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 ali