basic constructor that receives
only an existing DDS DomainParticipant and the name of the service:
Replier (DDSDomainParticipant * participant,
const std::string & service_name)

890

23.2.2 Destroying a Replier

Example:
Replier * replier =
new Replier(participant, "MyService");

To create a Replier with specific parameters you may use a different constructor that receives a ReplierParams structure:
Replier (const ReplierParams ¶ms)

Example:
Replier * replier = new Replier(
ReplierParams(participant).service_name("MyService")
.qos_profile("MyLibrary", "MyProfile"));

The service_name is used to generate the names of the request and reply Topics that the Requester and
Replier will use to communicate. For example, if the service name is “MyService”, the topic names for the
Requester and Replier will be “MyServiceRequest” and “MyServiceReply”, respectively. Therefore it is
important to use the same service_name when creating the Requester and the Replier.
If you need to specify different Topic names, you can override the default names by setting the actual
request and reply Topic names using request_topic_name() and reply_topic_name() accessors to the
ReplierParams structure prior to creating the Replier.

23.2.2 Destroying a Replier
To destroy a Replier and free its underlying entities:
virtual ~Replier ()

23.2.3 Setting Replier Parameters
To change the ReplierParams that are used to create a Replier, use the operations listed in Table 23.3
Operations to Set Replier Parameters.
Table 23.3 Operations to Set Replier Parameters
Operation

Description

datareader_qos

Sets the quality of service of the request DataReader.

datawriter_qos

Sets the quality of service of the reply DataWriter.

publisher

Sets a specific Publisher.

891

23.2.4 Summary of Replier Operations

Table 23.3 Operations to Set Replier Parameters
Operation

Description

qos_profile

Sets a QoS profile for the entities in this replier.

replier_listener

Sets a listener that is called when requests are available.

reply_topic_name

Sets a specific reply topic name.

reply_type_support

Sets the type support for the reply type.

request_topic_name

Sets a specific request topic name.

request_type_support

Sets the type support for the request type.

service_name

Sets the service name the Replier offers and Requesters use to match.

subscriber

Sets a specific Subscriber.

23.2.4 Summary of Replier Operations
There are four kinds of operations an application can perform using the Replier:
l

Waiting for requests to be received

l

Reading/taking the request data and associated information

l

l

Receiving requests (a convenience operation that combines waiting and getting the data into a single
operation)
Sending a reply for received request (i.e., publishing a reply sample on the reply Topic with special
meta-data so that the original Requester can identify it).

The Replier operations are summarized in Table 23.4 Replier Operations.
Table 23.4 Replier Operations
Operation
Waiting for
Requests

wait_for_
requests

Description
Waits for requests.

Reference
Waiting for Requests (Section 23.2.5.1 on
page 894)

892

23.2.5 Processing Incoming Requests with a Replier

Table 23.4 Replier Operations
Operation

Taking
Requests

Reading
Requests

Receiving
Requests

Sending
Replies

Getting
Underlying
Entities

Description

take_request

Copies the contents of a single request into a Sample and
removes it from the Replier cache.

take_
requests

Returns a LoanedSamples to access multiple requests and
removes the requests from the Replier cache.

read_request

Copies the contents of a single request into a Sample,
leaving it in the Replier cache

read_
requests

Returns a LoanedSamples to access multiple requests,
leaving them in the Replier cache.

receive_
request

Waits for a single request and copies its contents into a
Sample container.

receive_
requests

Waits for multiple requests and provides a LoanedSamples
container to access them.

send_reply

Sends a reply for a previous request.

get_request_
Retrieves the underlying DataReader.
datareader
get_reply_
datawriter

Reference

Reading and Taking Requests (Section
23.2.5.2 on the facing page)

Receiving Requests (Section 23.2.5.3 on page
895)

Sending Replies (Section 23.2.6 on page 896)

Accessing Underlying DataWriters and
DataReaders (Section 23.4 on page 898)

Retrieves the underlying DataWriter.

23.2.5 Processing Incoming Requests with a Replier
The Replier provides several operations that can be used to wait for and access the requests:
l
l

l

wait_for_requests(), see Waiting for Requests (Section 23.2.5.1 on the facing page)
take_request(), take_requests(), read_request(), and read_requests(), see Reading and Taking
Requests (Section 23.2.5.2 on the facing page)
receive_request() and receive_requests(), see Receiving Requests (Section 23.2.5.3 on page 895)

The wait_for_requests() operations are used to wait until requests arrive.
The take_request(), take_requests(), read_request(), and read_requests() operations access the
requests, once they have arrived.
The receive_request() and receive_requests() operations are convenience functions that combine waiting
for and accessing requests and are equivalent to calling the ‘wait’ operation followed by the corresponding
take_request() or take_requests() operations.

893

23.2.5.1 Waiting for Requests

Each of these operations has several variants, depending on the parameters that are passed in.
23.2.5.1 Waiting for Requests
Use the wait_for_requests() operation on the Replier to wait for requests. There are two variants of this
operation, depending on the parameters that are passed in. All these variants block the calling thread until
either there are replies or a timeout occurs.:
1. wait_for_requests (const DDS_Duration_t &max_wait)
2. wait_for_requests (int min_count, const DDS_Duration_t &max_wait)

The first variant (only passing in max_wait) blocks until one request is available or until max_wait time
has passed, whichever comes first.
The second variant blocks until min_count number of requests are available or until max_wait time has
passed.
Typically after waiting for requests, you will call take_request, take_requests, read_request, or read_
requests, see Sending Replies (Section 23.2.6 on page 896).
23.2.5.2 Reading and Taking Requests
You can use the following four operations to access requests: take_request, take_requests, read_
request, or read_requests.
As mentioned in Summary of Replier Operations (Section 23.2.4 on page 892), the difference between
the ‘take’ operations (take_request, take_requests) and the ‘read’ operations (read_request, read_
requests) is that ‘take’ operations remove the requests from the Replier cache. This means that future calls
to take_request, take_requests, read_request, or read_requests will not get the same request again.
The take_request and read_request operations access a single reply, whereas the take_requests and
read_requests can access a collection of replies.
There are two variants of the take_request and read_request operations, depending on the parameters
that are passed in:
1. take_request (connext::Sample & request)
read_request (connext::Sample & request)
2. take_request (connext::SampleRef & request,
const DDS_Duration_t & max_wait)
2. receive_request (connext::SampleRef request,
const DDS_Duration_t & max_wait)

The receive_request operation blocks until either a request is received or a timeout occurs. The contents
of the request are copied into the provided container (request). The first variant uses a Sample container,
whereas the second variant uses a SamepleRef container. A SampleRef can be used much like a Sample,
but it holds references to the request data and DDS SampleInfo, so there is no additional copy. In contrast,
using the Sample obtains a copy of both the data and the DDS SampleInfo.
There are two variants of the receive_requests() operation, depending on the parameters that are passed
in:
1. receive_requests (const DDS_Duration_t & max_wait)
2. receive_requests (int min_request_count,
int max_request_count,
const DDS_Duration_t & max_wait)

The receive_requests operation blocks until one or more requests are available, or a timeout occurs.

895

23.2.6 Sending Replies

The first variant (only passing in max_wait) blocks until one request is available or until max_wait time
has passed, whichever comes first. The contents of the request are copied into a LoanedSamples container
which is returned to the caller. An unlimited number of replies can be copied into the container.
The second variant blocks until min_request_count number of requests are available or until max_wait
time has passed, whichever comes first. Up to max_request_count number of requests will be copied into
a LoanedSamples container which is returned to the caller.
The resources for the LoanedSamples container must eventually be returned, either with return_loan() or
through the LoanedSamples destructor.

23.2.6 Sending Replies
There are three variants for send_reply(), depending on the parameters that are passed in:
1. send_reply
const
2. send_reply
const
3. send_reply
const

(const TReply & reply,
SampleIdentity_t & related_request_id)
(WriteSample & reply,
SampleIdentity_t & related_request_id)
(WriteSampleRef & reply,
SampleIdentity_t & related_request_id)

This operation sends a reply for a previous request. The related request ID can be retrieved from an existing request Sample.
The first variant is recommended if you do not need to change any of the default write parameters.
The other two variants allow you to set custom parameters for writing a reply. Unlike the Requester,
where retrieving the sample ID for correlation is common, on the Replier side using a WriteSample or
WriteSampleRef is only necessary when you need to overwrite the default write parameters. If that’s not
the case, use the first variant.
One reason to override the default write parameters is a multi-reply scenario in which a Replier generates
more than one reply for a request. In this case, all the intermediate replies (all but the last reply) should be
marked with the INTERMEDIATE_REPLY_SEQUENCE_SAMPLE bit-flag in the flag field within
WriteSample::info or WriteSampleRef::info.
A Requester can detect if a reply is the last reply in a sequence of replies by seeing if INTERMEDIATE_
REPLY_SEQUENCE_SAMPLE is NOT set in the flag field of Sample::info after receiving each reply.

23.3 SimpleRepliers
The SimpleReplier offers a simplified API to receive and process requests. The API is based on a userprovided object that implements the SimpleReplierListener interface. Requests are passed to the listener
operation implemented by the user-provided object, which processes the request and returns a reply.
The SimpleReplier is recommended if each request generates a single reply and computing the reply can
be done quickly with very little CPU resources and without calling any operations that may block the

896

23.3.1 Creating a SimpleReplier

processing thread. For example, looking something up in an internal memory-based data structure would
be a good use case for using a SimpleReplier.

23.3.1 Creating a SimpleReplier
To create a SimpleReplier with the minimum set of parameters, you can use the basic constructor:
SimpleReplier (DDSDomainParticipant *participant,
const std::string &service_name,
SimpleReplierListener &listener)

To create a SimpleReplier with specific parameters, you may use a different constructor that receives a SimpleReplierParams structure:
SimpleReplier (const SimpleReplierParams ¶ms)

23.3.2 Destroying a SimpleReplier
To destroy a SimpleReplier and free its resources use the destructor:
virtual ~SimpleReplier ()

23.3.3 Setting SimpleReplier Parameters
To change the SimpleReplierParams used to create a SimpleReplier, use the operations in Table 23.5 Operations to Set SimpleReplier Parameters.
Table 23.5 Operations to Set SimpleReplier Parameters
Operation

Description

datareader_qos

Sets the quality of service of the reply DataReader.

datawriter_qos

Sets the quality of service of the reply DataWriter.

publisher

Sets a specific Publisher.

qos_profile

Sets a QoS profile for the entities in this replier.

reply_topic_name

Sets a specific reply topic name.

reply_type_support

Sets the type support for the reply type.

request_topic_name

Sets a specific request topic name.

request_type_support

Sets the type support for the request type.

897

23.3.4 Getting Requests and Sending Replies with a SimpleReplierListener

Table 23.5 Operations to Set SimpleReplier Parameters
Operation

Description

service_name

Sets the service name the Replier offers and Requesters use to match.

subscriber

Sets a specific Subscriber.

23.3.4 Getting Requests and Sending Replies with a SimpleReplierListener
The on_request_available() operation on the SimpleReplierListener receives a request and returns a reply.
on_request_available(TRequest &request)

This operation gets called when a request is available. It should immediately return a reply. After calling
on_request_available(), Connext DDS will call the operation return_loan() on the SimpleReplierListener; this gives the application-defined listener an opportunity to release any resources
related to computing the previous reply.
retun_loan(TReply &reply)

23.4 Accessing Underlying DataWriters and DataReaders
Both Requester and Replier entities have underlying DDS DataWriter and DataReader entities. These are
created automatically when the Requester and Replier are constructed.
Accessing the DataWriter used by a Requester may be useful for a number of advanced use cases, such
as:
l

l
l

l

l

Finding matching subscriptions (e.g., Replier entities), see Finding Matching Subscriptions (Section
6.3.16.1 on page 309)
Setting a DataWriterListener, see Setting Up DataWriterListeners (Section 6.3.4 on page 269)
Getting DataWriter protocol or cache statuses, see Statuses for DataWriters (Section 6.3.6 on page
271)
Flushing a data batch after sending a number of request samples, see Flushing Batches of DDS Data
Samples (Section 6.3.9 on page 287)
Modifying the QoS

Accessing the reply DataReader may be useful for a number of advanced use cases, such as:
l

Finding matching publications (e.g., Requester entities), see Navigating Relationships Among Entities (Section 7.3.9 on page 489)

898

23.4 Accessing Underlying DataWriters and DataReaders

l

l

Getting DataReader protocol or cache statuses, see Checking DataReader Status and StatusConditions (Section 7.3.5 on page 468) and Statuses for DataReaders (Section 7.3.7 on page 470).
Modifying the QoS

To access these underlying objects:
RequestDataWriter
RequestDataReader
ReplyDataWriter *
ReplyDataReader *

* get_request_datawriter()
* get_request_datareader()
get_reply_datawriter()
get_reply_datareader()

899

Part 5: RTI Secure WAN Transport
The material in this part of the manual is only relevant if you have installed Secure WAN
Transport.
This feature is not installed as part of a Connext DDS package; it must be downloaded and
installed separately. It is only available on specific architectures. See the Secure WAN
Transport Release Notes and Installation Guide for details.
Secure WAN Transport is an optional package that enables participant discovery and data
exchange in a secure manner over the public WAN. Secure WAN Transport enables Connext
DDS to address the challenges in NAT traversal and authentication of all participants. By implementing UDP hole punching using the STUN protocol and providing security to channels by leveraging DTLS (Datagram TLS), you can securely exchange information between different sites
separated by firewalls.
This section includes:
l

Introduction to Secure WAN Transport (Section Chapter 24 on page 901)

l

Configuring RTI Secure WAN Transport (Section Chapter 25 on page 914)

900

Chapter 24 Introduction to Secure WAN
Transport
Secure WAN Transport provides transport plugins that can be used by developers of Connext
DDS applications. These transport plugins allow Connext DDS applications running on private networks to communicate securely over a Wide-Area Network (WAN), such the internet. There are
two primary components in the package which may be used independently or together: communication over Wide-Area Networks that involve Network Address Translators (NATs), and
secure communication with support for peer authentication and encrypted data transport.
The Connext DDS core is transport-agnostic. Connext DDS offers three built-in transports:
UDP/IPv4, UDP/IPv6, and inter-process shared memory. The implementation of NAT traversal
and secure communication is done at the transport level so that the Connext DDS core is not
affected and does not need to be changed, although there is additional on-the-wire traffic.
The basic problem to overcome in a WAN environment is that messages sent from an application
on a private local-area network (LAN) appear to come from the LAN's router address, not from the
internal IP address of the host running the application. This is due to the existence of a Network
Address Translator (NAT) at the gateway. This does not cause problems for client/server systems
because only the server needs to be globally addressable; it is only a problem for systems with
peer-to-peer communication models, such as Connext DDS. Secure WAN Transport solves this
problem, allowing communication between peers that are in separate LAN networks, using a UDP
hole-punching mechanism based on the STUN protocol (IETF RFC 3489bis) for NAT traversal.
This requires the use of an additional rendezvous server application, the RTI WAN Server.
Once the transport has enabled traffic to cross the NAT gateway to the WAN, it is flowing on network hardware that is shared (in some cases, over the public internet). In this context, it is important to consider the security of data transmission. There are three primary issues involved:
l

Authenticating the communication peer (source or destination) as a trusted partner;

l

Encrypting the data to hide it from other parties that may have access to the network;

901

24.1 WAN Traversal via UDP Hole-Punching

l

Validating the received data to ensure that it was not modified in transmission.

Secure WAN Transport addresses these problems by wrapping all RTPS-encoded data using the DTLS
protocol (IETF RFC 4347), which is a variant of SSL/TLS that can be used over a datagram networklayer transport such as UDP. The security features of the WAN Transport may also be used on an untrusted local-area network with the Secure Transport.
In summary, the package includes two transports:
l

l

The WAN Transport is for use on a WAN and includes security. It must be used with the WAN
Server, a rendezvous server that provides the ability to discover public addresses and to register and
look up peer addresses based on a unique WAN ID. The WAN Server is based on the STUN (Session Traversal Utilities for NAT) protocol [draft-ietfbehave-rfc3489bis], with some extensions.
Once information about public addresses for the application and its peers has been obtained and connections have been initiated, the server is no longer required to maintain communication with a peer.
(Note: security is disabled by default.)
The Secure Transport is an alternate transport that provides security on an untrusted LAN. Use of
the RTI WAN Server is not required.

Multicast communication is not supported by either of these transports.
This chapter provides a technical overview of:
l

WAN Traversal via UDP Hole-Punching (Section 24.1 below)

l

WAN Locators (Section 24.2 on page 907)

l

Datagram Transport-Layer Security (DTLS) (Section 24.3 on page 908)

l

Certificate Support (Section 24.4 on page 909)

For information on how to use Secure WAN Transport with your Connext DDS application, see Configuring RTI Secure WAN Transport (Section Chapter 25 on page 914).

24.1 WAN Traversal via UDP Hole-Punching
In order to resolve the problem of communication across NAT boundaries, the WAN Transport implements a UDP hole-punching solution for NAT traversal [draft-ietf-behave-p2p-state]. This solution uses a
rendezvous server, which provides the ability to discover public addresses, and to register and lookup peer
addresses based on a unique WAN ID. This server is based on the STUN (Session Traversal Utilities for
NAT) protocol [draft-ietf-behave-rfc3489bis], with some extensions. This protocol is a part of the solution
used for standards-based voice over IP applications; similar technology has be used by systems such as
Skype and has proven to be highly reliable. A key advantage of STUN is that it is based on UDP and
therefore is able to preserve the real-time characteristics of the DDS Interoperability Wire Protocol.

902

24.1.1 Protocol Details

Once information about public addresses for the application and its peers has been obtained, and connections have been initiated, the server is no longer required to maintain communication with a peer.
However, if communication fails, possibly due to changes in dynamically-allocated addresses, the server
will be needed to reopen new public channels.
Figure 24.1 RTI WAN Transport Architecture below shows the RTI WAN transport architecture.
Figure 24.1 RTI WAN Transport Architecture

24.1.1 Protocol Details
The UDP hole-punching algorithm implemented by the WAN transport has two different phases: registration and connection. This algorithm only works with cone or asymmetric NATs where the same public
address/port is assigned to all the sessions with the same private address/port address.
l

Registration Phase
The RTI WAN Server application runs on a machine that resides on the WAN network (i.e., not in
a private LAN). It has to be globally accessible to LAN applications. It is started by a script and acts
as a rendezvous point for LAN applications. During the registration phase, each transport locator is
registered with the RTI WAN Server using a STUN binding request message.
The RTI WAN Server associates RTPS locators with their corresponding public IPv4 transport
addresses (a combination of IP address and port) and stores that information in an internal table. Figure 24.2 Registration Phase on the next page illustrates the registration phase.

903

24.1.1 Protocol Details

Figure 24.2 Registration Phase

904

24.1.1 Protocol Details

l

Connection Phase
The connection phase starts when locator A wants to establish a connection with locator B. Locator
A obtains information about locator B via Connext DDS discovery traffic or the initial NDDS_
DISCOVERY_PEERS list. To establish a connection with locator B, locator A sends a STUN connect request to the RTI WAN server. The server sends a STUN connect response to locator A,
including information about the public IP transport address (IP address and port) of locator B. In parallel, the RTI WAN server contacts locator B using another STUN connect request to let it know
that locator A wants to establish a connection with it.
When locator A receives the public IP address of locator B, it will try to contact B using two STUN
binding request messages. The first message is sent to the public address of B and the second message is sent to the private address of B. The private address was obtained using the last 32 bits of the
locator address of B. The STUN binding request message directed to the public transport address of
B sent by locator A will open a hole in A's NAT to receive messages from B.
When locator B receives the public address of locator A, it will try to contact A sending a STUN
binding request message to that public address. This message will open a hole in B's NAT to receive
messages from A. When locator A receives the first STUN binding response from locator B, it starts
sending RTPS traffic.
The connection phase includes two processes: the connect process (Figure 24.3 Connect Process on
the next page) and the NAT hole punching process (Figure 24.4 NAT Hole Punching Process on
the next page).

905

24.1.1 Protocol Details

Figure 24.3 Connect Process

Figure 24.4 NAT Hole Punching Process

906

24.2 WAN Locators

l

STUN Liveliness
Finally, since bindings allocated by NAT expire unless refreshed, the clients (locators) must generate binding request messages for the server and other clients to refresh the bindings. The RTI
STUN protocol implementation uses the attribute LIVELINESS-PERIOD in the STUN binding
request to indicate the period in milliseconds at which a client will assert its liveliness. The WAN
Server will remove a locator from its mapping table when the liveliness contract is not met. Likewise, a transport instance will remove a STUN connection with a locator when this locator does not
assert its liveliness as indicated in the last binding request.

24.2 WAN Locators
The WAN transport does not use simple IP addresses to locate peers. A WAN transport locator consists of
a WAN ID, which is an arbitrary 12-byte value, and a bottom 4-byte value that specifies a fallback local
IPv4 address. Your peers list (NDDS_DISCOVERY_PEERS) must be configured to look for peers with
locators of the form:

l

The address is a 128-bit address in IPv6 notation.

l

The "wan://" part specifies that the address is for the WAN transport.

l

l

The next part, "::1", specifies the top 12 bytes of the address to be 11 zero bytes, followed by a byte
with value 1 (this corresponds to the peer's WAN ID).
The last part, "10.10.1.150" refers to the peers local IPv4 address, which will be used if the peers
are on the same local network.

A DomainParticipant using the WAN transport will have to initialize the DDS_DiscoveryQosPolicy’s initial_peers field with the WAN locator addresses corresponding to the peers to which it wants to connect
to. The value of initial_peers can be set using the environment variable NDDS_DISCOVERY_PEERS
or the NDDS_DISCOVERY_PEERS configuration file. (See Configuring the Peers List Used in Discovery (Section 14.2 on page 711).)

907

24.3 Datagram Transport-Layer Security (DTLS)

24.3 Datagram Transport-Layer Security (DTLS)
Data security is provided by wrapping all Connext DDS network traffic with the Datagram Transport
Layer Security (DTLS) protocol (IETF RFC 4347). DTLS is a relatively recent variant of the mature
SSL/TLS family of protocols which adds the capability to secure communication over a connectionless network-layer transport such as UDP. UDP is the preferred network layer transport for the DDS wire protocol RTPS, as well as for NAT traversal. Like SSL/TLS, the DTLS protocol provides capabilities for
certificate-based authentication, data encryption, and message integrity. The protocol specifies a number of
standard cryptographic algorithms that must be available; the base set is listed in the TLS 1.1 specification
(IETF RFC 4346).
Secure protocol support is provided by the open source OpenSSL library, which has supported the DTLS
protocol since the release of OpenSSL 0.9.8. Note however that many critical issues in DTLS were
resolved by the OpenSSL 0.9.8f release. For more detailed information about available ciphers, certificate
support, etc. please refer to the OpenSSL documentation. The DTLS protocol securely authenticates with
each individual peer; as such, multicast communication is not supported by the Secure Transport. There is
also a FIPS security-certified version of OpenSSL (OpenSSL-FIPS 1.1.1), but this does not yet support
DTLS.
The Secure Transport protocol stack is similar to the Secure WAN transport stack, but without the STUN
layer and server. See DTLS Architecture (Section Figure 24.5 below).
Figure 24.5 DTLS Architecture

908

24.3.1 Security Model

24.3.1 Security Model
In order to communicate securely, an instance of the secure plugin requires: 1) a certificate authority
(shared with all peers), 2) an identifying certificate which has been signed by the authority, 3) the private
key associated with the public key contained in the certificate.
The Certificate Authority (CA) is specified by using a PEM format file containing its public key or by
using a directory of PEM files following standard OpenSSL naming conventions. If a single CA file is
used, it may contain multiple CA keys. In order to successfully communicate with a peer, the CA keys that
are supplied must include the CA that has signed that peer's identifying certificate.
The identifying certificate is specified by using a PEM format file containing the chain of CAs used to
authenticate the certificate. The identifying certificate must be signed by a CA. It will either be directly
signed by a root CA (one of the CAs supplied above), by an authority whose certificate has been signed
by the root CA, or by a longer chain of certificate authorities. The file must be sorted starting with the certificate to the highest level (root CA). If the certificate is directly signed by a root CA, then this file will
only contain the root CA certificate followed by the identity certificate.
Finally, a private key is required. In order to avoid impersonation of an identity, this should be kept
private. It can be stored in its own PEM file specified in one of the private key properties, or it can be
appended to the certificate chain file.
One complication in the use of DTLS for communication by Connext DDS is that even though DTLS is a
connectionless protocol, it still has client/server semantics. The RTI Secure Transport maps a bidirectional
communication channel between two peer applications into a pair of unidirectional encrypted channels.
Both peers are playing the part of a client (when sending data) and a server (when receiving).

24.3.2 Liveliness Mechanism
When a peer shuts down cleanly, the DTLS protocol ensures that resources are released. If a peer crashes
or otherwise stops responding, a liveliness mechanism in the DTLS transport cleans up resources. You can
configure the DTLS handshake retransmission interval and the connection liveliness interval.

24.4 Certificate Support
Cryptographic certificates are required to use the security features of the WAN transport. This section
describes a mechanism to use the OpenSSL command line tool to generate a simple private certificate
authority. For more information, see the manual page for the openssl tool (http://www.openssl.org/docs/apps/openssl.html) or the book, "Network Security with OpenSSL" by Viega, Messier, & Chandra
(O'Reilly 2002), or other references on Public Key Infrastructure.
1. Initialize the Certificate Authority:
a. Create a copy of the openssl.cnf file and edit fields to specify the proper default names and
paths.

909

24.4 Certificate Support

b. Create the required CA directory structure:
mkdir myCA

mkdir myCA/certs

mkdir myCA/private

mkdir myCA/newcerts

mkdir myCA/crl

touch myCA/index.txt

c. Create a self-signed certificate and CA private key:
openssl req -nodes -x509 -days 1095 -newkey rsa:2048 \
-keyout myCA/private/cakey.pem -out myCA/cacert.pem \
-config openssl.cnf

2. For each identifying certificate:
a. You may want to create a copy of your customized openssl.cnf file with default identifying information to be used as a template for certificate request creation; the commands below refer to this file as
template.cnf.
b. Generate a certificate request and private key:
openssl req -nodes -new -newkey rsa:2048 -config template.cnf \
-keyout peer1key.pem -out peer1req.pem

c. Use the CA to sign the certificate request to generate certificate:
openssl ca -create_serial -config openssl.cnf -days 365 \
-in peer1req.pem -out myCA/newcerts/peer1cert.pem

d. Optionally, append the private key to the peer certificate:
cat myCA/newcerts/peer1cert.pem peer1key.pem \
$>${private location}/ peer1.pem

910

24.5 License Issues

24.5 License Issues
The OpenSSL toolkit stays under a dual license, i.e., both the conditions of the OpenSSL License and the
original SSLeay license apply to the toolkit. See below for the actual license texts. Actually both licenses
are BSD-style Open Source licenses. In case of any license issues related to OpenSSL please contact
openssl-core@openssl.org.
/*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*

====================================================================
Copyright (c) 1998-2007 The OpenSSL Project. All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
1. Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in
the documentation and/or other materials provided with the
distribution.
3. All advertising materials mentioning features or use of this
software must display the following acknowledgment:
"This product includes software developed by the OpenSSL Project
for use in the OpenSSL Toolkit. (http://www.openssl.org/)" *
4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
endorse or promote products derived from this software without
prior written permission. For written permission, please contact
openssl-core@openssl.org.
5. Products derived from this software may not be called "OpenSSL"
nor may "OpenSSL" appear in their names without prior written
permission of the OpenSSL Project.
6. Redistributions of any form whatsoever must retain the following
acknowledgment:
"This product includes software developed by the OpenSSL Project
for use in the OpenSSL Toolkit (http://www.openssl.org/)"
THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE OpenSSL PROJECT OR
ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
OF THE POSSIBILITY OF SUCH DAMAGE.
====================================================================
This product includes cryptographic software written by Eric Young
(eay@cryptsoft.com). This product includes software written by Tim
Hudson (tjh@cryptsoft.com).

911

24.5 License Issues

*/
Original SSLeay License
----------------------/* Copyright (C) 1995-1998 Eric Young (eay@cryptsoft.com)
* All rights reserved.
*
* This package is an SSL implementation written
* by Eric Young (eay@cryptsoft.com).
* The implementation was written so as to conform with Netscapes SSL.
*
* This library is free for commercial and non-commercial use as long as
* the following conditions are aheared to. The following conditions
* apply to all code found in this distribution, be it the RC4, RSA,
* lhash, DES, etc., code; not just the SSL code. The SSL documentation
* included with this distribution is covered by the same copyright terms
* except that the holder is Tim Hudson (tjh@cryptsoft.com).
*
* Copyright remains Eric Young's, and as such any Copyright notices in
* the code are not to be removed.
* If this package is used in a product, Eric Young should be given
* attribution
* as the author of the parts of the library used.
* This can be in the form of a textual message at program startup or
* in documentation (online or textual) provided with the package.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the copyright
*
notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
*
notice, this list of conditions and the following disclaimer in the
*
documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
*
must display the following acknowledgement:
*
"This product includes cryptographic software written by
*
Eric Young (eay@cryptsoft.com)"
*
The word 'cryptographic' can be left out if the routines from the
*
library
*
being used are not cryptographic related :-).
* 4. If you include any Windows specific code (or a derivative thereof)
*
from the apps directory (application code) you must include an
*
acknowledgement:
*
"This product includes software written by Tim Hudson
*
(tjh@cryptsoft.com)"
*
* THIS SOFTWARE IS PROVIDED BY ERIC YOUNG ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,

912

24.5 License Issues

*
*
*
*
*
*
*
*
*
*
*

STRICT
LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
SUCH DAMAGE.
The licence and distribution terms for any publicly available
version or
derivative of this code cannot be changed. i.e. this code cannot
simply be
copied and put under another distribution licence
[including the GNU Public Licence.] */

913

Chapter 25 Configuring RTI Secure
WAN Transport
The Secure WAN Transport package includes two transports:
l

l

The WAN Transport is for use on a WAN and includes security.1 It must be used with the
WAN Server, a separate application that provides additional services needed for Connext
DDS applications to communicate with each other over a WAN.
The Secure Transport is an alternate transport that provides security on an untrusted LAN.
Use of the RTI WAN Server is not required.

There are two ways in which these transports can be configured:
l

l

By setting up predefined strings in the Property QoS Policy of the DomainParticipant (on
UNIX, Solaris and Windows systems only). This process is described in Setting Up a Transport with the Property QoS (Section 25.2 on the next page).
By instantiating a new transport (Explicitly Instantiating a WAN or Secure Transport Plugin
(Section 25.5 on page 930)) and then registering it with the DomainParticipant, see
Installing Additional Builtin Transport Plugins with register_transport() (Section 15.7 on
page 765) (not available in Java API).

Refer to the API Reference HTML documentation for details on these two approaches.

25.1 Example Applications
A simple example is available to show how to configure the WAN transport. It includes example
settings to enable communication over WAN, and optional settings to enable security (along with

1Security is disabled by default.

914

25.2 Setting Up a Transport with the Property QoS

example certificate files to use for secure communication). The example is located in a/connext_dds//hello_world_wan.
As seen in the example, you can configure the properties of either transport by setting the appropriate
name/value pairs in the DomainParticipant’s PropertyQoS, as described in Setting Up a Transport with
the Property QoS (Section 25.2 below). This will cause Connext DDS to dynamically load the WAN or
Secure Transport libraries at run time and then implicitly create and register the transport plugin.
Another way to use the WAN or Secure transports is to explicitly create the plugin and use register_transport() to register the transport with Connext DDS (see Installing Additional Builtin Transport Plugins with
register_transport() (Section 15.7 on page 765)). This way is not shown in the example. See Explicitly
Instantiating a WAN or Secure Transport Plugin (Section 25.5 on page 930).

25.2 Setting Up a Transport with the Property QoS
The PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394) allows you to set up name/value pairs of data and attach them to an entity, such as a DomainParticipant. This will cause Connext
DDS to dynamically load the WAN or Secure Transport libraries at run time and then implicitly create and
register the transport plugin.
Please refer to Setting Builtin Transport Properties with the PropertyQosPolicy (Section 15.6 on page
748).
To assign properties, use the add_property() operation:
DDS_ReturnCode_t DDSPropertyQosPolicyHelper::add_property
(DDS_PropertyQosPolicy policy,
const char * name,
const char * value,
DDS_Boolean propagate)

For more information on add_property() and the other operations in the DDSPropertyQosPolicyHelper
class, please see Table 6.57 PropertyQoSPolicyHelper Operations, as well as the API Reference HTML
documentation.
The ‘name’ part of the name/value pairs is a predefined string, described in WAN Transport Properties
(Section 25.3 on page 917) and Secure Transport Properties (Section 25.4 on page 925).
Here are the basic steps, taken from the example Hello World application (for details, please see the
example application.)

aSee Paths Mentioned in Documentation (Section on page xxxviii).
915

25.2 Setting Up a Transport with the Property QoS

1. Get the default DomainParticipant QoS from the DomainParticipantFactory.
DDSDomainParticipantFactory::get_instance()->
get_default_participant_qos(participant_qos);

2. Disable the builtin transports.
participant_qos.transport_builtin.mask =
DDS_TRANSPORTBUILTIN_MASK_NONE;

3. Set up the DomainParticipant’s Property QoS.
a. Load the plugin.
DDSPropertyQosPolicyHelper::add_property (
participant_qos.property,
"dds.transport.load_plugins",
"dds.transport.wan_plugin.wan",
DDS_BOOLEAN_FALSE);

b. Specify the transport plugin library.
DDSPropertyQosPolicyHelper::add_property (
participant_qos.property,
"dds.transport.wan_plugin.wan.library",
"libnddstransportwan.so",
DDS_BOOLEAN_FALSE);

c. Specify the transport’s ‘create’ function.
DDSPropertyQosPolicyHelper::add_property (
participant_qos.property,
"dds.transport.wan_plugin.wan.create_function"
"NDDS_Transport_WAN_create",
DDS_BOOLEAN_FALSE);

d. Specify the WAN Server and instance ID.
DDSPropertyQosPolicyHelper::add_property (
participant_qos.property
"dds.transport.wan_plugin.wan.server",
"192.168.1.1",
DDS_BOOLEAN_FALSE);
DDSPropertyQosPolicyHelper::add_property (
participant_qos.property,

916

25.3 WAN Transport Properties

"dds.transport.wan_plugin.wan.transport_instance_id",
1,
DDS_BOOLEAN_FALSE);

e. Specify any other properties, as needed.
4. Create the DomainParticipant, using the modified QoS.
participant = DDSTheParticipantFactory->create_participant (
domainId,
participant_qos,
NULL /* listener */,
DDS_STATUS_MASK_NONE);

Property changes should be made before the transport is loaded: either before the
DomainParticipant is enabled, before the first DataWriter/DataReader is created, or before the
builtin topic reader is looked up, whichever one happens first.

25.3 WAN Transport Properties
Table 25.1 Properties for NDDS_Transport_WAN_Property_t lists the properties that you can set for the
WAN Transport.
Table 25.1 Properties for NDDS_Transport_WAN_Property_t
Property Name
(prefix with
‘dds.transport.WAN.wan1.’)

Property Value Description

1

Required
Comma-separated strings indicating the prefix names of all plugins that will be loaded by Connext
DDS. You will use this string as the prefix to the property names.
dds.transport.load_plugins
(note: this does not take a prefix)

For example: “dds.transport.WAN.wan1". (This assumes you used ‘dds.transport.WAN.wan1’ as
the alias to load the plugin. If not, change the prefix to match the string used with
dds.transport.load_plugins.)
This prefix must begin with 'dds.transport.'
Note: You can load up to 8 plugins.

1 Assuming you used ‘dds.transport.WAN.wan1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

917

25.3 WAN Transport Properties

Table 25.1 Properties for NDDS_Transport_WAN_Property_t
Property Name
(prefix with
‘dds.transport.WAN.wan1.’)

Property Value Description

1

Required

library

Must set to "libnddstransportwan.so" (for UNIX/Solaris systems) or "nddstransportwan.dll"
(for Windows system).
This library and the dependent OpenSSL libraries need to be in your library search path (pointed to
by the environment variable LD_LIBRARY_PATH on UNIX/Solaris systems, Path on Windows
systems, LIBPATH on AIX systems, DYLD_LIBRARY_PATH on Mac OS systems).
Required

create_function
Must be "NDDS_Transport_WAN_create"

aliases

Used to register the transport plugin returned by NDDS_Transport_WAN_create() (as specified
by .create_function) to the DomainParticipant. Aliases should be specified as a
comma-separated string, with each comma delimiting an alias.
If it is not specified, the prefix is used as the default alias for the plugin.
Specifies the verbosity of log messages from the transport.

verbosity

Possible values:
-1: silent
0 (default): errors only
1: errors and warnings
2: local status
5 or higher: all messages

parent.parent.address_bit_count

Number of bits in a 16-byte address that are used by the transport. Should be between 0 and 128.
For example, for an address range of 0-255, the address_bit_count should be set to 8.

parent.parent.properties_
bitmap

A bitmap that defines various properties of the transport to the Connext DDS core. Currently, the
only property supported is whether or not the transport plugin will always loan a buffer when
Connext DDS tries to receive a message using the plugin. This is in support of a zero-copy
interface.

1 Assuming you used ‘dds.transport.WAN.wan1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

918

25.3 WAN Transport Properties

Table 25.1 Properties for NDDS_Transport_WAN_Property_t
Property Name
(prefix with
‘dds.transport.WAN.wan1.’)

Property Value Description

1

Specifies the maximum number of buffers that Connext DDS can pass to the send() function of
the transport plugin.

parent.parent.gather_send_
buffer_count_max

The transport plugin send() API supports a gather-send concept, where the send() call can take
several discontiguous buffers, assemble and send them in a single message. This enables Connext
DDS to send a message from parts obtained from different sources without first having to copy the
parts into a single contiguous buffer.
However, most transports that support a gather-send concept have an upper limit on the number of
buffers that can be gathered and sent. Setting this value will prevent Connext DDS from trying to
gather too many buffers into a send call for the transport plugin.

Connext DDS requires all transport-plugin implementations to support a gather-send of least a
minimum number of buffers. This minimum number is defined as NDDS_TRANSPORT_PROPERTY_GATHER_SEND_BUFFER_COUNT_MIN.
The maximum size of a message in bytes that can be sent or received by the transport plugin.
parent.parent.message_size_max

This value must be set before the transport plugin is registered, so that Connext DDS can properly
use the plugin.
A list of strings, each identifying a range of interface addresses.
Interfaces must be specified as comma-separated strings, with each comma delimiting an interface.
If the list is non-empty, this "white" list is applied before the parent.parent.deny_interfaces (Section
on the facing page) list.

parent.parent.allow_interfaces

It is up to the transport plugin to interpret the list of strings passed in. Usually this interpretation will
be consistent with NDDS_Transport_String_To_Address_Fcn_cEA().
This property is not interpreted by the Connext DDS core; it is provided merely as a convenient
and standardized way to specify the interfaces for the benefit of the transport plugin developer and
user.
You must manage the memory of the list. The memory may be freed after the DomainParticipant is
enabled.

1 Assuming you used ‘dds.transport.WAN.wan1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

919

25.3 WAN Transport Properties

Table 25.1 Properties for NDDS_Transport_WAN_Property_t
Property Name
(prefix with
‘dds.transport.WAN.wan1.’)

Property Value Description

1

A list of strings, each identifying a range of interface addresses. If the list is non-empty, deny the
use of these interfaces.
Interfaces must be specified as comma-separated strings, with each comma delimiting an interface.
This "black" list is applied after the parent.parent.allow_interfaces (Section on the previous page) list
and filters out the interfaces that should not be used.
parent.parent.deny_interfaces

It is up to the transport plugin to interpret the list of strings passed in. Usually this interpretation will
be consistent with NDDS_Transport_String_To_Address_Fcn_cEA().
This property is not interpreted by the Connext DDS core; it is provided merely as a convenient
and standardized way to specify the interfaces for the benefit of the transport plugin developer and
user.
You must manage the memory of the list. The memory may be freed after the DomainParticipant is
enabled.
Size in bytes of the send buffer of a socket used for sending. On most operating systems,
setsockopt() will be called to set the SENDBUF to the value of this parameter.

parent.send_socket_buffer_size

This value must be greater than or equal to
parent.parent.message_size_max (Section on the previous page).
The maximum value is operating system-dependent.
If NDDS_TRANSPORT_UDPV4_SOCKET_BUFFER_SIZE_OS_DEFAULT, then setsockopt()
(or equivalent) will not be called to size the send buffer of the socket.
Size in bytes of the receive buffer of a socket used for receiving.
On most operating systems, setsockopt() will be called to set the RECVBUF to the value of this
parameter.

parent.recv_socket_buffer_size

This value must be greater than or equal to parent.parent.message_size_max (Section on the
previous page). The maximum value is operating system-dependent.
If NDDS_TRANSPORT_UDPV4_SOCKET_BUFFER_SIZE_OS_DEFAULT, then setsockopt()
(or equivalent) will not be called to size the receive buffer of the socket.

parent.unicast_enabled

Allows the transport plugin to use unicast UDP for sending and receiving. By default, it will be
turned on. Also by default, it will use all the allowed network interfaces that it finds up and running
when the plugin is instanced.

1 Assuming you used ‘dds.transport.WAN.wan1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

920

25.3 WAN Transport Properties

Table 25.1 Properties for NDDS_Transport_WAN_Property_t
Property Name
(prefix with
‘dds.transport.WAN.wan1.’)

Property Value Description

1

Prevents the transport plugin from using the IP loopback interface. Three values are allowed:

parent.ignore_loopback_interface

0: Enable local traffic via this plugin. This plugin will only use and report the IP loopback interface
only if there are no other network interfaces (NICs) up on the system.
1: Disable local traffic via this plugin. Do not use the IP loopback interface even if no NICs are
discovered. This is useful when you want applications running on the same node to use a more
efficient plugin like Shared Memory instead of the IP loopback.
Prevents the transport plugin from using a network interface that is not reported as RUNNING by
the operating system.

parent.ignore_nonrunning_
interfaces

The transport checks the flags reported by the operating system for each network interface upon
initialization. An interface which is not reported as UP will not be used. This property allows the
same check to be extended to the IFF_RUNNING flag implemented by some operating systems.
The RUNNING flag is defined to mean that "all resources are allocated", and may be off if there is
no link detected, e.g., the network cable is unplugged.
Two values are allowed:
0: Do not check the RUNNING flag when enumerating interfaces, just make sure the interface is
UP.
1: Check the flag when enumerating interfaces, and ignore those that are not reported as
RUNNING. This can be used on some operating systems to cause the transport to ignore interfaces
that are enabled but not connected to the network.
Prevents the transport plugin from doing a zero copy.
By default, this plugin will use the zero copy on OSs that offer it. While this is good for
performance, it may sometime tax the OS resources in a manner that cannot be overcome by the
application.

parent.no_zero_copy

The best example is if the hardware/device driver lends the buffer to the application itself. If the
application does not return the loaned buffers soon enough, the node may error or malfunction. In
case you cannot reconfigure the H/W, device driver, or the OS to allow the zero copy feature to
work for your application, you may have no choice but to turn off zero copy use.
By default this is set to 0, so Connext DDS will use the zero-copy API if offered by the OS.

1 Assuming you used ‘dds.transport.WAN.wan1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

921

25.3 WAN Transport Properties

Table 25.1 Properties for NDDS_Transport_WAN_Property_t
Property Name
(prefix with
‘dds.transport.WAN.wan1.’)

Property Value Description

1

Controls the blocking behavior of send sockets. CHANGING THIS FROM THE DEFAULT
CAN CAUSE SIGNIFICANT PERFORMANCE PROBLEMS.
Two values are defined:
l

parent.send_blocking
l

NDDS_TRANSPORT_UDPV4_BLOCKING_ALWAYS: Sockets are blocking (default socket options for Operating System).
NDDS_TRANSPORT_UDPV4_BLOCKING_NEVER: Sockets are modified
to make them non-blocking. THIS IS NOT A SUPPORTED
CONFIGURATION AND MAY CAUSE SIGNIFICANT
PERFORMANCE PROBLEMS.

Mask for the transport priority field. This is used in conjunction with transport_priority_mapping_
low/high to define the mapping from DDS transport priority to the IPv4 TOS field. Defines a
contiguous region of bits in the 32-bit transport priority value that is used to generate values for the
IPv4 TOS field on an outgoing socket.
parent.transport_priority_mask

For example, the value 0x0000ff00 causes bits 9-16 (8 bits) to be used in the mapping. The value
will be scaled from the mask range (0x0000 - 0xff00 in this case) to the range specified by low and
high.
If the mask is set to zero, then the transport will not set IPv4 TOS for send sockets.

parent.transport_priority_
mapping_low
parent.transport_priority_
mapping_high

Sets the low and high values of the output range to IPv4 TOS.
These values are used in conjunction with transport_priority_mask to define the mapping from DDS
transport priority to the IPv4 TOS field. Defines the low and high values of the output range for
scaling.
Note that IPv4 TOS is generally an 8-bit value.

enable_security

Required if you want to use security.

recv_decode_buffer_size

Size of buffer for decoding packets from wire. An extra buffer is required for storage of encrypted
data. The minimum value for this property is parent.parent.message_size_max (Section on page
919).

port_offset

Port offset to allow coexistence with non-secure UDP transport.

dtls_handshake_resend_interval

DTLS handshake retransmission interval in milliseconds

1 Assuming you used ‘dds.transport.WAN.wan1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

922

25.3 WAN Transport Properties

Table 25.1 Properties for NDDS_Transport_WAN_Property_t
Property Name
(prefix with
‘dds.transport.WAN.wan1.’)

Property Value Description

1

Liveliness interval (multiple of resend interval)
dtls_connection_liveliness_
interval

The connection will be dropped if no message from the peer is received in this amount of time. This
enables cleaning up state for peers that are no longer responding. A secure keep-alive message will
be sent every half-interval if no other sends have occurred for a given DTLS connection during that
time.
Default: 60 ms
A string that specifies the name of file containing Certificate Authority certificates. File should be in
PEM format. See the OpenSSL manual page for SSL_load_verify_locations for more information.

tls.verify.ca_file
If you want to use security, tls.verify.ca_file (Section above) or tls.verify.ca_path (Section below)
must be specified; both may be specified.

tls.verify.ca_path

A string that specifies paths to directories containing Certificate Authority certificates. Files should
be in PEM format, and follow the OpenSSL-required naming conventions. See the OpenSSL
manual page for SSL_CTX_load_verify_locations for more information.
If you want to use security, tls.verify.ca_file (Section above) or tls.verify.ca_path (Section above)
must be specified; both may be specified.

tls.verify.verify_depth

Maximum certificate chain length for verification.

tls.verify.verify_peer

If non-zero, use mutual authentication when performing TLS hand- shake (default). If zero, only the
reader side will present a certificate, which will be verified by the writer side.
This can be set to one of three values:
l

"default" selects NDDS_Transport_TLS_default_verify_callback()

l

"verbose" selects NDDS_Transport_TLS_verbose_verify_callback()

l

"none" requests no callback be registered

tls.verify.callback

tls.cipher.cipher_list

List of available (D)TLS ciphers. See the OpenSSL manual page for SSL_set_cipher_list for more
information on the format of this string.

1 Assuming you used ‘dds.transport.WAN.wan1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

923

25.3 WAN Transport Properties

Table 25.1 Properties for NDDS_Transport_WAN_Property_t
Property Name
(prefix with
‘dds.transport.WAN.wan1.’)

Property Value Description

1

List of available Diffie-Hellman (DH) key files. For example: "foo.h:2048,bar.h:1024" means:
dh_param_files[0].file = foo.pem,
tls.cipher.dh_param_files

dh_param_files[0].bits = 2048,
dh_param_files[1].file = bar.pem,
dh_param_files[1].bits = 1024

tls.cipher.engine_id

String ID of OpenSSL cipher engine to request.
Required if you want to use security.

tls.identity.certificate_chain_file

A string that specifies the name of a file containing an identifying certificate chain (in PEM format).
An identifying certificate is required for secure communication. The file must be sorted starting with
the certificate to the highest level (root CA). If no private key is specified, this file will be used to
load a non-RSA private key.

tls.identity.private_key_password

A string that specifies the password for private key.

tls.identity.private_key_file

A string that specifies that name of a file containing private key (in PEM format). If no private key is
specified (all values are NULL), this value will default to the same file as the specified certificate
chain file.

tls.identity.rsa_private_key_file

A string that specifies that name of a file containing an RSA private key (in PEM format).
Required

transport_instance_id[0] to
[NDDS_TRANSPORT_
WAN_TRANSPORT_
INSTANCE_ID_LENGTHb]

A set of comma-separated values to specify the elements of the array. This value must be unique for
all transport instances communicating with the same WAN Rendezvous Server.

interface_address

Locator, as a string

If less than the full array is specified, it will be right-aligned. For example, the string "01,02" results
in the array being set to:
{0,0,0,0,0,0,0,0,0,0,1,2}

Required
server
Server locator, as a string.

1 Assuming you used ‘dds.transport.WAN.wan1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'
bNDDS_TRANSPORT_WAN_TRANSPORT_INSTANCE_ID_LENGTH = 12

924

25.4 Secure Transport Properties

Table 25.1 Properties for NDDS_Transport_WAN_Property_t
Property Name
(prefix with
‘dds.transport.WAN.wan1.’)

Property Value Description

1

server_port

Server port number.

stun_retransmission_interval

STUN request messages requiring a response are resent with this interval. The interval is doubled
after each retransmission. Specified in msec.

stun_number_of_retransmissions

Maximum number of times STUN messages are resent unless a response is received.

stun_liveliness_period

Period at which messages are sent to peers to keep NAT holes open; and to the WAN server to
refresh bound ports. Specified in msec.

25.4 Secure Transport Properties
Table 25.2 Properties for NDDS_Transport_DTLS_Property_t lists the properties that you can set for the
Secure Transport.
Table 25.2 Properties for NDDS_Transport_DTLS_Property_t
Property Name
(prefix with
Property Value Description
‘dds.transport.DTLS.dtls1’)
b

Required
Comma-separated strings indicating the prefix names of all plugins that will be loaded by Connext
DDS. You will use this string as the prefix to the property names.
dds.transport.load_plugins
(note: this does not take a prefix)

For example: “dds.transport.DTLS.dtls1". (This assumes you used used ‘dds.transport.DTLS.dtls1’
as the alias to load the plugin. If not, change the prefix to match the string used with
dds.transport.load_plugins.)
This prefix must begin with 'dds.transport.'
Note: you can load up to 8 plugins.

1 Assuming you used ‘dds.transport.WAN.wan1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'
b Assuming you used ‘dds.transport.DTLS.dtls1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

925

25.4 Secure Transport Properties

Table 25.2 Properties for NDDS_Transport_DTLS_Property_t
Property Name
(prefix with
Property Value Description
‘dds.transport.DTLS.dtls1’)
a

Only required if linking dynamically
Must set to "libnddstransporttls.so" (for UNIX/Solaris) or "nddstransporttls.dll" (for Windows).
library

This library and the dependent OpenSSL libraries must be in your library search path (pointed to by
the environment variable LD_LIBRARY_PATH on UNIX/Solaris systems, Path on Windows
systems, LIBPATH on AIX systems, DYLD_LIBRARY_PATH on Mac OS systems).
Only required if linking dynamically

create_function
Must be "NDDS_Transport_DTLS_create"
Only required if linking statically
create_function_ptr

Defines the function pointer to the DTLS Transport Plugin creation function. Used for loading the
DTLS Transport plugin statically.
Must be set to the NDDS_Transport_DTLS_create function pointer.

aliases

Used to register the transport plugin returned by NDDS_Transport_DTLS_create() (as specified by
.create_function) to the DomainParticipant. Aliases should be specified as commaseparated strings, with each comma delimiting an alias. If it is not specified, the prefix is used as the
default alias for the plugin.
The network address at which to register this transport plugin.

network_address

The least significant transport_in.property.address_bit_count will be truncated. The remaining bits
are the network address of the transport plugin.
This value overwrites the value returned by the output parameter in NDDS_Transport_create_plugin
function as specified in ".create_function".
Specifies the verbosity of log messages from the transport.

verbosity

parent.address_bit_count

Possible values:
-1: silent
0 (default): errors only
1: errors and warnings
2: local status
5 or higher: all messages
Number of bits in a 16-byte address that are used by the transport. Should be between 0 and 128.
For example, for an address range of 0-255, the address_bit_count should be set to 8.

a Assuming you used ‘dds.transport.DTLS.dtls1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

926

25.4 Secure Transport Properties

Table 25.2 Properties for NDDS_Transport_DTLS_Property_t
Property Name
(prefix with
Property Value Description
‘dds.transport.DTLS.dtls1’)
a

parent.properties_bitmap

A bitmap that defines various properties of the transport to the Connext DDS core. Currently, the
only property supported is whether or not the transport plugin will always loan a buffer when
Connext DDS tries to receive a message using the plugin. This is in support of a zero-copy interface.

parent.gather_send_buffer_
count_max

Specifies the maximum number of buffers that Connext DDS can pass to the transport plugin’s send
() function.

parent.message_size_max

The maximum size of a message in bytes that can be sent or received by the transport plugin.
Maximum value: 16384.
A list of strings, each identifying a range of interface addresses.
Interfaces must be specified as comma-separated strings, with each comma delimiting an interface.

parent.allow_interfaces

If the list is non-empty, this "white" list is applied before the parent.deny_interfaces (Section below)
list.
You must manage the memory of the list. The memory may be freed after the DomainParticipant is
enabled.
A list of strings, each identifying a range of interface addresses.
Interfaces should be specified as comma-separated strings, with each comma delimiting an interface.

parent.deny_interfaces

This "black" list is applied after the parent.allow_interfaces (Section above) list and filters out the
interfaces that should not be used.
You must manage the memory of the list. The memory may be freed after the DomainParticipant is
enabled.

send_socket_buffer_size

Size in bytes of the send buffer of a socket used for sending.

recv_socket_buffer_size

Size in bytes of the receive buffer of a socket used for sending.

ignore_loopback_interface

Prevents the Transport Plugin from using the IP loopback interface.

a Assuming you used ‘dds.transport.DTLS.dtls1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

927

25.4 Secure Transport Properties

Table 25.2 Properties for NDDS_Transport_DTLS_Property_t
Property Name
(prefix with
Property Value Description
‘dds.transport.DTLS.dtls1’)
a

Prevents the transport plugin from using a network interface that is not reported as RUNNING by
the operating system.

ignore_nonrunning_interfaces

The transport checks the flags reported by the operating system for each network interface upon
initialization. An interface which is not reported as UP will not be used. This property allows the
same check to be extended to the IFF_RUNNING flag implemented by some operating systems.
The RUNNING flag is defined to mean that "all resources are allocated", and may be off if there is
no link detected, e.g., the network cable is unplugged.
Two values are allowed:
0: Do not check the RUNNING flag when enumerating interfaces, just make sure the interface is
UP.
1: Check the flag when enumerating interfaces, and ignore those that are not reported as RUNNING.
This can be used on some operating systems to cause the transport to ignore interfaces that are
enabled but not connected to the network.

transport_priority_mask

Mask for use of transport priority field.

transport_priority_mapping_low
Low and high values of output range to IPv4 TOS.
transport_priority_mapping_high

recv_decode_buffer_size

Size of buffer for decoding packets from wire. An extra buffer is required for storage of encrypted
data. The minimum value for this property is parent.message_size_max (Section on the previous
page).

port_offset

Port offset to allow coexistence with non-secure UDP transport.

dtls_handshake_resend_interval

DTLS handshake retransmission interval in milliseconds
Liveliness interval (multiple of resend interval)

dtls_connection_liveliness_
interval

The connection will be dropped if no message from the peer is received in this amount of time. This
enables cleaning up state for peers that are no longer responding. A secure keep-alive message will
be sent every half-interval if no other sends have occurred for a given DTLS connection during that
time.
Default: 60 ms

a Assuming you used ‘dds.transport.DTLS.dtls1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

928

25.4 Secure Transport Properties

Table 25.2 Properties for NDDS_Transport_DTLS_Property_t
Property Name
(prefix with
Property Value Description
‘dds.transport.DTLS.dtls1’)
a

A string that specifies the name of file containing Certificate Authority certificates. File should be in
PEM format. See the OpenSSL manual page for SSL_load_verify_locations for more information.
tls.verify.ca_file
tls.verify.ca_file (Section above) or tls.verify.ca_path (Section below) must be specified; both may
be specified.

tls.verify.ca_path

A string that specifies paths to directories containing Certificate Authority certificates. Files should
be in PEM format, and follow the OpenSSL-required naming conventions. See the OpenSSL
manual page for SSL_CTX_load_verify_locations for more information.
tls.verify.ca_file (Section above) or tls.verify.ca_path (Section above) must be specified; both may be
specified.

tls.verify.verify_depth

Maximum certificate chain length for verification.

tls.verify.verify_peer

If non-zero, use mutual authentication when performing TLS hand- shake (default). If zero, only the
reader side will present a certificate, which will be verified by the writer side.
This can be set to one of three values:
"default" selects NDDS_Transport_TLS_default_verify_callback()

tls.verify.callback
"verbose" selects NDDS_Transport_TLS_verbose_verify_callback()
"none" requests no callback be registered
tls.cipher.cipher_list

List of available (D)TLS ciphers. See the OpenSSL manual page for SSL_set_cipher_list for more
information on the format of this string.
List of available Diffie-Hellman (DH) key files. For example: "foo.h:2048,bar.h:1024" means:
dh_param_files[0].file = foo.pem,

tls.cipher.dh_param_files

dh_param_files[0].bits = 2048,
dh_param_files[1].file = bar.pem,
dh_param_files[1].bits = 1024

tls.cipher.engine_id

String ID of OpenSSL cipher engine to request.

a Assuming you used ‘dds.transport.DTLS.dtls1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

929

25.5 Explicitly Instantiating a WAN or Secure Transport Plugin

Table 25.2 Properties for NDDS_Transport_DTLS_Property_t
Property Name
(prefix with
Property Value Description
‘dds.transport.DTLS.dtls1’)
a

tls.identity.certificate_chain_file

RequiredA string that specifies the name of a file containing an identifying certificate chain (in PEM
format). An identifying certificate is required for secure communication. The file must be sorted
starting with the certificate to the highest level (root CA). If no private key is specified, this file will
be used to load a non-RSA private key.

tls.identity.private_key_password

A string that specifies the password for private key.

tls.identity.private_key_file

A string that specifies that name of a file containing private key (in PEM format). If no private key is
specified (all values are NULL), this value will default to the same file as the specified certificate
chain file.

tls.identity.rsa_private_key_file

A string that specifies that name of a file containing an RSA private key (in PEM format).

25.5 Explicitly Instantiating a WAN or Secure Transport Plugin
As described on Page 914, there are two ways to instantiate a transport plugin. This section describes the
mechanism that includes calling NDDSTransportSupport::register_transport(). (The other way is to
use the Property QoS mechanism, described in Setting Up a Transport with the Property QoS (Section
25.2 on page 915)).
Notes:
l

l

This way of instantiating a transport is not supported in the Java API. If you are using Java, use the
Property QoS mechanism, described in Setting Up a Transport with the Property QoS (Section 25.2
on page 915).
To use this mechanism, there are extra libraries that you must link into your program and an
additional header file that you must include. Please see the Additional Header Files and Include
Directories (Section 25.5.1 on the next page) and Additional Libraries (Section 25.5.2 on the next
page) for details.

To instantiate a WAN or Secure Transport prior to explicitly registering it with NDDSTransportSupport::register_transport(), use one of the following functions:
NDDS_Transport_Plugin* NDDS_Transport_WAN_new (
const struct NDDS_Transport_WAN_Property_t * property_in)

a Assuming you used ‘dds.transport.DTLS.dtls1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

930

25.5.1 Additional Header Files and Include Directories

NDDS_Transport_Plugin* NDDS_Transport_DTLS_new (
const struct NDDS_Transport_DTLS_Property_t * property_in)

See the API Reference HTML documentation for details on these functions.

25.5.1 Additional Header Files and Include Directories
l

To use the Secure WAN Transport API, you must include an extra header file (in addition to those
in Table 9.1 Header Files to Include for Connext DDS (All Architectures)).
#include "ndds/ndds_transport_secure_wan.h"

Assuming that Secure WAN Transport is installed in the same directory as Connext DDS (see
Table 9.2 Include Paths for Compilation (All Architectures)), no additional include paths need to be
added for the Secure WAN Transport API. If this is not the case, you will need to specify the appropriate include path.
l

If you want to access OpenSSL data structures, add the OpenSSL include directory, //include, and include the OpenSSL headers before ndds_transport_secure_
wan.h:
#include 
#include  (if accessing certificate functions)
etc.

On Windows systems, if you are loading statically: you should also include the OpenSSL file,
applink.c, in your application. It can be found in the OpenSSL include directory, or included as
.

25.5.2 Additional Libraries
To use the Secure WAN Transport API, you must link in additional libraries, which are listed in the RTI
Connext DDS Core Libraries Platform Notes (in the appropriate section for your architecture). Refer to
Required Libraries (Section 9.3.1 on page 625) for the differences between shared and static libraries.

25.5.3 Compiler Flags
No additional compiler flags are required.

931

Part 6: RTI Persistence Service
Persistence Service is only available with the Connext DDS Professional, Basic, and
Evaluation packages types.
The material in this part of the manual describes Persistence Service. It saves DDS data samples so
they can be delivered to subscribing applications that join the system at a later time—even if the
publishing application has already terminated.
This section includes:
l

Introduction to RTI Persistence Service (Section Chapter 26 on page 933)

l

Configuring Persistence Service (Section Chapter 27 on page 934)

l

Running RTI Persistence Service (Section Chapter 28 on page 962)

l

l

Administering Persistence Service from a Remote Location (Section Chapter 29 on page
966)
Advanced Persistence Service Scenarios (Section Chapter 30 on page 972)

932

Chapter 26 Introduction to RTI Persistence
Service
Persistence Service is a Connext DDS application that saves DDS data samples to transient or permanent storage, so they can be delivered to subscribing applications that join the system at a later
time—even if the publishing application has already terminated.
Persistence Service runs as a separate application; you can run it on the same node as the publishing application, the subscribing application, or some other node in the network.
When configured to run in PERSISTENT mode, Persistence Service can use the filesystem or a
relational database that provides an ODBC driver. For each persistent topic, it collects all the data
written by the corresponding persistent DataWriters and stores them into persistent storage. See the
RTI Persistence Service Release Notes for the list of platforms and relational databases that have
been tested.
When configured to run in TRANSIENT mode, Persistence Service stores the data in memory.
The following chapters assume you have a basic understanding of DDS terms such as DomainParticipants, Publishers, DataWriters, Topics, and Quality of Service (QoS) policies. For an overview
of DDS terms, please see Data-Centric Publish-Subscribe Communications (Section Chapter 2 on
page 10). You should also have already read Mechanisms for Achieving Information Durability
and Persistence (Section Chapter 12 on page 675).

933

Chapter 27 Configuring Persistence
Service
To use Persistence Service:
1. Modify your Connext DDS applications.
l

The DURABILITY QosPolicy (Section 6.5.7 on page 368) controls whether or not,
and how, published DDS samples are stored by Persistence Service for delivery to
late-joining DataReaders. See Data Durability (Section 12.5 on page 692).
l

l

l

For each DataWriter whose data must be stored, set the Durability QosPolicy’s
kind to DDS_PERSISTENT_DURABILITY_QOS or DDS_TRANSIENT_
DURABILITY_QOS.
For each DataReader that needs to receive stored data, set the Durability
QosPolicy’s kind to DDS_PERSISTENT_DURABILITY_QOS or DDS_
TRANSIENT_DURABILITY_QOS.

Optionally, modify the DURABILITY SERVICE QosPolicy (Section 6.5.8 on page
372), which can be used to configure Persistence Service.

By default, the History and ResourceLimits QosPolicies for a Persistence Service
DataReader (PRSTDataReader) and Persistence Service DataWriter (PRSTDataWriter)
with topic 'A' will be configured using the values specified in the XML file (unless you use
the tag  in the persistence group definition, see Creating Persistence
Groups (Section 27.8 on page 947)). Setting the  tag to true will
cause the History and ResourceLimits QosPolicies for a PRSTDataReader and
PRSTDataWriter to be configured using the DURABILITY SERVICE QosPolicy (Section
6.5.8 on page 372) of the first-discovered DataWriter publishing 'A'. (For more information
on the PRSTDataReader and PRSTDataWriter, see RTI Persistence Service (Section 12.5.1
on page 692).)

934

27.1 How to Load the Persistence Service XML Configuration

2. Create a configuration file or edit an existing file, as described in XML Configuration File (Section
27.2 on the facing page).
3. Start Persistence Service with your configuration file, as described in Starting Persistence Service
(Section 28.1 on page 962). You can start it on either application’s node, or even an entirely different node (provided that node is included in one of the applications’ NDDS_DISCOVERY_
PEERS lists).

27.1 How to Load the Persistence Service XML Configuration
Persistence Service loads its XML configuration from multiple locations. This section presents the various
approaches, listed in load order.
The first three locations only contain QoS Profiles and are inherited from Connext DDS (see Configuring
QoS with XML (Section Chapter 17 on page 791)).
l

$NDDSHOME/resource/xml/NDDS_QOS_PROFILES.xml
This file contains the DDS default QoS values; it is loaded automatically if it exists. (First to be
loaded.)

l

File specified in the NDDS_QOS_PROFILES Environment Variable
The files (or XML strings) separated by semicolons referenced in this environment variable are
loaded automatically.

l

/USER_QOS_PROFILES.xml
This file is loaded automatically if it exists.

The next locations are specific to Persistence Service.
l

/resource/xml/RTI_PERSISTENCE_SERVICE.xml

l

This file contains the default Persistence Service configurations; it is loaded if it exists. There are
two default configurations: default and defaultDisk. The default configuration persists all the topics into memory. The defaultDisk configuration persists all the topics into files located in the current
working directory.
/USER_PERSISTENCE_SERVICE.xml

l

This file is loaded automatically if it exists.
File specified using the command line option, -cfgFile
The command-line option -cfgFile (see Table 28.1 Persistence Service Command-Line Options) can
be used to specify a configuration file.

935

27.2 XML Configuration File

27.2 XML Configuration File
The configuration file uses XML format. Let's look at a very basic configuration file, just to get an idea of
its contents. You will learn the meaning of each line as you read the rest of this section:
l

QoS Configuration (Section 27.3 on page 939)

l

Configuring the Persistence Service Application (Section 27.4 on page 940)

l

Configuring Remote Administration (Section 27.5 on page 942)

l

Configuring Persistent Storage (Section 27.6 on page 943)

l

Configuring Participants (Section 27.7 on page 946)

l

Creating Persistence Groups (Section 27.8 on page 947)

l

Enabling Distributed Logger in RTI Services (Section Chapter 39 on page 1049)

l

Enabling RTI Monitoring Library in Persistence Service (Section 27.12 on page 958)

Example Configuration File








DDS_KEEP_ALL_HISTORY_QOS




DDS_RELIABLE_RELIABILITY_QOS


DDS_KEEP_ALL_HISTORY_QOS








72

true


936

27.2.1 Configuration File Syntax





/tmp
PS




71


true
true







27.2.1 Configuration File Syntax
The configuration file must follow these syntax rules:
l

The syntax is XML and the character encoding is UTF-8.

l

Opening tags are enclosed in <>; closing tags are enclosed in .

l

A value is a UTF-8 encoded string. Legal values are alphanumeric characters. All leading and trailing spaces are removed from the string before it is processed.

l

For example, "  value " is the same as "value".
All values are case-sensitive unless otherwise stated.

l

Comments are enclosed as follows: .

l

The root tag of the configuration file must be  and end with .

l

The primitive types for tag values are specified in Table 27.1 Supported Tag Values.

Table 27.1 Supported Tag Values
Type

Format

Notes

yes, 1, true, BOOLEAN_TRUE or DDS_BOOLEAN_TRUE: these all
mean TRUE
DDS_Boolean

Not case-sensitive
no, 0, false, BOOLEAN_FALSE or DDS_BOOLEAN_FALSE: these
all mean FALSE

937

27.2.2 XML Validation

Table 27.1 Supported Tag Values
Type
DDS_Enum

Format
A string. Legal values are those listed in the C or Java API Reference
HTML documentation.

Notes
Must be specified as a string. (Do not use
numeric values.)

-2147483648 to 2147483647
or 0x80000000 to 0x7fffffff
DDS_Long

A 32-bit signed integer
or LENGTH_UNLIMITED
or DDS_LENGTH_UNLIMITED

0 to 4294967296
DDS_
or
UnsignedLong
0 to 0xffffffff

A 32-bit unsigned integer

String

All leading and trailing spaces are ignored
between two tags

UTF-8 character string

27.2.2 XML Validation
27.2.2.1 Validation at Run Time
Persistence Service validates the input XML files using a builtin Document Type Definition (DTD). You
can find a copy of the builtin DTD in a/resource/schema/rti_persistence_service.dtd.
(This is only a copy of what the Persistence Service core uses. Changing this file has no effect unless you
specify its path with the DOCTYPE tag, described below.)
You can overwrite the builtin DTD by using the XML tag, . For example, the following
indicates that Persistence Service must use a different DTD file to perform validation:


If you do not specify the DOCTYPE tag in the XML file, the builtin DTD is used.
The DTD path can be absolute, or relative to the application's current working directory.
27.2.2.2 Validation During Editing
Persistence Service provides DTD and XSD files that describe the format of the XML content. We recommend including a reference to one of these documents in the XML file that contains the persistence service’s configuration—this provides helpful features in code editors such as Visual Studio and Eclipse,

aSee Paths Mentioned in Documentation (Section on page xxxviii).

938

27.3 QoS Configuration

including validation and auto-completion while you are editing the XML file. Including a reference to the
XSD file in the XML documents provides stricter validation and better auto-completion than the corresponding DTD file.
The DTD and XSD definitions of the XML elements are in
/resource/schema (rti_persistence_service.dtd and
rti_persistence_service.xsd, respectively).
To include a reference to the XSD document in your XML file, use the attribute xsi:noNamespaceSchemaLocation in the  tag. For example (in the following, replace  with the Connext DDS installation directory, see Paths Mentioned in Documentation (Section on page xxxviii)):


...


To include a reference to the DTD document in your XML file, use the  tag. For example
(in the following, replace  with the Connext DDS installation directory):

/resource/schema/rti_persistence_service.dtd">

...


27.3 QoS Configuration
Each persistence group and participant has a set of DDS QoSs. There are six tags:
l



l



l



l



l



l



Each QoS is identified by a name. The QoS can inherit its values from other QoSs described in the XML
file. For example:


DDS_KEEP_ALL_HISTORY_QOS

939

27.4 Configuring the Persistence Service Application




In the above example, the writer QoS named 'DerivedWriterQos' inherits the values from the writer QoS
'BaseWriterQos' contained in the library 'Lib'. The HistoryQosPolicy kind is set to DDS_KEEP_ALL_
HISTORY_QOS.
Each XML tag with an associated name can be uniquely identified by its fully qualified name in C++
style. For more information on tags, see Configuring QoS with XML (Section Chapter 17 on page 791)
The persistence groups and participants can use QoS libraries and profiles to configure their QoS values.
For example:






DDS_KEEP_ALL_HISTORY_QOS






...


true
true





For more information about QoS libraries and profiles see Configuring QoS with XML (Section Chapter
17 on page 791).

27.4 Configuring the Persistence Service Application
Each execution of the Persistence Service application is configured using the content of a tag: . When you start Persistence Service (described in Starting Persistence Service (Section
28.1 on page 962)), you must specify which  tag to use to configure the service.
For example:


...



940

27.4 Configuring the Persistence Service Application

If you do not specify a service name when you start Persistence Service, the service will print the list of
available configurations and then exit.
Because a configuration file may contain multiple  tags, one file can be used to configure multiple Persistence Service executions.
Table 27.2 Persistence Service Application Tags lists the tags you can specify for a persistence service.
Notice that  is required. For default values, please see the API Reference HTML documentation.
Table 27.2 Persistence Service Application Tags
Tags within



Description

Enables and configures remote administration. See Configuring Remote Administration (Section 27.5 on
the facing page).

Number
of Tags
Allowed
0 or 1

Provides a description for the persistence service configuration.
Example:




Persists in the file system all topics
published with PERSISTENT durability



0 or 1

A DDS_Boolean that indicates whether or not a PRSTDataWriter will purge a DDS sample from its cache

See Configuring Durable Subscriptions in Persistence Service (Section 27.9 on page 955).
For each  tag, Persistence Service creates two DomainParticipants on the same domain ID:
one to subscribe to changes and one to publish changes. There may be more Participant pairs created when
there are multiple versions of a type (see Support for Extensible Types (Section 27.13 on page 959)).
The QoS values used to configure both DomainParticipants are the same, except for:
l


l

The participant_id in the WIRE_PROTOCOL QosPolicy (DDS Extension) (Section 8.5.9 on page 610)).
If participant_id is not -1 (the default value, which means automatic selection),Persistence Service uses participant_id for the first DomainParticipant and
participant_id+1 for the second DomainParticipant.

1 or more
(required)

The TCP server ports are configured with the properties dds.transport.tcp.server_bind_port and
dds.transport.tcp.public_address. See TCP/TLS Transport Properties (Section 35.1.6 on page 1002).

941

27.5 Configuring Remote Administration

Table 27.2 Persistence Service Application Tags
Tags within



Number
of Tags
Allowed

Description

When this tag is present, the topic data will be persisted to disk. You can select between file storage and
relational database storage. See Configuring Persistent Storage (Section 27.6 on the next page).

0 or 1

Enables synchronization in redundant persistence service instances.
 See Synchronizing of Persistence Service Instances (Section 27.10 on page 956).

0 or 1

Default: Synchronization is not enabled

27.5 Configuring Remote Administration
You can create a Connext DDS application that can remotely control Persistence Service. The  tag is used to enable remote administration and configure its behavior.
By default, remote administration is turned off in Persistence Service.
When remote administration is enabled, Persistence Service will create a DomainParticipant, Publisher,
Subscriber, DataWriter, and DataReader. These Entities are used to receive commands and send
responses. You can configure these entities with QoS tags within the  tag.
Table 27.3 Remote Administration Tags lists the tags allowed within  tag. Notice that
the  tag is required.
For more details, please see Administering Persistence Service from a Remote Location (Section Chapter
29 on page 966).
Note: The command-line options used to configure remote administration take precedence over the XML
configuration (see Table 28.1 Persistence Service Command-Line Options).
Table 27.3 Remote Administration Tags
Tags within


Description

Number of Tags
Allowed

Configures the DataReader QoS for remote administration.
If the tag is not defined, Persistence Service will use the DDS defaults with the
following changes:


reliability.kind = DDS_RELIABLE_RELIABILITY_QOS (this value cannot
be changed)

0 or 1

history.kind = DDS_KEEP_ALL_HISTORY_QOS
resource_limits.max_samples = 32

942

27.6 Configuring Persistent Storage

Table 27.3 Remote Administration Tags
Tags within


Description

Number of Tags
Allowed

Configures the DataWriter QoS for remote administration.



If the tag is not defined, Persistence Servicewill use the DDS defaults with the
following changes:

0 or 1

history.kind = DDS_KEEP_ALL_HISTORY_QOS
resource_limits.max_samples = 32
Configures RTI Distributed Logger.


0 or 1
See



Specifies which domain ID Persistence Service will use to enable remote
administration.

1 (required)

Configures the DomainParticipant QoS for remote administration.


0 or 1
If the tag is not defined, Persistence Service will use the DDS defaults.
Configures the Publisher QoS for remote administration.



0 or 1
If the tag is not defined, Persistence Service will use the DDS defaults.
Configures the Subscriber QoS for remote administration.



0 or 1
If the tag is not defined, Persistence Service will use the DDS defaults.

27.6 Configuring Persistent Storage
The  tag is used to persist DDS samples into permanent storage. If the  tag is not specified, the service will operate in TRANSIENT mode and all the data will
be kept in memory. Otherwise, the persistence service will operate in PERSISTENT mode and all the
topic data will be stored into the filesystem or into a relational database that provides an ODBC driver.
Table 27.4 Persistent Storage tags lists the tags that you can specify in .
Relational Database Limitations: The ODBC storage does not support BLOBs. The maximum size for
a serialized DDS sample is 65535 bytes in MySQL.

943

27.6 Configuring Persistent Storage

Table 27.4 Persistent Storage tags
Tags within


Description

Number
of Tags
Allowed

When this tag is present, the topic data will be persisted in a relational database.


This tag is required if  is not specified.

0 or 1

See Table 27.5 External Database Tags.
When this tag is present, the topic data will be persisted into files.



This tag is required if  is not specified.

0 or 1

See Table 27.6 Filesystem tags.



This DDS_Boolean (see Table 27.1 Supported Tag Values) indicates if the topic data associated with a
previous execution of the persistence service must be restored or not. If the topic data is not restored, it
will be deleted from the persistent storage.

0 or 1

Default: 1



Defines the length in bytes of the database column used to store the TypeObjects associated with
PRSTDataWriters and PRSTDataReader.
For additional information on TypeObjects, see the RTI Connext DDS Core Libraries Getting Started
Guide Addendum for Extensible Types.

0 or 1

Default: 10488576

Table 27.5 External Database Tags
Tags within




Description

DSN used to connect to the database using ODBC. You should create this DSN through the ODBC
settings on Windows systems, or in your .odbc.ini file on UNIX/Linux systems.

Number of
Tags
Allowed

1 (required)

This tag is required.



Specifies the ODBC driver to load. By default, Connext DDS will try to use the standard ODBC
driver manager library (UnixOdbc on UNIX/Linux systems, the Windows ODBC driver manager on
Windows systems).

0 or 1

Password to connect to the database.


0 or 1
Default: no username is used
Username to connect to the database.



0 or 1
Default: no username is used

944

27.6 Configuring Persistent Storage

Table 27.6 Filesystem tags
Tags within


Description

Number of
Tags
Allowed

Specifies the directory of the files in which topic data will be persisted. There will be one file per
PRSTDataWriter/PRSTDataReader pair.


The directory must exist; otherwise the service will report an error upon start up.

0 or 1

Default: current working directory
A name prefix associated with all the files created by Persistence Service.


0 or 1
Default: PS
Sets the journal mode of the persistent storage. This tag can take these values:
l

DELETE: Deletes the rollback journal at the conclusion of each transaction.
l

TRUNCATE: Commits transactions by truncating the rollback journal to zero-length instead
of deleting it.
l

PERSIST: Prevents the rollback journal from being deleted at the end of each transaction.
Instead, the header of the journal is overwritten with zeros.


0 or 1

l

MEMORY: Stores the rollback journal in volatile RAM. This saves disk I/O.
l

WAL: Uses a write-ahead log instead of a rollback journal to implement transactions.
l

OFF: Completely disables the rollback journal. If the application crashes in the middle of a
transaction when the OFF journaling mode is set, the files containing the DDS samples will
very likely be corrupted.
Default: DELETE
Determines the level of synchronization with the physical disk.
This tag can take three values:
l

FULL: Every DDS sample is written into physical disk as Persistence Service receives it.


l

NORMAL: DDS samples are written into disk at critical moments.

0 or 1

l

OFF: No synchronization is enforced. Data will be written to physical disk when the OS
flushes its buffers.
Default: OFF

945

27.7 Configuring Participants

Table 27.6 Filesystem tags
Tags within




Description

Specifies the name of the trace file for debugging purposes. The trace file contains information about
all SQL statements executed by the persistence service.

Number of
Tags
Allowed

0 or 1

Default: No trace file is generated
Sets the auto-vacuum status of the storage. This tag can take these values:
l

NONE: When data is deleted from the storage files, the files remain the same size.


0 or 1

l

FULL: The storage files are compacted every transaction.
Default: FULL

27.7 Configuring Participants
An XML  tag will contain a set of . The persistence service will persist topics published in the domainIDs associated with these participants. For example:


71
...


72
...



Using the above example, the persistence service will create two pairs of DomainParticipants on
DDS domains 71 and 72, respectively. In each pair, one DomainParticipant is used to receive data and the
other to publish.
After the DomainParticipants are created, the persistence service will monitor the discovery traffic, looking for topics to persist.
Notice that in some cases there may be more than one pair of DomainParticipants per domain when there
are multiple versions of a type for a given topic. (See Support for Extensible Types (Section 27.13 on page
959).)
The  tag can be specified alternatively as an attribute of . For example:



946

27.8 Creating Persistence Groups

...



Table 27.7 Participant Tagsdescribes the participant tags. Notice that  is required.
Table 27.7 Participant Tags
Tags within




Description

Domain ID associated with the Participant. The domain ID can be specified as an attribute of the
participant tag.

Number of
Tags
Allowed

0 or 1

Default: 0
Configures a set of Durable Subscriptions for a given topic. This is a sequence of  tags,
each of which has a , a , and a . For example:





DurSub1
Example MyType
2


DurSub2
Example MyType



0 or 1

Default: Empty list
See Configuring Durable Subscriptions in Persistence Service (Section 27.9 on page 955) for
additional information


Participant QoS.



A persistence group describes a set of topics whose data that must be persisted by the persistence
service.

0 or 1
Default: DDS defaults
1 or more
(required)

27.8 Creating Persistence Groups
The topics that must be persisted in a specific domain ID are specified using  tags. A
 tag defines a set of topics identified by a POSIX expression.
For example:

71


947

27.8 Creating Persistence Groups

...



In the above example, the persistence group 'PerGroup1' is associated with all the topics published in DDS
domain 71 whose name starts with 'H'.
When a participant discovers a topic that matches a persistence group, it will create a PRSTDataReader
and a PRSTDataWriter. The PRSTDataReader and PRSTDataWriter will be configured using the QoS
policies associated with the persistence group. The DDS samples received by the PRSTDataReader will
be persisted in the queue of the corresponding PRSTDataWriter.
A  tag can contain multiple persistence groups; the set of topics that each one represents can
intersect.
Table 27.8 Persistence Group Tags further describes the persistence group tags. For default values, please
see the API Reference HTML documentation.
Table 27.8 Persistence Group Tags
Tags within


Number
of Tags
Allowed

A DDS_Boolean (see Table 27.1 Supported Tag Values) that enables support for durable subscriptions in
the PRSTDataWriters created in a persistence group.


When Durable Subscriptions are not required, setting this property to 0 will increase performance.

0 or 1

Default: 1
Content filter topic expression. A persistence group can subscribe to a specific set of data based on the value
of this expression.


A filter expression is similar to the WHERE clause in SQL. For more information on the syntax, please see
0 or 1
the API Reference Documentation (from the Modules page, select RTI Connext DDS DDS API Reference,
Queries and Filters Syntax).
Default: no expression




PRSTDataReader QoS1. See QoSs (Section 27.8.1 on page 952).
0 or 1
Default: DDS defaults
PRSTDataWriter QoS2. See QoSs (Section 27.8.1 on page 952).
0 or 1
Default: DDS defaults

1These fields cannot be set and are assigned automatically: protocol.virtual_guid, protocol.rtps_object_id, durability.kind.
2These fields cannot be set and are assigned automatically: protocol.virtual_guid, protocol.rtps_object_id, durability.kind.

948

27.8 Creating Persistence Groups

Table 27.8 Persistence Group Tags
Tags within


Number
of Tags
Allowed

Specifies a list of POSIX expressions separated by commas that describe the set of topics to be denied in
the persistence group.


This "black" list is applied to the topics that pass the filter specified with the  tag

0 or 1

Default: *
Specifies a list of POSIX expressions separated by commas that describe the set of topics associated with
the persistence group.


The filter can be specified as an attribute of  as well.

0 or 1

Default: *



This flag configures the memory allocation policy for DDS samples in PRSTDataReaders and
PRSTDataWriters.

0 or 1

See Memory Management (Section 27.8.5 on page 954).



A DDS_Boolean (see Table 27.1 Supported Tag Values) that controls whether or not the persistence
service propagates dispose messages from DataWriters to DataReaders.

0 or 1

Default: 1



A DDS_Boolean (see Table 27.1 Supported Tag Values). When this tag is 1, the DDS data samples sent by
the PRSTDataWriters preserve the source timestamp that was associated with them when they were
0 or 1
published by the original DataWriter.
Default: 0



A DDS_Boolean (see Table 27.1 Supported Tag Values) that controls whether or not the persistence
service propagates unregister messages from DataWriters to DataReaders.

0 or 1

Default: 0
Publisher QoS. See QoSs (Section 27.8.1 on page 952).


0 or 1
Default: DDS defaults
This property controls how often (expressed as a number of DDS samples) the PRSTDataReader state is
stored in the database. The PRSTDataReaders are the DataReaders created by the persistence service.



A high frequency will provide better performance. However, if the persistence service is restarted, it may
receive some duplicate DDS samples. The persistence service will send these duplicates DDS samples on
the wire but they will be filtered by the DataReaders and they will not be propagated to the application.

0 or 1

This property is only applicable when the persistence service operates in persistent mode (the  tag is present).
Default: 1

949

27.8 Creating Persistence Groups

Table 27.8 Persistence Group Tags
Tags within


Number
of Tags
Allowed

A DDS_Boolean (see Table 27.1 Supported Tag Values) that indicates if the persistence service will create
an independent database connection per PRSTDataWriter in the group (0) or per Publisher (1) in the group.
When  is 0 and  is 1, there is a single database connection
per group. All the PRSTDataWriters will share the same connection.


When  is 1 or  is 0, there is a database connection per
PRSTDataWriter.

0 or 1

This parameter is only applicable to configurations persisting the data into a relational database using the tag
 in .
See Sharing a Database Connection (Section 27.8.4 on page 954)
Default: 0



A DDS_Boolean (see Table 27.1 Supported Tag Values) that indicates if the persistence service should
create one Publisher per persistence group or one Publisher per PRSTDataWriter inside the persistence
group. See Sharing a Publisher/Subscriber (Section 27.8.3 on page 953).

0 or 1

Default: 1



A DDS_Boolean (see Table 27.1 Supported Tag Values) that indicates if the persistence service should
create one Subscriber per persistence group or one Subscriber per PRSTDataReader in the persistence
group.

0 or 1

See Sharing a Publisher/Subscriber (Section 27.8.3 on page 953).
Default: 1



Subscriber QoS. See QoSs (Section 27.8.1 on page 952).
0 or 1
Default: DDS defaults
Topic QoS. See QoSs (Section 27.8.1 on page 952).



0 or 1
Default: DDS defaults

A DDS_Boolean (see Table 27.1 Supported Tag Values) that indicates if the HISTORY and RESOURCE_
LIMITS QoS policy of the PRSTDataWriters and PRSTDataReaders should be configured based on the

See DurabilityService QoS Policy (Section 27.8.2 on page 953)
Default: 0



Controls how often (expressed in milliseconds) DDS samples are marked as ACK'd in the database by the
PRSTDataWriter.

0 or 1

Default: 0

950

27.8 Creating Persistence Groups

Table 27.8 Persistence Group Tags
Tags within


Number
of Tags
Allowed

Controls how often (expressed in milliseconds) transactions are committed for a PRSTDataWriter.



A value of 0 indicates that transactions will be committed immediately. This is the recommended setting to
avoid losing data in the case of an unexpected error in Persistence Service and/or the underlying
hardware/software infrastructure.

0 or 1

For applications that can tolerate some data losses, setting this tag to a value greater than 0 will increase
performance.
Default: 0
Controls how often (expressed as a number of DDS samples) transactions are committed for a
PRSTDataWriter.



A value of 1 indicates that DDS samples will be persisted by the PRSTDataWriters immediately. This is the
recommended setting to avoid losing data in the case of an unexpected error in persistence service and/or the
0 or 1
underlying hardware/software infrastructure.
For application that can tolerate some data losses, setting this tag to a value greater than 1 will increase
performance.
Default: 1
Defines how many DDS samples will be pre-fetched by a PRSTDataWriter to satisfy requests from latejoiners.



When a DataReader requests DDS samples from a PRSTDataWriter by sending a NACK message, the
PRSTDataWriter may retrieve additional DDS samples from the database to minimize disk access.

0 or 1

This paramater determines that amount of DDS samples that will be retrieved preemptively from the
database by the PRSTDataWriter.
Default: 20000



This tag can be used to enable and configure a DDS sample log for the PRSTDataWriters in a persistence
group. A DDS sample log is a buffer of DDS samples on disk that, when used in combination with
delegate reliability, allow decoupling the original DataWriters from slow DataReaders.
For additional information on the DDS sample log, see Scenario: Slow Consumer (Section 30.3 on page
975).

0 or 1

Default: DDS sample log is disabled

951

27.8.1 QoSs

Table 27.8 Persistence Group Tags
Tags within


Number
of Tags
Allowed

A DDS_Boolean (see Table 27.1 Supported Tag Values) that determines how much state will be kept in
memory by the PRSTDataWriters in order to avoid accessing the persistent storage.
The property is only applicable when the persistence service operates in persistent mode (the  tag is present).



If this property is 1, the PRSTDataWriters will keep a copy of all the instances in memory. They will also
keep a fixed state overhead of 24 bytes per DDS sample. This mode provides the best performance.
However, the restore operation will be slower and the maximum number of DDS samples that a
PRSTDataWriter can manage will be limited by the available physical memory.

0 or 1

If this property is 0, all the state will be kept in the underlying persistent storage. In this mode, the maximum
number of DDS samples that a PRSTDataWriter can manage will not be limited by the available physical
memory.
Default: If the HistoryQosPolicy‘s kind is KEEP_LAST or the ResourceLimitsQosPolicy’s max_samples
!= DDS_UNLIMITED_LENGTH, the default is 1. Otherwise, the default is 0.
A DDS_Boolean (see Table 27.1 Supported Tag Values) that indicates if Persistence Service will use
Waitsets or Listeners to read data from the PRSTDataReaders of the group.
By default, the usage of Waitsets is disabled. With this configuration, Persistence Service uses the on_data_
available() listener callback to take the data from the PRSTDataReaders within the persistence group. The
write operation in a PRSTDataWriter is called within the listener callback.
When Waitsets are enabled, Persistence Service will use them to read the data:


If  is set to 1, there will be a single Waitset and a read thread shared across all the
PRSTDataReaders in the group.

0 or 1

If  is set to 0, there will be a Waitset and a read thread per PRSTDataReader in the
group.
The write operation in a PRSTDataWriter is called by the read thread associated with the PRSTDataReader.
Default: 0

27.8.1 QoSs
When a persistence service discovers a topic 'A' that matches a specific persistence group, it creates a
reader (known as ‘PRSTDataReader’) and writer (‘PRSTDataWriter’) to persist that topic. The QoSs associated with these readers and writers, as well as the corresponding publishers and subscribers, can be configured inside the persistence group using QoS tags.
For example:

71

...


952

27.8.2 DurabilityService QoS Policy




...



For instance, the number of DDS samples saved by Persistence Service is configurable through the
HISTORY QosPolicy (Section 6.5.10 on page 376) of the PRSTDataWriters.
If a QoS tag is not specified the persistence service will use the corresponding DDS default values (DurabilityService QoS Policy (Section 27.8.2 below) describes an exception to this rule).

27.8.2 DurabilityService QoS Policy
The DURABILITY SERVICE QosPolicy (Section 6.5.8 on page 372) associated with a DataWriter is
used to configure the HISTORY and the RESOURCE_LIMITS associated with the PRSTDataReaders
and PRSTDataWriters.
By default, the HISTORY and RESOURCE_LIMITS of a PRSTDataReader and PRSTDataWriter with
topic 'A' will be configured using the values specified in the XML file used to configure Persistence Service. To overwrite those values and use the values in the DURABILITY SERVICE QosPolicy (Section
6.5.8 on page 372) of the first discovered DataWriter publishing 'A', you can use the tag  in the persistence group definition:

71

...
1
...



27.8.3 Sharing a Publisher/Subscriber
By default, the PRSTDataWriters and PRSTDataReaders associated with a persistence group will share
the same Publisher and Subscriber.
To associate a different Publisher and Subscriber with each PRSTDataWriter and PRSTDataReader, use
the tags  and , as follows:

71

...
0
0
...

953

27.8.4 Sharing a Database Connection




27.8.4 Sharing a Database Connection
By default, the persistence service will share a single ODBC database connection to persist the topic data
received by each PRSTDataReader.
To associate an independent database connection to the PRSTDataReaders created by the persistence service, use the tag , as follows:

71

...
0
...



Sharing a database connection optimizes the resource usage. However, the concurrency of the system
decreases because the access to the database connection must be protected.

27.8.5 Memory Management
The DDS samples received and stored by the PRSTDataReaders and PRSTDataWriters are in serialized
form.
The serialized size of a DDS sample is the number of bytes required to send the DDS sample on the wire.
The maximum serialized size of a DDS sample is the number of bytes that the largest DDS sample for a
given type requires on the wire.
By default, the PRSTDataReaders and PRSTDataWriters created by the persistence service try to allocate
multiple DDS samples to their maximum serialized size. This may cause memory allocation issues when
the maximum serialized size is significantly large.
For PRSTDataReaders, the number of DDS samples in the DataReader’s queues can be controlled using
the QoS values resource_qos.resource_limits.max_samples and resource_qos.resource_limits.initial_
samples.
The PRSTDataWriters keep a cache of DDS samples so that they do not have to access the database every
time. The minimum size of this cache is 32 DDS samples.
In addition, each PRSTDataWriter keeps an additional DDS sample called the DB sample, which is used
to move information from the DataWriter cache to the database and vice versa
The  tag in a persistence group can be used to control the memory allocation
policy for the DDS samples created by PRSTDataReaders and PRSTDataWriters in the persistence group.

954

27.9 Configuring Durable Subscriptions in Persistence Service

Table 27.9 Memory Management Tags describes the memory management tags.
Table 27.9 Memory Management Tags
Tags within



Description

This tag is used to control the memory associated with the DB sample in a PRSTDataWriter. The
persistence service will not be able to store a DDS sample into persistent storage if the serialized size is
greater than this value. Therefore, this parameter must be used carefully.

Number
of Tags
Allowed

0 or 1

Default: LENGTH_UNLIMITED (DB sample is allocated to the maximum size).
This tag applies to both PRSTDataReaders and PRSTDataWriters. Its value determines the maximum size
(in bytes) of the buffers that will be pre-allocated to store the DDS samples. If the space required for a new
DDS sample is greater than this size, the persistence service will allocate the memory dynamically to the
exact size required by the DDS sample.



This parameter is used to control the memory allocated for the DDS samples in the PRSTDataReaders
queues and the PRSTDataWriters caches.

0 or 1

The size of the DB sample in the PRSTDataWriters is controlled by the value of the tag .
Default: LENGTH_UNLIMITED (DDS samples are allocated to the maximum size).

27.9 Configuring Durable Subscriptions in Persistence Service
This section assumes you are familiar with the concept of Required Subscriptions (Section 6.3.13 on page
294).
A Durable Subscription is a Required Subscription where DDS samples are stored and forwarded by Persistence Service.
There are two ways to create a Durable Subscriptions:
1. Programmatically using a DomainParticipant API:
A subscribing application can register a Durable Subscription by providing the topic name and the
endpoint group information, consisting of the Durable Subscription role_name and the quorum. To
register or delete a Durable Subscription, use the DomainParticipant’s register_durable_subscription() and delete_durable_subscription() operations, respectively (see Table 8.3 DomainParticipant Operations). The Durable Subscription information is propagated via a built-in topic to
Persistence Service.
2. Preconfigure Persistence Service with a set of Durable Subscriptions:
Persistence Service can be (pre-)configured with a list of Durable Subscriptions using the  XML tag under .

955

27.9.1 DDS Sample Memory Management With Durable Subscriptions


...


Logger
Track
2


Processor
Track
1




After registering or configuring the persistence service with specific Durable Subscriptions, the persistence
service will keep DDS samples until they are acknowledged by all the required Durable Subscriptions. In
the above example, the DDS samples must be acknowledged by two DataReaders that belong to the “Logger” Durable Subscription and one DataReader belonging to the “Processor” Durable Subscription.

27.9.1 DDS Sample Memory Management With Durable Subscriptions
The maximum number of DDS samples that will be kept in a PRSTDataWriter queue is determined by the
value of  in the  used to configure the PRSTDataWriter.
By default, a PRSTDataWriter configured with KEEP_ALL  will keep the DDS
samples in its cache until they are acknowledged by all the Durable Subscriptions associated with the
PRSTDataWriter. After the DDS samples are acknowledged by the Durable Subscriptions, they will be
marked as reclaimable but they will not be purged from the PRSTDataWriter’s queue until the DataWriter
needs these resources for new DDS samples. This may lead to inefficient resource utilization, especially
when  is high or UNLIMITED.
The PRSTDataWriter behavior can be changed to purge DDS samples after they have been acknowledged by all the active/matching DataReaders and all the Durable Subscriptions configured for the . To do so, set the tag  under  to TRUE. Notice that this setting is global to the service and applies to all the PRSTDataWriters
created by each .

27.10 Synchronizing of Persistence Service Instances
By default, different Persistence Service instances do not synchronize with each other. For example, in a
scenario with two Persistence Service instances, the first persistence service could receive a DDS sample
‘S1’ from the original DataWriter that is not received by the second persistence service. If the disk where
the first persistence service stores its DDS samples fails, ‘S1’ will be lost.

956

27.11 Enabling RTI Distributed Logger in Persistence Service

To enable synchronization between Persistence Servic instances, use the tag  under
. When it comes to synchronization, there are two different kinds of information
that can be synchronized independently:
l

l

Information about Durable Subscriptions and their states (see Configuring Durable Subscriptions in
Persistence Service (Section 27.9 on page 955))
DDS data samples

Table 27.10 Synchronization Tags
Tags within


Description

Number
of Tags
Allowed

Enables synchronization of DDS data samples in redundant Persistence Service instances.
When set to 1, DDS samples lost on the way to one service instance can be repaired by another
without impacting the original publisher of that message.


To synchronize the instances, the tag  must be set to 1 in every instance involved
in the synchronization.

0 or 1

Note: This DDS sample synchronization mechanism is not equivalent to database replication. The
extent to which database instances have identical contents depends on the destination ordering and
other QoS settings for the Persistence Service instances.
Default: 0
Enables synchronization of Durable Subscriptions in redundant Persistence Service instances.


When set to 1, the different Persistence Service instances will synchronize their Durable Subscription
information. This information includes the set of Durable Subscriptions as well as information about
0 or 1
the Durable Subscription’s state, such as the DDS samples that have already been received by the
Durable Subscriptions.
Default: 0



The period (in milliseconds) at which the information about Durable Subscriptions is synchronized.
0 or 1
Default: 5000 milliseconds

27.11 Enabling RTI Distributed Logger in Persistence Service
Persistence Service provides integrated support for RTI Distributed Logger (see Part 10: RTI Distributed
Logger (Section on page 1039)).
Distributed Logger is included in Connext DDS but it is not supported on all platforms; see the RTI Connext DDS Core Libraries Platform Notes to see which platforms support Distributed Logger.

957

27.12 Enabling RTI Monitoring Library in Persistence Service

When you enable Distributed Logger, Persistence Service will publish its log messages to Connext DDS.
Then you can use RTI Monitor1 to visualize the log message data. Since the data is provided in a Connext
DDS topic, you can also use rtiddsspy or even write your own visualization tool.
To enable Distributed Logger, modify the Persistence Service XML configuration file. In the  section, add the  tag as shown in the example below.

...

...

true

...

...


There are more configuration tags that you can use to control Distributed Logger’s behavior. For example,
you can specify a filter so that only certain types of log messages are published. For details, see Enabling
Distributed Logger in RTI Services (Section Chapter 39 on page 1049)

27.12 Enabling RTI Monitoring Library in Persistence Service
Persistence Service provides integrated support for RTI Monitoring Library (see Part 9: RTI Monitoring
Library (Section on page 1022)).
To enable monitoring in Persistence Service, you must specify the property rti.monitor.library for the participants that you want to monitor. For example:


54




rti.monitor.library
rtimonitoring
false




1RTI Monitor is a separate GUI application that can run on the same host as your application or on a different host.

958

27.13 Support for Extensible Types



...




Since Persistence Service is statically linked with RTI Monitoring Library, you do not need to have it in
your library search path.
For details on how to configure the monitoring process, see Configuring Monitoring Library (Section
Chapter 37 on page 1034).

27.13 Support for Extensible Types
Persistence Service includes partial support for the "Extensible and Dynamic Topic Types for DDS" specification from the Object Management Group (OMG)1. This section assumes that you are familiar with
Extensible Types and you have read the RTI Connext DDS Core Libraries Getting Started Guide
Addendum for Extensible Types.
Persistence groups can publish and subscribe to topics associated with final and extensible types.
The service will automatically create different pairs (PRSTDataReader, PRSTDataWriter) for each version
of a type discovered for a topic in a persistence group. In Connext DDS 5.0, it is not possible to associate
more than one type with a topic within a single DomainParticipant, therefore each version of a type
requires its own DomainParticipant.
The TYPE_CONSISTENCY_ENFORCEMENT QosPolicy (Section 7.6.6 on page 532) kind for each
PRSTDataReader is set to DISALLOW_TYPE_COERCION. This value cannot be overwritten by the
user.
For example:
struct A
long
};
struct B
long
long
};

{
x;
{
x;
y;

Let’s assume that Persistence Service is configured as follows and we have two DataWriters on Topic “T”
publishing type “A” and type “B” and sending TypeObject information.

1http://www.omg.org/spec/DDS-XTypes/

959

27.13.1 Type Version Discrimination




T




When Persistence Service discovers the first DataWriter with type “A”, it will create a DataReader
(PRSTDataReader) to read DDS samples from that DataWriter, and a DataWriter (PRSTDataWriter) to
publish and store the received DDS samples so they can be available to late-joiners.
When Persistence Service discovers the second DataWriter with type “B”, it will see that type “B” is not
equal to type “A”; then it will create a new pair (PRSTDataReader, PRSTDataWriter) to receive and store
DDS samples from the second DataWriter.
Since the PRSTDataReaders are created with the TypeConsistencyEnforcementQosPolicy’s kind set to
DISALLOW_TYPE_COERCION, the PRSTDataReader with type “A” will not match the DataWriter
with type “B”. Likewise, the PRSTDataReader with type “B” will not match the DataWriter with type
“A”.

27.13.1 Type Version Discrimination
Persistence Service uses the rules described in the RTI Connext DDS Core Libraries Getting Started
Guide Addendum for Extensible Types to decide whether or not to create a new pair (PRSTDataReader,
PRSTDataWriter) when it discovers a DataWriter for a topic “T”.
For DataWriters created with previous Connext DDS releases, Persistence Service will select the first pair
(PRSTDataReader, PRSTDataWriter) with a registered type name equal to the discovered registered type
name since DataWriters created with previous Connext DDS releases (before 5.0) do not send TypeObject information.

27.14 TCP Transport Support in Persistence Service
You can configure Persistence Service's Participants to use the TCP Transport. To do so, enable the TCP
Transport under the proper XML Persistence Service's  tag.
Make sure the string prefix passed in the property dds.transport.load_plugins is
"dds.transport.tcp". For more information about how to enable the TCP Transport, please see
TCP/TLS Transport Properties (Section 35.1.6 on page 1002).
Note that the Persistence Service's participant_qos will be used at least by two Participants: one for sending data and another for receiving data. Consequently, at least two TCP Transport plugins will be instantiated when enabling the TCP Transport. In order to avoid port collisions, Persistence Service will
automatically assign consecutive ports. For a base, it will use the values set for dds.transport.tcp.server_

960

27.14 TCP Transport Support in Persistence Service

bind_port (only when it is non-zero) and dds.transport.tcp.public_address (only if it is set). Consequently, the Participants creating a TCP Transport running as a server will open a minimum of two TCP
ports.

961

Chapter 28 Running RTI Persistence
Service
This chapter describes how to start and stop Persistence Service.
You can run Persistence Service on any node in the network. It does not have to be run on the
same node as the publishing or subscribing applications for which it is saving/delivering data. If
you run it on a separate node, make sure that the other applications can find it during the discovery
process—that is, it must be in one of the NDDS_DISCOVERY_PEERS lists.

28.1 Starting Persistence Service
The script to run Persistence Service’s executable is located in 1/bin.
RTI Persistence Service
Usage: rtipersistenceservice [options]
Options:
-cfgFile
 Configuration file. This parameter is optional
since the configuration can be loaded from
other locations
-cfgName
 Configuration name. This parameter is required
and it is used to find a 
matching tag in the configuration files
-appName
 Application name. Used to identify this
execution
for remote administration and to name the
DomainParticipants
Default: -cfgName
-identifyExecution
Appends the host name and process ID to the
appName to help
ensure unique names
-domainId

domain ID for the DomainParticipants created
by the service
Default: Use XML value

1See Paths Mentioned in Documentation (Section on page xxxviii)

962

28.1 Starting Persistence Service

-remoteAdministrationDomainId

-restore

<0|1>

-noAutoStart
-infoDir



-maxObjectsPerThread



-serviceThreadStackSize



-verbosity

[0-6]

-version
-help


Enables remote administration and sets
the domain ID for the communication
Default: Use XML value
Indicates whether or not persistence service
must restore its state from the persistent
storage
Default: Use XML value
Use this option if you plan to start RTI
Persistence Service remotely
The info directory of the running persistence
service. The service writes a ps.pid file into
this directory when is started. When the
service finalizes the file is deleted
Default: None
Sets the maximum number of objects that can
be stored per thread for a
DomainParticipantFactory
Default: Connext DDS default
Service thread stack size
Default: OS default
RTI Persistence Service verbosity
* 0 - silent
* 1 - exceptions (Core Libraries and Service)
* 2 - warnings (Service)
* 3 - information (Service)
* 4 - warnings (Core Libraries and Service)
* 5 - tracing (Service)
* 6 - tracing (Core Libraries and Service)
Default: 1 (exceptions)
Prints RTI Persistence Service version
Displays this information

The command-line options are described with more detail in Table 28.1 Persistence Service CommandLine Options
Table 28.1 Persistence Service Command-Line Options
Command-line Option

Description
Assigns a name to the execution of Persistence Service.
Remote commands will refer to the persistence service using this name.

-appName 

In addition, the name of the DomainParticipants created by Persistence Service will be based on this
name as follows:
RTI Persistence Service: : ()
Default: The name given with -cfgName if present, otherwise it is “RTI_Persistence_Service”
Specifies an XML configuration file for the Persistence Service.

-cfgFile 

The parameter is optional since the Persistence Service configuration can be loaded from other locations.
See How to Load the Persistence Service XML Configuration (Section 27.1 on page 935) for further
details.

963

28.1 Starting Persistence Service

Table 28.1 Persistence Service Command-Line Options
Command-line Option

Description
Required.
Selects a Persistence Service configuration.

-cfgName 

The same configuration files can be used to configure multiple persistence services. Each Persistence
Service instance will load its configuration from a different  tag based on the name
specified with this option.
If not specified, Persistence Service will print the list of available configurations and then exit.

-identifyExecution

Appends the host name and process ID to the service name provided with the -appName option. This
helps ensure unique names for remote administration.
Sets the domain ID for the DomainParticipants created by Persistence Service.

-domainId 
If not specified, the value in the  XML tag (see Table 27.7 Participant Tags) is used.
Enables remote administration and sets the domain ID for remote communication.
When remote administration is enabled, Persistence Service will create a DomainParticipant, Publisher,
remoteAdministrationDomainId Subscriber, DataWriter, and DataReader in the designated DDS domain.

This option overwrites the value of the tag  within .
Default: Use the value  under .
-help

Prints the Persistence Service version and list of command-line options.
Specifies the license file (path and filename). Only applicable to licensed versions of Persistence Service.

-licenseFile 

If not specified, Persistence Service looks for the license as described in the RTI Connext DDS Core
Libraries Getting Started Guide.
Indicates whether or not Persistence Service must restore its state from the persistent storage. 0 = do not
restore; 1 = do restore.

-restore <0|1>
If this option is not specified, the corresponding XML value in the  tag (see Table
27.4 Persistent Storage tags) is used.
Indicates that Persistence Service will not be started when the process is executed.
-noAutoStart

Use this option if you plan to start Persistence Service remotely, as described in Administering
Persistence Service from a Remote Location (Section Chapter 29 on page 966).

964

28.2 Stopping Persistence Service

Table 28.1 Persistence Service Command-Line Options
Command-line Option

Description
The info directory of the running Persistence Service.
Using this command line option, Persistence Service can be configured to create a file used to monitor the
status of the last shutdown.
At startup, the Persistence Service instance will create a file called ps.pid into the directory specified by infoDir.
If Persistence Service is shutdown gracefully, the file will be deleted before the process exists.

-infoDir 

If Persistence Service is not shutdown gracefully, the file will not be deleted.
You can detect the shutdown state of Persistence Service by checking for the presence of the ps.pid file.
If the file is present and Persistence Service is no longer running, the previous shutdown was not
graceful.
If Persistence Service is started and a ps.pid file exists, Persistence Service will immediately shutdown.
In this case, you must remove the file before Persistence Service can be restarted again.
Default: The file ps.pid will not be generated.

-maxObjectsPerThread 

Parameter used to configure the maximum objects per thread in the DomainParticipantFactory created by
Persistence Service.
Default: DDS default
Service thread stack size.

-serviceThreadStackSize 
Default: DDS default
Persistence Service verbosity:
0 - No verbosity
1 - Exceptions (Core Libraries and Persistence Service) (default)
2 - Warning (Persistence Service)
-verbosity

3 - Information (Persistence Service)
4 - Warning (Core Libraries and Persistence Service)
5 - Tracing (Persistence Service)
6 - Tracing (Core Libraries and Persistence Service)
Each verbosity level, n, includes all the verbosity levels smaller than n.

-version

Prints the Persistence Service version.

28.2 Stopping Persistence Service
To stop Persistence Service: press Ctrl-C. Persistence Service will close all files and perform a clean
shutdown. Persistence Service can also be stopped and shutdown remotely (see Administering Persistence
Service from a Remote Location (Section Chapter 29 on page 966)).

965

Chapter 29 Administering Persistence
Service from a Remote Location
Persistence Service can be controlled remotely by sending commands through a special Topic.
Any Connext DDS application can be implemented to send these commands and receive the corresponding responses. A shell application that sends/receives these commands is provided with Persistence Service.
The script for the shell application is $NDDSHOME/bin/rtipssh.
Entering rtipssh -help will show you the command-line options:
RTI Persistence Service Shell v5.2.0
Usage: rtipssh [options]...
Options:
-domainId 
Domain ID for the remote configuration
-timeout

Max time to wait a remote response
-cmdFile

Run commands in this file
-help
Displays this information

29.1 Enabling Remote Administration
By default, remote administration is disabled in Persistence Service.
To enable remote administration you can use the  tag (see Configuring Remote
Administration (Section 27.5 on page 942)) or the -remoteAdministrationDomainId commandline parameter (see Table 28.1 Persistence Service Command-Line Options), which enables remote
administration and sets the domain ID for remote communication.
When remote administration is enabled, Persistence Service will create a DomainParticipant, Publisher, Subscriber, DataWriter, and DataReader in the designated DDS domain. (The QoS values
for these entities are described in Configuring Remote Administration (Section 27.5 on page 942).)

966

29.2 Remote Commands

29.2 Remote Commands
This section describes the remote commands using the shell interface; Accessing Persistence Service from
a Connext DDS Application (Section 29.3 on the facing page) explains how to use remote administration
from a Connext DDS application.
Remote commands:
start (Section 29.2.1 below) 
stop (Section 29.2.2 below) 
shutdown (Section 29.2.3 on the facing page) 
status (Section 29.2.4 on the facing page) 

Parameters:
 can be:
l

l

The application name of a persistence service, such as “MyPersistenceService1”, as specified at
start-up with the command-line option -appName
A wildcard expression1 for a persistence service name, such as
“MyPersistenceService*”

29.2.1 start
start 

The start command starts the persistence service instance. DDS samples will not be persisted until the persistence service is started.
By default, the persistence service is started automatically when the process is executed. To start the service remotely use the command line option -noAutoStart (see Table 28.1 Persistence Service CommandLine Options).

29.2.2 stop
stop 

The stop command stops the persistence service instance.
An instance that has been stopped can be started again using the command start.

1As defined by the POSIX fnmatch API (1003.2-1992 section B.6)

967

29.2.3 shutdown

29.2.3 shutdown
shutdown 

The command shutdown stops the persistence service instance and finalizes the process

29.2.4 status
status 

The status command gets the status of a running persistence service instance. Possible values are
STARTED and STOPPED.

29.3 Accessing Persistence Service from a Connext DDS Application
You can send commands to control an Persistence Service instance from your own Connext DDS application.
You will need to create a DataWriter for a specific topic and type. Then, you can send a DDS sample that
contains a command and its parameters. Optionally, you can create a DataReader for a specific topic to
receive the results of the execution of your commands.
The topics are:
l

rti/persistence_service/administration/command_request

l

rti/persistence_service/administration/command_response

The types are:
l

RTI::PersistenceService::Administration::CommandRequest

l

RTI::PersistenceService::Administration::CommandResponse

You can find the IDL definitions for these types in /resource/idl/PersistenceServiceAdministration.idl.
The QoS configuration of your DataWriter and DataReader must be compatible with the one used by the
persistence service (see how this QoS is configured in Configuring Remote Administration (Section 27.5
on page 942)).
The following example in C shows how to send a command to shutdown a persistence service instance:
/***********************************************************/
/*** Create the Entities needed to send command request ****/
/***********************************************************/
participant = DDS_DomainParticipantFactory_create_participant(
DDS_TheParticipantFactory, domainId,
&DDS_PARTICIPANT_QOS_DEFAULT, NULL,
DDS_STATUS_MASK_NONE);

968

29.3 Accessing Persistence Service from a Connext DDS Application

if (participant == NULL)
{ /* Error */ }
if (publisher == NULL)
{ /* Error */ }
subscriber = DDS_DomainParticipant_create_subscriber(
participant, &DDS_SUBSCRIBER_QOS_DEFAULT,
NULL, DDS_STATUS_MASK_NONE);
publisher = DDS_DomainParticipant_create_publisher(
participant, &DDS_PUBLISHER_QOS_DEFAULT,
NULL, DDS_STATUS_MASK_NONE);
if (publisher == NULL)
{ /* Error */ }
typeName =
RTI_PersistenceService_Administration_CommandRequestTypeSupport_get_type_name();
retcode =
RTI_PersistenceService_Administration_CommandRequestTypeSupport_register_type(
participant, typeName);
if (retcode != DDS_RETCODE_OK)
{ /* Error */ }
topicCmd = DDS_DomainParticipant_create_topic(
participant,
"rti/persistence_service/administration/command_request",
typeName, &DDS_TOPIC_QOS_DEFAULT,
NULL, DDS_STATUS_MASK_NONE);
if (topicCmd == NULL)
{ /* Error */ }
typeName =
RTI_PersistenceService_Administration_CommandResponseTypeSupport_get_type_name();
retcode =
RTI_PersistenceService_Administration_CommandResponseTypeSupport_register_type(
participant, typeName);
if (retcode != DDS_RETCODE_OK)
{ /* Error */ }
topicResponse = DDS_DomainParticipant_create_topic(
participant,
"rti/persistence_service/administration/command_response",
typeName, &DDS_TOPIC_QOS_DEFAULT, NULL,
DDS_STATUS_MASK_NONE);
if (topicResponse == NULL)
{ /* Error */ }
writerQos.reliability.kind = DDS_RELIABLE_RELIABILITY_QOS;
writerQos.history.kind = DDS_KEEP_ALL_HISTORY_QOS;
writer = DDS_Publisher_create_datawriter(
publisher, topicCmd, &writerQos,
NULL /* listener */,
DDS_STATUS_MASK_NONE);
if (writer == NULL)
{ /* Error */ }
readerQos.reliability.kind = DDS_RELIABLE_RELIABILITY_QOS;

969

29.3 Accessing Persistence Service from a Connext DDS Application

readerQos.history.kind
= DDS_KEEP_ALL_HISTORY_QOS;
reader = DDS_Subscriber_create_datareader(
subscriber,
DDS_Topic_as_topicdescription(topicResponse),
&readerQos, NULL, DDS_STATUS_MASK_NONE);
if (reader == NULL)
{ /* Error */ }
/*******************************************************************/
/*** Wait for discovery ********************************************/
/*******************************************************************/
/* Wait until we discover one reader and one writer matching
* with the command request DataWriter and the command response
* DataReader */
while (count < maxPollPeriods)
{
retcode = DDS_DataWriter_get_publication_matched_status(
writer, &pubMatchStatus);
if (retcode != DDS_RETCODE_OK)
{ /* Error */ }
retcode = DDS_DataReader_get_subscription_matched_status(
reader, &subMatchStatus);
if (retcode != DDS_RETCODE_OK) { /* Error */ }
if (pubMatchStatus.total_count == 1 &&
subMatchStatus.total_count == 1)
{ break; }
count++;
NDDS_Utility_sleep(&pollPeriod);
}
if (count == maxPollPeriods)
{ /* Error */ }
/*******************************************************************/
/*** Send the command request **************************************/
/*******************************************************************/
request =
RTI_PersistenceService_Administration_CommandRequestTypeSupport_create_data();
if (request == NULL)
{ /* Error */ }
/* request->id provides an unique way to identify a request so that
* it can be correlated with a response. Although one of the fields is
* called host it does not necessarily has to contain the IP address of
* the host. Same applies to app */
request->id.host = 0;
request->id.app = 0;
request->id.invocation = 0;
strcpy(request->target_ps, "MyPersistenceService");
request->command._d = RTI_PERSISTENCE_SERVICE_COMMAND_SHUTDOWN;
retcode = RTI_PersistenceService_Administration_CommandRequestDataWriter_write(
(RTI_PersistenceService_Administration_CommandRequestDataWriter *) writer,
request, &instance_handle);
if (retcode != DDS_RETCODE_OK)
{ /* Error */ }

970

29.3 Accessing Persistence Service from a Connext DDS Application

/*******************************************************************/
/*** Wait for response ********************************************/
/*******************************************************************/
response =
RTI_PersistenceService_Administration_CommandResponseTypeSupport_create_data();
if (response == NULL)
{ /* Error */ }
count = 0;
while (count < maxPollPeriods) {
retcode =
RTI_PersistenceService_Administration_CommandResponseDataReader_take_next_sample(
(RTI_PersistenceService_Administration_CommandResponseDataReader*) reader,
response, &sampleInfo);
if (retcode == DDS_RETCODE_OK) {
break;
} else if (retcode != DDS_RETCODE_NO_DATA) {
/* Error */
}
NDDS_Utility_sleep(&pollPeriod);
count++;
}
if (count == maxPollPeriods) {
printf("No response received\n");
} else {
printf("Response received: %s\n",response->message);
}

971

Chapter 30 Advanced Persistence Service
Scenarios
This section covers several advanced scenarios for using Persistence Service.

30.1 Scenario: Load-balanced Persistence Services
Each running instance of the Persistence Service executes as a single process in a single computer.
In high-throughput scenarios the Persistence Service may become a bottleneck. The main reasons
are:
l

l

l

If the Persistence Service is configured to persist its DDS samples to durable storage (a disk
or a database) this will further limit the throughput of DDS samples that can be persisted to
what the database and/or disk can handle. Depending on computer hardware, the disk or
database this limit may be in the order of tens of thousands of DDS samples per second
which is far less than what could be communicated system-wide.
Depending on the CPU there will be limits on the throughput of DDS samples that can be
received by a single process.
The computer running the Persistence Service is typically connected to the network via a
single network interface so the data that can be persisted will be limited to the throughput
that flows though a single interface which is typically far less that the aggregated throughput
that can flow on the complete network.

To overcome these limits multiple instances of the RTI Persistence Service can be run in parallel.
These instances may run in multiple machines and be configured in a “load balancing” fashion
such that each Persistence Service process is only responsible for persisting a subset of the data
published on the DDS domain.
Multiple strategies for partitioning the data stored by each Persistence Service instance are possible:

972

30.1 Scenario: Load-balanced Persistence Services

l

l

Balance Persistence Services by Topic name. This strategy configures each persistence service to
persist different Topic names. This is accomplished by associating a filter expression with the declaration of the persistent groups used to configure each Persistence Service (see Creating Persistence
Groups (Section 27.8 on page 947)). The filter expression is applied to the Topic names, so for
example one Persistence Service could be configured with the filter “[A-Z]*” filter in the name of
the Topics that it will persist and the second with the filter “[a-z]*”. With this configuration the first
Persistence Service will persist data produced by DataWriters that specify durability TRANSIENT
or PERSISTENT and have a Topic name that starts with a capital letter and the second Persistence
Service will do the same for Topics that start with a lower-case letter.
Balance Persistence Services by data content. In some scenarios the data published on a single
Topic is too much for a single Persistence Service to handle. In this case the Persistence Services
can also be configured with filter expressions based on the content of the data. This is accomplished
by associating a content filter with the declaration of the persistent groups used to configure each
Persistence Service (see Creating Persistence Groups (Section 27.8 on page 947)).

When multiple instances of Persistence Service are used to store data on the same Topic, it becomes possible for DDS samples from the same original DataWriter to be stored in separate instances of Persistence
Service. In this situation, Connext DDS DataReaders automatically merge the data from the multiple Persistence Services such that the relative order of the DDS samples from the original DataWriter is preserved. This Connext DDS capability is called Collaborative Datawriters because multiple DataWriters,
in this case the ones for different Persistence Services, collaborate to reconstruct the original stream. (See
Collaborative DataWriters (Section Chapter 11 on page 670)).

973

30.2 Scenario: Delegated Reliability

Figure 30.1 Load-Balanced Persistence Services Scenario

30.2 Scenario: Delegated Reliability
The DDS-RTPS reliability protocol requires the DataWriter to periodically send HeartBeat messages to
the DataReaders, process their ACKs and NACK messages, keep track of the DataReader state, and send
the necessary repairs. The additional load caused by the reliability protocol increases with the number of
reliable DataReaders matched with the DataWriter. Even if the data is sent via multicast the number of
ACKs and NACKs will increase with the number of DataReaders.
In situations where there many DataReaders are subscribing to the same Topic, the reliability and repair
traffic may become too much for the DataWriter to handle and negatively impact its performance. To
address this situation, Connext DDS provides the ability to configure the DataWriter so that it delegates
the reliability task to a separate service. This approach is known as delegated reliability.
To take advantage of delegated reliability, both the original DataWriter and DataReader must be configured to enable an external service to ensure the reliability on their behalf. This is done by setting both
the dds.data_writer.reliability.delegate_reliability property on the DataWriter and the dds.data_reader.reliability.delegate_reliability property on the DataReader to 1.
With this configuration, the DataWriter creates a reliable channel to Persistence Service, yet sends data
using ‘best-effort’ reliability to the DataReaders directly. If a DDS sample is dropped, Persistence Service
will repair the DDS sample. Persistence Service is configured with push_on_write (in the DATA_
WRITER_PROTOCOL QosPolicy (DDS Extension) (Section 6.5.3 on page 347)) set to false. This way,

974

30.3 Scenario: Slow Consumer

DDS samples will only be sent from Persistence Service to the DataReaders when they are explicitly
NACKed by the DataReader.
Figure 30.2 Delegated Reliability Scenario

30.3 Scenario: Slow Consumer
Unless special measures are taken, the presence of slow consumers can impact the overall behavior of the
system. If a DataReader is not keeping up with the DDS samples being sent by the DataWriter, it will
apply back-pressure to the DataWriter to slow the rate at which the DataWriter can write DDS samples.
With delegated reliability (see Scenario: Delegated Reliability (Section 30.2 on the previous page)), the original DataWriter can offload the processing of the ACK/NACK messages generated by the DataReaders
to a PRSTDataWriter. However, the original DataWriter still has a reliable channel with the
PRSTDataReader that can slow it down.
By default, Persistence Service uses the Connext DDS receive thread to read DDS samples from the
PRStDataReaders, write the DDS samples to the PRSTDataWriters history, and send ACKs to the original DataWriter. With this configuration, a PRSTDataReader does not ACK DDS samples to the original
DataWriter until they are written into the corresponding PRSTDataWriter’s history. Since multiple
DataReaders may be accessing the PRSTDataWriter history at the same time that the persistence service is
trying to write new DDS samples, the PRSTDataWriter history becomes a contention point that can indirectly slow down the original DataWriter (see Slow-Consumer Scenario with Delegated Reliability (Section Figure 30.3 on the facing page)).

975

30.3 Scenario: Slow Consumer

Figure 30.3 Slow-Consumer Scenario with Delegated Reliability

To remove this contention point and decouple the slow consumer from the original DataWriter, Persistence Service supports a mode where DDS samples can be buffered prior to being added to the
PRSTDataWriter’s queue (see Slow Consumer Scenario with Delegated Reliability and DDS Sample Log
(Section Figure 30.4 on the next page)).

976

30.3 Scenario: Slow Consumer

Figure 30.4 Slow Consumer Scenario with Delegated Reliability and DDS Sample Log

If the PRSTDataWriter slows down due to the presence of slow consumers, the buffer will hold DDS
samples such that the original DataWriter and the rest of the system are not impacted. This buffer is called
the Persistence Service sample log. The persistence service creates a separate DDS sample log per
PRSTDataWriter in the group. In addition to the DDS sample log, the persistence service creates a thread
(write thread) whose main function is to read DDS samples from the log and write them to the associated
PRSTDataWriter. There is one thread per PRSTDataWriter.
Persistence Service currently does not allow multiple DDS sample logs to share the same write
thread.
Persistence Service can be configured to enable DDS sample logging per persistence group using the
 XML tag to specify the log’s configuration parameters—see Table 30.1 Sample Logging Tags..

977

30.3 Scenario: Slow Consumer

Table 30.1 Sample Logging Tags
Tags
within




Description

A DDS_Boolean (see Table 27.1 Supported Tag Values) that indicates whether or not DDS sample
logging is enabled in the container persistence group.

Number of
Tags Allowed

0 or 1

Default: 0



Specifies the maximum size of a DDS sample log file in Mbytes. When a log file becomes full,
Persistence Service creates a new log file.

0 or 1

Default: 60 MB
The period (in milliseconds) at which Persistence Service removes DDS sample log files whose full

Default: 10000 milliseconds



Determines how many DDS samples should be read and processed at once by the DDS sample log write
thread.

0 or 1

0 or 1

Default: 100 DDS samples
DDS samples in the DDS sample log are identified by two attributes:
l

The file ID


l

The row ID (position within the file)

0 or 1

The read bookmark indicates the most recently processed DDS sample.
This tag indicates how often (in milliseconds) the read bookmark is persisted into disk.
Default: 1000 milliseconds

Enabling DDS sample logging in a persistence group is expensive. For every PRSTDataWriter,
Persistence Service will create a write thread and an event thread that will be in charge of flushing
the log files and storing the read bookmark. Therefore, DDS sample logging should be enabled
only for the persistence groups where it is needed based on the potential presence of slow
consumers and/or the expected data rate in the persistence group. Small data rates will likely not
require a DDS sample log.

978

Part 7: RTI CORBA Compatibility Kit
The material in this part of the manual is only relevant if you have purchased the CORBA Compatibility Kit, an optional package that allows Connext DDS’s code generator, RTI Code Generator, to output type-specific code that is compatible with OCI’s distribution of TAO and the
JacORB distribution.
This section includes:
l

Introduction to RTI CORBA Compatibility Kit (Section Chapter 31 on page 980)

l

Generating CORBA-Compatible Code (Section Chapter 32 on page 982)

l

Supported IDL Types (Section Chapter 33 on page 985)

979

Chapter 31 Introduction to RTI CORBA
Compatibility Kit
RTI CORBA Compatibility Kit is an optional package that allows the RTI Code Generator to output type-specific code that is compatible with OCI’s or DOC’s distribution of TAO and the
JacORB distribution.
By having compatible data types, your applications can use CORBA and Connext DDS APIs,
with no type conversions required.
For more information about OCI's or DOC’s distribution of TAO and JacORB, please refer to the
documentation included with those distributions. Additional information can be found on OCI’s
TAO website (www.theaceorb.com), DOC’s TAO website (www.dre.vanderbilt.edu), and
JacORB’s website (www.jacorb.org). TAO and JacOrb distributions that are compatible with this
version of Connext DDS are available from the RTI Support Portal, accessible from https://support.rti.com.
This figure shows the process of using IDL files and types that are shared with CORBA:

980

Chapter 31 Introduction to RTI CORBA Compatibility Kit

CORBA Compatibility Kit is designed to be installed on top of Connext DDS; this kit enables RTI Code
Generator to support these CORBA-specific command-line options:
[-corba [CORBA Client header file]]
[-dataReaderSuffix ]
[-dataWriterSuffix ]
[-orb ]
[-typeSequenceSuffix ]

The above options are described in the RTI Code Generator User’s Manual.
On the wire, the serialized version of the code for types generated using the -corba option is identical to
the serialized version of the code for types generated without the option. As result, endpoints (DataReaders or DataWriters) using type-support code generated with -corba can fully communicate with endpoints
using type-support code generated without -corba.

981

Chapter 32 Generating CORBACompatible Code
The CORBA Compatibility Kit enables RTI Code Generator to produce type-specific code that is
compatible with OCI’s distribution of TAO for C++ and with JacORB for Java.
When using RTI Code Generator, specify the -corba option on the command line to generate compatible code. The -corba option enables the use of data structures for both CORBA and Connext
DDS API calls without requiring any translation: the IDL-to-language mapping is the same for
both.
There are some trade-offs to consider:
l

l

l

While the -corba option provides the benefit of CORBA-compatible type-specific code, it
does not provide support for bit fields, pointers and ValueTypes.
For complex types such as sequences and strings, the memory management is different when
the -corba option is used. When code is generated without the option, the memory needed
for the type is pre-allocated at system initialization. When code is generated with the option,
the memory is allocated when it is needed, so memory allocation system calls may occur
while the system is in steady state.
Without the -corba option, access to data fields within types may be faster under some circumstances. CORBA-compatible types require the use of accessor methods. When
-corba is not used, while the accessor methods are provided for convenience, they can be
bypassed and the data can be accessed directly. This direct access is available to the user as
well as to the Connext DDS internal implementation code. As a result, depending on the
complexity of the types used, overall system latency could be lower when using non-compatible types (that is, when -corba is not used).

The following sections describe how to use the CORBA Compatibility Kit. In addition to these
instructions, a simple example is available.

982

32.1 Generating C++ Code

By default, examples are copied into your home directory the first time you run RTI Launcher or any script
in /bin. This document refers to the location of the copied examples as .
Wherever you see , replace it with the appropriate path.
Default path to the examples:
l

Mac OS X systems: /Users/your user name/rti_workspace/5.2.3/examples

l

UNIX-based systems: /home/your user name/rti_workspace/5.2.3/examples

l

Windows systems: your Windows documents folder\rti_workspace\5.2.3\examples
Where 'your Windows documents folder' depends on your version of Windows. For example, on
Windows 7, the folder is C:\Users\your user name\Documents; on Windows Server 2003, the
folder is C:\Documents and Settings\your user name\Documents.

Note: You can specify a different location for rti_workspace. You can also specify that you do not want
the examples copied to the workspace. For details, see Controlling Location for RTI Workspace and Copying of Examples in the Connext DDS Core Libraries Getting Started Guide.
l

C++ using TAO:
l Generating Java Code (Section 32.2 on the facing page)
l

l

See the example in /corba/c++ and read Instructions.pdf.

Java using JacORB:
l Generating Java Code (Section 32.2 on the facing page)
l

See the example in /corba/java and read Instructions.pdf.

32.1 Generating C++ Code
To generate CORBA-compatible type-specific code, first run TAO’s code generator, tao_idl, on the IDL
file containing your data types. If you followed the TAO distribution compilation instructions contained in
this document, the tao_idl compiler executable will be in the TAO install directory under /bin.
/bin/tao_idl .idl

This will generate CORBA support files for your data types. The generated file will have a name matching
the pattern C.h and will contain the type definitions. Pass this header file as a parameter
to rtiddsgen to generate the Connext DDS support code for the data types.

983

32.2 Generating Java Code

rtiddsgen -language C++ -corba C.h -example \
 .idl

The optional -example  flag will generate code for a publisher and a subscriber. It will also
generate an .mpc file (and an .mwc file for Windows) that can be used with TAO's Makefile, Project and
Workspace Creator (MPC) to generate a makefile or a Visual Studio project file for your DDS-CORBA
application. The .mpc file is meant to work out-of-the-box with the DDS-CORBA C++ Message example
only, so you will have to modify it to compile your custom application. Please refer to the DDS-CORBA
C++ example for more information about using MPC (see the Instructions document).

32.2 Generating Java Code
To generate Java CORBA-compatible type specific code, first run the JacORB code generator on the IDL
file containing your data types.
/bin/idl .idl

After generating the CORBA code for the IDL types run rtiddsgen as follows:
rtiddsgen -language Java -corba -example  \
.idl

The optional -example  flag will generate code for a DDS publisher and a DDS subscriber. It will also generate a makefile specific to your architecture that can be used to compile the
example using the publisher and subscriber code generated.
To form a complete code set, use the type class generated by the CORBA IDL compiler and the files generated by RTI Code Generator.

984

Chapter 33 Supported IDL Types
Table 33.1 Supported IDL Types when Using rtiddsgen -corba lists the IDL types supported when
using the –corba option.
Table 33.1 Supported IDL Types when Using rtiddsgen -corba
IDL
Construct

Support

Modules

Supported

Interfaces

Ignored

Constants

Supported

Basic Data Types Supported
Enums

Supported

String Types

Supported

Wide String
Types

Supported
Supported

Struct Types

Note: In-line nested structures are not supported (whether using -corba or not). See Note 1 (Section on the
next page).

Fixed Types

Ignored

Union Types

Supported
Supported

Sequence Types
Note: Sequences of anonymous sequences are not supported. See Note 2 (Section on the next page).

985

Chapter 33 Supported IDL Types

Table 33.1 Supported IDL Types when Using rtiddsgen -corba
IDL
Construct

Support

Array Types

Supported

Typedefs

Supported
Not Supported.

Any
Note that RTI Code Generator does not ignore them. This construct cannot be in the IDL file.
Value Types

Ignored

Exception Types

Ignored
Supported

Type Code

RTI Code Generator generates Connext DDS TypeCodes
CORBA TypeCodes are generated by the CORBA IDL compiler

Note 1
Inline nested structures, such as the following example, are not supported.
struct Outer {
short outer_short;
struct Inner {
char inner_char;
short inner_short;
} outer_nested_inner;
};

Note 2
Sequences of anonymous Sequences are not supported. This kind of type will be banned in future revisions of CORBA. For example, the following is not supported:
sequence,4> MySequence;

Instead, sequences of sequences can be supported using typedef definitions. For example, this is supported:
typedef sequence MyShortSequence;
sequence MySequence;

986

Part 8: RTI TCP Transport
RTI TCP Transportis only available on specific architectures. See the RTI Connext DDS Core
Libraries Platform Notes for details.
Out of the box, Connext DDS uses the UDPv4 and Shared Memory transport to communicate
with other DDS applications. This configuration is appropriate for systems running within a single
LAN. However, using UDPv4 introduces some problems when Connext DDS applications in different LANs need to communicate:
l

UDPv4 traffic is usually filtered out by the LAN firewalls for security reasons.

l

Forwarded ports are usually TCP ports.

l

Each LAN may run in its own private IP address space and use NAT (Network Address
Translation) to communicate with other networks.

TCP Transport enables participant discovery and data exchange using the TCP protocol (either on
a local LAN, or over the public WAN). TCP Transport allows Connext DDS to address the challenges of using TCP as a low-level communication mechanism between peers and limits the number of ports exposed to one. (When using the default UDP transport, a Connext DDS application
uses multiple UDP ports for communication, which may make it unsuitable for deployment across
firewalled networks).

987

Chapter 34 TCP Communication
Scenarios
TCP Transport can be used to address multiple communication scenarios—from simple communication within a single LAN, to complex communication scenarios across LANs where NATs
and firewalls may be involved. This section describes these scenarios:
l

Communication Within a Single LAN (Section 34.1 below)

l

Symmetric Communication Across NATs (Section 34.2 on the next page)

l

Asymmetric Communication Across NATs (Section 34.3 on page 990)

34.1 Communication Within a Single LAN
TCP Transport can be used as an alternative to UDPv4 to communicate Connext DDS applications running inside the same LAN. Figure 34.1 Communication within a Single LAN on the
next page shows how to configure the TCP transport in this scenario.

988

34.2 Symmetric Communication Across NATs

Figure 34.1 Communication within a Single LAN

l

l

parent.classid (Section on page 1004) and server_bind_port (Section on page 1009) are transport
properties configured using the PropertyQosPolicy of the participant. (Note: When the TCP transport is instantiated, by default it is configured to work in a LAN environment using symmetric communication and binding to port 7400 for incoming connections.) For additional information about
these properties, see Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t.
Initial Peers represents the peers to which the participant will be announced to. Usually, these peers
are configured using the DiscoveryQosPolicy of the participant or the environment variable NDDS_
DISCOVERY_PEERS. For information on the format of initial peers, see Choosing a Transport
Mode (Section 35.1.1 on page 993).

Unlike the UDPv4 transport, you must specify the initial peers, because multicast cannot be used
with TCP.

34.2 Symmetric Communication Across NATs
In NAT communication scenarios, each one of the LANs has a private IP address space. The communication with other LANs is done through NAT routers that translate private IP addresses and ports into
public IP addresses and ports.
In symmetric communication scenarios, any Connext DDS application can initiate TCP connections with
other applications. Figure 34.2 Symmetric Communication Across NATs on the facing page shows how
to configure the TCP transport in this scenario.

989

34.3 Asymmetric Communication Across NATs

Figure 34.2 Symmetric Communication Across NATs

Notice that initial peers refer to the public address of the remote LAN where the Connext DDS application
is deployed and not the private address of the node where the application is running. In addition, the transport associated with a Connext DDS instance will have to be configured with its public address (public_
address (Section on page 1008)) so that this information can be propagated as part of the discovery process.
Because the public address and port of the Connext DDS instances must be known before the communication is established, the NAT Routers will have to be configured statically to translate (forward) the
private server_bind_port (Section on page 1009) into a public port. This process is known as static NAT
or port forwarding; it allows traffic originating in outer networks to reach designated peers in the LAN
behind the NAT router. You will need to refer to your router’s configuration manual to understand how to
correctly set up port forwarding.

34.3 Asymmetric Communication Across NATs
This scenario is similar to the previous one, except in this case the TCP connections can be initiated only
by the Connext DDS instance in LAN1. For security reasons, incoming connections to LAN1 are not
allowed. In this case, the peer in LAN1 is considered ‘unreachable.’ Unreachable peers can publish and
subscribe just like any other peer, but communication can occur only to a ‘reachable’ peer.

990

34.3 Asymmetric Communication Across NATs

Figure 34.3 Asymmetric Communication Across NATs below shows how to configure the TCP transport
in this scenario. Notice that the transport property server_bind_port is set to 0 to configure the node as
unreachable.
Figure 34.3 Asymmetric Communication Across NATs

In an asymmetric configuration, an unreachable peer (that is behind a firewall or NAT without port forwarding) can still publish and subscribe like a reachable peer, but with some important limitations:
l

l

An unreachable peer can only communicate with reachable peers: two unreachable peers cannot
establish a direct communication since they are both behind a firewall and/or NAT.
Note that since Connext DDS always relies on a direct connection between peers (even if there is a
third node that can be reachable by both unreachable peers), communication can never occur
between unreachable peers. For example, suppose Peers A and B are unreachable and Peer C is
reachable. Communication can take place between A and C, and between B and C, but not between
A and B. For this configuration, you should consider using RTI Federation Service (available for
purchase as a separate product).
It can take longer to discover unreachable peers than reachable ones. This is because a reachable
peer has to wait for the unreachable peer to establish the communication first.
For example, suppose Peer A (unreachable) starts before Peer B (reachable). The discovery mechanism of A attempts to connect to the (not-yet existing) Peer B. Since it fails, it will retry after n

991

34.3 Asymmetric Communication Across NATs

seconds. Right after that, B starts. If A would be reachable (and in B’s peer list), the discovery mechanism will immediately contact A. In this case, since A cannot be reached, B needs to wait until the
discovery process of A decides to retry.
This effect can be minimized by modifying the QoS that controls the discovery mechanism used by
A. In particular, you should set the DomainParticipant’s
DiscoveryConfig QoS policy’s min_initial_participant_announcement_period to a small value.
Note that the concept of symmetric/asymmetric configuration is a local concept that only describes the communication mechanism between two peers. A reachable peer can be involved in symmetric communication
with another reachable peer, and at the same time have asymmetric communication with a unreachable
peer. When a peer attempts to communicate with a remote peer, it knows if the remote peer is reachable or
not by looking at the transport address provided.

992

35.1 Configuring the TCP Transport
TCP Transport is distributed as a both shared and static library in /lib/. The library is called nddstransporttcp.
Mechanisms for Configuring the Transport:
By explicitly instantiating a new transport (see Explicitly Instantiating the TCP Transport Plugin (Section 35.1.2 on the next page)) and then registering it with the DomainParticipant (see
Installing Additional Builtin Transport Plugins with register_transport() (Section 15.7 on page
765)). (Not available in the Java and .NET APIs.)
Through the Property QoS policy of the DomainParticipant (on UNIX, Solaris and Windows
systems only). This process is described in Configuring the TCP Transport with the Property
QosPolicy (Section 35.1.3 on page 996).
This section describes:
Choosing a Transport Mode (Section 35.1.1 below)
Explicitly Instantiating the TCP Transport Plugin (Section 35.1.2 on the next page)
Configuring the TCP Transport with the Property QosPolicy (Section 35.1.3 on page 996)
Setting the Initial Peers (Section 35.1.4 on page 999)
Support for External Hardware Load Balancers in TCP Transport Plugin (Section 35.1.5 on page
1000)
TCP/TLS Transport Properties (Section 35.1.6 on page 1002)

35.1.1 Choosing a Transport Mode
When you configure the TCP transport, you must choose one of the following types of communication:

993

35.1.2 Explicitly Instantiating the TCP Transport Plugin

TCP over LAN — Communication between the two peers is not encrypted (data is written directly to a
TCP socket). Each node can use all the possible interfaces available on that machine to receive connections. The node can only receive connections from machines that are on a local LAN.
TCP over WAN — Communication is not encrypted (data is written directly to a TCP socket). The node
can only receive connections from a specific port, which must be configured in the public router of the
local network (WAN mode).
TLS over LAN — This is similar to the TCP over LAN, where the node can use all the available network interfaces to TX/RX data (LAN nodes only), but in this mode, the data being written on the physical
socket is encrypted first (through the openssl library). Performance (throughput and latency) may be less
than TCP over LAN since the data needs to be encrypted before going on the wire. Discovery time may
be longer with this mode because when the first connection is established, the two peers exchange handshake information to ensure line protection. For more general information on TLS, see Datagram Transport-Layer Security (DTLS) (Section 24.3 on page 908).
TLS over WAN — The data is encrypted just like TLS over LAN, but it can be sent and received only
from a specific port of the router.
Note: To use either TLS mode, you also need RTI TLS Support, which is available for purchase as a separate package.
An instance of the transport can only communicate with other nodes that use the same transport mode.
You can specify the transport mode in either the NDDS_Transport_TCPv4_Property_t structure (see
TCP/TLS Transport Properties (Section 35.1.6 on page 1002)) or in the parent.classid (Section on page
1004) field of the Properties QoS (see Configuring the TCP Transport with the Property QosPolicy
(Section 35.1.3 on page 996)). Your choice of transport mode will also be reflected in the prefix you
use for setting the initial peers (see Setting the Initial Peers (Section 35.1.4 on page 999)).

35.1.2 Explicitly Instantiating the TCP Transport Plugin
As described on Page 993, there are two ways to configure a transport plugin. This section describes the
way that includes explicitly instantiating and registering a new transport. (The other way is to use the Property QoS mechanism, described in Configuring the TCP Transport with the Property QosPolicy (Section
35.1.3 on page 996)).
Notes:
This way of instantiating a transport is not supported in the Java and .NET APIs. If you are using
Java or .NET, use the Property QoS mechanism described in Configuring the TCP Transport with the
Property QosPolicy (Section 35.1.3 on page 996).
To use this mechanism, there are extra libraries that you must link into your program and an additional header file that you must include. Please see Additional Header Files and Include Directories (Sec-

994

35.1.2.1 Additional Header Files and Include Directories

tion 35.1.2.1 on the facing page) and Additional Libraries and Compiler Flags (Section 35.1.2.2 below)
for details.
To instantiate a TCP transport:
Include the extra header file described in Additional Header Files and Include Directories (Section
35.1.2.1 below).
Instantiate a new transport by calling NDDS_Transport_TCPv4_new():
NDDS_Transport_Plugin* NDDS_Transport_TCPv4_new (
const struct NDDS_Transport_TCPv4_Property_t * property_in)

Register the transport by calling NDDSTransportSupport::register_transport().
See the API Reference HTML documentation for details on these functions and the contents of the
NDDS_Transport_TCPv4_Property_t structure.
35.1.2.1 Additional Header Files and Include Directories
To use the TCP Transport API, you must include an extra header file (in addition to those in Table 9.1
Header Files to Include for Connext DDS (All Architectures)):
#include "ndds/transport_tcp/transport_tcp_tcpv4.h"

Since TCP Transport is in the same directory as Connext DDS (see Table 9.2 Include Paths for Compilation (All Architectures)), no additional include paths need to be added for the TCP Transport API. If
this is not the case, you will need to specify the appropriate include path.
35.1.2.2 Additional Libraries and Compiler Flags
To use the TCP Transport, you must add the nddstransporttcp library to the link phase of your application. There are four different kind of libraries, depending on if you want a debug or release version, and
static or dynamic linking with Connext DDS.
For UNIX- based systems, the libraries are:
l

libnddstransporttcp.a — Release version, dynamic libraries

l

libnddstransporttcpd.a — Debug version, dynamic libraries

l

libnddstransporttcpz.a — Release version, static libraries

l

libnddstransporttcpzd.a — Debug version, static libraries

For Windows-based systems, the libraries are:

995

35.1.3 Configuring the TCP Transport with the Property QosPolicy

l

NDDSTRANSPORTTCP.LIB — Release version, dynamic libraries

l

NDDSTRANSPORTTCPD.LIB — Debug version, dynamic libraries

l

NDDSTRANSPORTTCPZ.LIB — Release version, static libraries

l

NDDSTRANSPORTTCPZD.LIB — Debug version, static libraries

Notes for using TLS:
To use either TLS mode (see Choosing a Transport Mode (Section 35.1.1 on page 993)), you also need
RTI TLS Support, which is available for purchase as a separate package. The TLS library (libnddstls.so
or NDDSTLS.LIB, depending on your platform) must be in your library search path (pointed to by the
environment variable LD_LIBRARY_PATH on UNIX/Solaris systems, Path on Windows systems,
LIBPATH on AIX systems, DYLD_LIBRARY_PATH on Mac OS systems).
If you already have $NDDSHOME/lib/ in your library search path, no extra steps are
needed to use TLS once TLS Support is installed.
Even if you link everything statically, you must make sure that the location for $NDDSHOME/lib/ (or wherever the TLS library is located) is in your search path. When the TCP Transport Plugin is explicitly instantiated, the TLS library is loaded dynamically, even if you use static linking for
everything else. To load TLS libraries statically, please see Configuring the TCP Transport with the Property QosPolicy (Section 35.1.3 below).
Your search path must also include the location for the OpenSSL library, which is used by the TLS library.

35.1.3 Configuring the TCP Transport with the Property QosPolicy
The PROPERTY QosPolicy (DDS Extension) (Section 6.5.17 on page 394) allows you to set up name/value pairs of data and attach them to an entity, such as a DomainParticipant.
Like all QoS policies, there are two ways to specify the Property QoS policy:
Programmatically, as described in this section and Getting, Setting, and Comparing QosPolicies (Section
4.1.7 on page 158). This includes using the add_property() operation to attach name/value pairs to the
Property QosPolicy and then configuring the DomainParticipant to use that QosPolicy (by calling set_qos
() or specifying QoS values when the DomainParticipant is created).
With an XML QoS Profile, as described in Configuring QoS with XML (Section Chapter 17 on page
791). This causes Connext DDS to dynamically load the TCP transport library at run time and then implicitly create and register the transport plugin.
To add name/value pairs to the Property QoS policy, use the add_property() operation:
DDS_ReturnCode_t DDSPropertyQosPolicyHelper::add_property
(DDS_PropertyQosPolicy policy, const char * name,
const char * value, DDS_Boolean propagate)

996

35.1.3 Configuring the TCP Transport with the Property QosPolicy

For more information on add_property() and the other operations in the DDSPropertyQosPolicyHelper
class, see Table 6.57 PropertyQoSPolicyHelper Operations, as well as the API Reference HTML documentation.
The ‘name’ part of the name/value pairs is a predefined string. The property names for the TCP Transport
are described in Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t.
Here are the basic steps, taken from the example Hello World application (for details, please see the
example application.)
Get the default DomainParticipant QoS from the DomainParticipantFactory.
DDSDomainParticipantFactory::get_instance()->
get_default_participant_qos(participant_qos);

Disable the builtin transports.
participant_qos.transport_builtin.mask =
DDS_TRANSPORTBUILTIN_MASK_NONE;

Set up the DomainParticipant’s Property QoS.
Load the plugin.
DDSPropertyQosPolicyHelper::add_property (
participant_qos.property,
"dds.transport.load_plugins",
"dds.transport.TCPv4.tcp1",
DDS_BOOLEAN_FALSE);

Specify the transport plugin library.
DDSPropertyQosPolicyHelper::add_property (
participant_qos.property,
"dds.transport.TCPv4.tcp1.library",
"nddstransporttcp",
DDS_BOOLEAN_FALSE);

Specify the transport’s ‘create’ function.
DDSPropertyQosPolicyHelper::add_property (
participant_qos.property,
"dds.transport.TCPv4.tcp1.create_function",
"NDDS_Transport_TCPv4_create", DDS_BOOLEAN_FALSE);

Set the transport to work in a WAN configuration with a public address:

997

35.1.3.1 Configuring the TCP Transport to be Loaded Statically

DDSPropertyQosPolicyHelper::add_property (
participant_qos.property,
"dds.transport.TCPv4.tcp1.parent.classid",
”NDDS_TRANSPORT_CLASSID_TCPV4_WAN”, DDS_BOOLEAN_FALSE);
DDSPropertyQosPolicyHelper::add_property (
participant_qos.property,
"dds.transport.TCPv4.public_address",
"182.181.2.31",
DDS_BOOLEAN_FALSE);

Specify any other properties, as needed.
Create the DomainParticipant using the modified QoS.
participant =
DDSTheParticipantFactory->create_participant (
domainId,
participant_qos,
NULL /* listener */,
DDS_STATUS_MASK_NONE);

Property changes should be made before the transport is loaded—either before the
DomainParticipant is enabled, before the first DataWriter/DataReader is created, or before the
builtin topic reader is looked up, whichever one happens first.
35.1.3.1 Configuring the TCP Transport to be Loaded Statically
Similar to the previous example, here are the basic steps to load the TCP Transport Plugin statically.
1. Get the default DomainParticipant QoS from the DomainParticipantFactory.
DDSDomainParticipantFactory::get_instance()->
get_default_participant_qos(participant_qos);

2. Disable the builtin transports.
participant_qos.transport_builtin.mask =
DDS_TRANSPORTBUILTIN_MASK_NONE;

3. Set up the DomainParticipant’s Property QoS.
a. Load the plugin.
DDSPropertyQosPolicyHelper::add_property
(participant_qos.property,
"dds.transport.load_plugins",
"dds.transport.TCPv4.tcp1",DDS_BOOLEAN_FALSE);

998

35.1.3.2 Loading TLS Support Libraries Statically

b. Specify the transport’s ‘create’ function pointer.
DDSPropertyQosPolicyHelper::add_pointer_property
(participant_qos.property,
"dds.transport.TCPv4.tcp1.create_function_ptr",
(void*)NDDS_Transport_TCPv4_create);

c. Set the transport to work in a WAN configuration with a public address:
DDSPropertyQosPolicyHelper::add_property
(participant_qos.property,
"dds.transport.TCPv4.tcp1.parent.classid",
”NDDS_TRANSPORT_CLASSID_TCPV4_WAN”,
DDS_BOOLEAN_FALSE);
DDSPropertyQosPolicyHelper::add_property
(participant_qos.property,
"dds.transport.TCPv4.tcp1.public_address",
"182.181.2.31",
DDS_BOOLEAN_FALSE);

d. Specify any other properties, as needed.
4. Create the DomainParticipant using the modified QoS.
participant = DDSTheParticipantFactory->create_participant
(domainId, participant_qos,
NULL /* listener */, DDS_STATUS_MASK_NONE);

35.1.3.2 Loading TLS Support Libraries Statically
The process to load TLS Support library statically is similar, but in this case both the tls_create_function_
ptr and tls_delete_function_ptr properties need to be set.
DDSPropertyQosPolicyHelper::add_pointer_property
(participant_qos.property,
"dds.transport.TCPv4.tcp1.tls_create_function_ptr",
(void*)RTITLS_ConnectionEndpointFactoryTLSv4_create);
DDSPropertyQosPolicyHelper::add_pointer_property
(participant_qos.property,
"dds.transport.TCPv4.tcp1.tls_delete_function_ptr",
(void*)RTITLS_ConnectionEndpointFactoryTLSv4_delete);

35.1.4 Setting the Initial Peers
Note: You must specify the initial peers (you cannot use the defaults because multicast cannot be used
with TCP).
For TCP Transport, the addresses of the initial peers (NDDS_DISCOVERY_PEERS) that will be contacted during the discovery process have the following format:

999

35.1.5 Support for External Hardware Load Balancers in TCP Transport Plugin

l

For WAN communication using TCP: tcpv4_wan://:

l

For WAN communication using TLS: tlsv4_wan://:

l

For LAN communication using TCP: tcpv4_lan://:

l

For LAN communication using TLS: tlsv4_lan://:

For example:
setenv NDDS_DISCOVERY_PEERS tcpv4_wan://10.10.1.165:7400,
tcpv4_wan://10.10.1.111:7400,tcpv4_lan://192.168.1.1:7500

When the TCP transport is configured for LAN communication (with the parent.classid (Section on
page 1004) property), the IP address is the LAN address of the peer and the port is the server port
used by the transport (the server_bind_port (Section on page 1009) property).
When the TCP transport is configured for WAN communication (with the parent.classid (Section on
page 1004) property), the IP address is the WAN or public address of the peer and the port is the
public port that is used to forward traffic to the server port in the TCP transport.

35.1.5 Support for External Hardware Load Balancers in TCP Transport
Plugin
For two Connext DDS applications to communicate, the TCP Transport Plugin needs to establish 4-6 connections between the two communicating applications. The plugin uses these connections to exchange
DDS data (discovery or user data) and TCP Transport Plugin control messages.
With the default configuration, the TCP Transport Plugin does not support external load balancers. This is
because external load balancers do not forward the traffic to a unique TCP Transport Plugin server, but
they divide the connections among multiple servers. Because of this behavior, when an application running a TCP Transport Plugin client tries to establish all the connections to an application running a TCP
Transport Plugin server, the server may not receive all the required connections.
In order to support external load balancers, the TCP Transport Plugin provides a session-ID negotiation
feature. When session-ID negotiation is enabled (by setting the negotiate_session_id property to true), the
TCP Transport Plugin will perform the negotiation depicted in Session-ID Negotiation (Section Figure
35.1 on the facing page).

1000

35.1.5 Support for External Hardware Load Balancers in TCP Transport Plugin

Figure 35.1 Session-ID Negotiation

During the session-ID negotiation, the TCP Transport Plugin exchanges three types of messages:
Session-ID Request: This message is sent from the client to the server. The server must respond with a
session-ID response.
Session-ID Response: This message is sent from the server to the client as a response to a session-ID
request. The client will store the session ID contained in this message.
Session-ID Indication: This message is sent from the client to the server; it does not require a response
from the server.
The negotiation consists of the following steps:
1. The TCP client sends a session-ID request with the session ID set to zero.
2. The TCP server sends back a session-ID response with the session ID set to zero.
3. The external load balancer modifies the session-ID response, setting the session ID with a value that
is meaningful to the load balancer and identifies the session.
4. The TCP client receives the session-ID response and stores the received session ID.

1001

35.1.5.1 Session-ID Messages

5. For each new connection, the TCP client sends a session-ID indication containing the stored session
ID. This will allow the load balancer to redirect to the same server all the connections with the same
session ID.
35.1.5.1 Session-ID Messages
TCP Payload for Session-ID Message (Section below) depicts the TCP payload of a session-ID message.
The payload consists of 48 bytes. In particular, your load balancer needs to read/modify the following two
fields:
CTRLTYPE: This field allows a load balancer to identify session-ID messages. Its value (two bytes) varies according to the session-ID message type: 0x0c05 for a request, 0x0d05 for a response, or 0x0c15 for
an indication.
SESSION-ID: This field consists of 16 bytes that the load balancer can freely modify according to its
requirements.
TCP Payload for Session-ID Message
00

01

02

03

04

05

06

07

08

0xDD

0x54

0xDD

0x55

CTRLTYPE

RTI reserved

19

20

21

22

23

24

25

26

27

35

36

37

38

39

40

41

42

43

RTI reserved

16

17

18

09

10

11

12

13

14

15

28

29

30

31

44

45

46

47

RTI reserved

32

33

34

SESSION-ID

To ensure all the TCP connections within the same session are directed to the same server, you must configure your load balancer to perform the two following actions:
Modify the SESSION-ID field in the session-id response with a value that identifies the session within the
load balancer.
Make the load-balancing decision according to the value of the SESSION-ID field in the session-ID indication.

35.1.6 TCP/TLS Transport Properties
Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t describes the TCP and TLS transport
properties.

1002

35.1.6 TCP/TLS Transport Properties

Note: To use TLS, you also need RTI TLS Support, which is a separate component.
Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Required
dds.transport.load_plugins
(Note: this does not take a prefix)

Comma-separated strings indicating the prefix names of all plugins that will be loaded
by Connext DDS. For example: “dds.transport.TCPv4.tcp1". You will use this
string as the prefix to the property names.
Note: you can load up to 8 plugins.
Only required if linking dynamically
Must be "nddstransporttcp".

library

This library must be in your library search path (pointed to by the environment variable
LD_LIBRARY_PATH on UNIX/Solaris systems, Path on Windows systems,
LIBPATH on AIX systems, DYLD_LIBRARY_PATH on OS X systems).
Only required if linking dynamically

create_function
Must be “NDDS_Transport_TCPv4_create”.
Only required if linking statically
create_function_ptr

Defines the function pointer to the TCP Transport Plugin creation function. Used for
loading TCP Transport Plugin statically.
Must be set to the NDDS_Transport_TCPv4_create function pointer.
Defines the function pointer to the TLS Support creation function. Used for loading
TLS Support libraries statically.

tls_create_function_ptr

Must be set to the RTITLS_ConnectionEndpointFactoryTLSv4_create function
pointer.
Note: In order to have effect, the tls_delete_function_ptr property must also be set.
Defines the function pointer to the TLS Support deletion function. Used for loading
TLS Support libraries statically.

tls_delete_function_ptr

Must be set to the RTITLS_ConnectionEndpointFactoryTLSv4_delete function
pointer.
Note: In order to have effect, the tls_create_function_ptr property must also be set.

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1003

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

aliases

Used to register the transport plugin returned by NDDS_Transport_TCPv4_create()
(as specified by .create_function) to the DomainParticipant. Aliases
should be specified as a comma-separated string, with each comma delimiting an alias.
Default: the transport prefix
Must be set to one of the following values:
NDDS_TRANSPORT_CLASSID_TCPV4_LAN
for TCP communication within a LAN
NDDS_TRANSPORT_CLASSID_TLSV4_LAN
for TLS communication within a LAN

parent.classid

NDDS_TRANSPORT_CLASSID_TCPV4_WAN
for TCP communication across LANs and firewalls
NDDS_TRANSPORT_CLASSID_TLSV4_WAN
for TLS communication across LAN and firewalls
Default: NDDS_TRANSPORT_CLASSID_TCPV4_LAN
Note: To use either TLS mode, you also need RTI TLS Support which is available for
purchase as a separate package.
Specifies the maximum number of buffers that Connext DDS can pass to the send()
function of the transport plugin.
The transport plugin send() API supports a gather-send concept, where the send() call
can take several discontiguous buffers, assemble and send them in a single message.
This enables Connext DDS to send a message from parts obtained from different
sources without first having to copy the parts into a single contiguous buffer.

parent.gather_send_
buffer_count_max

However, most transports that support a gather-send concept have an upper limit on
the number of buffers that can be gathered and sent. Setting this value will prevent
Connext DDS from trying to gather too many buffers into a send call for the
transport plugin.

Connext DDS requires all transport-plugin implementations to support a gather-send
of least a minimum number of buffers. This minimum number is defined as NDDS_
TRANSPORT_PROPERTY_GATHER_SEND_BUFFER_COUNT_MIN.
Default: 128

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1004

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

parent.message_size_max

The maximum size of a message in bytes that can be sent or received by the transport
plugin.
Default: 65536
A list of strings, each identifying a range of interface addresses that can be used by the
transport.
Interfaces must be specified as comma-separated strings, with each comma delimiting
an interface.

parent.allow_interfaces_list

For example: 10.10.*, 10.15.*
If the list is non-empty, this "white" list is applied before parent.deny_interfaces_list
(Section below).
Default: All available interfaces are used.
A list of strings, each identifying a range of interface addresses that will not be used by
the transport.
If the list is non-empty, deny the use of these interfaces.

parent.deny_interfaces_list

Interfaces must be specified as comma-separated strings, with each comma delimiting
an interface.
For example: 10.10.*
This "black" list is applied after parent.allow_interfaces_list (Section above) and
filters out the interfaces that should not be used.
Default: No interfaces are denied
Size, in bytes, of the send buffer of a socket used for sending. On most operating
systems, setsockopt() will be called to set the SENDBUF to the value of this
parameter.

send_socket_buffer_size

This value must be greater than or equal to parent.message_size_max (Section
above), or -1.
When set to -1, setsockopt() (or equivalent) will not be called to size the send buffer
of the socket.
The maximum value is operating system-dependent.
Default: 131072

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1005

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Size, in bytes, of the receive buffer of a socket used for receiving.
On most operating systems, setsockopt() will be called to set the RECVBUF to the
value of this parameter.

recv_socket_buffer_size

This value must be greater than or equal to parent.message_size_max (Section on
the previous page), or -1.
When set to -1, setsockopt() (or equivalent) will not be called to size the receive buffer
of the socket.
The maximum value is operating-system dependent.
Default: 131072
Prevents the transport plugin from using the IP loopback interface.
This property is ignored when parent.classid (Section on page 1004) is NDDS_
TRANSPORT_CLASSID_TCPV4_WAN or NDDS_TRANSPORT_
CLASSID_TLSV4_WAN.
Two values are allowed:

ignore_loopback_interface

0: Enable local traffic via this plugin. The plugin will only use and report the IP
loopback interface only if there are no other network interfaces (NICs) up on the
system.
1: Disable local traffic via this plugin. This means “do not use the IP loopback
interface, even if no NICs are discovered.” This setting is useful when you want
applications running on the same node to use a more efficient plugin like shared
memory instead of the IP loopback.
Default: 1

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1006

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Prevents the transport plugin from using a network interface that is not reported as
RUNNING by the operating system.
The transport checks the flags reported by the operating system for each network
interface upon initialization. An interface which is not reported as UP will not be used.
This property allows the same check to be extended to the IFF_RUNNING flag
implemented by some operating systems. The RUNNING flag is defined to mean that
"all resources are allocated" and may be off if no link is detected (e.g., the network
cable is unplugged).
ignore_nonrunning_interfaces
Two values are allowed:
0: Do not check the RUNNING flag when enumerating interfaces, just make sure the
interface is UP.
1: Check the flag when enumerating interfaces, and ignore those that are not reported
as RUNNING. This can be used on some operating systems to cause the transport to
ignore interfaces that are enabled but not connected to the network.
Default: 1
Mask for the transport priority field. This is used in conjunction with transport_
priority_mapping_low (Section below)/transport_priority_mapping_high (Section
below) to define the mapping from DDS transport priority to the IPv4 TOS field.
Defines a contiguous region of bits in the 32-bit transport priority value that is used to
generate values for the IPv4 TOS field on an outgoing socket.
transport_priority_mask

For example, the value 0x0000ff00 causes bits 9-16 (8 bits) to be used in the mapping.
The value will be scaled from the mask range (0x0000 -0xff00 in this case) to the
range specified by low and high.
If the mask is set to zero, then the transport will not set IPv4 TOS for send sockets.
Default: 0
Sets the low and high values of the output range to IPv4 TOS.

transport_priority_mapping_low

These values are used in conjunction with transport_priority_mask (Section above) to
define the mapping from DDS transport priority to the IPv4 TOS field. Defines the
low and high values of the output range for scaling.
Note that IPv4 TOS is generally an 8-bit value.

transport_priority_mapping_high

Default transport_priority_mapping_low: 0
Default transport_priority_mapping_high: 0xFF

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1007

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

server_socket_backlog

The backlog parameter determines what is the maximum length of the queue of
pending connections.
Default: 5
Required for WAN communication (see note below)
Public IP address and port (WAN address and port) (separated with ‘:’ ) associated
with the transport instantiation.
For example: 10.10.9.10:4567
This field is used only when parent.classid (Section on page 1004) is NDDS_
TRANSPORT_CLASSID_TCPV4_WAN or NDDS_TRANSPORT_
CLASSID_TLSV4_WAN.

public_address

The public address and port are necessary to support communication over WAN that
involves Network Address Translators (NATs). Typically, the address is the public
address of the IP router that provides access to the WAN. The port is the IP router port
that is used to reach the private server_bind_port (Section on the facing page) inside
the LAN from the outside. This value is expressed as a string in the form: ip[:port],
where ip represents the IPv4 address and port is the external port number of the router.
Host names are not allowed in the public_address because they may resolve to an
internet address that is not what you want (i.e., ‘localhost’ may map to your local IP or
to 127.0.0.1).
Note: If you are using an asymmetric configuration, public_address does not have to
be set for the non-public peer.

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1008

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Private IP port (inside the LAN) used by the transport to accept TCP connections.
If this property is set to zero, the transport will disable the internal server socket,
making it impossible for external peers to connect to this node. In this case, the node is
considered unreachable and will communicate only using the asymmetric mode with
other (reachable) peers.
server_bind_port

For WAN communication, this port must be forwarded to a public port in the NATenabled router that connects to the outer network.
The server_bind_port cannot be shared among multiple participants on a common host.
On most operating systems, attempting to reuse the same server_bind_port for multiple
participants on a common host will result in a "port already in use" error. However,
Windows systems will not recognize if the server_bind_port is already in use;
therefore care must be taken to properly configure Windows systems.
Default: 7400
Allocation settings applied to read buffers.
These settings configure the initial number of buffers, the maximum number of buffers
and the buffers to be allocated when more buffers are needed.
Default:

read_buffer_allocation
read_buffer_allocation.initial_count = 2
read_buffer_allocation.max_count = -1 (unlimited)
read_buffer_allocation.incremental_count = -1 (number of buffers will keep
doubling on each allocation until it reaches max_count)

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1009

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Allocation settings applied to buffers used for asynchronous (non-blocking) write.
These settings configure the initial number of buffers, the maximum number of buffers
and the buffers to be allocated when more buffers are needed.
Default:
write_buffer_allocation.initial_count = 4
write_buffer_allocation.max_count = 1000
write_buffer_allocation

write_buffer_allocation.incremental_count = 10
Note that for the write buffer pool, the max_count is not set to unlimited. This is to
avoid having a fast writer quickly exhaust all the available system memory, in case of a
temporary network slowdown. When this write buffer pool reaches the maximum, the
low-level send command of the transport will fail; at that point Connext DDS will
take the appropriate action (retry to send or drop it), according to the application’s QoS
(if the transport is used for reliable communication, the data will still be sent
eventually).
Allocation settings applied to buffers used to serialize and send control messages.
These settings configure the initial number of buffers, the maximum number of buffers
and the buffers to be allocated when more buffers are needed.
Default:

control_buffer_allocation
control_buffer_allocation.initial_count = 2
control_buffer_allocation.max_count = -1 (unlimited)
control_buffer_allocation.incremental_count = -1 (number of buffers will keep
doubling on each allocation until it reaches max_count)
Allocation settings applied to control messages.
These settings configure the initial number of messages, the maximum number of
messages and the messages to be allocated when more messages are needed.
Default:
control_message_allocation
control_message_allocation.initial_count = 2
control_message_allocation.max_count = -1 (unlimited)
control_message_allocation.incremental_count = -1 (number of messages will keep
doubling on each allocation until it reaches max_count)

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1010

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Allocation settings applied to control messages attributes.
These settings configure the initial number of attributes, the maximum number of
attributes and the attributes to be allocated when more attributes are needed.
Default:
control_attribute_allocation
control_attribute_allocation.initial_count = 2
control_attribute_allocation.max_count = -1 (unlimited)
control_attribute_allocation.incremental_count = -1 (number of attributes will keep
doubling on each allocation until it reaches max_count)
Forces asynchronous send. When this parameter is set to 0, the TCP transport will
attempt to send data as soon as the internal send() function is called. When it is set to
1, the transport will make a copy of the data to send and enqueue it in an internal send
buffer. Data will be sent as soon as the low-level socket buffer has space.

force_asynchronous_send

Normally setting it to 1 delivers better throughput in a fast network, but will result in a
longer time to recover from various TCP error conditions. Setting it to 0 may cause the
low-level send() function to block until the data is physically delivered to the lower
socket buffer. For an application writing data at a very fast rate, it may cause the caller
thread to block if the send socket buffer is full. This could produce lower throughput in
those conditions (the caller thread could prepare the next packet while waiting for the
send socket buffer to become available).
Default: 0
The maximum size of a TCP segment.
This parameter is only supported on Linux architectures.

max_packet_size

By default, the maximum size of a TCP segment is based on the network MTU for
destinations on a local network, or on a default 576 for destinations on non-local
networks. This behavior can be changed by setting this parameter to a value between 1
and 65535.
Default: -1 (default behavior)

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1011

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Configures the sending of KEEP_ALIVE messages in TCP.
Setting this value to 1, causes a KEEP_ALIVE packet to be sent to the remote peer if a
long time passes with no other data sent or received.
This feature is implemented only on architectures that provide a low-level
implementation of the TCP keep-alive feature.
enable_keep_alive

On Windows systems, the TCP keep-alive feature can be globally enabled through the
system’s registry: \HKEY_LOCAL_MACHINE\SYSTEM\
CurrentControlSet\Tcpip\Parameters.
Refer to MSDN documentation for more details.
On Solaris systems, most of the TCP keep-alive parameters can be changed though the
kernel properties.
Default: 0
Specifies the interval of inactivity in seconds that causes TCP to generate a KEEP_
ALIVE message.

keep_alive_time

This parameter is only supported on Linux and Mac architectures.
Default: -1 (OS default value)
Specifies the interval in seconds between KEEP_ALIVE retries.

keep_alive_interval

This parameter is only supported on Linux architectures.
Default: -1 (OS default value)
The maximum number of KEEP_ALIVE retries before dropping the connection.

keep_alive_retry_count

This parameter is only supported on Linux architectures.
Default: -1 (OS default value)

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1012

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Changes the default OS TCP User Timeout configuration. If set to a value greater than
0, it specifies the maximum amount of time in seconds that transmitted data may remain
unacknowledged before TCP will forcibly close the corresponding connection and
return ETIMEDOUT to the application.
user_timeout
If set to 0, TCP Transport plugin will use the system default.
Currently this feature is supported only on Linux 2.6.37 and higher platforms.
Default: 0 (use system's default).
Configures the connection liveliness feature. See Connection Liveliness (Section
35.1.6.1 on page 1020).
Defaults:
connection_liveliness

connection_liveliness.enable: 0
connection_liveliness.lease_duration: 10
connection_liveliness.assertions_per_lease_duration: 3
Configures the event thread used by the TCP Transport plugin for providing some
features.
Defaults:

event_thread

event_thread.priority: THREAD_PRIORITY_DEFAULT
event_thread.stack_size: THREAD_STACK_SIZE_DEFAULT
event_thread.mask: PRIORITY_ENFORCE | STDIO
Disables the TCP nagle algorithm.

disable_nagle

When this property is set to 1, TCP segments are always sent as soon as possible,
which may result in poor network utilization.
Default: 0

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1013

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Bitmap that specifies the verbosity of log messages from the transport.
Logging values:
-1 (0xffffffff): do not change the current verbosity
0x00: silence
0x01: errors
0x02: warnings
0x04: local
0x08: remote
0x10: period
logging_verbosity_bitmap

0x80: other (used for control protocol tracing)
0x9F: all (errors, warnings, local, remote, period, and other)
You can combine these values by logically ORing them together.
Default: -1
Note: the logging verbosity is a global property shared across multiple instances of the
TCP transport. If you create a new TCP Transport instance with logging_verbosity_
bitmap different than -1, the change will affect all the other instances as well.
The default TCP transport verbosity is errors and warnings.
Note: The option of 0x80 (other) is used only for tracing the internal control protocol.
Since the output is very verbose, this feature is enabled only in the debug version of
the TCP Transport library
(libnddstransporttcpd.so / LIBNDDSTRANSPORTD.LIB).

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1014

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Maximum number of outstanding connection cookies allowed by the transport when
acting as server.
A connection cookie is a token provided by a server to a client; it is used to establish a
data connection. Until the data connection is established, the cookie cannot be reused
by the server.
outstanding_connection_cookies

To avoid wasting memory, it is good practice to set a cap to the maximum number of
connection cookies (pending connections).
When the maximum value is reached, a client will not be able to connect to the server
until new cookies become available.
Range: 1 or higher, or -1 (which means an unlimited number).
Default: 100
Maximum lifespan (in seconds) of the cookies associated with pending connections.

outstanding_connection_
cookies_life_span

If a client does not connect to the server before the lifespan of its cookie expires, it will
have to request a new cookie.
Range: 1 second or higher, or -1
Default: -1, which means an unlimited amount of time (effectively disabling the
feature).

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1015

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Controls the maximum time (in seconds) the low-level sendto() function is allowed to
block the caller thread when the TCP send buffer becomes full.
If the bandwidth used by the transport is limited, and the sender thread tries to push
data faster than the OS can handle, the low-level sendto() function will block the caller
until there is some room available in the queue. Limiting this delay eliminates the
possibility of deadlock and increases the response time of the internal DDS thread.
This property affects both CONTROL and DATA streams. It only affects
SYNCHRONOUS send operations. Asynchronous sends never block a send
operation.
send_max_wait_sec

For synchronous send() calls, this property limits the time the DDS sender thread can
block for a full send buffer. If it is set too large, Connext DDS not only won't be able
to send more data, it also won't be able to receive any more data because of an internal
resource mutex.
Setting this property to 0 causes the low-level function to report an immediate failure if
the TCP send buffer is full.
Setting this property to -1 causes the low-level function to block forever until space
becomes available in the TCP buffer.
Default: 3 seconds.
Configures the socket monitoring API used by the transport. This property can have
the following values:
SELECT: The transport uses the POSIX select API to monitor sockets.
WINDOWS_IOCP: The transport uses Windows I/O completion ports to monitor
sockets. This value only applies to Windows systems.

socket_monitoring_kind

WINDOWS_WAITFORMULTIPLEOBJECTS: The transport uses the API
WaitForMultipleObjects to monitor sockets. This value only applies to Windows
systems.
Default: SELECT
Note: The value selected for this property may affect transport performance and
scalability. On Windows systems, using WINDOWS_IOCP provides the best
performance and scalability.

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1016

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Configures I/O completion ports when socket_monitoring_kind (Section on the
previous page) is set to WINDOWS_IOCP.

windows_iocp

This setting configures the number of threads the plugin creates to process I/O
completion packets (thread_pool_size) and the number of those threads that the
operating system can allow to concurrently run (concurrency_value).
Defaults:
windows_iocp.thread_pool_size: 2
windows_iocp.concurrency_value: 1
When set to 1, the TCP Transport Plugin will perform a session negotiation that will
help external load balancers identify all the connections associated with a particular
session between two Connext DDS applications. This keeps the connections from
being divided among multiple servers and ensures proper communication.

negotiate_session_id

For more information about this property, see Support for External Hardware Load
Balancers in TCP Transport Plugin (Section 35.1.5 on page 1000).
Default: 0
Note: The value of this property must be consistent among all the applications running
the TCP Transport Plugin. If two applications have a different value for this property,
they may not communicate.
Specifies a timeout for the negotiation of a new connection accepted by the server.
When the TCP Transport plugin accepts a new connection, some TCP Transport
plugin-specific negotiation is exchanged between the client and the server.

server_connection_negotiation_
timeout

This property controls the maximum time (in seconds) the negotiation for a server
connection can remain in progress. If the negotiation has not completed after the
specified timeout, the connection will be closed. Then the TCP Transport plugin can
restart the process of establishing that connection.
Range: 1 second or higher.
Default: 10 seconds

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1017

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

Specifies a timeout for the initial handshake for a connection.
Some of the TCP Transport plugin configurations (e.g., when using TLS over TCP)
require an initial handshake for each established connection.
initial_handshake_timeout :

This property controls the maximum time (in seconds) the initial handshake for a
connection can remain in progress. If the handshake has not completed after the
specified timeout, the connection will be closed. Then the TCP Transport plugin can
restart the process of establishing and handshaking that connection.
Range: 1 second or higher.
Default: 10 seconds

tls.verify.ca_file

A string that specifies the name of file containing Certificate Authority certificates. File
should be in PEM format. See the OpenSSL manual page for SSL_load_verify_
locations for more information.
To enable TLS, ca_file or ca_path is required; both may be specified (at least
one is required).

tls.verify.ca_path

A string that specifies paths to directories containing Certificate Authority certificates.
Files should be in PEM format and follow the OpenSSL-required naming conventions.
See the OpenSSL manual page for SSL_CTX_load_verify_locations for more
information.
To enable TLS, ca_file or ca_path is required; both may be specified (at least
one is required).

tls.verify.verify_depth

Maximum certificate chain length for verification.
Name of the file containing the Certificate Revocation List.

tls.verify.crl_file
File should be in PEM format.
String containing an identifying certificate (in PEM format) or certificate chain
(appending intermediate CA certs in order).
tls.identity.certificate_chain

An identifying certificate is required for secure communication. The string must
be sorted starting with the certificate to the highest level (root CA). If this is specified,
certificate_chain_file must be empty.

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1018

35.1.6 TCP/TLS Transport Properties

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

File containing identifying certificate (in PEM format) or certificate chain (appending
intermediate CA certs in order).
tls.identity.certificate_chain_file

An identifying certificate is required for secure communication. The file must be
sorted starting with the certificate to the highest level (root CA). If this is specified,
certificate_chain must be empty.
Optionally, a private key may be appended to this file. If no private key option is
specified, this file will be used to load a private key.

tls.identity.private_key_password

A string that specifies the password for private key.
String containing private key (in PEM format).

tls.identity.private_key

At most one of private_key and private_key_file may be specified. If no private key
is specified (all values are NULL), the private key will be read from the certificate
chain file.
File containing private key (in PEM format).

tls.identity.private_key_file

At most one of private_key and private_key_file may be specified. If no private key
is specified (all values are NULL), the private key will be read from the certificate
chain file.
String containing additional RSA private key (in PEM format).

tls.identity.rsa_private_key

For use if both an RSA and non-RSA key are required for the selected cipher. At most
one of rsa_private_key and rsa_private_key_file may be specified.
At most one of rsa_private_key and rsa_private_key_file may be specified.
File containing additional RSA private key (in PEM format).

tls.identity.rsa_private_key_file

For use if both an RSA and non-RSA key are required for the selected cipher. At most
one of rsa_private_key and rsa_private_key_file may be specified.
At most one of rsa_private_key and rsa_private_key_file may be specified.

tls.cipher.cipher_list

List of available (D)TLS ciphers. See the OpenSSL manual page for SSL_set_cipher_
list for more information on the format of this string.

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1019

35.1.6.1 Connection Liveliness

Table 35.1 Properties for NDDS_Transport_TCPv4_Property_t
Property Name
(prefix with
‘dds.transport.TCPv4.tcp1.’)

Description

1

List of available Diffie-Hellman (DH) key files. For example: "foo.h:2048,bar.h:1024"
means:
dh_param_files[0].file = foo.pem,
tls.cipher.dh_param_files

dh_param_files[0].bits = 2048,
dh_param_files[1].file = bar.pem,
dh_param_files[1].bits = 1024

tls.cipher.engine_id

ID of OpenSSL cipher engine to request.

35.1.6.1 Connection Liveliness
The connection_liveliness property configures the connection liveliness feature. When enabled, the TCP
Transport plugin will periodically exchange some additional control traffic (liveliness requests/responses)
over one of the connections between the TCP Client and Server. This traffic allows determining if a that
connection is not alive anymore, and thus proceed to its close. This avoids depending on the OS notification about the status of the connection, potentially decreasing the time to reestablish lost connections.
The following parameters can be configured:
l
l

l

connection_liveliness.enable: Enables or disables the feature.
connection_liveliness.lease_duration: In seconds, the timeout by which the connection liveliness
must be asserted or the connection will be considered not alive. It is also used also as the period
between connection liveliness checks. Therefore, the maximum time before a connection is marked
as not alive is 2*connection_liveliness.lease_duration.
connection_liveliness.assertions_per_lease_duration: The number of liveliness requests send per
each lease duration. Increasing this value will increase the overhead send into the network, but it
will also make the connection liveliness mechanism more robust.

This feature relies on the creation on an additional thread in the TCP Transport Plugin (the event thread).
For more information about how to configure this thread, see the event_thread in Table 35.1 Properties
for NDDS_Transport_TCPv4_Property_t.

1Assuming you used ‘dds.transport.TCPv4.tcp1’ as the alias to load the plugin. If not, change the prefix to match the string

used with dds.transport.load_plugins. This prefix must begin with 'dds.transport.'

1020

35.1.6.1 Connection Liveliness

Enabling this feature breaks backwards compatibility with TCP Transport plugins that do not
include this feature.

1021

Part 9: RTI Monitoring Library
RTI Monitoring Library is a plug-in that enables RTI Connext DDS™ applications to provide monitoring data. The monitoring data can be visualized with RTI Monitor, a separate GUI application
that can run on the same host as Monitoring Library or on a different host.
Connext DDS notifies Monitoring Library every time an entity is created/deleted or a QoS is
changed. Monitoring Library periodically queries the status of all Connext DDS entities. You can
enable/disable monitoring by setting values in the DomainParticipant’s PropertyQosPolicy (programmatically or through an XML QoS profile).

1022

Part 9: RTI Monitoring Library

This part of the User’s Manual includes:
l

Using Monitoring Library in Your Application (Section Chapter 36 on page 1024)

l

Configuring Monitoring Library (Section Chapter 37 on page 1034)

1023

Chapter 36 Using Monitoring Library in
Your Application
36.1 Enabling Monitoring
There are two ways to enable monitoring in your application:
l

l

Method 1—Change the Participant QoS to Automatically Load the Dynamic Monitoring
Library (Section 36.1.1 on the next page)
Method 2—Change the Participant QoS to Specify the Monitoring Library Create Function
Pointer and Explicitly Load the Monitoring Library (Section 36.1.2 on the next page)

Notes:
l

l

l

The libraries that you will need for Monitoring are listed in the RTI Connext DDS Core
Libraries Platform Notes.
If your original application has made modifications to either the ParticipantQos resource_
limits.type_code_max_serialized_length or any of the transport's default settings to enable
large type code or large data, refer to What Monitoring Topics are Published? (Section 36.3
on page 1031) for additional QoS modifications that may be needed.
Monitoring Library creates internal DataWriters to publish monitoring data by making modifications based on the default DataWriter QoS settings. If you have made changes to the
default DataWriter QoS, especially if you have increased/decreased the initial or maximum
DDS sample/instance values, Monitoring Library may have trouble creating DataWriters to
publish monitoring data, or it may limit the number of statistics that you can publish through
the internal monitoring writers. If this is true for your case, you may want to specify the qos_
library and qos_profile that will be used to create these internal writers for publishing monitoring data, to avoid being impacted by default DataWriter QoS settings. See Configuring
Monitoring Library (Section Chapter 37 on page 1034) for details.

1024

36.1.1 Method 1—Change the Participant QoS to Automatically Load the Dynamic Monitoring Library

36.1.1 Method 1—Change the Participant QoS to Automatically Load the
Dynamic Monitoring Library
If all of the following are true, you can enable monitoring simply by changing your participant QoS (otherwise, use Method 2—Change the Participant QoS to Specify the Monitoring Library Create Function
Pointer and Explicitly Load the Monitoring Library (Section 36.1.2 below)):
l

Your application is linked to dynamic Connext DDS libraries, or you are using Java or .Net, and

l

You will run your application on a Linux, Windows, Solaris, AIX or Mac OS platform, and

l

You are NOT linking in an additional monitoring library into your application at link time (you let
the middleware load the monitoring library for you automatically as needed).

If you change the QoS in an XML file as shown below, you can enable/disable monitoring without recompiling. If you change the QoS in your source code, you may need to recompile every time you enable/disable monitoring.
If you need to change the participant QoS by hand, refer to the definition of BuiltinQosLib::Generic.Monitoring.Common in /resource/xml/BuiltinProfiles.documentationONLY.xml for the values you should set.
Example XML to enable monitoring:




rti.monitor.library
rtimonitoring


rti.monitor.create_function
RTIDefaultMonitor_create





36.1.2 Method 2—Change the Participant QoS to Specify the Monitoring
Library Create Function Pointer and Explicitly Load the Monitoring
Library
If any of the following are true, you must change the Participant QoS to enable monitoring and explicitly
load the correct version of Monitoring Library at compile time:

1025

36.1.2.1 Method 2-A: Change the Participant QoS by Specifying the Monitoring Library Create Function

l

Your application is linked to the static version of Connext DDS libraries.

l

You are NOT running your application on Linux, Windows, Solaris, AIX or Mac OS platforms.

l

You want to explicitly link in the monitoring library (static or dynamic) into your application.

There are two ways to do this:
l

l

Method 2-A: Change the Participant QoS by Specifying the Monitoring Library Create Function
Pointer in Source Code (Section 36.1.2.1 below): Applies to most users who cannot use Method 1
and do not mind changing/recompiling source code every time you enable/disable monitoring, or
whose system does not support setting environment variables programmatically. Participant QoS
must be defined in source code with this approach.
Method 2-B: Change the Participant QoS by Specifying the Monitoring Library Create Function
Pointer in an Environment Variable (Section 36.1.2.2 on page 1029): Applies to users who cannot
use Method 1 and want to specify the create function pointer via an environment variable. This
approach allows the Participant QoS to be defined in an XML file or in source code.

36.1.2.1 Method 2-A: Change the Participant QoS by Specifying the Monitoring Library
Create Function Pointer in Source Code
1. Modify your Connext DDS application based on the following examples.
Traditional C++ Example:
#include "ndds/ndds_cpp.h"
#include "monitor/monitor_common.h"
extern "C" int publisher_main(int domainId, int sample_count)
{
...
DDSDomainParticipant *participant = NULL;
DDS_DomainParticipantQos participant_qos;
char valueBuffer[17];
/* Get default QoS */
retcode =
DDSTheParticipantFactory->get_default_participant_qos(
participant_qos);
if (retcode != DDS_RETCODE_OK) {
/*Error*/
}
/* This property indicates that the DomainParticipant
has monitoring turned on. The property name MUST be
"rti.monitor.library". The value can be anything.*/
retcode = DDSPropertyQosPolicyHelper::add_property(
participant_qos.property,
"rti.monitor.library", "rtimonitoring", DDS_BOOLEAN_FALSE);

1026

36.1.2.1 Method 2-A: Change the Participant QoS by Specifying the Monitoring Library Create Function

if (retcode != DDS_RETCODE_OK) {
/*Error*/
}
/* The property name "rti.monitor.create_function"
indicates the entry point for the monitoring library.
The value MUST be the value of the function pointer of
RTIDefaultMonitor_create */
sprintf(valueBuffer, "%p", RTIDefaultMonitor_create);
retcode = DDSPropertyQosPolicyHelper::add_property(
participant_qos.property,
"rti.monitor.create_function_ptr",
valueBuffer, DDS_BOOLEAN_FALSE);
if (retcode!= DDS_RETCODE_OK) {
/* Error */
}
/* Create DomainParticipant with participant_qos */
participant = DDSTheParticipantFactory->create_participant(
domainId, participant_qos,NULL /* listener */,
DDS_STATUS_MASK_NONE);
if (participant == NULL) {
/* Error */
}
...

Modern C++ Example:
#include "rti/rti.hpp" // include all the modern C++ API
#include "monitor/monitor_common.h" // for RTIDefaultMonitor_create
//...
using rti::core::policy::Property;
// Get the property policy from the default DomainParticipantQos
auto participant_qos = dds::core::QosProvider::Default().participant_qos();
auto property_policy = participant_qos.policy();
// This property turns monitoring on
property_policy.set(Property::Entry("rti.monitor.library",
"rtimonitoring"));
// This property specifies the entry point (function pointer) for the
// monitoring library.
std::ostringstream monitor_function_to_str;
monitor_function_to_str << RTIDefaultMonitor_create;
property_policy.set(Property::Entry(
"rti.monitor.create_function_ptr", monitor_function_to_str.str()));

1027

36.1.2.1 Method 2-A: Change the Participant QoS by Specifying the Monitoring Library Create Function

C Example:
#include "ndds/ndds_c.h"
#include "monitor/monitor_common.h"
...
extern "C" int publisher_main(int domainId, int sample_count)
{
DDS_DomainParticipantFactory *factory = NULL;
struct DDS_DomainParticipantQos participantQos =
DDS_DomainParticipantQos_INITIALIZER;
char valueBuffer[17];
DDS_DomainParticipant *participant = NULL;
factory = DDS_DomainParticipantFactory_get_instance();
if (factory == NULL) {
/* error */
}
if (DDS_DomainParticipantFactory_get_default_participant_qos(
factory, &participantQos) != DDS_RETCODE_OK) {
/* error */
}
/* This property indicates that the DomainParticipant has
monitoring turned on. The property name MUST be
“rti.monitor.library”. The value can be anything.*/
if (DDS_PropertyQosPolicyHelper_add_property(
&participantQos.property,
"rti.monitor.library", "rtimonitoring",
DDS_BOOLEAN_FALSE) != DDS_RETCODE_OK) {
/* error */
}
/* The property name "rti.monitor.create_function_ptr"
indicates the entry point for the monitoring library.
The value MUST be the value of the function pointer
of RTIDefaultMonitor_create */
sprintf(valueBuffer, "%p", RTIDefaultMonitor_create);
if (DDS_PropertyQosPolicyHelper_add_property(
&participantQos.property,
"rti.monitor.create_function_ptr",valueBuffer,
DDS_BOOLEAN_FALSE) != DDS_RETCODE_OK) {
/* error */
}
/* create DomainParticipant with participantQos */
participant=
DDS_DomainParticipantFactory_create_participant(
factory, domainId, &participantQos,
NULL /* listener */,
DDS_STATUS_MASK_NONE);
if (participant == NULL) {
/* error */

1028

36.1.2.2 Method 2-B: Change the Participant QoS by Specifying the Monitoring Library Create Function

}
DDS_DomainParticipantQos_finalize(&participantQos);
...

Note:
l

In the above code, you may notice that valueBuffer is initialized to 17 characters. This is
because a pointer (RTIDefaultMonitor_create) is at most 8 bytes (on a 64-bit system) and
it takes two characters to represent a byte in hex. So the total size must be:
(2 * 8 characters) + 1 null-termination character = 17 characters.

2. Link the Monitoring Library for your platform into your application at compile time (the Monitoring
libraries are listed in the RTI Connext DDS Core Libraries Platform Notes).
The kind of monitoring library that you link into your application at compile time must be consistent
with the kind of Connext DDS libraries that you are linking into your application (static/dynamic,
release/debug version of the libraries).
On Windows systems:If you are linking a static monitoring library, you will also need to link in
Psapi.lib at compile time.
36.1.2.2 Method 2-B: Change the Participant QoS by Specifying the Monitoring Library
Create Function Pointer in an Environment Variable
This is similar to Method 2-A, but if you specify the function pointer value for rti.monitor.create_function_ptr in an environment variable that is set programmatically, you can specify your QoS either in an
XML file or in source code. If you specify the QoS in an XML file, you can enable/disable monitoring
without recompiling. If you change the QoS in your source code, you may need to recompile every time
you enable/disable monitoring.
1. In XML, enable monitoring by setting the rti.monitor.create_function_ptr property to an environment variable. In our example, the variable is named RTIMONITORFUNCPTR.

1029

36.1.2.2 Method 2-B: Change the Participant QoS by Specifying the Monitoring Library Create Function





rti.monitor.library
rtimonitoring


rti.monitor.create_function_ptr
$(RTIMONITORFUNCPTR)





2. In the DDS application that links in the monitoring library, get the function pointer of RTIDefaultMonitor_create and write it to the same environment variable you named in Step 1 and create
a DomainParticipant by using the XML profile specified in Step 1. (Setting of the environment variable must appear in the application before it creates the DomainParticipant using the profile from
Step 1.)
Here is an example in C:
#include 
#include 
#include "monitor/monitor_common.h"
...
char putenvBuffer[34];
int putenvReturn;
putenvBuffer[0] = '\0';
sprintf(putenvBuffer, "RTIMONITORFUNCPTR=%p",
RTIDefaultMonitor_create);
putenvReturn = putenv(putenvBuffer);
if (putenvReturn) {
printf(
"Error: couldn't set env variable for RTIMONITORFUNCPTR. "
"error code: %d\n", putenvReturn );
}
...
/* create DomainParticipant using XML profile from Step 1 */
...

Note: In the above code, you may notice that putenvBuffer is initialized to 34 characters. This is
because a pointer (RTIDefaultMonitor_create) is at most 8 bytes (on a 64-bit system) and it takes 2
characters to represent a byte in hex. So the total size must be: strlen(RTIMONITORFUNCPTR) +
(2 * 8 characters) + 1 null-termination character = 17 + 16 + 1 = 34 characters
3. Link the Monitoring Library for your platform into your application at compile time (the Monitoring
libraries are listed in the RTI Connext DDS Core Libraries Platform Notes).

1030

36.2 How does Monitoring Library Work?

The kind of monitoring library that you link into your application at compile time must be consistent
with the kind of Connext DDS libraries that you are linking into your application (static/dynamic,
release/debug version of the libraries).
On Windows systems: If you are linking a static monitoring library, you will also need to link in
Psapi.lib at compile time.

36.2 How does Monitoring Library Work?
Monitoring Library works by creating DDS Topics that publish information about the other DDS entities
contained in the same operating system process. The Topics can be created inside of the first DomainParticipant that enables the library (the default). Or they may be created in a separate DomainParticipant if
the rti.monitor.config.new_participant_domain_id property is used. Use cases for this latter configuration include controlling the domain ID on which this information is exchanged (for example to
ensure that this data does not interfere with production topics) as well as the ability to specify the QoS that
is used for the DomainParticipant (through the rti.monitor.config.qos_library and rti.monitor.config.qos_profile properties). It may be desirable to specify the QoS for Monitoring Library's
DomainParticipant if the information will be consumed on a different transport or simply to enable the feature but keep it as isolated from the production system as possible.

36.3 What Monitoring Topics are Published?
Two categories of predefined monitoring topics are sent out:
l

l

Descriptions are published when an entity is created or deleted, or there are QoS changes (see Table
36.1 Descriptions (QoS and Other Static System Information)).
Entity Statistics are published periodically (see Table 36.2 Entity Statistics (Statuses, Aggregated
Statuses, CPU and Memory Usage)).

Table 36.1 Descriptions (QoS and Other Static System Information)
Topic Name

Topic Contents

rti/dds/monitoring/domainParticipantDescription

DomainParticipant QoS and other static information

rti/dds/monitoring/topicDescription

Topic QoS and other static information

rti/dds/monitoring/publisherDescription

Publisher QoS and other static information

rti/dds/monitoring/subscriberDescription

Subscriber QoS and other static information

rti/dds/monitoring/dataReaderDescription

DataReader QoS and other static information

rti/dds/monitoring/dataWriterDescription

DataWriter QoS and other static information

1031

36.4 Enabling Support for Large Type-Code (Optional)

Table 36.2 Entity Statistics (Statuses, Aggregated Statuses, CPU and Memory Usage)
Topic Name

Topic Contents

rti/dds/monitoring/domainParticipantEntityStatistics

Number of entities discovered in the system, CPU and memory usage of the
process

rti/dds/monitoring/dataReaderEntityStatistics

DataReader statuses

rti/dds/monitoring/dataWriterEntityStatistics

DataWriter statuses

rti/dds/monitoring/topicEntityStatistics

Topic statuses

rti/dds/monitoring/
dataReaderEntityMatchedPublicationStatistics

DataReader statuses calculated on a per discovered matching writer basis

rti/dds/monitoring/
dataWriterEntityMatchedSubscriptionStatistics

DataWriter statuses calculated on a per discovered matching reader basis

rti/dds/monitoring/
DataWriter statuses calculated on a per sending destination basis
dataWriterEntityMatchedSubscriptionWithLocatorStatistics

All monitoring data are sent out using specially created DataWriters with the above topics.
You can configure some aspects of Monitoring Library’s behavior, such as which monitoring topics to
turn on, which user topics to monitor, how often to publish the statistics topics, and whether to publish
monitoring data using (a) the participant created in the user’s application that has monitoring turned on or
(b) a separate participant created just for publishing monitoring data. See Configuring Monitoring Library
(Section Chapter 37 on page 1034).

36.4 Enabling Support for Large Type-Code (Optional)
Some monitoring topics have large type-code (larger than the default maximum type code serialized size
setting). If you use Monitor to display all the monitoring data, it already has all the monitoring types builtin and therefore it uses the default maximum type-code serialized size in the Connext DDS application and
there is no problem. However, if you are using any other tools to display monitoring data (such as RTI
Spreadsheet Add-in for Microsoft Excel, rtiddsspy, or writing your own application to subscribe to monitoring data), or if your user data-type has large type-code, you may need to increase the maximum typeobject serialized size setting in the DomainParticipantResourceLimitsQosPolicy.

1032

36.5 Troubleshooting Monitoring

36.5 Troubleshooting Monitoring
36.5.1 Buffer Allocation Error
Monitoring Library obtains the default DataWriter QoS from the Connext DDS application’s DomainParticipant. If the application has changed the default QoS Profile, either through application code or in an
XML file, Monitoring Library will use this new default QoS. In specific scenarios, the new default QoS
may cause your Connext DDS application to run out of memory and report error messages similar to these:
REDAFastBufferPool_growEmptyPoolEA: !allocate buffer of 1210632000 bytes
[D0012|ENABLE]REDAFastBufferPool_newWithNotification:!create fast buffer pool buffers
[D0012|ENABLE]PRESTypePluginDefaultEndpointData_createWriterPool:!create writer buffer pool
[D0012|ENABLE]WriterHistorySessionManager_new:!create newAllocator
[D0012|ENABLE]WriterHistoryMemoryPlugin_createHistory:!create sessionManager
[D0012|ENABLE]PRESWriterHistoryDriver_new:!create _whHnd
[D0012|ENABLE]PRESPsService_enableLocalEndpointWithCursor:!create WriterHistoryDriver
[D0012|ENABLE]PRESPsService_enableAllLocalEndpointsInGroupWithCursor:!enable endpoint
[D0012|ENABLE]PRESPsService_enableGroupWithCursor:!enableAllLocalEndpointsInGroupWithCursor
[D0012|ENABLE]PRESPsService_enableGroup:!enableGroupWithCursor
[D0012|ENABLE]RTIDefaultMonitorPublisher_enableEntitiesAndStartThreadI:!create enable
publisher
[D0012|ENABLE]RTIDefaultMonitorPublisher_onEventNotify:!create enable entities

To resolve this problem, either:
l

l

Configure Monitoring Library to use a non-default QoS Profile. For details, see Configuring Monitoring Library (Section Chapter 37 on page 1034).
Change the default QoS to have a lower value for DataWriter’s initial_samples; this field is part of
the ResourceLimitsQosPolicy.

1033

Chapter 37 Configuring Monitoring Library
You can control some aspects of Monitoring Library’s behavior by setting the PropertyQosPolicy
of the DomainParticipant, either via an XML QoS profile or in your application’s code prior to creating the DomainParticipant.
Two example QoS profiles are provided in
/connext_dds/qos/MONITORING_LIBRARY_QOS_PROFILES.xml:
l

CustomerExampleMonitoringLibrary::CustomerExampleMonitoringProfile
This is an example of how to enable Monitoring Library for your applications. It can be used
as a guide to enabling Monitoring Library quickly in your applications.

l

RTIMonitoringQosLibrary::RTIMonitoringQosProfile

This profile documents the QoS used by Monitoring Library. It can also be used as a starting point
if you want to tune QoS for Monitoring Library (normally not necessary). Use cases for this
include customizing DomainParticipant QoS (often the transports) to accommodate preferences or
environment. This same profile can also be used to subscribe to the Monitoring Library Topics.
This is useful in situations where the Monitoring Library information can be used directly by system components or it is not possible to use the RTI Monitor tool.
See the qos_library (Section on page 1036) and qos_profile (Section on page 1036) properties in
Table 37.1 Configuration Properties for Monitoring Library for further information on when to use
the example profiles in MONITORING_LIBRARY_QOS_PROFILES.xml.
Table 37.1 Configuration Properties for Monitoring Library lists the configuration properties that
you can set for Monitoring Library. These properties are immutable; they cannot be changed after
the DomainParticipant is created.

aSee Paths Mentioned in Documentation (Section on page xxxviii)

1034

Chapter 37 Configuring Monitoring Library

Table 37.1 Configuration Properties for Monitoring Library
Property Name
(all must be
prepended with
“rti.monitor.config.”)

Property Value

This boolean value specifies whether or not Monitoring Library should collect CPU and memory usage
statistics for the process in the topic rti/dds/monitoring/domainParticipantDescription.

get_process_statistics

This property is only applicable to Linux and Windows systems—obtaining CPU and memory usage on other
architectures is not supported.
CPU usage is reported in terms of time spent since the process has been started. It can be longer than the
actual running time of the process on a multi-core machine.
Default: true if unspecified
To create a separate participant that will be used to publish monitoring information in the application, set this
to the domain ID that you want to use for the newly created participant.

new_participant_domain_id

This property can be used with the qos_library (Section on the facing page) and qos_profile (Section on the
facing page) properties to specify the QoS that will be used to create a new participant.
Default: Not set (means you want to reuse the participant in your application that has monitoring turned on to
publish statistics information for that participant)
Period of time to sample and publish all monitoring topics, in units of seconds.

publish_period
Default: 5 if unspecified
Priority of the thread used to sample and publish monitoring data.
This value is architecture dependent.
publish_thread_priority

Default if unspecified: same as the default used in Connext DDS for the event thread:
Windows systems: -2
Linux systems: -999999 (meaning use OS-default priority)
Stack size used for the thread that samples and publishes monitoring data. This value is architecture
dependent.

publish_thread_stacksize

Default if unspecified: same as the default used in Connext DDS for the event thread:
Windows systems: 0 (meaning use the default size for the executable).
Linux systems: -1 (meaning use OS’s default value).

1035

Chapter 37 Configuring Monitoring Library

Table 37.1 Configuration Properties for Monitoring Library
Property Name
(all must be
prepended with
“rti.monitor.config.”)

Property Value

Describes the type of thread.
Supported values (may be combined with by OR’ing with ‘|’ as seen in the default below):
l

publish_thread_options

FLOATING_POINT: Code executed within the thread may perform floating point
operations

l

STDIO: Code executed within the thread may access standard

l

I/O REALTIME_PRIORITY: The thread will be scheduled on a real-time basis

l

PRIORITY_ENFORCE: Strictly enforce this thread's priority

Default: FLOATING_POINT|STDIO (same as the default used in Connext DDS for the event thread)
Specifies the name of the QoS library that you want to use for creating entities in the monitoring library (if
you do not want to use default QoS values as set by the monitoring library).
qos_library

The QoS values used for internally created entities can be found in the library RTIMonitoringQosLibrary in
/connext_dds/qos/MONITORING_LIBRARY_QOS_PROFILES.xml.
Default: Not set (means you want to use default Monitoring Library QoS values)
Specifies the name of the QoS profile that you want to use for creating entities in the monitoring library (if
you do not want to use the default QoS values).

qos_profile

The QoS values used for internally created entities can be found in the profile
RTIMonitoringPublishingQosProfile in /connext_dds/qos/MONITORING_
LIBRARY_QOS_PROFILES.xml.
Default: Not set (means you want to use default Monitoring Library QoS values)
Monitoring Library obtains all statuses of all entities in the Connext DDS application. This boolean value
controls whether or not the change counts in those statuses are reset by Monitoring Library.
If set to true, the change counts are reset each time Monitoring Library is done accessing them.

reset_status_change_counts
If set to false, the change counts truly reflect what users will see in their application and are unaffected by the
access of the monitoring library.
Default: false
This boolean value controls whether or not the entities created internally by Monitoring Library should be
included in the entity counts published by the participant entity statistics topic.
skip_monitor_entities

If set to true, the internal monitoring entities will not be included in the count. (Thirteen internal writers are
created by the monitoring library by default.)
Default: true

1036

Chapter 37 Configuring Monitoring Library

Table 37.1 Configuration Properties for Monitoring Library
Property Name
(all must be
prepended with
“rti.monitor.config.”)

Property Value

If set to true, DomainParticipant PropertyQosPolicy name and value pairs will not be sent out through the
domainParticipantDescriptionTopic. This is necessary if you are linking with Monitoring Library and any of
these conditions occur:
skip_participant_
properties

l

l

The PropertyQosPolicy of a DomainParticipant has more than 32 properties.
Any of the properties in PropertyQosPolicy of a DomainParticipant has a name
longer than 127 characters or a value longer than 511 characters.

Default: false if unspecified
If set to true, DataReader PropertyQosPolicy name and value pairs will not be sent out through the
dataReaderDescriptionTopic. This is necessary if you are linking with Monitoring Library and any of these
conditions occur:
skip_reader_
properties

l

l

The PropertyQosPolicy of a DataReader has more than 32 properties.
Any of the properties in PropertyQosPolicy of a DataReader has a name longer than
127 characters or a value longer than 511 characters.

Default: false if unspecified
If set to true, DataWriter PropertyQosPolicy name and value pairs will not be sent out through the
dataWriterDescriptionTopic. This is necessary if you are linking with Monitoring Library and any of these
conditions occur:
l

skip_writer_properties
l

The PropertyQosPolicy of a DataWriter has more than 32 properties.
Any of the properties in PropertyQosPolicy of a DataWriter has a name longer than
127 characters or a value longer than 511 characters.

Default: false if unspecified

topics

Filter for monitoring topics, with regular expression matching syntax as specified in the Connext DDS
documentation (similar to the POSIX fnmatch syntax). For example, if you only want to send description
topics and the entity statistics topics, but NOT the matching statistics topics, you can specify
“*Description,*EntityStatistics”.
Default: * if unspecified

usertopics

Filter for user topics, with regular expression matching syntax as specified in the Connext DDS
documentation (similar to the POSIX fnmatch syntax). For example, if you only want to send monitoring
information for reader/writer/topic entities for topics that start with Foo or Bar, you can specify “Foo*,Bar*”.
Default: * if unspecified

1037

Chapter 37 Configuring Monitoring Library

Table 37.1 Configuration Properties for Monitoring Library
Property Name
(all must be
prepended with
“rti.monitor.config.”)

Property Value

Sets the verbosity on the monitoring library for debugging purposes (does not affect the topic/data that is sent
out).

verbosity

l

-1: Silent

l

0: Exceptions only

l

1: Warnings

l

2 and up: Higher verbosity level

Default: 1 if unspecified
Controls the threshold at which dynamic memory allocation is used, expressed as a number of bytes.
If the serialized size of the data to be sent is smaller than this size, a pre-allocated writer buffer pool is used to
obtain the memory.
writer_pool_buffer_max_
size

If the serialized size of the data is larger than this value, the memory is allocated dynamically.
This setting can be used to control memory consumption of the monitoring library, at the cost of performance,
when the maximum serialized size of the data type is large (which is the case for some description topics’ data
types) or if you have several participants on the same machine.
The default setting is -1, meaning memory is always obtained from the writer buffer pool, whose size is
determined by the maximum serialized size.

1038

Part 10: RTI Distributed Logger
RTI® Distributed Logger is a library that enables applications to publish their log messages to Connext DDS. The log message data can be visualized with RTI Monitor, a separate GUI application
that can run on the same host as your application or on a different host. Since the data is provided
in a Topic, you can also use rtiddsspy or even write your own visualization tool.
Distributed Logger can send Connext DDS errors, warnings and other internal messages as a DDS
Topic. In fact, Distributed Logger also provides a remote command topic so that its behavior can
be remotely controlled at run time.

1039

Part 10: RTI Distributed Logger

This part of the User’s Manual includes:
l

Using Distributed Logger in a Connext DDS Application (Section Chapter 38 on page 1041)

l

Enabling Distributed Logger in RTI Services (Section Chapter 39 on page 1049)

1040

Chapter 38 Using Distributed Logger in a
Connext DDS Application
There are two ways to use Distributed Logger: directly through its API or by attaching it to an
existing logging framework as an ‘appender’ or a ‘handler.’ Using the API directly is straightforward, but keep in mind that Distributed Logger is not intended to be a full-featured logging library. In particular, it does not contain the ability to log messages to standard out/error. Rather, it is
primarily intended to be integrated into third-party logging infrastructures.
The libraries that you will need for Distributed Logger are listed in the RTI Connext DDS Core
Libraries Platform Notes.
Distributed Logger comes with third-party integrations for the open-source project log4j (http://logging.apache.org/log4j/) as well as Java’s built-in logging library (java.util.logging). Please see
Examples (Section 38.2 on the next page) for examples that illustrate these integrations.
Distributed Logger captures and forwards Connext DDS internal information, warning, and error
messages using a DDS topic. It monitors these messages using the same mechanism as user log
messages.
These Connext DDS log messages are sent over DDS automatically as soon as you initialize Distributed Logger (by calling RTI_DL_DistLogger_getInstance() in C or C++, or Logger.getLogger(...) in Java; see the API Reference HTML documentation for details).

38.1 Using the API Directly
Details on using the Distributed Logger APIs are provided in the API Reference HTML documentation: /doc/api/connext_dds/distributed_logger/. Start by
opening index.html.

1See Paths Mentioned in Documentation (Section on page xxxviii)

1041

38.2 Examples

If you plan to use the Distributed Logger’s API directly, please be aware of the following notes. To configure the options, create an options object and update its fields. Once your updates are complete, set the
options on Distributed Logger. It is important that this be done before Distributed Logger is instantiated.
Distributed Logger acts as a singleton and there is no way to change the options after it has been created.
When your application is ready to exit, use the ‘delete’ method. This will delete all Entities and threads
associated with Distributed Logger.

38.2 Examples
Distributed Logger includes several examples in /distributed_logger:
l

c/hello_distributed_logger
This is a simple example of how to use the API directly and does not publish or subscribe to any
Topics except the ones related to Distributed Logger.

l

c++/hello_distributed_logger
This is a simple example of how to use the API directly and does not publish or subscribe any Topics except the ones related to Distributed Logger.

l

java/hello_direct_usage
This is a simple example of how to use the API directly and does not publish or subscribe any Topics except the ones related to Distributed Logger.

l

java/hello_file_logger
This example shows how an application can use the information provided by Distributed Logger.
As the name suggests, this example subscribes to log messages and writes them to a file. Multiple
DDS domains can be subscribed to simultaneously if desired. The example is meant to strike a balance between simplicity and function. Certainly more features could be added to make it a production-ready application but that would obscure the goal of the example.

l

java/hello_java_util_logging
This is an adaptation of the Hello_idl example which replaces all System.{out/err} invocations with
Java logging library equivalents. It adds Distributed Logger through a configuration file.

1See Paths Mentioned in Documentation (Section on page xxxviii)

1042

38.3 Data Type Resource

l

java/hello_log4j_logging
This is an adaptation of the Hello_idl example which replaces all System.{out/err} invocations with
log4j library equivalents. It adds Distributed Logger through a configuration file.

Each example has a READ_ME.txt file which explains how to build and run it.

38.3 Data Type Resource
You can find the data types used by Distributed Logger in /resource/idl/distog.idl.
If you want to generate code and interact with Distributed Logger through Topics, you can use this file to
do so. You will need to provide extra command-line arguments to RTI Code Generator (rtiddsgen). (This
allows us to accommodate multiple language bindings within the same file. As a consequence, we’ve used
preprocessor definitions to achieve this functionality.) The command-line options which must be added to
rtiddsgen are as follows:
l

For C or C++: –D LANGUAGE_C

l

For Java: –D LANGUAGE_JAVA

l

For .Net: –D LANGUAGE_DOTNET

If you plan to use the generated code in your application (to subscribe to log messages, for
instance) be aware that the type names used might not match the default ones. Do not use the
generated type names obtained when calling get_type_name() or found in distlogSupport.h. Use
the variables in Table 38.1 Registration Names for each Distributed Logger Type instead.
Table 38.1 Registration Names for each Distributed Logger Type
Type

Registered Typename

Variable
C/C++:
RTI_DL_LOG_MESSAGE_TYPE_NAME

Log Message

com::rti::dl::LogMessage
Java:
LOG_MESSAGE_TYPE_NAME.VALUE
C/C++:
RTI_DL_STATE_TYPE_NAME

Administration State

com::rti::dl::admin::State
Java:
STATE_TYPE_NAME.VALUE

1See Paths Mentioned in Documentation (Section on page xxxviii)

1043

38.4 Distributed Logger Topics

Table 38.1 Registration Names for each Distributed Logger Type
Type

Administration Command Request

Administration Command Response

Registered Typename

Variable
C/C++:
RTI_DL_COMMAND_REQUEST_TYPE_NAME

com::rti::dl::admin::
CommandRequest

Java:
COMMAND_REQUEST_TYPE_NAME.VALUE
C/C++:
RTI_DL_COMMAND_RESPONSE_TYPE_NAME

com::rti::dl::admin::
CommandResponse

Java:
COMMAND_RESPONSE_TYPE_NAME.VALUE

For instance, to subscribe to log messages in C you will need to do the following:
retcode = RTI_DL_LogMessageTypeSupport_register_type(
participant, RTI_DL_LOG_MESSAGE_TYPE_NAME);

38.4 Distributed Logger Topics
Distributed Logger uses four Topics to publish log messages, state, and command responses and one topic
to subscribe to command requests. These are detailed in Table 38.2 Topics Used by Distributed Logger.
Table 38.2 Topics Used by Distributed Logger
Topic

Type Name

Quality of Service
Reliable

rti/distlog

com::rti::dl::LogMessage
Transient Local
Reliable

rti/distlog/administration/state

com::rti::dl::admin::State
Transient Local

rti/distlog/administration/command_request

com::rti::dl::admin::CommandRequest

Reliable

rti/distlog/administration/command_response

com::rti::dl::admin::CommandResponse

Reliable

38.5 Distributed Logger IDL
The IDL describing the types used for Topics created by Distributed Logger are in 1/resource/idl/distlog.idl. You can use this IDL to create custom applications that use the data provided by

1See Paths Mentioned in Documentation (Section on page xxxviii)
1044

38.6 Viewing Log Messages

Distributed Logger and/or to remotely control any Distributed Logger instances that are running in your
system. The IDL has been designed to take advantage of the latest type-support features in Connext DDS.

38.6 Viewing Log Messages
One way to see the messages from Distributed Logger is to use RTI Monitor.
Figure 38.1 Viewing Log Messages with RTI Monitor

Other ways to see the log messages include using rtiddsspy or writing your own visualization tool. If you
want to write your own application that interacts with Distributed Logger, you can find the IDL in
1/resource/idl/distlog.idl.

38.7 Logging Levels
Log levels in Distributed Logger are organized as follows (ordered by importance). This table also shows
the mapping between logging levels in the Connext DDS middleware and Distributed Logger.

1See Paths Mentioned in Documentation (Section on page xxxviii)

1045

38.8 Distributed Logger Quality of Service Settings

Connext DDS Logger Log Level

Distributed Logger Log Level

NDDS_CONFIG_LOG_LEVEL_ERROR

RTI_DL_ERROR_LEVEL

NDDS_CONFIG_LOG_LEVEL_WARNING

RTI_DL_WARNING_LEVEL

NDDS_CONFIG_LOG_LEVEL_STATUS_LOCAL

RTI_DL_NOTICE_LEVEL

NDDS_CONFIG_LOG_LEVEL_STATUS_REMOTE

RTI_DL_INFO_LEVEL

NDDS_CONFIG_LOG_LEVEL_DEBUG

RTI_DL_DEBUG_LEVEL

38.8 Distributed Logger Quality of Service Settings
To ensure that Distributed Logger works correctly with other RTI tools, some QoS settings are hard-coded
and cannot be modified by customized profiles. Table 38.3 QoS Values Used by Distributed Logger lists
the QoS values that are set in Distributed Logger. Values in bold are hard-coded; therefore even if they
appear in an XML profile, they remain as noted in the table.
Table 38.3 QoS Values Used by Distributed Logger
Entity

Property

Value

Presentation.access_scope

PRES_INSTANCE_PRESENTATION_QOS

Presentation.coherent_access

false

Presentation.ordered_access

false

Presentation.access_scope

PRES_INSTANCE_PRESENTATION_QOS

Presentation.coherent_access

false

Presentation.ordered_access

false

Reliability.kind

DDS_RELIABLE_RELIABILITY_QOS

Durability.kind

DDS_TRANSIENT_LOCAL_DURABILITY_
QOS

Reliability.kind

DDS_RELIABLE_RELIABILITY_QOS

Durability.kind

DDS_TRANSIENT_LOCAL_DURABILITY_
QOS

Administration Command Request Topic

Reliability.kind

DDS_RELIABLE_RELIABILITY_QOS

Administration Command Response Topic

Reliability.kind

DDS_RELIABLE_RELIABILITY_QOS

Subscriber

Publisher

Log Message Topic

Administration State Topic

1046

38.8 Distributed Logger Quality of Service Settings

Table 38.3 QoS Values Used by Distributed Logger
Entity

Property

Value

Ownership.kind

DDS_SHARED_OWNERSHIP_QOS

Latency_budget.duration.sec

0

Latency_budget.duration.nanosec 0

Log Message DataWriter

Liveliness.kind

DDS_AUTOMATIC_LIVELINESS_QOS

Destination_order.kind

DDS_BY_RECEPTION_TIMESTAMP_
DESTINATIONORDER_QOS

Reliability.kind

DDS_RELIABLE_RELIABILITY_QOS

Durability.kind

DDS_TRANSIENT_LOCAL_DURABILITY_QOS

History.kind

DDS_KEEP_LAST_HISTORY_QOS

History.depth

10

Ownership.kind

DDS_SHARED_OWNERSHIP_QOS

Latency_budget.duration.sec

0

Latency_budget.duration.nanosec 0

Administration State DataWriter

Liveliness.kind

DDS_AUTOMATIC_LIVELINESS_QOS

Destination_order.kind

DDS_BY_RECEPTION_TIMESTAMP_
DESTINATIONORDER_QOS

Reliability.kind

DDS_RELIABLE_RELIABILITY_QOS

Durability.kind

DDS_TRANSIENT_LOCAL_DURABILITY_QOS

History.kind

DDS_KEEP_LAST_HISTORY_QOS

History.depth

1

1047

38.8 Distributed Logger Quality of Service Settings

Table 38.3 QoS Values Used by Distributed Logger
Entity

Property

Value

Ownership.kind

DDS_SHARED_OWNERSHIP_QOS

Latency_budget.duration.sec

0

Latency_budget.duration.nanosec 0

Administration Command Response
DataWriter

Liveliness.kind

DDS_AUTOMATIC_LIVELINESS_QOS

Destination_order.kind

DDS_BY_RECEPTION_TIMESTAMP_
DESTINATIONORDER_QOS

Reliability.kind

DDS_RELIABLE_RELIABILITY_QOS

History.kind

DDS_KEEP_LAST_HISTORY_QOS

History.depth

10

Ownership.kind

DDS_SHARED_OWNERSHIP_QOS

Latency_budget.duration.sec

DDS_DURATION_INFINITE_SEC

Latency_budget.duration.nanosec DDS_DURATION_INFINITE_NSEC

Administration Command Request DataReader

Deadline.period.sec

DDS_DURATION_INFINITE_SEC

Deadline.period.nanosec

DDS_DURATION_INFINITE_NSEC

Liveliness.kind

DDS_AUTOMATIC_LIVELINESS_QOS

Destination_order.kind

DDS_BY_RECEPTION_TIMESTAMP_
DESTINATIONORDER_QOS

Reliability.kind

DDS_RELIABLE_RELIABILITY_QOS

History.kind

DDS_KEEP_LAST_HISTORY_QOS

History.depth

10

1048

Chapter 39 Enabling Distributed Logger in
RTI Services
Many RTI components provide integrated support for Distributed Logger (check the component’s
Release Notes) and include the Distributed Logger library in their distribution. To enable Distributed Logger in these components, modify their XML configuration file. In the 
section, add the  tag as shown in this example:


10

true
DEBUG
2048


THREAD_PRIORITY_BELOW_NORMAL

8192

0
1


THREAD_SETTINGS_CPU_NO_ROTATION




...


The tags supported within the  tag are described in Table 39.1 Distributed
Logger Tags.

1049

Chapter 39 Enabling Distributed Logger in RTI Services

Table 39.1 Distributed Logger Tags
Tags within


Description

Controls whether or not Distributed Logger should be enabled at start up. This field is required.

Allowed values: TRUE or FALSE

Number
of Tags
Allowed
1
(required)

The filter level for the log messages to be sent. Distributed Logger uses the filter level to discard log
messages before they can be sent from the application/service. This is the minimum log level that will be
sent out over the network. For example, when using the NOTICE level, any INFO, DEBUG and TRACElevel log messages will be filtered out and not sent from the application/service to Connext DDS.

See important information in Relationship Between Service Verbosity and Filter Level
(Section 39.1 on page 1052).
Can be set to these values:



l

SILENT

l

FATAL

l

SEVERE

l

ERROR

l

WARNING

l

NOTICE

l

INFO

l

DEBUG

l

TRACE (most verbose level, default)

0 or 1

The size of an internal message queue used to store log messages before they are written to DDS.


0 or 1
Default, 128 log messages.



See Table 39.2 Distributed Logger Thread Tags.

0 or 1

1050

Chapter 39 Enabling Distributed Logger in RTI Services

Table 39.2 Distributed Logger Thread Tags
Tags within
/


Description

Number of
Tags Allowed

Each  specifies a processor on which the Distributed Logger thread may run.




value


0 or 1

Only applies to platforms that support controlling CPU core affinity (see the RTI Connext
DDS Core Libraries Platform Notes).
Determines how the CPUs in  will be used by the Distributed Logger thread.
The value can be either:
l

THREAD_SETTINGS_CPU_NO_ROTATION
The thread can run on any listed processor, as determined by OS scheduling.



0 or 1
l

THREAD_SETTINGS_CPU_RR_ROTATION
The thread will be assigned a CPU from the list in round-robin order.

Only applies to platforms that support controlling CPU core affinity (see the RTI Connext
DDS Core Libraries Platform Notes).
A collection of flags used to configure threads of execution. Not all of these options may
be relevant for all operating systems. May include these bits:



l

STDIO

l

FLOATING_POINT

l

REALTIME_PRIORITY

l

PRIORITY_ENFORCE

0 or 1

It can also be set to a combination of the above bits by using the “or” symbol (|), such as
STDIO|FLOATING_POINT.
Default: MASK_DEFAULT

1051

39.1 Relationship Between Service Verbosity and Filter Level

Table 39.2 Distributed Logger Thread Tags
Tags within
/


Description

Number of
Tags Allowed

Thread priority. The value can be specified as an unsigned integer or one of the following
strings.



l

THREAD_PRIORITY_DEFAULT

l

THREAD_PRIORITY_HIGH

l

THREAD_PRIORITY_ABOVE_NORMAL

l

THREAD_PRIORITY_NORMAL

l

THREAD_PRIORITY_BELOW_NORMAL

l

THREAD_PRIORITY_LOW

0 or 1

When using an unsigned integer, the allowed range is platform-dependent.


Thread stack size, specified as an unsigned integer or set to the string THREAD_STACK_
0 or 1
SIZE_DEFAULT. The allowed range is platform-dependent.

39.1 Relationship Between Service Verbosity and Filter Level
A service’s verbosity influences the way the log messages reach Distributed Logger and their quantity. If a
service (such as RTI Persistence Service, RTI Routing Service, or another service that is integrated with
Distributed Logger) is configured with a low verbosity, it will not pass a lot of messages to Distributed
Logger, even if the Distributed Logger filter level is set to a very verbose one (such as TRACE). On the
contrary, a high verbosity will work better, because it will pass more messages to Distributed Logger; in
this case the filter level will have more effect.
Note: Since Distributed Logger uses a separate thread to send log messages, there is little impact on performance with more verbose filter levels. However, there is some performance penalty in services that use
a higher verbosity.

1052


Source Exif Data: 
File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.4
Linearized                      : No
Page Count                      : 1093
Page Mode                       : UseOutlines
Language                        : en-us
Producer                        : madbuild
Create Date                     : 2016:04:25 12:11:57-04:00
Modify Date                     : 2016:04:25 12:11:57-04:00
Title                           : RTI Connext DDS Core Libraries User's Manual
Author                          : RTI
Subject                         : 


EXIF Metadata provided by EXIF.tools









 






Navigation menu