0% found this document useful (0 votes)

274 views

Programming Multiplayer Games

Introduction to multiplayer games

Uploaded by

Biris Horea

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

274 views

Programming Multiplayer Games

Introduction to multiplayer games

Uploaded by

Biris Horea

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 577

Programming Multiplayer Games

Andrew Mulholland and Teijo Hakala

Wordware Publishing, Inc.

Library of Congress Cataloging-in-Publication Data Mulholland, Andrew. Programming multiplayer games / by Andrew Mulholland and Teijo Hakala. p. cm. Includes index. ISBN 1-55622-076-6 (pbk.; companion cd-rom) 1. Computer games--Programming. I. Hakala, Teijo. II. Title. QA76.76.C672M855 2004 794.8'1711dc22 2003027637 CIP

No part of this book may be reproduced in any form or by any means without permission in writing from Wordware Publishing, Inc. Printed in the United States of America

ISBN 1-55622-076-6 10 9 8 7 6 5 4 3 2 1 0403

UNIX is a registered trademark of The Open Group in the United States and other countries. All brand names and product names mentioned in this book are trademarks or service marks of their respective companies. Any omission or misuse (of any kind) of service marks or trademarks should not be regarded as intent to infringe on the property of others. The publisher recognizes and respects all marks used by companies, manufacturers, and developers as a means to distinguish their products. This book is sold as is, without warranty of any kind, either express or implied, respecting the contents of this book and any disks or programs that may accompany it, including but not limited to implied warranties for the books quality, performance, merchantability, or fitness for any particular purpose. Neither Wordware Publishing, Inc. nor its dealers or distributors shall be liable to the purchaser or any other person or entity with respect to any liability, loss, or damage caused or alleged to have been caused directly or indirectly by this book.

All inquiries for volume purchases of this book should be addressed to Wordware Publishing, Inc., at the above address. Telephone inquiries may be made by calling: (972) 423-0090

Contents
About the Authors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv

Part I Theory
Chapter 1 Introduction to Windows Programming . . . . . . . . . . 3 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Windows Messaging System . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Creating a Window . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Sending Information to Your Window . . . . . . . . . . . . . . . . . . . . . 8 Static Link Libraries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Creating a Static Link Library . . . . . . . . . . . . . . . . . . . . . 10 Using a Static Link Library. . . . . . . . . . . . . . . . . . . . . . . 13 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Chapter 2 Using Databases . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . What Is MySQL? . . . . . . . . . . . . . . . . . Installing MySQL . . . . . . . . . . . . . . . . SQL Statements . . . . . . . . . . . . . . . . . Data Definition Language . . . . . . . . . . . . Creating and Dropping Databases . . . . Creating a Database . . . . . . . . Dropping a Database . . . . . . . . Column (Field) Types in MySQL . . . . Adding, Modifying, and Dropping Tables Creating Tables. . . . . . . . . . . Modifying Tables . . . . . . . . . . Dropping (Removing) Tables . . . Data Manipulation Language (DML) . . . . . . Inserting Data . . . . . . . . . . . . . . . Modifying Data . . . . . . . . . . . . . . Removing (Deleting) Data . . . . . . . . Using Select Statements . . . . . . . . . Relational Databases. . . . . . . . . . . . . . . Data Import Methods . . . . . . . . . . . . . . Importing from a Text File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 . . . 15 . . . 15 . . . 16 . . . 19 . . . 19 . . . 19 . . . 20 . . . 21 . . . 22 . . . 24 . . . 24 . . . 26 . . . 28 . . . 29 . . . 30 . . . 32 . . . 34 . . . 35 . . . 40 . . . 44 . . . 44

iii

Contents

Importing from a Native Source . . . . . . . . . . . . . . . . Backing Up and Restoring Data . . . . . . . . . . . . . . . . . . . . Backing up a Database to a File . . . . . . . . . . . . . . . . Restoring a Backed-Up Database . . . . . . . . . . . . . . . MySQL C++ Interface . . . . . . . . . . . . . . . . . . . . . . . . Example 1 Connecting and Retrieving Data from MySQL Example 2 Updating Data in MySQL from an Application Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

45 47 47 49 50 50 53 55

Chapter 3 Creating Web-Based Server Interfaces . . . . . . . . . 57 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Setting Up an Apache 1.3.x Web Server . . . . . . . . . . . . . . . . . . . 57 Installing PHP4 for Apache 1.3.x . . . . . . . . . . . . . . . . . . . . . . . 60 Using PHP: Hypertext Preprocessor . . . . . . . . . . . . . . . . . . . . . 63 The Basics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 Example 1 index.php . . . . . . . . . . . . . . . . . . . . . 64 Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Example 2 index2.php . . . . . . . . . . . . . . . . . . . . 65 Operators and Loops . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Example 3 index3.php . . . . . . . . . . . . . . . . . . . . 67 Conditional Statements. . . . . . . . . . . . . . . . . . . . . . . . . 68 Example 4 index4.php . . . . . . . . . . . . . . . . . . . . 68 Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Example 5 index5.php . . . . . . . . . . . . . . . . . . . . 70 User Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 Example 6a input.php. . . . . . . . . . . . . . . . . . . . . 71 Example 6b output.php . . . . . . . . . . . . . . . . . . . . 72 The Command System . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 Example 7a core.php . . . . . . . . . . . . . . . . . . . . . 74 Example 7b welcome.php . . . . . . . . . . . . . . . . . . 74 Example 8a core.php . . . . . . . . . . . . . . . . . . . . . 76 Example 8b welcome.php . . . . . . . . . . . . . . . . . . 76 Example 8c page1.php . . . . . . . . . . . . . . . . . . . . 77 Example 8d page2.php . . . . . . . . . . . . . . . . . . . . 77 Accessing MySQL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 MySQL Example 1 Connecting and Disconnecting . . . . . . . . 78 MySQL Example 2 Storing and Retrieving Data . . . . . . . . . 79 MySQL Example 3 Updating and Removing Data . . . . . . . . . 84 Using FastTemplate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Multiple Templates . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 Converting the Command Parser Example to FastTemplate . . . . 95 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 Chapter 4 Introduction to TCP/IP. . . . . . . . . . . . . . . . . . 103 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

Contents

What Is a Protocol? . . . . . . . . . . OSI Model. . . . . . . . . . . . . . . OSI Model Layers . . . . . . . Internet Protocol . . . . . . . . . . . Introduction to the Transport Layer Transmission Control Protocol User Datagram Protocol . . . Ports . . . . . . . . . . . . . . . . . . Introduction to Sockets . . . . . . . Socket Types . . . . . . . . . . Address. . . . . . . . . . . . . Platforms . . . . . . . . . . . . History of WinSock . . . . . . Summary . . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

103 104 104 106 108 108 109 109 110 111 112 112 113 113

Chapter 5 Basic Sockets Programming . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . WinSock Initialization . . . . . . . . . . . . . . . . . . . . . WSAStartup Function (Win32) . . . . . . . . . . . . . WSACleanup Function (Win32). . . . . . . . . . . . . WSAEnumProtocols Function (Win32) . . . . . . . . WinSock Initialization Function . . . . . . . . . . . . Error Handling . . . . . . . . . . . . . . . . . . . . . . . . . WSAGetLastError Function (Win32) . . . . . . . . . Sockets Data Types. . . . . . . . . . . . . . . . . . . . . . . Platform-specific Data Types . . . . . . . . . . . . . . Address Structures . . . . . . . . . . . . . . . . . . . IPv4 Address Structure . . . . . . . . . . . . . IPv6 Address Structure . . . . . . . . . . . . . Generic Address Structure. . . . . . . . . . . . Basic Sockets Functions . . . . . . . . . . . . . . . . . . . . socket Function (Unix, Win32) . . . . . . . . . . . . . bind Function (Unix, Win32) . . . . . . . . . . . . . . connect Function (Unix, Win32) . . . . . . . . . . . . listen Function (Unix, Win32) . . . . . . . . . . . . . accept Function (Unix, Win32) . . . . . . . . . . . . . close Function (Unix)/closesocket Function (Win32) . Input/Output Functions . . . . . . . . . . . . . . . . . . . . send Function (Unix, Win32) . . . . . . . . . . . . . . recv Function (Unix, Win32) . . . . . . . . . . . . . . sendto Function (Unix, Win32) . . . . . . . . . . . . . recvfrom Function (Unix, Win32). . . . . . . . . . . . Address Data Conversion Functions . . . . . . . . . . . . . inet_aton Function (Unix, Win32) . . . . . . . . . . . Client/Server Programming . . . . . . . . . . . . . . . . . . Server Methods . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. 115 . . 115 . . 115 . . 115 . . 116 . . 117 . . 117 . . 120 . . 120 . . 121 . . 121 . . 121 . . 121 . . 122 . . 123 . . 124 . . 124 . . 125 . . 126 . . 127 . . 128 . . 129 . . 129 . . 129 . . 131 . . 132 . . 133 . . 133 . . 134 . . 134 . . 134

Contents

Clients . . . . . . . . . . . . . . Byte Ordering. . . . . . . . . . . . . . Creating a Server . . . . . . . . . . . . TCP. . . . . . . . . . . . . . . . UDP . . . . . . . . . . . . . . . Simple Echo TCP Server . . . . main Function . . . . . . . InitSockets Function . . . ServerProcess Function . Simple Echo UDP Server . . . . InitSockets Function . . . ServerProcess Function . Creating a Client . . . . . . . . . . . . TCP. . . . . . . . . . . . . . . . UDP . . . . . . . . . . . . . . . Simple Echo TCP Client . . . . main Function . . . . . . . InitSockets Function . . . ClientProcess Function. . Simple Echo UDP Client . . . . InitSockets Function . . . ClientProcess Function. . Running the Simple Echo Application. Summary . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

135 136 136 137 138 139 142 142 145 146 148 149 150 150 151 151 153 154 155 157 159 159 160 161

Chapter 6 I/O Operations . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . Detecting Network Events . . . . . . . . . . . select (Unix, Win32). . . . . . . . . . . . Macros . . . . . . . . . . . . . . . . . . . WSAAsyncSelect (Win32) . . . . . . . . WSAEventSelect (Win32). . . . . . . . . WSAWaitForMultipleEvents (Win32) . . Event Object . . . . . . . . . . . . . . . . Multithreading . . . . . . . . . . . . . . . . . . What Is Multithreading? . . . . . . . . . CreateThread (Win32) . . . . . . . . . . pthread_create (Unix) . . . . . . . . . . . I/O Strategy . . . . . . . . . . . . . . . . . . . . Blocking I/O . . . . . . . . . . . . . . . . Non-blocking I/O . . . . . . . . . . . . . Signal-driven I/O . . . . . . . . . . . . . Multiplexing I/O . . . . . . . . . . . . . . I/O Control . . . . . . . . . . . . . . . . . . . . ioctl (Unix)/ioctlsocket (Win32) . . . . . setsockopt/getsockopt (Unix, Win32) . .

. . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. 163 . . 163 . . 163 . . 163 . . 164 . . 165 . . 165 . . 166 . . 166 . . 167 . . 167 . . 168 . . 169 . . 169 . . 169 . . 170 . . 170 . . 171 . . 172 . . 172 . . 173

Contents

shutdown (Unix, Win32) Broadcasting . . . . . . . . . . Searching for Servers . . Broadcast Function . . . Summary . . . . . . . . . . . .

. . . . .

174 175 175 176 177

Part II Tutorials
Tutorial 1 Using 2DLIB . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Configuring Visual Studio . . . . . . . . . . . . . . . . . . . . . . Creating a Skeleton Project . . . . . . . . . . . . . . . . . . . . . Creating the Workspace. . . . . . . . . . . . . . . . . . . . Adding the Static Libraries . . . . . . . . . . . . . . . . . . Adding the Source File . . . . . . . . . . . . . . . . . . . . Creating a Basic Windowed Application with 2DLIB . . . . The WinMain Function . . . . . . . . . . . . . . . . . The Windows Procedure . . . . . . . . . . . . . . . . The Complete Code . . . . . . . . . . . . . . . . . . Using the 2DLIB Graphics Routines . . . . . . . . . . . . . . . . 2D Positions on the Screen . . . . . . . . . . . . . . . . . . Use of Colors . . . . . . . . . . . . . . . . . . . . . . . . . Plotting a Single Pixel. . . . . . . . . . . . . . . . . . . . . Drawing a Line . . . . . . . . . . . . . . . . . . . . . . . . Drawing a Rectangle/Filled Rectangle . . . . . . . . . . . . Drawing a Triangle/Filled Triangle . . . . . . . . . . . . . . Graphic Loading Functions . . . . . . . . . . . . . . . . . . Graphics Display (Blitting) Function . . . . . . . . . . . . . Keyboard Input Method . . . . . . . . . . . . . . . . . . . . 2DLIB Example 1 Moving Primitives with the Cursor Keys . Complete Code Listing for Example 1 . . . . . . . . . . . . 2DLIB Example 2 Loading and Rotating Graphics . . . . . . . Complete Code Listing for Example 2 . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tutorial 2 Creating Your Network Library . . Introduction . . . . . . . . . . . . . . . . . . . . Why Create a Network Library of Our Own? . . Planning the Structure . . . . . . . . . . . . . . Planning the Functionality . . . . . . . . . . . . Identifying Hosts . . . . . . . . . . . . . Sending Data to Hosts . . . . . . . . . . Pinging Calculating Network Latency Timing Out . . . . . . . . . . . . . . . . . Building the Library . . . . . . . . . . . . . . . Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 . . 181 . . 181 . . 182 . . 182 . . 183 . . 183 . . 184 . . 184 . . 185 . . 187 . . 190 . . 190 . . 190 . . 191 . . 191 . . 191 . . 191 . . 192 . . 192 . . 193 . . 195 . . 197 . . 200 . . 201 . . 203 . 205 . . 205 . . 206 . . 206 . . 207 . . 207 . . 208 . . 209 . . 209 . . 209 . . 210

. . . . . . . . . .

vii

Contents

Unix/Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . Creating Independent Code . . . . . . . . . . . . . . . . . . . . . Creating Definitions for Data Types . . . . . . . . . . . . . Log System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . StartLogConsole Function . . . . . . . . . . . . . . . . . . dreamConsole Constructor . . . . . . . . . . . . . . . . . . dreamConsole Destructor . . . . . . . . . . . . . . . . . . println Function . . . . . . . . . . . . . . . . . . . . . . . . StartLog Function . . . . . . . . . . . . . . . . . . . . . . . LogString Function . . . . . . . . . . . . . . . . . . . . . . StopLog Function . . . . . . . . . . . . . . . . . . . . . . . Getting Started . . . . . . . . . . . . . . . . . . . . . . . . . . . . Setting Up Source and Header Files . . . . . . . . . . . . . dreamSock.h File . . . . . . . . . . . . . . . . . . . . dreamMessage Class . . . . . . . . . . . . . . . . . . dreamClient Class . . . . . . . . . . . . . . . . . . . dreamServer Class . . . . . . . . . . . . . . . . . . . Global Setup Functions . . . . . . . . . . . . . . . . . . . . dreamSock_Initialize . . . . . . . . . . . . . . . . . . dreamSock_InitializeWinSock . . . . . . . . . . . . . dreamSock_Shutdown . . . . . . . . . . . . . . . . . Global Socket Functions . . . . . . . . . . . . . . . . . . . dreamSock_Socket Function . . . . . . . . . . . . . . dreamSock_SetNonBlocking Function . . . . . . . . dreamSock_SetBroadcasting Function . . . . . . . . dreamSock_StringToSockaddr Function. . . . . . . . dreamSock_OpenUDPSocket Function . . . . . . . . dreamSock_CloseSocket Function. . . . . . . . . . . dreamSock_GetPacket Function . . . . . . . . . . . . dreamSock_SendPacket Function . . . . . . . . . . . dreamSock_Broadcast Function . . . . . . . . . . . . dreamSock_GetCurrentSystemTime Function . . . . dreamSock_Linux_GetCurrentSystemTime Function dreamSock_Win_GetCurrentSystemTime Function . Retrieving Error Values . . . . . . . . . . . . . . . . . . . . Summary of Global Functions . . . . . . . . . . . . . . . . Creating dreamSock Network Library . . . . . . . . . . . . . . . dreamMessage Class Member Variables. . . . . . . . . . . dreamMessage Class Functionality . . . . . . . . . . . . . Init Function . . . . . . . . . . . . . . . . . . . . . . Clear Function . . . . . . . . . . . . . . . . . . . . . GetNewPoint Function . . . . . . . . . . . . . . . . . AddSequences Function . . . . . . . . . . . . . . . . Write Function . . . . . . . . . . . . . . . . . . . . . WriteByte Function. . . . . . . . . . . . . . . . . . . WriteShort Function . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

211 212 212 213 214 215 215 215 216 217 218 219 219 220 225 226 228 229 229 230 231 232 238 239 239 240 241 243 243 244 245 246 246 247 248 248 248 249 249 253 253 254 255 255 255 256

viii

Contents

WriteLong Function . . . . . . . . . . . . WriteFloat Function . . . . . . . . . . . . WriteString Function . . . . . . . . . . . . BeginReading Function. . . . . . . . . . . Read Function. . . . . . . . . . . . . . . . ReadByte Function . . . . . . . . . . . . . ReadShort Function . . . . . . . . . . . . ReadLong Function . . . . . . . . . . . . . ReadFloat Function . . . . . . . . . . . . . ReadString Function . . . . . . . . . . . . dreamMessage Summary . . . . . . . . . dreamClient Class Member Variables . . . . . . dreamClient Class Functionality . . . . . . . . . dreamClient Constructor. . . . . . . . . . dreamClient Destructor . . . . . . . . . . Initialize Function. . . . . . . . . . . . . . Uninitialize Function . . . . . . . . . . . . Reset Function . . . . . . . . . . . . . . . DumpBuffer Function . . . . . . . . . . . System Messages vs. User Messages. . . SendConnect Function . . . . . . . . . . . SendDisconnect Function . . . . . . . . . SendPing Function . . . . . . . . . . . . . ParsePacket Function. . . . . . . . . . . . GetPacket Function. . . . . . . . . . . . . SendPacket Function (Internal Message) . SendPacket Function (External Message) dreamClient Summary . . . . . . . . . . . dreamServer Class Member Variables . . . . . . dreamServer Class Functionality . . . . . . . . . dreamServer Constructor . . . . . . . . . dreamServer Destructor . . . . . . . . . . Initialize Function. . . . . . . . . . . . . . Uninitialize Function . . . . . . . . . . . . SendAddClient Function . . . . . . . . . . SendRemoveClient Function. . . . . . . . SendPing Function . . . . . . . . . . . . . AddClient Function . . . . . . . . . . . . . RemoveClient Function . . . . . . . . . . ParsePacket Function. . . . . . . . . . . . CheckForTimeout Function . . . . . . . . GetPacket Function. . . . . . . . . . . . . SendPackets Function . . . . . . . . . . . dreamServer Summary. . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

256 256 257 257 257 258 258 259 259 259 260 260 262 268 268 268 269 270 270 270 271 272 272 273 275 276 278 279 280 280 289 289 290 290 290 292 293 293 295 297 299 301 303 304 304

Contents

Tutorial 3 Creating a Basic Network Application with dreamSock . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . Planning the Functionality . . . . . . . . . . . . . . . . . . Catching Exceptions . . . . . . . . . . . . . . . . . Creating a Basic Client Application . . . . . . . . . . . . . signin.h File . . . . . . . . . . . . . . . . . . . . . . CSignin Class . . . . . . . . . . . . . . . . . . . . . network.h File . . . . . . . . . . . . . . . . . . . . . main.h File . . . . . . . . . . . . . . . . . . . . . . . common.h File . . . . . . . . . . . . . . . . . . . . . main.cpp File. . . . . . . . . . . . . . . . . . . . . . Global Variables . . . . . . . . . . . . . . . . . . . . CreateAccountDialogProc Function . . . . . . . . . WinMain Function . . . . . . . . . . . . . . . . . . . signin.cpp File CSignin Class Methods . . . . . . CSignin Constructor . . . . . . . . . . . . . . CSignin Destructor . . . . . . . . . . . . . . . ReadPackets Function . . . . . . . . . . . . . AddClient Function . . . . . . . . . . . . . . . RemoveClient Function . . . . . . . . . . . . RemoveClients Function . . . . . . . . . . . . SendSignIn Function . . . . . . . . . . . . . . SendKeepAlive Function . . . . . . . . . . . . Connect Function . . . . . . . . . . . . . . . . Disconnect Function . . . . . . . . . . . . . . RunNetwork Function . . . . . . . . . . . . . Creating a Basic Server Application. . . . . . . . . . . . . signin.h File . . . . . . . . . . . . . . . . . . . . . . CSigninServer Class . . . . . . . . . . . . . . . . . network.h File . . . . . . . . . . . . . . . . . . . . . common.h File . . . . . . . . . . . . . . . . . . . . . main.cpp File. . . . . . . . . . . . . . . . . . . . . . Global Variables . . . . . . . . . . . . . . . . . . . . WindowProc Function . . . . . . . . . . . . . . . . . WinMain Function . . . . . . . . . . . . . . . . . . . daemonInit Function . . . . . . . . . . . . . . . . . keyPress Function . . . . . . . . . . . . . . . . . . . main Function . . . . . . . . . . . . . . . . . . . . . signin.cpp File CSigninServer Class Methods . . CSigninServer Constructor . . . . . . . . . . CSigninServer Destructor . . . . . . . . . . . InitNetwork Function . . . . . . . . . . . . . ShutdownNetwork Function . . . . . . . . . . ReadPackets Function . . . . . . . . . . . . . SendExitNotification Function . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. 305 . . 305 . . 306 . . 306 . . 306 . . 307 . . 308 . . 309 . . 309 . . 310 . . 310 . . 316 . . 316 . . 319 . . 323 . . 329 . . 329 . . 329 . . 333 . . 334 . . 335 . . 336 . . 336 . . 337 . . 337 . . 337 . . 338 . . 338 . . 339 . . 340 . . 340 . . 341 . . 347 . . 347 . . 347 . . 351 . . 352 . . 353 . . 355 . . 361 . . 361 . . 362 . . 362 . . 362 . . 366

Contents

AddClient Function . . . . RemoveClient Function . RemoveClients Function . RunNetwork Function . . Summary . . . . . . . . . . . . . . . .

. . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . .

366 367 368 369 369

Tutorial 4 Creating the Game Lobby . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . Creating the Lobby Client Application . . . . . . . . . . Creating the Dialogs . . . . . . . . . . . . . . . . . . . . Lobby Dialog . . . . . . . . . . . . . . . . . . . . . Create Game Dialog . . . . . . . . . . . . . . . . . Create View Players Dialog. . . . . . . . . . . . . Join Game Dialog . . . . . . . . . . . . . . . . . . Lobby System Code . . . . . . . . . . . . . . . . . . . . Lobby Client Code . . . . . . . . . . . . . . . . . . . . . lobby.h File . . . . . . . . . . . . . . . . . . . . . . network.h File . . . . . . . . . . . . . . . . . . . . main.h File . . . . . . . . . . . . . . . . . . . . . . main.cpp File. . . . . . . . . . . . . . . . . . . . . CreateViewPlayersDialogProc Function . . CreateGameDialogProc Function . . . . . . JoinGameDialogProc Function . . . . . . . . LoginDialogProc Function . . . . . . . . . . LobbyDialogProc Function . . . . . . . . . . WinMain Function . . . . . . . . . . . . . . lobby.cpp File CLobby Class Methods . . . . . RefreshPlayerList Function . . . . . . . . . ReadPackets Function . . . . . . . . . . . . RequestGameData Function . . . . . . . . . SendChat Function . . . . . . . . . . . . . . SendCreateGame Function. . . . . . . . . . SendRemoveGame Function . . . . . . . . . SendStartGame Function. . . . . . . . . . . Connect Function . . . . . . . . . . . . . . . Disconnect Function . . . . . . . . . . . . . RunNetwork Function . . . . . . . . . . . . Unimplemented Functions . . . . . . . . . . Lobby Server Code . . . . . . . . . . . . . . . . . . . . . lobby.h File . . . . . . . . . . . . . . . . . . . . . . network.h File . . . . . . . . . . . . . . . . . . . . main.cpp File. . . . . . . . . . . . . . . . . . . . . lobby.cpp File CLobbyServer Class Methods . ReadPackets Function . . . . . . . . . . . . Unimplemented Functions . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. 371 . . 371 . . 371 . . 372 . . 372 . . 375 . . 376 . . 376 . . 377 . . 377 . . 377 . . 379 . . 379 . . 380 . . 380 . . 380 . . 382 . . 382 . . 384 . . 386 . . 389 . . 398 . . 399 . . 401 . . 402 . . 402 . . 402 . . 403 . . 403 . . 403 . . 404 . . 405 . . 406 . . 406 . . 407 . . 407 . . 408 . . 418 . . 425 . . 425

Contents

Tutorial 5 Creating Your Online Game . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . Designing the Functionality . . . . . . . . . . . . . . . . . Frame Time . . . . . . . . . . . . . . . . . . . . . . Compressing Messages . . . . . . . . . . . . . . . . Dead Reckoning . . . . . . . . . . . . . . . . . . . . Frame History . . . . . . . . . . . . . . . . . . . . . Handling Messages . . . . . . . . . . . . . . . . . . Game Server Code . . . . . . . . . . . . . . . . . . . . . . server.h File . . . . . . . . . . . . . . . . . . . . . . network.h File . . . . . . . . . . . . . . . . . . . . . main.cpp File. . . . . . . . . . . . . . . . . . . . . . network.cpp File CArmyWarServer Class Part 1 InitNetwork Function . . . . . . . . . . . . . ReadPackets Function . . . . . . . . . . . . . SendCommand Function . . . . . . . . . . . . ReadDeltaMoveCommand Function . . . . . . BuildMoveCommand Function. . . . . . . . . BuildDeltaMoveCommand Function. . . . . . server.cpp File CArmyWarServer Class Part 2 . GenerateRandomMap Function . . . . . . . . CalculateVelocity Function . . . . . . . . . . . CalculateHeading Function. . . . . . . . . . . CalculateBulletVelocity Function . . . . . . . MovePlayer Function. . . . . . . . . . . . . . CheckFlagCollisions Function . . . . . . . . . Frame Function . . . . . . . . . . . . . . . . . lobby.cpp File . . . . . . . . . . . . . . . . . . . . . AddGame Function . . . . . . . . . . . . . . . RemoveGame Function . . . . . . . . . . . . RemoveGames Function . . . . . . . . . . . . Summary of Server Code . . . . . . . . . . . . . . . Game Client Code . . . . . . . . . . . . . . . . . . . . . . client.h File . . . . . . . . . . . . . . . . . . . . . . network.h File . . . . . . . . . . . . . . . . . . . . . common.h File . . . . . . . . . . . . . . . . . . . . . main.cpp File. . . . . . . . . . . . . . . . . . . . . . VectorLength and VectorSubtract Functions . ApplicationProc Function. . . . . . . . . . . . Dialog Procedures . . . . . . . . . . . . . . . Main Loop. . . . . . . . . . . . . . . . . . . . network.cpp File . . . . . . . . . . . . . . . . . . . . StartConnection Function . . . . . . . . . . . SendCommand Function . . . . . . . . . . . . SendStartGame Function. . . . . . . . . . . . SendRequestNonDeltaFrame Function . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. 427 . . 427 . . 428 . . 428 . . 428 . . 429 . . 430 . . 430 . . 430 . . 431 . . 436 . . 436 . . 437 . . 449 . . 449 . . 451 . . 452 . . 453 . . 454 . . 458 . . 467 . . 468 . . 469 . . 470 . . 472 . . 475 . . 477 . . 478 . . 478 . . 479 . . 481 . . 481 . . 481 . . 482 . . 487 . . 488 . . 488 . . 488 . . 489 . . 489 . . 489 . . 490 . . 502 . . 502 . . 504 . . 504

xii

Contents

Connect Function . . . . . . . . . . . . . . . . . . . Disconnect Function . . . . . . . . . . . . . . . . . ReadMoveCommand Function . . . . . . . . . . . . ReadDeltaMoveCommand Function . . . . . . . . . BuildDeltaMoveCommand Function. . . . . . . . . RunNetwork Function . . . . . . . . . . . . . . . . client.cpp File . . . . . . . . . . . . . . . . . . . . . . . . CArmyWar Constructor and Destructor Functions. InitializeEngine Function. . . . . . . . . . . . . . . Shutdown Function . . . . . . . . . . . . . . . . . . DrawMap Function . . . . . . . . . . . . . . . . . . Frame Function . . . . . . . . . . . . . . . . . . . . CheckVictory Function . . . . . . . . . . . . . . . . KillPlayer Function . . . . . . . . . . . . . . . . . . GetClientPointer Function . . . . . . . . . . . . . . CheckKeys Function . . . . . . . . . . . . . . . . . CheckPredictionError Function . . . . . . . . . . . CheckBulletPredictionError Function. . . . . . . . CalculateVelocity Function . . . . . . . . . . . . . . CalculateHeading Function. . . . . . . . . . . . . . PredictMovement Function . . . . . . . . . . . . . MoveObjects Function . . . . . . . . . . . . . . . . lobby.cpp File . . . . . . . . . . . . . . . . . . . . . . . . RefreshGameList Function. . . . . . . . . . . . . . RefreshJoinedPlayersList Function . . . . . . . . . Other Unimplemented Functions . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . .

504 505 505 506 508 509 511 525 526 527 528 530 531 532 532 532 533 534 535 536 537 539 541 541 541 542 542

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543

xiii

About the Authors

Andrew Mulholland has a BSc (Hons) in Computer Games Technology and is a partner in a games development company based in Scotland called Hunted Cow Studios (www.huntedcow.com). The companys current project is an online gaming web site called CowPlay.com, which currently offers free multiplayer games.

Andrew Mulholland

Teijo Hakala is a software engineer from Jyvskyl, Finland, who specializes in network programming, game programming, and optimization. He also has wide work experience with computer technology.

Teijo Hakala

xiv

Introduction
With Internet technology developing rapidly and the use of broadband Internet services increasingly common, Internet computer gaming has become ever more popular, while documentation on how to develop Internet games remains inadequate. Programming Multiplayer Games provides in-depth coverage of all the major topics associated with online game programming, as well as giving the programmer easy to follow, step-by-step tutorials on how to create a fully functional network library, back-end MySQL database, and a complete, working online game. The book contains two main parts. The first explains practical theory on how to utilize MySQL, PHP4, sockets, and basic Windows programming. The second part consists of five extensive tutorials, leading you through the stages of creating a working online game, which you can both learn from and expand upon. After reading this book, you will have a solid knowledge of online game programming and you will also be able to start making your own online games. Also note that the companion CD contains all the source code from the book and a ready-to-use version of the network library you will create in the tutorial section. We hope you enjoy reading and learning from this book as much as we have enjoyed writing it!

This page intentionally left blank.

Part I

Theory
The theory section of this book is full of practical information that will help you understand how to make functional online games. We recommend that you read through this section thoroughly before attempting the tutorial section, as there is a lot of knowledge that will benefit you here. This section first covers the basics of dialog-based Windows programming, which we will utilize in the tutorial section to create our login and lobby system for the sample online game. Then we cover how to use MySQL and PHP to create a back-end database for your game, allowing you to interact with game data directly from a web browser. We also give an introduction to TCP/IP and sockets, followed by how to get started with sockets programming. Finally, we learn about different ways to send data and how to modify the behavior of our sockets.

This page intentionally left blank.

Chapter 1

Introduction to Windows Programming

Introduction
The most essential knowledge anyone can have is the basics. If you already know how to create dialog-based Windows applications, you can skip this chapter, but if you do not, this chapter will give you a quick and easy introduction so that you will find the rest of this book more accessible.

Windows Messaging System

Windows controls everything through the use of its messaging system. This is a fundamental idea to grasp if you wish to create any Windowsbased applications. Tasks to be processed by the operating system are stored in a queue. For example, when a user clicks a button in a window, a message is added to the queue and is then sent to the appropriate window to inform it that the button has been pressed.

Chapter 1 / Introduction to Windows Programming

When the operating system creates a window, the window continually checks for messages being sent to it. If it receives a relevant message, it will react accordingly; otherwise, it will send it back to the operating system to be reprocessed. Each window created is assigned a unique handle that is used to control and determine which messages are relevant to that window. In code, this is defined as the HWND, or window handle. The main reason behind this system is to allow for multitasking and multithreading. This means that the operating system can run more than one application in one instance, even though a processor can only handle one task at a time. There is a lot more to windows than this, but this should give you a reasonable overview of how the system works.

Creating a Window
Load up Microsoft Visual Studio and select File, New The following dialog box is now visible in the center of the screen.
Figure 1-1

Select the Projects tab at the top of the dialog and then choose the Win32 Application option on the main display. Select the location for your project, enter your projects name, and click OK. Next, select the type of project you wish to create. Leave it on the default option (An empty project) and click the Finish button. A project information box is now visible; simply click OK in this box. Now we are working with the Visual Studio main interface. Currently the ClassView is active, but we are interested in the FileView, so select this tab.

Chapter 1 / Introduction to Windows Programming

Figure 1-2

The FileView is a list of all the C and C++ source and header files that are active in your project. Currently we do not have any files in our project, so we need to add our main C++ source file. Select File, New as you did before, but this time we will be using the Files tab instead of the Projects tab. The following dialog will be visible.
Figure 1-3

Select the C++ Source File option as shown in Figure 1-3 and type in the filename as main.cpp. Now click the OK button to add this empty file to your project. You now have your main source file in your project and it is visible in the Visual Studio editor. There are two main items required in a standard Windows program: the entry point to your program, which is named WinMain, and the Windows callback procedure, commonly named WndProc, which is used to keep your Windows application up to date. For what we require though, it is best to take the dialog approach, making it even simpler to design and code. First, we need to add our dialog, so click File, New again, but this time you want to add a resource script. Type in the filename as resource and click OK. Once this is done, you will notice another tab has appeared between the ClassView and FileView tabs. This tab is called the ResourceView; it allows you to visually create and edit dialogs for use within your program.

Chapter 1 / Introduction to Windows Programming

Figure 1-4

Once you select the ResourceView tab, you will be presented with the resource editor. Right-click on resource.rc in the main view and then left-click on the Insert option. You will then be presented with the following dialog box.
Figure 1-5

Select the Dialog option and click the New button. Now you will see a sample dialog box in front of you. For now, we will not do much to it except change the name of the title bar and its identifier, which I explain in the following code. Double-click on the sample dialog box that Visual Studio created. Now a dialog properties box can be seen. All we are interested in here is the ID, which will probably be set to IDD_DIALOG1, and the Caption, which should be Dialog. Lets change the ID to IDD_CLIENT and the Caption to Window Example. Its time to go back and do some code now. We have our dialog template that we can call from our code, so lets do it. Here is the code required to make your dialog window appear on the screen. The OK button on the dialog can be pressed but will have no action, whereas the Cancel button will close the dialog.
// Simple Windows Code #include <windows.h> #include "resource.h" LRESULT CALLBACK ClientDlgProc(HWND DialogWindow, UINT Message, WPARAM wParam, LPARAM lParam) { // Process Messages switch(Message) { case WM_INITDIALOG: return FALSE; case WM_COMMAND:

Chapter 1 / Introduction to Windows Programming

switch(wParam) { case IDCANCEL: EndDialog(DialogWindow, FALSE); return FALSE; default: break; } break; default: break; } return FALSE; } int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, LPSTR lpCmdLine, int nCmdShow) { DialogBox((HINSTANCE) hInstance, MAKEINTRESOURCE(IDD_CLIENT), NULL, (DLGPROC) ClientDlgProc); return 0; }

NOTE If you get an error that tells you it cant find afxres.h, you need to install MFC support for Visual Studio, which comes with the Visual Studio package.

If you have never seen Windows code before, the above code may look complex and a little confusing. Welcome to the world of Windows! Well, its not that bad, honest. Lets start with the WinMain function. This is simply the point at which Windows starts executing your code. Do not worry about the variables that are passed in; they are controlled by Windows and are beyond the scope of this book. The main issue here is the DialogBox function and the ClientDlgProc callback procedure that creates our dialog window on the screen. The first parameter is the instance of the application that you simply take from the first parameter of the WinMain function. Next is the identifier that we set when we created the template for our dialog. The third parameter is of no interest to us so we set it to NULL, but the final one is. This is a pointer to the update function for the dialog. Each dialog you create requires this update function (basically the same idea as a Windows procedure). In this update function is where you set the actions for buttons and other useful tools. So we set this update function to our callback function for the dialog (ClientDlgProc).

Chapter 1 / Introduction to Windows Programming

For example, the identifier for the Cancel button is IDCANCEL. As you can see in the code, there is a case statement for the Cancel button so when it is clicked, it will close the dialog window. Other buttons can be easily added to the template using the toolbox on the template editor. Just remember that each button must contain a unique ID so you can reference it from within your code.

Sending Information to Your Window

In addition to being useful for debugging, being able to update information to a window is essential knowledge and can be used in many situations, such as displaying how many players are connected to the game server. First you have to add a static text string to the dialog window. To do this you need to go back to the template editor by selecting the ResourceView tab as before. Then you simply double-click on the IDD_CLIENT text as seen in Figure 1-6 to bring up your dialog in the main area. Next, select the Aa button from the Controls toolbox and place it somewhere on your dialog as seen in the figure. Now double-click on the text you added to the dialog box to display its properties.
Figure 1-6

The following dialog will now be visible on the screen. All we need to change here is the ID. Change the text IDC_STATIC to IDC_ SERVERSTATUS. This will give it more meaning when it comes to adding it into the code.
Figure 1-7

Chapter 1 / Introduction to Windows Programming

Now that we have some text, we want to be able to set it to a value from within our code. For example, if we want the text to read Server Online, add this line of code after the line that contains case WM_INITDIALOG:
SendDlgItemMessage(DialogWindow,IDC_SERVERSTATUS, WM_SETTEXT,NULL,(long)"Server Online");

Then when the dialog box is initialized, Windows will send a message to the dialog box to tell it to update the IDC_SERVERSTATUS text with the string you supplied in the function. In this case, it would update the text from Static to Server Online. The first parameter is the handle to the dialog, which is the first variable that is passed into the dialog update procedure. The second is the identifier for what you want to change or update. Next comes the command that you wish to send. There are many commands and the best way to figure out how they work is just to experiment with them. For now we are using WM_SETTEXT, which tells Windows to change a static text string in a dialog box. The fourth variable is not used for the WM_SETTEXT command, so we simply set it to NULL. The final variable is used to declare the string that we want to update the static text with, in this case Server Online. Also note that the string must be typecast to a long.
TIP Try experimenting with editable text. It works on the same principles, and you simply send a WM_GETTEXT message to retrieve what the user entered.

Static Link Libraries

Later when we create our online tutorial game, we will be using static link libraries to encapsulate all our network and graphics code, which makes it easier to reuse for future projects. As well as the reusability factor, static link libraries also protect your source code while allowing others to use the functionality of your code. When creating a static link library, a WinMain function or a Windows update function (WndProc) are not required. This is because we are not actually creating a program but rather just a collection of functions that we can use from within our programs. The library is created using standard C/C++ source and header files, but the output is a library rather than a Windows executable. When you compile your library, the compiler will generate a library file with the extension .lib. To use the functions in your library, you are required to include it in your project and include the header file that

Chapter 1 / Introduction to Windows Programming

was used to create the library (which contains all the external variables, types, and function prototypes). The easiest way to use your library is to create Lib and Include folders for your library and include those directories in the Visual Studio directory settings that are explained in the sections to follow.

Creating a Static Link Library

When creating a static link library, the best thing to do is to create a directory structure on your hard drive for the library to be stored in. From our experience, we recommend the structure shown in Figure 1-8.
Figure 1-8

n n

Examples All example programs that display how to use your library are stored in this folder. This is probably one of the most useful things that can accompany your library since the source code is not visible to any other programmer using it. Formats This is where you store any file formats specific to your static link library (i.e., your own 3D model format). Include This stores the entire collection of C/C++ header files that are needed to use your library. This one of the directories that you must set up in Visual Studio to make your library work. Lib This is where you actually store your complete library file. It is a good idea to include both the release and debug versions of your library here so that it is easier to debug programs created with it. This is the other directory required by Visual Studio. Source All source code related to your library must be kept in this folder. Tools Any programming tools that are used alongside your library are stored here (such as file format converters).

n n

Now that we have our structure, we need to create a static link library project. This is done by selecting File, New in Visual Studio. The following dialog is now visible.

Chapter 1 / Introduction to Windows Programming

Figure 1-9

Select the Win32 Static Library option and enter a name for your library. Next, select the location and press the OK button. In the dialog that appears, leave both the Pre-Compiled header and MFC support options unchecked as we will not be using either in this book, and then click Finish. A project information box is now visible; simply click OK. Next, we need to add our source and header files to the project as you did when creating the window. Remember, this time we do not require the WinMain or update procedure, just functions that we wish to reuse. Lets call the source file library.cpp and the header file library.h for this example. If you now press F7, it will build your library and put it in the Release or Debug folder depending on your project configuration. The library will have the same name as your project. For example, if your project is named GamePhysics your library will be gamephysics.lib. It is a good idea to make Visual Studio automatically copy your header file and library to the correct directories in your structure to assure that you are always using the latest version. Selecting Project, Settings from the main menu makes the following dialog visible.

Chapter 1 / Introduction to Windows Programming

Figure 1-10

Set the Output file name for your library and choose where you want it to go relative to the directory your source files are in. This must be done for both release and debug configurations. In the previous dialog the Debug settings are currently active.
TIP It is a good idea to have both debug and release versions of your library. Call your release version the correct name (i.e., GamePhysics.lib) and add a suffix to the debug version (i.e., GamePhysicsDebug.lib) to distinguish them easily.

Now the library file is created in the correct directory in our structure when the project is built. We also want to copy our header file to the Include directory of our structure. As can be seen in Figure 1-11, you simply add a post-build command that tells the compiler to copy the header file(s) to the Include directory.

Chapter 1 / Introduction to Windows Programming

Figure 1-11

We have now covered the basics on how to create a static link library. The most important thing to remember is to prototype all functions and extern all global variables you wish to be accessible outside of your library. In the next section, we discuss how to set up Visual Studio to find your library.

Using a Static Link Library

Once you have built your library and it is in the correct directories, you need to tell Visual Studio where to look for it. Select Tools, Options from the main Visual Studio menu to open the Options dialog box. If the Directories tab is unselected, select it now to display a list of directories as shown in Figure 1-12. Only the top three should be visible unless you have previously added other static libraries.
Figure 1-12

Chapter 1 / Introduction to Windows Programming

You must now add the Include directory of your library for the Include files and the Lib directory for the Library files. You select which one you wish to add by changing the top-right drop-down box. Now Visual Studio is able to find and recognize your static link library. To use it in a program you have written, you must first load your project into Visual Studio, then select Project, Settings Next, select the Link tab and add your library name before kernel32.lib. Also, remember to include the header file for your library in your main code.

Summary
This chapter discussed the basics to creating dialog-based applications in Visual Studio. The best thing to do is experiment by creating dialogs, adding buttons, and making the buttons set strings to different values when you press them. Also, learn to use editable text as this is highly useful and allows the user to give feedback to your program. If you are interested in learning more, there are entire books written on Windows programming and there is also an excellent resource available on the Internet (http://msdn.microsoft.com), but the quick introduction provided here will give you enough knowledge to understand the concepts used in this book.
TIP Windows 98 Programming from the Ground Up by Herbert Schildt (ISBN: 0-07-882306-4) provides an excellent way of understanding Windows programming.

Chapter 2

Using Databases
Introduction
This chapter covers how to create a stable and fast database system for your game server. Although it is possible to store player information on the players local machine, it makes much more sense to store the data on the server for many reasons. The most important reason is to prevent players from running hacks on their computers to change their character data in the game or even back the data up to another file. If the data is stored on the server, it makes it impossible for the player to run a local game hack that will modify the character data. Another reason for this is to allow players to play from different locations and machines without having to copy their character data to another computer. In this chapter, we cover how to install and use MySQL and then look at how to access a database from within a C++ application (aka a game).

What Is MySQL?
MySQL is an open source relational database management system. Its purpose is to store and allow easy and fast access to data.

Chapter 2 / Using Databases

Installing MySQL
Installing MySQL is relatively painless. Firstly, you need the installation program for MySQL, which is available on the CD that accompanies this book. Alternatively, you can download the latest version from http://www.mysql.com.
NOTE If you choose to install MySQL in a folder other than the default (C:\MYSQL) or you wish to start MySQL on NT/Win2000 as a service, you need to create a file named MY.CNF in the root of your C:\ drive with the following information in it (or append the following information to either \Windows\my.ini or \winnt\my.ini, depending on whether you are using Windows 98 or 2000/XP , respectively):
[mysqld] basedir=E:/installation-path/ datadir=E:/data-path/

After you have installed MySQL, the installation directory will contain a file named my-example.cnf. You can use this as a template to create your own my.cnf file.

When you start the installation, you will be asked which type of setup you would prefer: Typical, Compact, or Custom. Here we select the Typical option, as it will install all the components we require in order to work with MySQL.
Figure 2-1 Selecting the Typical install option

After selecting Typical, simply press the Next button to automatically complete the installation process.

Chapter 2 / Using Databases

Now that we have MySQL installed, first we will browse the directory so we can see what it has installed for us. If we open up the directory to which we installed MySQL to (typically c:\mysql), we can see the following directory structure.
Figure 2-2 The MySQL directory structure

All we really have use for here is the bin directory, which contains the MySQL server and client executables, and the docs directory, which contains the HTML version of the MySQL manual.
NOTE An Adobe PDF version of the MySQL manual is supplied on the companion CD. We find this easier to read, but you will also require the free program Adobe Acrobat Reader to view this manual, which is also on the CD and available to download at http://www.adobe.com/products/acrobat/readermain.html.

Lets now take a look in the bin directory and see what is of use to us there. The contents can be seen in the following image.

Chapter 2 / Using Databases

Figure 2-3 The contents of the bin directory

As you can see, there are many executables in this directory; some are daemons (i.e., the MySQL server) and some are console-based clients for accessing the MySQL server.
NOTE A daemon is simply a program or process that sits idly in the background until it is invoked to perform its task.

The executable mysql.exe is a console-based client, which is used to interact with the MySQL server using the SQL language. We cover SQL in great depth later in this chapter. If we try to run the console client (mysql.exe) now, the following screen will appear for a couple of seconds and then disappear.
Figure 2-4 A blank window?!

Why? The reason the window appears and promptly disappears is because there is currently no MySQL server to connect to i.e., there is no MySQL daemon running for the client to interact with.

Chapter 2 / Using Databases

So the obvious thing to do now is to run a MySQL server so we can access it via the client. The best way to do this is to run the winmysqladmin.exe file located in the bin directory. The first time this is run it will ask for a username and password. This isnt really important for testing, so just enter something like admin for the username and the password. After that, you will see a small traffic light icon appear in your system tray with the green light indicating the MySQL server is running. Note that MySQL will now automatically run every time Windows is booted up.

SQL Statements
Now that we have the MySQL server running, it is time to load up the MySQL console client. This is done the same way as we loaded it before (i.e., run the mysql.exe executable that is located in the c:\mysql\bin\ directory). In SQL (structured query language), there are two types of statements we can execute. These are DDL (data definition language) and DML (data manipulation language). DDL is used to affect the structure of the database, such as adding databases, tables, etc., whereas DML is used to add and modify data in an existing database and retrieve information. The following sections cover these statements.

Data Definition Language

First we take a look at DDL, as we require a database to work with before we can manipulate any data within it.

Creating and Dropping Databases

First, lets see what databases already exist in the MySQL server. We can view this information by using the SHOW command. Enter the following (except the mysql> part) into the MySQL console client, followed by a Return:
mysql> SHOW DATABASES;

When you press Return, the following should be visible in the console display:

Chapter 2 / Using Databases

Figure 2-5 Dialog box

As you can see, there are two databases already created in the MySQL server. The mysql database contains administration information for the MySQL server and should not be modified. The other database, test, is exactly what it is called a test for the MySQL server. Again it is not a good idea to remove it.

Creating a Database
So how do we add our own database to the MySQL server? By using the CREATE command. To create a database called mydata, we would use the following syntax.
mysql> CREATE DATABASE mydata;

Notice that the semicolon is added after every command in the SQL language. When we press Return after entering this command, the console informs us that the query was okay. This can be seen in Figure 2-6.
NOTE The following length and character restrictions are imposed on the names of databases, tables, columns, and aliases. Table 2-1 Naming restrictions Identifier Database Table Column Alias Max Length 64 64 64 255 Valid Characters All valid directory name characters except . and /. All valid directory name characters except . and /. All are valid All are valid

Chapter 2 / Using Databases

Figure 2-6 Creating a database

Now that we have created the database, we can ensure it is on the server by again using the SHOW command as follows:
mysql> SHOW DATABASES;

When we press Return after entering this command, we can see that our database has been added to the list (note that the list is in alphabetical order, not the order in which they were created).
Figure 2-7 The mydata database has been added to the list.

Dropping a Database
Now we will remove a database from the server. Note that when we do this all data (if any) will be lost. To remove a database, we drop it from the server by using the DROP command. So to drop our new mydata database we would use the following command:
mysql> DROP DATABASE mydata;

When we execute this command by pressing Return, the query will be reported as OK, as shown in the following screenshot.
Figure 2-8 Dropping a database

Lets again list the databases using the SHOW command:

mysql> SHOW DATABASES;

Chapter 2 / Using Databases

Notice that our mydata database is no longer visible in the list.

Figure 2-9 Database listing after the DROP command

Column (Field) Types in MySQL

Before we discuss the creation of tables within databases, now is a good time to discuss the different column types we can have in tables. Each column in a table must be assigned a type that represents the kind of information that field is going to hold. Table 2-2 lists the available types you can use.
Table 2-2 MySQL column types Type TINYINT SMALLINT MEDIUMINT INT BIGINT Description A very small integer. Signed range is 128 to 127. Unsigned range is 0 to 255. A small integer. Signed range is 32768 to 32767. Unsigned range is 0 to 65535. A medium-sized integer. Signed range is 8388608 to 8388607. Unsigned range is 0 to 16777215. A normal-sized integer. Signed range is 2147483648 to 2147483647. Unsigned range is 0 to 4294967295. A large integer. Signed range is 9223372036854775808 to 9223372036854775807. Unsigned range is 0 to 18446744073709551615. A small, single-precision floating-point number that cannot be unsigned. Signed range is 3.402823466E+38 to 1.175494351E38, 0, and 1.175494351E38 to 3.402823466E+38. A double-precision floating-point number that cannot be unsigned. Signed range is 1.7976931348623157E+308 to 2.2250738585072014E308, 0, and 2.2250738585072014E308 to 1.7976931348623157E+308.

FLOAT

DOUBLE

Chapter 2 / Using Databases

Type DECIMAL

Description An unpacked floating-point number that cannot be unsigned. Works like a char column in that the number is stored as a string (i.e., each number uses one character in the string). A date. Range is 1000-01-01 to 9999-12-31 and is in the format YYYY-MM-DD. A time. Range is 838:59:59 to 838:59:59 and is in the format HH:MM:SS. A combination of date and time. Range is 1000-01-01 00:00:00 to 9999-12-31 23:59:59 and is in the format YYYY-MM-DD HH:MM:SS. A year in two- or four-digit format (default is four-digit). Range is 1901 to 2155 and also 0000. A timestamp. Range is 1970-01-01 00:00:00 to sometime in the year 2037 in the format of YYYYMMDDHHMMSS. A fixed-length string that is always right-padded with spaces to the specified length when stored. The range is 1 to 255 characters depending on the length specified. A variable-length string. A tiny binary object with a maximum length of 255 characters. *See Note below table. A binary object with a maximum length of 65,535 characters. *See Note below table. A medium binary object with a maximum length of 16,777,215 characters. *See Note below table. A large binary object with a maximum length of 429,496,295 characters. *See Note below table. An enumeration. A list of string values of which only one can be selected. Maximum of 65,535 distinct values. A set. A string object that can have zero or more values, each of which must be chosen from the list (i.e., 'val1', 'val2', etc.). Maximum of 64 members.

DATE TIME DATETIME

YEAR[(2|4)] TIMESTAMP CHAR(length)

VARCHAR TINYBLOB/ TINYTEXT BLOB/TEXT MEDIUMBLOB/ MEDIUMTEXT LONGBLOB/ LONGTEXT ENUM('val1','val2'...) SET('val1','val2'...)

NOTE The only difference between the BLOB and TEXT types is that for sorting and comparisons, a BLOB is case sensitive whereas the TEXT type is not.

Chapter 2 / Using Databases

Adding, Modifying, and Dropping Tables

The following sections discuss how to manipulate tables in databases.

Creating Tables
Now that we know the possible types for the columns in our tables, lets look at how we actually go about creating a table. Lets say we wish to create a table to hold some user details within a database. In fact, we will be using a similar table later in the tutorial section. We want to store the users title, first name, last name, age, email address, and the date they were added to the database. This will require the following columns: Title Firstname Surname Age Email DateAdded Before we get into how to actually add the information, lets first think about how we are going to store it. Or, more to the point, what types we require for each of the columns. For the title, firstname, and surname, we can use the TEXT type as it contains plenty of characters to allow for all possibilities. For age, an unsigned TINYINT would be an obvious choice as these are numerical and no one has ever been known to live past 255. For email address, we can again use a TEXT type, as it will give us substantial storage space for the address. And finally, for the date that the user was added to the table, we can use a TIMESTAMP. This is very useful in that the time and date can be retrieved automatically into the database; this is discussed later in the chapter. So now that we know what types we want for our columns, we first need to create a database to add the table into. This goes back to what we learned in the previous section. Lets create a database called myinfo with the following command:
mysql> CREATE DATABASE myinfo;

When we execute this command, the console should report that the query was okay. We can now check that our database has been created by using the following command:
mysql> SHOW DATABASES;

Chapter 2 / Using Databases

When we execute this command, the following should be visible in the console:
Figure 2-10 The myinfo database is now visible in the console after using the SHOW DATABASES command.

Now we need to tell MySQL that we wish to perform actions on the myinfo database. This is accomplished by using the USE command:
mysql> USE myinfo;

After executing this command, any DDL (data definition language) and DML (data manipulation language) statements that are executed will affect the database in use, which in this case is our myinfo database. With the database set up and ready to accept commands, we can create our table (which we will call userinfo) with the following statement:
mysql> -> -> -> -> -> -> -> -> CREATE TABLE userinfo ( id INT auto_increment, title TEXT, firstname TEXT, surname TEXT, age TINYINT, email TEXT, dateadded TIMESTAMP, PRIMARY KEY(id));

Lets break this up a little so we can see what is going on. First, we declare that we wish to create a table by entering CREATE TABLE. Next we specify the name we wish to call the table, in this case userinfo. Then within parentheses we list all of the column names and types we require. Note how we have added an extra field named id. This makes it easier to handle data in a relational way, as we discuss later in this chapter. Finally, note the addition of the primary key as the last parameter. This is used to determine how the table is optimized within the database. Again, we discuss the use of keys later in this chapter. We can now see if our table was created successfully by executing the following command:
mysql> SHOW TABLES;

Chapter 2 / Using Databases

When this is executed, the following output should be visible in the console:
Figure 2-11 Here the userinfo table can be seen as part of our database.

Note that you can also view the columns in a table by using the following command:
mysql> DESCRIBE userinfo;

When you execute this command, the console will display all the details for each of the columns in the userinfo table, as shown in Figure 2-12.
Figure 2-12 Describing the userinfo table

This information can be useful to ensure the table was created as you envisioned and to recap the columns a table contains at a later date.

Modifying Tables
Now lets see how to modify a table. Modifying a table can range from simply changing the type of one of the columns to adding a complete new column or removing an existing column. Lets first look at how we can change the name of an existing column. In our userinfo table we have a column called firstname, but lets now change this to read forename instead. To make this change we need to use the following syntax:
mysql> ALTER TABLE userinfo CHANGE firstname forename TEXT;

Note how we also must supply the data type for the column as well as its old and new names. Here is how this should look in the MySQL console client:

Chapter 2 / Using Databases

Figure 2-13 Modifying a column name

If we now describe the userinfo table with the following command:

mysql> DESCRIBE userinfo;

we can see that the firstname column has been renamed to forename:
Figure 2-14 Description of the updated userinfo table

We can also change the data types of columns in tables. To change the age column from a TINYINT to an INT, we would use the following command:
mysql> ALTER TABLE userinfo MODIFY age INT;

After executing this command and then using the DESCRIBE command on this table, we can see the type has changed to INT.
Figure 2-15 Now the age column is of type INT rather than TINYINT.

Finally, it is good to know how to remove fields from a table when they are no longer required. Lets say we no longer require the email field in our userinfo table. What we want to do is drop the field from our table,

Chapter 2 / Using Databases

just as we did earlier in the chapter when we dropped the database. Here is the syntax for removing the email field:
mysql> ALTER TABLE userinfo DROP email;

Here is how this looks in the MySQL console:

Figure 2-16 Dropping a field from a table

Once this command is executed, if we describe the table with the following command:
mysql> DESCRIBE userinfo;

we can see that the email field has been removed from our userinfo table.
Figure 2-17 As you can see, the email field has now been removed.

Dropping (Removing) Tables

Removing tables from a database is very simple, but without careful use can have disastrous effects. The main thing to note when you drop a table is that you also lose all the data contained within the table. Therefore, it is always wise to back up a database before executing any drop commands. We will look at backing up a database later in this chapter. Lets now look at how we drop the userinfo table from our myinfo database. To do this, we need to execute the following command in the MySQL console client:
mysql> DROP TABLE userinfo;

Chapter 2 / Using Databases

Once this command is executed, we can verify that the table has been removed by using the SHOW command to see what tables are currently in our myinfo database:
mysql> SHOW TABLES;

As you can see from Figure 2-18, the table no longer exists in the database.
Figure 2-18 After dropping the userinfo table, we have an empty database.

Data Manipulation Language (DML)

In this section, we look at how to add, modify, and remove data from tables in the database using DML (data manipulation language). Without this knowledge, we would not really have any use for a database; it would simply be a static entity with no purpose. Lets first create a database and table to work with in this section, using the DDL we learned in the previous section. Our database will be called dmlexample, so lets create that with the following statement:
mysql> CREATE DATABASE dmlexample;

We then need to specify that we wish to use the new database by executing the following statement:
mysql> USE dmlexample;

The console should now inform us that the database has changed, as seen in Figure 2-19.
Figure 2-19 Creating the dmlexample database

Chapter 2 / Using Databases

Now we need to create a table called sampletable. This table will contain the following fields: Username Password Age Email DateCreated This is accomplished using the knowledge we learned in the section on DDL statements. First we need to specify that we wish to use the dmlexample database with the following command:
mysql> USE dmlexample;

Once this is done, we can create our table with the following DDL statement:
mysql> -> -> -> -> -> CREATE TABLE sampletable ( username TEXT, password TEXT, age INT, email TEXT, datecreated TIMESTAMP);

Figure 2-20 shows how this should look in the MySQL console client:
Figure 2-20 Creating our sample table

Inserting Data
To add rows to the table, we use the INSERT command. Here is how we would add a single row to our sampletable:
mysql> INSERT INTO sampletable VALUES (andrew, qwerty, 20, andrew@huntedcow.com, NULL);

Figure 2-21 shows how this looks when we enter it into the MySQL console. Notice how the feedback from the console tells us that one row has been affected. This means we have added one row to our sampletable table.

Chapter 2 / Using Databases

Figure 2-21 Inserting a single row of data

We can then use the SELECT command to view the data in the table. We go into more detail about this command later in this chapter as it is very important, but for now we just use it without further explanation. Lets use SELECT to see what data is in our dmlexample table:
mysql> SELECT * FROM sampletable;

When we execute this statement, the following will be visible in the MySQL console:
Figure 2-22 Viewing the new row in the table

NOTE Notice how the datecreated field reflects the time and date when we added the row to the table. This is because we specified NULL when we added the row; doing this will make a TIMESTAMP field grab the current date and time from the system by default.

It is also possible to add several rows of data in a single command. Lets try this now by adding another three rows to our table in a single INSERT command, as shown below:
mysql> -> -> -> INSERT INTO sampletable VALUES ('teijo', 'mrt', 21, 'teijo@kanetti.fi', NULL), ('jim', 'letmein', 23, 'jim@email.net', NULL), ('wes', 'opensesame', 31, 'wes@email.net', NULL);

Chapter 2 / Using Databases

When we execute this command, the following can be seen in the MySQL console client:
Figure 2-23 Inserting multiple rows in a single statement

As you can see, this time the feedback from the console indicates that three rows have been affected; thus, we have added three rows to our table. We can verify this by again using the SELECT command:
mysql> SELECT * FROM sampletable;

When this is executed, you will see that the table contains four rows (or records, if you like) of information. Figure 2-24 shows the MySQL console after the SELECT statement has been executed.
Figure 2-24 Now we have four rows in the table.

Modifying Data
To modify existing data in a table we use the UPDATE command. First lets try to change all the passwords in all the rows in the table to changeme. This can be accomplished with the following statement:
mysql> UPDATE sampletable SET password = changeme;

When we execute this statement, the console will inform us that four rows have been affected as we have changed the password for every row in that table. We can see the effect on the table by using the SELECT command as follows:
myql> SELECT * FROM sampletable;

Chapter 2 / Using Databases

Here is a screenshot of this command being executed in the console:

Figure 2-25 The password field has been updated in all of the rows.

But what if you only want to update a single row? Lets say we wish to change the password for Teijo from changeme back to mrt. We would use the following statement to do this:
mysql> UPDATE sampletable SET password = mrt WHERE username = teijo;

When we execute this command in the console, it informs us that one row has been affected. This is because it will only update the password field if the username field is equal to teijo. If we use the SELECT command on the table now, we can see that only Teijos password has changed, as shown in Figure 2-26.
Figure 2-26 Updating only a single row

We can also use this technique to update only certain fields. For example, we could change all the passwords of the people who are 30 years old or younger. Here is the command to do this:
mysql> UPDATE sampletable SET password = young WHERE age <= 30;

When we execute this command, it will inform us that three rows have been affected, as three of the four records in our table have an age field that is 30 or less. If we then use the SELECT command we can see the following output in the console:

Chapter 2 / Using Databases

Figure 2-27 Conditional updates

TIP

A useful idea is to update a TIMESTAMP field with NULL. This will retrieve the latest time from the system the database is running on. A practical use would be to note the last time a player logged in.

Removing (Deleting) Data

Removing data from a table is done very similarly to updating data. Lets look at how to delete a single row of data from the database using the following statement:
mysql> DELETE FROM sampletable WHERE username = teijo;

When we execute this command, the MySQL console client will inform us that one row was affected. If we now use the SELECT command on the table, the following can be seen in the console:
Figure 2-28 Deleting a single row

Again, as with the update statements, we can specify conditions to allow us to delete, for example, everyone who is younger than 30. Lets do this now with the following statement:
mysql> DELETE FROM sampletable WHERE age < 30;

Chapter 2 / Using Databases

When we execute this statement, the client will inform us that two rows have been affected, or in this case deleted. If we now use the SELECT command on our table, we will see that only one row is left in the table.
Figure 2-29 Conditional deleting

Finally, it is also possible to delete all the rows from a table in a single statement. All we need to do for this is not specify any condition, as we did when we updated all the password fields to changeme earlier. Here is the statement we require to delete all the rows in a table (i.e., empty the table):
mysql> DELETE FROM sampletable;

After executing this, if we select all the information in the table using the SELECT command, the following will be shown in the console:
Figure 2-30 Deleting all the data from a table

As you can see in Figure 2-30, the table now contains no information as it has all been deleted.

Using SELECT Statements

So far we have simply used the following command to show all the data in our sampletable table:
mysql> SELECT * FROM sampletable;

Chapter 2 / Using Databases

What this actually does is it fetches all the fields from the sampletable and returns them. The * is a wildcard which basically means it represents anything (or in this case, any field). Before we discuss the SELECT statement further, lets add some data to experiment with to our sampletable table. Use the following statement to insert some data:
mysql> -> -> -> -> -> INSERT INTO sampletable VALUES ('andrew', 'qwerty', 21, 'andrew@huntedcow.com, NULL), ('andrew', 'letmein', 27, 'andrew@email.net', NULL), ('george', 'paper', 19, 'george@email.net', NULL), ('jenny', 'jen999', 27, 'jen@email.net', NULL), ('sandra', 'sdra2', 27, 'sandra@email.net', NULL);

Figure 2-31 shows how this should look when we enter it into the console and execute it.
Figure 2-31 Inserting our new data into the sampletable table

Now that we have added our data into the table, if we use the SELECT statement with the wildcard (*) as we were doing before, it will retrieve and display all of the information from the table to the console. Lets try this with the following statement:
mysql> SELECT * FROM sampletable;

Here is a screenshot of the output from the console:

Figure 2-32 Using the wildcard with a SELECT statement

Chapter 2 / Using Databases

As you can see, the statement has retrieved all of the information from the table, that is, all of the rows, and all of the columns contained in each of the rows. Lets say all we want to retrieve is the password field. To get all of the passwords from the sampletable table, we would use this statement:
mysql> SELECT password FROM sampletable;

When we execute this statement, we can expect the following output from the console:
Figure 2-33 Retrieving only a single column

Notice how we simply replace the wildcard (*) with the column we wish to retrieve. We can also retrieve multiple columns by using a comma to delimit them. To select both the username column and password column only, use this statement:
mysql> SELECT username, password FROM sampletable;

When we execute this statement, we can see in the console that only the username and password fields have been selected from the table, as shown in Figure 2-34.
Figure 2-34 Retrieving multiple columns

We know how to retrieve individual fields from the tables, but how do we retrieve a single row? We can easily apply a condition to a SELECT statement, just as we did when we were updating the table and deleting

Chapter 2 / Using Databases

from the table. Using a conditional SELECT statement, lets only display Jennys information from the database. Here is the statement to do this:
mysql> SELECT * FROM sampletable WHERE username = jenny;

When we execute this statement, only Jennys details will be displayed in the MySQL client console, as shown in Figure 2-35.
Figure 2-35 Selecting a single row

We can also select specified fields, such as finding the password that related to a username. Here is how we would retrieve the password that belongs to George:
mysql> SELECT password FROM sampletable WHERE username = george;

When we execute this statement, we can see that only a single field is displayed Georges password, as shown in Figure 2-36.
Figure 2-36 Selecting a single row with specified columns

In our sample data, there are two rows with the username andrew. If we use a conditional statement to get the password for Andrew, we will in fact get two passwords, one for each andrew entry in the database. Here is the statement that will give us that result:
mysql> SELECT password FROM sampletable WHERE username = andrew;

When we execute this statement, we can see that we have two passwords showing in the console. Here is a screenshot of this result:

Chapter 2 / Using Databases

Figure 2-37 The twopassword problem!

Later in this chapter we discuss a way around this problem with the use of relational databases and keys, but lets not go into that just yet. Instead, lets have a look at how the LIKE command can help us find needed information. Using LIKE is ideal for finding strings in databases, especially if you only have a part of the complete string (i.e., for a search engine). For example, lets say we wish to find someone in the database who has a name starting with the letter J. To accomplish this, we would use the following statement:
mysql> SELECT * FROM sampletable WHERE username LIKE j%;

When we execute this statement, we can expect the following output from the MySQL console:
Figure 2-38 Using LIKE with a SELECT statement

Notice here how Jenny was retrieved, as her username is the only one to start with a J. The % represents a wildcard when used with LIKE, so if we used the following statement instead:
mysql> SELECT * FROM sampletable WHERE username LIKE %j%;

it would mean that the letter J could appear anywhere in the string. Also note that you can have more than a single character, such as:
mysql> SELECT * FROM sampletable WHERE username LIKE %nny;

This would retrieve all of the people who have names that end with the text nny.

Chapter 2 / Using Databases

Relational Databases
We have been looking mainly at how to create database structures and do simple data manipulation within them. However, there are a lot of ideas and theories that make databases even more useful to us. Lets now look at what sort of structure we would want for a relational database. Think of a database that related players in a game to one another, for example to determine who was a friend of each player and, conversely, who was an enemy of each player. Lets first create a table to store the data for each of the players, with the addition of a primary key, which optimizes the database for searches on that particular column. Note also that every row of data in the primary key must be unique to one another. Here is the statement required to create our gamedata database and our playerdata table:
mysql> CREATE DATABASE gamedata; mysql> USE gamedata; mysql> -> -> -> -> -> CREATE TABLE playerdata ( username CHAR(255) UNIQUE NOT NULL, password CHAR(255), age INT, datecreated TIMESTAMP, PRIMARY KEY(username));

Notice here how we set the username column to be UNIQUE and also NOT NULL. In simple terms, this means that it must contain a value and that value must not be the same as any other username in any other record in the table. Note also that we have set the primary key of the table to be the username field, as we will be mainly searching on this field. Additionally, we need some way of storing a players friends and enemies. This is done by means of a link table. A link table is really just a normal database table, but its main purpose is to relate data in some way or another to conserve space and also optimize the way the database accesses the information. To create two link tables, one for relating friends and one for relating enemies to each other, use these statements:
mysql> CREATE TABLE relatefriends ( -> player CHAR(255), -> friend CHAR(255));

And also
mysql> CREATE TABLE relateenemies ( -> player CHAR(255), -> enemy CHAR(255));

Chapter 2 / Using Databases

If we show the tables in the database with the following command:

mysql> SHOW TABLES;

we can see from Figure 2-39 that our database now contains three different tables, our playerdata table and the two link tables.
Figure 2-39 Our three tables in the gamedata database

Lets add some sample data to the playerdata table so we can experiment with the link tables and understand how to use them effectively. Here is the statement to add our sample data into the playerdata table:
mysql> -> -> -> -> -> INSERT INTO playerdata VALUES ('Andrew', 'qwerty', 20, NULL), ('Henry', 'letmein', 34, NULL), ('Sandra', 'dra33', 19, NULL), ('John', 'j12d', 23, NULL), ('Jenny', 'jen123', 34, NULL);

If we select all the information from the playerdata table now using the following command:
mysql> SELECT * FROM playerdata

we can see in Figure 2-40 that all of our data is now in the playerdata table.
Figure 2-40 Our data in the playerdata table

Chapter 2 / Using Databases

Now that we have some sample data, lets create some relations between the players in the database. First, add to the relatefriends link table the fact that Henry is friends with Sandra. Here is the statement required to add this to the link table:
mysql> INSERT INTO relatefriends VALUES -> ('Henry', 'Sandra');

If we now show all of the data from the relatefriends link table, the following will be visible in the MySQL console:
Figure 2-41 Our data in the friends table

Lets add some more sample data into both the relatefriends and relateenemies link tables and then see how we can manipulate the data. Here are the two statements required to add the sample data:
mysql> -> -> -> -> INSERT INTO relatefriends VALUES ('Andrew', 'Henry'), ('Andrew', 'John'), ('Andrew', 'Jenny'), ('Sandra', 'Jenny');

and also
mysql> -> -> -> INSERT INTO relateenemies VALUES ('Andrew', 'Sandra'), ('Henry', 'Jenny'), ('Henry', 'John');

Lets see if we can find out who Andrew is friends with by using the following statement:
mysql> SELECT friend FROM relatefriends WHERE player = Andrew;

When we execute this statement, the console displays a list of all of the players with which Andrew is friends, as shown in Figure 2-42.

Chapter 2 / Using Databases

Figure 2-42 Finding out a players friend list

When we start implementing databases into Java in the next chapter, we could use this data to find out more information about each of Andrews friends. Again, we can do exactly the same with the relateenemies link table. For example, we could find out all of Henrys enemies with the following statement:
mysql> SELECT enemy FROM relateenemies WHERE player = Henry

When we execute this statement, the following console output can be expected:
Figure 2-43 Finding out a players enemy list

If we then wanted to find out more information about Henrys enemy who has a username of Jenny, we would use the following statement:
mysql> SELECT * FROM playerdata WHERE username = Jenny;

Here is a screenshot of our expected console output:

Figure 2-44 Finding more data about an enemy

Chapter 2 / Using Databases

Data Import Methods

Here we learn how to import data into our tables.

Importing from a Text File

Lets now create a text file that contains several records to be added to our playerdata table. We do this simply by denoting each column with a tab and each row by a new line. Figure 2-45 shows five lines of data to be added to the database in Windows Notepad:
Figure 2-45 The data to import in Windows Notepad

Note how we use \N to specify a field that contains NULL and also that an extra tab is required after each row of data to signify the end of that row. We have saved this file in the MySQL bin directory (i.e., c:\mysql\bin) with the filename import.txt. Now go to the MySQL console client and enter the following:
mysql> LOAD DATA LOCAL INFILE 'import.txt' INTO TABLE playerdata;

The console will inform us that five rows have been affected, or in this case added to our database. This can be seen in the following screenshot of the console.
Figure 2-46 Importing data from a text file

Chapter 2 / Using Databases

If we now select all the information from the playerdata table, we can see that our five rows of data have been imported correctly into the database. Here is a screenshot of the client that shows our imported data in the table.
Figure 2-47 The imported data in our playerdata table

Importing from a Native Source

Another method for importing data is using the Microsoft Excel spreadsheet program or basically any other application that can export data as tab-delimited data. For this example, however, we will use Microsoft Excel. Lets now enter another five rows of data we wish to add to our playerdata table into Excel. Once this is done, it should look roughly the same as Figure 2-48.
Figure 2-48 Entering the data into Microsoft Excel

Chapter 2 / Using Databases

Once our data is entered, we need to save it in a format MySQL can understand. In this case we will use tab-delimited values and save them in a text file called excel.txt in the MySQL bin directory.
Figure 2-49 Saving as a tab-delimited text file

Now the process is the same as importing a text file as we did in the last section. In fact, if you open up the text file in Windows Notepad you will see the file format is identical to what we created in the previous section. Figure 2-50 shows how the file looks when we open it up in Notepad.
Figure 2-50 The excel.txt file in Notepad

Chapter 2 / Using Databases

Backing Up and Restoring Data

Lets turn our attention to how we can export and import a database in MySQL. This is a relatively simple process but is extremely important for backing up data.

Backing up a Database to a File

When we back up a database from MySQL, it is written to a text file and is simply a list of the SQL statements that are required to recreate the database. Lets export the gamedata database that we created earlier to a text file called gamedata.txt. First we need to open up a command line by clicking Run on the Start button in Windows. When the Run dialog appears, type in command and press the OK button.
NOTE On Window 2000/XP , instead of typing in command, you need to type in cmd.

A command-line window will appear. Next you need to go to the bin directory of MySQL using the following command:
cd C:\mysql\bin

Note that you may have to change the above line if you modified the default MySQL installation directory. Figure 2-51 shows how this should look.
Figure 2-51 The command-line window (MS-DOS)

Now that we are in the correct directory, we will use a utility called mysqldump, which exports a specified database to a file of our choice. Here is how we would export our gamedata database to a text file called gamedata.txt:
mysqldump gamedata > gamedata.txt

If we open up the text file (which is now located the mysql\bin\ directory), we can see that it contains many SQL statements and comments added by the mysqldump utility. Here is a listing of our exported database text file:
# MySQL dump 8.16 #

Chapter 2 / Using Databases

# Host: localhost Database: gamedata #-------------------------------------------------------# Server version 3.23.47 # # Table structure for table 'playerdata' # CREATE TABLE playerdata ( username char(255) NOT NULL default '', password char(255) default NULL, age int(11) default NULL, datecreated timestamp(14) NOT NULL, PRIMARY KEY (username), UNIQUE KEY username (username) ) TYPE=MyISAM; # # Dumping data for table 'playerdata' # INSERT INSERT INSERT INSERT INSERT INTO INTO INTO INTO INTO playerdata playerdata playerdata playerdata playerdata VALUES VALUES VALUES VALUES VALUES ('Andrew','qwerty',20,20020209203741); ('Henry','letmein',34,20020209203741); ('Sandra','dra33',19,20020209203741); ('John','j12d',23,20020209203741); ('Jenny','jen123',34,20020209203741);

# # Table structure for table 'relateenemies' # CREATE TABLE relateenemies ( player char(255) default NULL, enemy char(255) default NULL ) TYPE=MyISAM; # # Dumping data for table 'relateenemies' # INSERT INTO relateenemies VALUES ('Andrew','Sandra'); INSERT INTO relateenemies VALUES ('Henry','Jenny'); INSERT INTO relateenemies VALUES ('Henry','John'); # # Table structure for table 'relatefriends' # CREATE TABLE relatefriends ( player char(255) default NULL, friend char(255) default NULL ) TYPE=MyISAM; #

Chapter 2 / Using Databases

# Dumping data for table 'relatefriends' # INSERT INSERT INSERT INSERT INSERT INTO INTO INTO INTO INTO relatefriends relatefriends relatefriends relatefriends relatefriends VALUES VALUES VALUES VALUES VALUES ('Henry','Sandra'); ('Andrew','Henry'); ('Andrew','John'); ('Andrew','Jenny'); ('Sandra','Jenny');

Restoring a Backed-Up Database

Now its time to look at how we would go about restoring a database that has been backed up. First, lets drop our gamedata database from MySQL using the following statement:
mysql> DROP DATABASE gamedata;

Now that our gamedata database has been removed, we need to create a new, empty database to import our data into. So lets do this with the following statement:
mysql> CREATE DATABASE newgamedata;

Next, we need to open up a command-line window again (by using the Run dialog and entering command or cmd). Change to the mysql\bin\ directory as we did previously when we were exporting the data and then type in the following command to import the data from our gamedata.txt text file into our newgamedata database:
mysql newgamedata < gamedata.txt

Figure 2-52 shows a screenshot of the command-line window:

Figure 2-52 Importing a text file into MySQL

Chapter 2 / Using Databases

MySQL C++ Interface

The C++ interface for MySQL, MySQL++, gives the database system a practical use from a game developers point of view. It allows us to access a MySQL database via the IP address in which the database is stored. This means we can easily retrieve and update data from our database directly within our server applications. First, you need to copy the MySQL++ libraries. They are on the companion CD in a file called mysql++-1.7.1-win32-vc++.zip. Extract the libraries with a utility such as WinZip (a trial version is available on the CD and at http://www.winzip.com). You must then include the lib and include directories in Visual Studio 6. In addition, you must include the lib and include directories from the MySQL directory. Setting up and using static libraries (such as MySQL++) is explained in detail in Chapter 1. Once the library has been set up, we can use it to write an application to retrieve data from a database that is stored in MySQL. Lets create a simple console application so we can understand the basics of how to access, retrieve, and update information in a database from a C++ application. Remember to include the mysql++.lib in your project settings (see Chapter 1 for more detail on how to do this). You must also copy the libmySQL.dll from the MySQL++ libraries example directory into your applications directory (or the windows/system directory) in order for your application to execute correctly.

Example 1 Connecting and Retrieving Data from MySQL

Here is the C++ code we require to create a connection to our gamedata database and list all the information from our playerdata table. After the code listing, each segment of the code is explained.
#include <iostream> #include <iomanip> #include <mysql++>

int main(void) { // -> Create a connection to the database Connection con("gamedata","127.0.0.1"); // -> Create a query object that is bound to our connection Query query = con.query(); // -> Assign the query to that object

Chapter 2 / Using Databases

query << "SELECT * FROM playerdata"; // -> Store the results from the query Result res = query.store(); // -> Display the results to the console

In the code, we first create a connection to the server on which the database is stored. We use the following code segment to achieve this:
Connection con("gamedata","127.0.0.1");

Connection is simply a class whose constructor takes in the parameters to establish a connection to a MySQL database. The first parameter is the name of the database you wish to connect to. The second is the IP address of the server the database is located on. Notice here that the IP address is 127.0.0.1; this is a special IP address that represents the local machine, i.e., the machine that the C++ application is running on. Next, we create a query object to allow us to pass queries into the connection we have established with the database. This is done with the following code segment:
Query query = con.query();

We can now use any standard MySQL query that we have used in the MySQL console with this query variable. We process a query using the following code:
query << "SELECT * FROM playerdata";

This code does the same as selecting all the information in the playerdata table in the MySQL console.

Chapter 2 / Using Databases

Next, we store the results from the query in a Result class, which contains a random access iterator for cycling through all the records of data that the query returned. Here is how we assign the query results into the Result class:
Result res = query.store();

Next, we print the field headings to the screen. This is not essential, but it makes the output data easier to understand. Now that we have the results in the Result class, we can use the iterator to cycle through all the records that the query returned. We declare the iterator as i in this code:
Result::iterator i;

Then we also want to create a Row class, which will hold each record of data as we cycle through the records with the iterator. We create the Row class as follows:
Row row;

Finally, we cycle through the data, outputting each record to the screen on a new line. We declare which field we wish to print from the current result by accessing the correct part of the Row class as follows:
row["fieldname"];

Here is the code we use to cycle through each record contained in the Result class:
for (i = res.begin(); i != res.end(); i++) { row = *i; }

Each time through the loop, the current record is assigned to the Row class so we can access individual fields from each record. Therefore, we can print each record using the following code within the for loop:
for (i = res.begin(); i != res.end(); i++) { row = *i; cout << setw(10) << row["username"] << setw(10) << row["password"] << setw(10) << row["age"] << endl; }

Here is the expected output from our C++ application:

Chapter 2 / Using Databases

Figure 2-53

Example 2 Updating Data in MySQL from an Application

In this example, we will modify a field of data from our playerdata table in the gamedata database. First we will display the field, then change the value, and display the field again so we can check that the data has been updated. We are going to change the password of the player with the username Katy. Here is the code for the C++ console application we require to do this. After the code, we explain how it works.
#include <iostream> #include <iomanip> #include <mysql++>

int main(void) { // -> Create a connection to the database Connection con("gamedata","127.0.0.1"); // -> Create a query object that is bound to our connection Query query = con.query();

//// DISPLAY BEFORE UPDATE // -> Assign the query to that object query << "SELECT * FROM playerdata WHERE username = 'Katy'"; // -> Store the results from the query Result res = query.store(); // -> Display the results to the console cout << "Before Update" << endl; cout << "-------------" << endl;

Chapter 2 / Using Databases

// -> Show the field headings cout.setf(ios::left); cout << setw(10) << "username" << setw(10) << "password" << setw(10) << "age" << endl; Result::iterator i; Row row; // The Result class has a read-only random access iterator for (i = res.begin(); i != res.end(); i++) { row = *i; cout << setw(10) << row["username"] << setw(10) << row["password"] << setw(10) << row["age"] << endl; } //// UPDATE THE INFORMATION // Send an execute and update query in MySQL query << "UPDATE playerdata SET password = 'qwerty' WHERE username = 'Katy'"; query.execute(); //// DISPLAY AFTER UPDATE // -> Assign the query to that object query << "SELECT * FROM playerdata WHERE username = 'Katy'"; // -> Store the results from the query res = query.store(); // -> Display the results to the console cout << "Before Update" << endl; cout << "-------------" << endl; // -> Show the field headings cout.setf(ios::left); cout << setw(10) << "username" << setw(10) << "password" << setw(10) << "age" << endl; // The Result class has a read-only random access iterator for (i = res.begin(); i != res.end(); i++) { row = *i; cout << setw(10) << row["username"] << setw(10) << row["password"] << setw(10) << row["age"] << endl; } return 1; }

Chapter 2 / Using Databases

The only major difference between this code and the code in Example 1 is the small segment in the middle that updates the field in the playerdata table. The code before and after that simply displays the record from the table we are modifying (in the same way as the last example). Lets look at the middle segment to see how it works.
query << "UPDATE playerdata SET password = 'qwerty' WHERE username = Katy";

First, we set the query to what we want it to be, just as we would enter it in the MySQL console. Once we have our query set, we then need to execute the query using the following command:
query.execute();

We did not require this command before as the store command that we used to store the results in the Result class automatically executes the query. When we then execute the application, we see the values of the record before and after we update the data. Here is a snapshot of the output from our application:
Figure 2-54

As you can see, the password field has been changed successfully from ka42 to qwerty as we intended.

Summary
In this chapter, you learned how to create and use a MySQL database from both the MySQL console and a C++ application. In the next chapter, we move on to learning how to create web-based interfaces for our game servers using the PHP language.

This page intentionally left blank.

Chapter 3

Creating Web-Based Server Interfaces

Introduction
In this chapter we are going to look at the PHP4 language and then learn how to use it to create web-based interfaces for our game servers. First, however, we will look at how to set up an Apache web server in Windows XP (so we can test our code locally), and then we will install and provide the PHP4 functionality within it. After this, we will look at the foundations of the PHP4 language, moving from accessing a game database to developing a complete web interface for a game server.

Setting Up an Apache 1.3.x Web Server

The first step is to get the Apache package, either by downloading it from http://www.apache.org or installing it directly from the companion CD. The Apache package on the CD has a file named apache_1.3.28win32-x86-no_src.msi. Start the installation by double-clicking this file. 57

Chapter 3 / Creating Web-Based Server Interfaces

You should then be presented with the following installation splash screen.
Figure 3-1 Installing Apache 1.3.x

Click Next in the splash screen, read and accept the license, then click Next twice, so that the following dialog is visible.
Figure 3-2 Setting up the server information

In this dialog, since it is simply a testing server, we can enter 127.0.0.1 (the localhost) in both the Network Domain and the Server Name fields. Then set the Administrators Email Address field to something like andrew@127.0.0.1, as this again is irrelevant here. Next, ensure the Run as a service for All Users radio button is selected and click Next to continue. You will then be asked if you wish to install the complete package or perform a custom installation.

Chapter 3 / Creating Web-Based Server Interfaces

Figure 3-3 Installation type

Leave this as a Complete install, unless you have reason to do otherwise, and click Next to proceed. Next, you will be asked to select the location to which Apache should be installed, which will have a subfolder to hold the web site data. The dialog for this is shown in Figure 3-4.
Figure 3-4 Installation folder

As can be seen in the dialog, the default installation folder is C:\Program Files\Apache Group, which again should be left alone unless you have reason to change it. After accepting the installation folder by clicking Next, you will be prompted to start the installation process. Do this now by clicking the Install button. Apache will install and start running automatically after the installation is complete and also each time Windows starts up. You can test to see if your installation was successful by entering http://127.0.0.1/ in your web browser (such as Internet Explorer).

Chapter 3 / Creating Web-Based Server Interfaces

When you do this, you should see a web page displayed as follows:
Figure 3-5 Testing the installation

Installing PHP4 for Apache 1.3.x

Once our local testing web server is set up, we can add PHP support to it, allowing pages ending with the extension .php and .php3 to be first parsed by PHP4 before being sent to the browser. To install PHP4, first grab the PHP4 package php-4.3.4-Win32.zip from either the PHP web site (www.php.net) or the companion CD. Once you have this zip file, extract it to the folder C:\php, using a utility such as WinZip (the shareware version is available on the CD as well as a download from www.winzip.com). When the files are extracted, the directory structure should look as follows:
Figure 3-6 The extracted PHP zip file

Chapter 3 / Creating Web-Based Server Interfaces

We then want to copy two dynamic link libraries (DLLs) into the Windows system folder (usually C:\windows\system or C:\winnt\ system). Copy the following two files, located within the C:\php and C:\php\sapi folders respectively, into the Windows system folder now: php4ts.dll and php4apache.dll. Now we want to copy the standard PHP configuration file, php.ini-dist (located in the C:\php folder), into the Windows directory (i.e., C:\windows or C:\winnt). Once there, rename the file to php.ini and open it in Notepad. We need to make one minor adjustment to the php.ini file to specify the correct temporary folder in which to store session data (well learn about this later in the chapter, but basically its a very easy and safe way of using cookies). So, once the file is open in Notepad, perform a search for the string /tmp until you find the following line:
session.save_path = /tmp

Once this is found, change it to read as follows:

session.save_path = c:\temp

Then ensure that you create a folder on your C:\ drive called temp. Save the changes to the php.ini file and close Notepad. The final step is to inform Apache about PHP by editing the Apache configuration file. To do this, first open the configuration file called httpd.conf in Notepad, which is located in the C:\Program Files\Apache Group\Apache\conf folder (providing you have installed Apache to the default directory). Once the file is open in Notepad, add the following two lines to it:
LoadModule php4_module c:/php/sapi/php4apache.dll AddType application/x-httpd-php .php

After this change is made, we need to restart the Apache web server in order to load in the PHP4 module. To restart the server, go into the Start menu and then into the Programs submenu. Select the Apache HTTP Server folder, then Control Apache Server, and then Restart. After this is done, we need to make sure the PHP integration was successful. The main web folder is located at the following location: C:\Program Files\Apache Group\Apache\htdocs. So any file placed within this folder would be accessible from the following URL: http://127.0.0.1/. Therefore, if we created a subfolder within the htdocs folder, such as mysite, the entire path would look as follows: C:\Program Files\Apache Group\Apache\htdocs\mysite. We could then access this from the URL http://127.0.0.1/mysite/. As you can see, this is a useful way to organize your different web projects.

Chapter 3 / Creating Web-Based Server Interfaces

Anyway, back to the point. We need to test PHP , so to do this we can write a very simple PHP script that displays all the information about the PHP installation. So lets now create a file called test.php that will be placed within the htdocs folder. Once created, open this file in Notepad, then enter the following into it. test.php
<?php print phpinfo(); ?>

If you then access the URL http://127.0.0.1/test.php, you should see the following page visible in the browser:
Figure 3-7 A successful PHP installation

If when you access the URL, it does not look like Figure 3-7, but instead looks similar to Figure 3-8.
Figure 3-8 A failed PHP installation

check that you have: n Followed the previous installation instructions precisely n Named the file test.php and not test.php.txt n Restarted the Apache server

Chapter 3 / Creating Web-Based Server Interfaces

Now that you have successfully installed Apache and PHP4, we are going to look at the basics of the PHP language.

Using PHP: Hypertext Preprocessor

Now that PHP is installed successfully, the next step is to grab a decent editor, as Notepad doesnt quite suffice for coding PHP . The best free IDE (integrated development environment) in our opinion is PHPEdit (http://www.phpedit.net); however, if you have some money to burn, you may wish to take a look at Dreamweaver MX (http://www.macromedia.com). Which IDE you use is really personal preference, so experiment with different ones and see which you prefer. To get started, lets create a folder in the htdocs folder called testsite, then, assuming youre using our editor of choice (PHPEdit), browse to the folder using the File Explorer window. When this is done, depending on how you lay out the IDE, it should look something similar to the following.
Figure 3-9 PHPEdit IDE

Chapter 3 / Creating Web-Based Server Interfaces

The Basics
In the previous version of PHP (version 3), PHP files had the file extension .php3, but since the fourth release all PHP files should have the extension .php. This indicates to the Apache web server that the file should be parsed by the preprocessor before it is output to the clients web browser. Just as index.html and index.htm are recognized as default pages that should be loaded when someone types in a URL, so can index.php be used as the default page. All PHP code with a .php file extension needs to be placed within special tags. Before you write any PHP code, you must first tell the preprocessor that you are going to do so. This is indicated by specifying the following tag:
<?php

After you have finished writing PHP code, you close the tag with the following:
?>

Within the PHP tags, you can also add comments to your code in exactly the same way as C++:
// single line comment

/* multi line comment */

Lets try an example. Save a file within the IDE called index.php to the testsite folder.

Example 1 index.php
<html> <head> <title>Example PHP File</title> </head> <body> This is an example PHP file </body> </html>

When we call the example in the web browser (http://127.0.0.1/testsite/ index.php), we should see the output shown in Figure 3-10.

Chapter 3 / Creating Web-Based Server Interfaces

Figure 3-10 Example 1 output

Wait! Thats just a standard HTML file with no PHP , yet its in a PHP file! Well, basically anything outside of the PHP tags is interpreted as standard HTML and is output to the browser as normal. This is very useful as we can turn PHP on and off as we require even hundreds of times within the same source file. Lets look at a more exciting example that uses variables.

Variables
Although PHP does have types of variables (such as INT, FLOAT, etc.), they are all handled internally so there is little need to worry about the actual type of your variables. Note, however, that it is possible to directly set and get the types. We will look into this in more detail later. As well as types not being overly important, there is also no need to declare a variable; you can just initialize it and use it directly. Lets expand upon our previous example to show how to assign and print variables to the screen.

Example 2 index2.php
<?php $name = "Andrew"; $website = "www.huntedcow.com"; $number = 5; ?>

<html> <head> <title>Example PHP File</title> </head> <body> This is an example PHP file <br><br> <?php print "Name was $name"; print "<br>Website was <a href=\"http://".$website."\"> $website</a>"; print "<br>The number was <b>$number</b>"; ?> </body> </html>

Chapter 3 / Creating Web-Based Server Interfaces

When we load Example 2 in the browser (http://127.0.0.1/testsite/ index2.php), it should look somewhat like the following:
Figure 3-11 Example 2 output

At the top of the file, we start with the following block of code:
<?php $name = "Andrew"; $website = "www.huntedcow.com"; $number = 5; ?>

All we are doing here is entering a PHP block and assigning three variables. Note that variables are defined by the use of the $ sign and we use the " to denote strings, whereas we just assign numbers directly. We then turn PHP off using the ?> tag and continue by printing out HTML as we did in the previous example. Then, we enter PHP again and print out the $name variable using the following line of code:
print "Name was $name";

Notice in this line how we actually have the $name variable within another string that we wish to print. Before the string is output to the browser, the value of $name is replaced with the actual value, which in this example is Andrew. The next line creates a web link using HTML and places the $website variable within the href and also after it to display it to the browser. This can be seen in the following line of code:
print "<br>Website was <a href=\"http://".$website."\">$website</a>";

Notice how we have used " within the string. This is possible as we have escaped them by adding a backslash \ before each occurrence. Finally, we printed the number variable $number using the following line of code:
print "<br>The number was <b>$number</b>";

Note in this final line how we have placed a bold HTML text tag in front of the $number variable so that when it is printed it comes out bold.

Chapter 3 / Creating Web-Based Server Interfaces

Operators and Loops

Next in our brief overview we are going to take a look at simple mathematical operators and loops. PHP supports all the basic operators as well as the shortcuts that C++ contains (such as the += operator). Lets look at the next example where we print out a multiplication table.

Example 3 index3.php
<html> <head> <title>Example PHP File</title> </head> <body> <center> Multiplication Table <br><br> <table border=1> <?php for($i = 1; $i <= 10; $i++) { ?><tr><?php for($j = 1; $j <= 10; $j++) { print "<td align=\"right\">"; print ($i*$j); print "</td>"; } ?></tr><?php } ?> </table> </body> </html>

When you run Example 3 in the browser (http://127.0.0.1/testsite/ index3.php), you should see the multiplication table shown in Figure 3-12.

Chapter 3 / Creating Web-Based Server Interfaces

Figure 3-12 Example 3 output

To display the multiplication table, we set up two for loops. The first is:
for($i = 1; $i <= 10; $i++)

This initializes a variable called $i to the value of 1, then increments it by 1 each time through the loop. Inside this loop, a table row is initialized in HTML and another loop is created to fill in the row with values. This can be seen here:
for($j = 1; $j <= 10; $j++) { print "<td align=\"right\">"; print ($i*$j); print "</td>"; }

Then, before the outer loop is terminated, the table row is ended.

Conditional Statements
Next, we will look at how to construct a simple if statement. For an example, we will expand upon the previous multiplication table example and make the top and left-hand numbers bold. Take a look at Example 4.

Example 4 index4.php
<html> <head> <title>Example PHP File</title> </head> <body> <center> Multiplication Table <br><br> <table border=1> <?php for($i = 1; $i <= 10; $i++)

Chapter 3 / Creating Web-Based Server Interfaces

{ ?><tr><?php for($j = 1; $j <= 10; $j++) { print "<td align=\"right\">"; if($i == 1 || $j == 1) print "<b>"; print ($i*$j); print "</td>"; } ?></tr><?php } ?> </table> </body> </html>

When this is executed within the browser (http://127.0.0.1/testsite/ index4.php), it should look like the following:
Figure 3-13 Example 4 output

As you can see from the code, all we have added in this example is the following two lines:
if($i == 1 || $j == 1) print "<b>";

All this if statement does is print the <b> tag to the browser if the $i or $j variables are equal to 1. Simple stuff! The other important conditional statement is the switch statement, which works in the same manner as it does in C++. However, in PHP it is also possible to use strings within the case. For example:
$mystring = andrew; switch($mystring) {

Chapter 3 / Creating Web-Based Server Interfaces

case andrew: // do something break; case teijo: // do something else break; }

Arrays
As with variables, there is no need to initialize arrays and they are treated in exactly the same manner as variables. For example, we could set three values in an array as follows:
$myarray[0] = Andrew; $myarray[1] = Teijo; $myarray[2] = Wes;

Then we could display them with the following code segment:

for($i=0; $i<count($myarray); $i++) print $myarray[$i];

Functions
When we start thinking about functions, we get into the area of variable scope. In PHP , when a variable is declared within a function, its value is local to that function unless the global keyword is specified before it. Lets look at a simple example of how to create a function to print a string of text that is passed into it.

Example 5 index5.php
<html> <head> <title>Example PHP File</title> </head> <body> <?php WriteString("Hello!"); ?> </body> </html> <?php function WriteString($theString) { print $theString; } ?>

Chapter 3 / Creating Web-Based Server Interfaces

When this is executed in the browser (http://127.0.0.1/testsite/ index5.php), the following should be shown:
Figure 3-14 Example 5 output

User Input
Now we get into the territory where PHP differs from C++ in many ways, purely due to the integration with HTML. To gather input from the user, we use HTML forms. For an example of user input, we are going to create two PHP scripts. The first, input.php, will simply display an HTML form. The second, output.php, will display the information that was gathered by the form. Here are the two files that are required to make this example work.

Example 6a input.php
<html> <head> <title>Example PHP File</title> </head> <body> <center> <form method="post" action="output.php"> <table border=1> <tr> <td>Your name:</td><td><input type="text" name="YourName"></td> </tr> <tr> <td>Your favourite color:</td> <td> <select name="YourColor"> <option value="Red">Red</ option> <option value="Blue">Blue</ option> <option value="Green"> Green</option> </select> </td> </tr> <tr> <td>Over 18?:</td><td><input type= "checkbox" name="Over18"></td>

Chapter 3 / Creating Web-Based Server Interfaces

</tr> <tr> <td colspan=2> <center> <input type="submit" value="Send Data"> </center> </td> </tr> </table> </form> </center> </body> </html>

Example 6b output.php
<html> <head> <title>Example PHP File</title> </head> <body> <center> <?php print "Name was ".$_POST["YourName"]."<br>"; print "Color was ".$_POST["YourColor"]."<br>"; if($_POST["Over18"] == "on") { print "Over 18 was ticked"; } else { print "The user was under 18"; } ?> </center> </body> </html>

When we load the input.php file into the web browser and fill in some information, it will look similar to the following (http://127.0.0.1/ testsite/input.php):
Figure 3-15 Example 6 input.php

Chapter 3 / Creating Web-Based Server Interfaces

When the Send Data submit button is clicked, we should expect something similar to the following to be visible.
Figure 3-16 Example 6 output.php after being passed form data

So how did the information get sent? When using a form, there are two methods of sending information to another page. The best method is the POST method. If you look at where we have declared the form in input.php, you will see this.
<form method="post" action="output.php">

Note also that we have specified the action as output.php, which means that any data collected by the form will be sent to the output.php script upon the user clicking the Submit button. When the data arrives in output.php, it is stored within a global array called $_POST, which is accessed associatively via the name the input was given on the form. For example, the name input was declared in the form as follows:
<input type="text" name="YourName">

Which means that when this is passed to output.php, it can be accessed by referencing:
$_POST[YourName]

Similarly, if the form method is specified as GET (which means the data is attached to the end of the URL), the global $_GET array can be used to access the data in the same way.

The Command System

Now that we have skimmed the basics of PHP , lets start to look at how to correctly structure our code and keep it maintainable for larger projects. What youre about to learn is the command system. (Im not sure if it has an official name, but I think its an okay name for it J.) The idea behind this is to give our code a central processing point that then references methods to actually deal with the functionality. Lets take a look at the following core.php and welcome.php files:

Chapter 3 / Creating Web-Based Server Interfaces

Example 7a core.php
<?php // include libraries here... include("welcome.php"); // command processing... switch($cmd) { default: showWelcome(); break; } ?>

Example 7b welcome.php
<?php function showWelcome() { ?> <html> <head> <title>The command processor example</title> </head> <body> This is the welcome page </body> </html> <?php } ?>

When we execute this example in the browser (http://127.0.0.1/testsite/ core.php), we should expect the following to be visible:
Figure 3-17 Example 7 core.php

Chapter 3 / Creating Web-Based Server Interfaces

In our core.php file, we first call the include function and pass in the filename welcome.php, which basically includes the welcome.php file when executing the code. Then we create a switch statement that examines the value of a $cmd variable (which, as it is not contained within a function, is global). We then make the default of the switch statement call the showWelcome method, which is defined within the welcome.php file and simply prints This is the welcome page to the browser. Lets expand upon this to create multiple pages. Before this, however, we are going to create a header.php and a footer.php file which will contain the methods showHeader and showFooter, respectively. Well then call the header method before and the footer method after the switch statement in core.php so we can put any standard look and feel of the page there without rewriting it for each page. These files will look as follows: header.php
<?php function showHeader() { ?> <html> <head> <title>Command Processor</title> </head> <body bgcolor="#0000AA"> <center> <table width="600" border=1 bgcolor="#FFFFFF"> <tr> <td> <?php } ?>

footer.php
<?php function showFooter() { ?> </td> </tr> </table> </center> </body> </html>

Chapter 3 / Creating Web-Based Server Interfaces

<?php } ?>

If we then create the files page1.php and page2.php, adding some text and hyperlinks into them, we will have the following four files (excluding the header and footer):

Example 8a core.php
<?php // include libraries here... include("header.php"); include("footer.php"); include("welcome.php"); include("page1.php"); include("page2.php"); // command processing... showHeader(); switch($cmd) { case "page1": showPage1(); break; case "page2": showPage2(); break; default: showWelcome(); break; } showFooter(); ?>

Example 8b welcome.php
<?php function showWelcome() { ?> <center> <b>Welcome!</b> <br><br>

Chapter 3 / Creating Web-Based Server Interfaces

Please click one of the following pages to visit it... </center> <ul> <li><a href="core.php?cmd=page1">Page 1</a></li> <li><a href="core.php?cmd=page2">Page 2</a></li> </ul> <?php } ?>

Example 8c page1.php
<?php function showPage1() { ?> <center> <b>Page 1</b> <br><br> Welcome to page 1! Why not <a href="core.php?cmd=page2">visit page 2?</a> </center> <?php } ?>

Example 8d page2.php
<?php function showPage2() { ?> <center> <b>Page 2</b> <br><br> Welcome to page 2! Why not either <a href="core.php?cmd=page1"> go back to page 1</a> or <a href="core.php">visit the welcome page again.</a> </center> <?php } ?>

Chapter 3 / Creating Web-Based Server Interfaces

When you load up the core.php file in the browser, you should see the following:
Figure 3-18 Example 8 core.php

Clicking on either of the Page 1 or Page 2 links will display the respective pages. Notice the following in the address bar of the browser when you click the Page 1 link: http://127.0.0.1/testsite/core.php?cmd= page1. As you can see, the variable cmd has been set to the value page1. This is generated automatically when you use a form (with the GET method) and is done without placing the information in the URL with the POST method. To add multiple variables to the URL you can delimit them with the & character. For example:
http://127.0.0.1/testsite/core.php?cmd=page1&myothervar=blah

Accessing MySQL
Lets see how we can utilize the SQL language we discovered in the previous chapter with PHP . MySQL works exceptionally well with PHP and there are in fact built-in functions for interacting with MySQL. In this section we are going to look at these functions and see how to store, retrieve, and modify information in the database directly from the web browser.

MySQL Example 1 Connecting and Disconnecting

In this first example we are going to find out how to connect and terminate a connection to a database. Lets first create a database in MySQL by executing the following at the MySQL command prompt:
mysql> CREATE DATABASE phptest;

Chapter 3 / Creating Web-Based Server Interfaces

After that is created, we can make a PHP script that will connect to, use, and disconnect from the database. Lets take a look at this script now, which we have called mysql1.php. mysql1.php
<?php // connect... $dbh = mysql_connect("localhost", "root", ""); // select the database... if(!mysql_select_db("phptest", $dbh)) print "Unable to connect to db"; // disconnect... mysql_close($dbh); ?>

In this example, we first make a call to mysql_connect, which takes in the host the database is running on (which 99 percent of the time will be the computer/server the script is executing on). The second parameter is the username and the third is the password required to connect to the database server. If the connection fails, PHP will output descriptive errors to the browser, which will look something like the following (weve changed the password to be incorrect):
Figure 3-19 A connection failure

MySQL Example 2 Storing and Retrieving Data

Now that we know how to connect to the database, lets look at how we can store and display information stored within the database. For this, we are going to be using the command processor setup we learned earlier in this chapter. Before we delve into the PHP , lets first create a table called users within the database for storing players names, ages, aliases, and passwords. Here is the MySQL query we require to do this:

Chapter 3 / Creating Web-Based Server Interfaces

mysql> -> -> -> -> -> ->

create table users ( id int auto_increment, name tinytext, age int, alias tinytext, password tinytext, primary key(id));

Now lets look at the complete source code for this example, then we will look into detail at how it works. (Note that we also require the header.php and footer.php files we created earlier.) mysqlcore.php
<?php // include libraries here... include("header.php"); include("footer.php"); include("users.php"); // special commands (no header/footer) switch($cmd) { case "doadduser": doAddUser(); exit(); } // command processing... showHeader(); switch($cmd) { case "adduser": addUser(); break; default: showUsers(); break; } showFooter(); ?>

Chapter 3 / Creating Web-Based Server Interfaces

users.php
<?php function showUsers() { ?> <center> <br><b>Registered Users</b><br><br> <a href="mysqlcore.php?cmd=adduser">Add User</a> <br><br> <table width="450" border="1" cellpadding="4"> <tr bgcolor="#BBBBBB"> <td width="25%"><b>Name</td> <td width="25%"><b>Age</td> <td width="25%"><b>Alias</td> <td width="25%"><b>Password</td> </tr> <?php // connect... $dbh = mysql_connect("localhost", "root", ""); // select the database... mysql_select_db("phptest", $dbh); // fetch all the users... $results = mysql_query("SELECT * FROM users ORDER BY name", $dbh); while($row = mysql_fetch_array($results)) { print "<tr>"; print "<td>".htmlspecialchars($row["name"])."</td>"; print "<td>".htmlspecialchars($row["age"])."</td>"; print "<td>".htmlspecialchars($row["alias"])."</td>"; print "<td>".htmlspecialchars($row["password"])."</td>"; print "</tr>"; } mysql_close($dbh); ?> </table> </center> <?php } function addUser() { ?> <center> <br><b>Add User Form</b><br><br> <form method="POST" action="mysqlcore.php">

Chapter 3 / Creating Web-Based Server Interfaces

<input type="hidden" name="cmd" value="doadduser"> <table width="450" border="1" cellpadding="4"> <tr> <td>Name:</td> <td><input type="text" name="name"></td> </tr> <tr> <td>Age:</td> <td><input type="text" name="age" size=3></td> </tr> <tr> <td>Alias:</td> <td><input type="text" name="alias"></td> </tr> <tr> <td>Password:</td> <td><input type="password" name= "password"></td> </tr> </table> <br><br> <input type="submit" value="Add User"> <input type="button" value="Cancel" onClick= "window.location = 'mysqlcore.php'"> </form> <?php } function doAddUser() { // connect... $dbh = mysql_connect("localhost", "root", ""); // select the database... mysql_select_db("phptest", $dbh); // add the new user... mysql_query("INSERT INTO users VALUES (NULL, \"".$_POST["name"]."\", \"".$_POST["age"]."\", \"".$_POST["alias"]."\", \"".$_POST["password"]."\")", $dbh); // close the connection... mysql_close($dbh); // Redirect the page... header("Location: mysqlcore.php"); } ?>

When we execute this in the browser (http://127.0.0.1/testsite/ mysqlcore.php), we should have two screens available that look as follows (note that we added two users before taking the screen shots).

Chapter 3 / Creating Web-Based Server Interfaces

Figure 3-20 Registered users

Figure 3-21 The Add User screen

Lets now see how this works. First, we have the showUsers function in users.php, which displays a list of all the users stored in the users table, ordered by the name. In this method, we first connect to the database as we did in the previous example using the following two lines of code:
$dbh = mysql_connect("localhost", "root", ""); mysql_select_db("phptest", $dbh);

After we are connected, we execute a SELECT query by calling the mysql_query method, passing in the query we wish to execute and the handle to the database (which we obtained with the mysql_connect method).
$results = mysql_query("SELECT * FROM users ORDER BY name", $dbh);

We then have a handle to the result set in a variable called $results. We can use this to actually retrieve the results into an associative array using the mysql_fetch_array function.
while($row = mysql_fetch_array($results)) {

Chapter 3 / Creating Web-Based Server Interfaces

The array $row can be accessed by passing the string names of the field names in the database into it (such as name and alias). Within this while loop, a new table row (<tr>) is created for each entry in the database and it is printed out to the browser. The addUser function simply displays the HTML form for gathering the information; however, note the following line within the form:
<input type="hidden" name="cmd" value="doadduser">

By adding a hidden input to the form, we can specify a command for our command processor to interpret; in this case we have specified the doadduser command, which will in turn call the doAddUser function. The doAddUser function is called in the mysqlcore.php before the showHeader function is called as it is not intended to display any information. Basically, the doAddUser function will insert the data into the MySQL database and send the browser to a different page without the user ever knowing. The reason behind this is simple: If the browser was not redirected, the user would be inserted into the database again if he hit Refresh on the page. So, in the doAddUser function, once the database connection is established, it is then a simple case of performing an INSERT query, feeding the form data into the SQL statement. This can be seen in the following line of code:
mysql_query("INSERT INTO users VALUES (NULL, \"".$_POST["name"]."\", \"".$_POST["age"]."\", \"".$_POST["alias"]."\", \"".$_POST["password"]."\")", $dbh);

After this, the connection is closed and a call to the PHP header method is made. Into that is passed the location to which the browser should be redirected. This can be seen here:
header("Location: mysqlcore.php");

MySQL Example 3 Updating and Removing Data

In this final MySQL example, we look at how to modify and remove data directly from PHP by expanding upon the previous example. Lets first look at the complete modified code and then we will look into the required changes.

Chapter 3 / Creating Web-Based Server Interfaces

mysqlcore.php
<?php // include libraries here... include("header.php"); include("footer.php"); include("users.php"); // special commands (no header/footer) switch($cmd) { case "doadduser": doAddUser(); exit(); case "deleteuser": deleteUser(); exit(); } // command processing... showHeader(); switch($cmd) { case "adduser": addUser(); break; default: showUsers(); break; } showFooter(); ?>

Chapter 3 / Creating Web-Based Server Interfaces

<td <td <td <td <td <td

width="25%"><b>Name</td> width="25%"><b>Age</td> width="25%"><b>Alias</td> width="25%"><b>Password</td> width="25%"><b>Edit</td> width="25%"><b>Delete</td>

</tr> <?php // connect... $dbh = mysql_connect("localhost", "root", ""); // select the database... mysql_select_db("phptest", $dbh); // fetch all the users... $results = mysql_query("SELECT * FROM users ORDER BY name", $dbh); while($row = mysql_fetch_array($results)) { print "<tr>"; print "<td>".htmlspecialchars($row["name"])."</td>"; print "<td>".htmlspecialchars($row["age"])."</td>"; print "<td>".htmlspecialchars($row["alias"])."</td>"; print "<td>".htmlspecialchars($row["password"])."</td>"; print "<td><input type=\"button\" value=\"Edit\" onClick=\"window.location = 'mysqlcore.php?cmd= adduser&id=".$row["id"]."';\"></td>"; print "<td><input type=\"button\" value=\"Delete\" onClick=\"window.location = 'mysqlcore.php?cmd= deleteuser&id=".$row["id"]."';\"></td>"; print "</tr>"; } mysql_close($dbh); ?> </table> </center> <?php } function addUser() { $id = $_GET["id"]; if($id != "") { $dbh = mysql_connect("localhost", "root", ""); mysql_select_db("phptest", $dbh); $results = mysql_query("SELECT * FROM users WHERE id = '$id'", $dbh); if($row = mysql_fetch_array($results)) {

Chapter 3 / Creating Web-Based Server Interfaces

$name = $row["name"]; $age = $row["age"]; $alias = $row["alias"]; $password = $row["password"]; } mysql_close($dbh); } ?> <center> <br><b> <?php if($id == "") print "Add User Form"; else print "Edit User Form"; ?> </b><br><br> <form method="POST" action="mysqlcore.php"> <input type="hidden" name="cmd" value="doadduser"> <input type="hidden" name="id" value="<?php print $id; ?>"> <table width="450" border="1" cellpadding="4"> <tr> <td>Name:</td> <td><input type="text" name="name" value= "<?php print $name; ?>"></td> </tr> <tr> <td>Age:</td> <td><input type="text" name="age" size=3 value= "<?php print $age; ?>"></td> </tr> <tr> <td>Alias:</td> <td><input type="text" name="alias" value= "<?php print $alias; ?>"></td> </tr> <tr> <td>Password:</td> <td><input type="password" name="password" value="<?php print $password; ?>"></td> </tr> </table> <br><br> <input type="submit" value="<?php if($id == "") print "Add User"; else print "Save Changes"; ?>"> <input type="button" value="Cancel" onClick= "window.location = 'mysqlcore.php'"> </form>

Chapter 3 / Creating Web-Based Server Interfaces

<?php } function doAddUser() { $id = $_POST["id"]; // connect... $dbh = mysql_connect("localhost", "root", ""); // select the database... mysql_select_db("phptest", $dbh); // add the new user... if($id == "") mysql_query("INSERT INTO users VALUES (NULL, \"".$_POST["name"]."\", \"".$_POST["age"]."\", \"".$_POST["alias"]."\", \"".$_POST["password"]."\")", $dbh); else { // update the details... mysql_query("UPDATE users SET name = \"".$_POST["name"]."\", age = \"".$_POST["age"]."\", alias = \"".$_POST["alias"]."\", password = \"".$_POST["password"]."\" WHERE id = '$id'", $dbh); } // close the connection... mysql_close($dbh); // Redirect the page... header("Location: mysqlcore.php"); } function deleteUser() { $id = $_GET["id"]; // connect... $dbh = mysql_connect("localhost", "root", ""); // select the database... mysql_select_db("phptest", $dbh); // delete the user... mysql_query("DELETE FROM users WHERE id = '$id'", $dbh); // close the connection... mysql_close($dbh); // Redirect the page... header("Location: mysqlcore.php"); } ?>

Chapter 3 / Creating Web-Based Server Interfaces

When this is run in the browser, the main screen should look similar to the following screenshot:
Figure 3-22 Edit and Delete buttons

As you can see from Figure 3-22, we have added two extra columns to the table to allow the editing and deletion of any of the users added to the system. When the user clicks the Edit button, an onClick JavaScript call to change the URL to the following occurs:
mysqlcore.php?cmd=adduser&id=[UID];

[UID] is replaced in the above line with the users unique ID within the database. As you can see, however, we are sending the same adduser command as if the user clicked the Add User link at the top of the page. This is so we can reuse the same form we created for adding a user to allow the user to edit the stored information. In the addUser function, we now first obtain the id variable passed in the URL and check whether it has been assigned a value using the following few lines of code:
$id = $_GET["id"]; if($id != "") {

If an id has been passed in the URL, we then attempt to retrieve the users details from the database using the following segment of code:
$dbh = mysql_connect("localhost", "root", ""); mysql_select_db("phptest", $dbh); $results = mysql_query("SELECT * FROM users WHERE id = '$id'", $dbh); if($row = mysql_fetch_array($results)) { $name = $row["name"]; $age = $row["age"]; $alias = $row["alias"]; $password = $row["password"];

Chapter 3 / Creating Web-Based Server Interfaces

} mysql_close($dbh);

We can then print these values to the form (since if there was no data, the variables contain no data). For example, here is the name input box:
<input type="text" name="name" value="<?php print $name; ?>">

Note in the form that we also need to pass the id to the doAddUser function by including it as a hidden field in the form, as follows:
<input type="hidden" name="id" value="<?php print $id; ?>">

In the doAddUser function, we can once again check the value of the id variable (note that it comes through as a POST variable from the form, however). If the id variable is not assigned, we know it is a new user and add it to the database. If it is assigned a value, we simply create an UPDATE statement and amend the details stored in the database to the new values passed in from the form. The other part of this example is where the user clicks the Delete button. When this happens, in the same way as the Edit button, we call the following URL:
mysqlcore.php?cmd=deleteuser&id=[UID]

Again, [UID] is substituted with the users id field within the database. As can be seen from the URL, the deleteuser command is passed to mysqlcore.php. When this happens, it makes a call to the deleteUser function we have created, which then redirects back to the main screen after the SQL to remove the user has been executed.

Using FastTemplate
Now that we have looked at how to access MySQL from within PHP , you are probably looking at the code and thinking it is getting messy and you would be correct. The best way to keep our code tidy is to separate the logic from the design elements (i.e., the HTML). In this final section we are going to look at the FastTemplate class (http://www.thewebmasters.net/ php/FastTemplate.phtml and also available on the CD) to see how it can solve our messy code problem. The idea of FastTemplate is in its name templates! Basically the plan is to create templates of what your pages, rows of tables, etc., should look like with placeholders for information to be filled in. The code will then assign values to these placeholders and eventually print it all to the screen.

Chapter 3 / Creating Web-Based Server Interfaces

Lets first look at a very simple example where we create a template file with two placeholders (defined by the curly brackets). Here is our template file now: simple.tpl.html
<html> <title>{PAGE_TITLE}</title> <body> The name was specified as {NAME} </body> </html>

This file should be saved in a folder called tpl, which should be a subfolder of the main testsite directory. If we open up this file in the browser directly (http://127.0.0.1/testsite/tpl/simple.tpl.html), it will look something like the following.
Figure 3-23 simple.tpl.html

As you can see, we have added the following two placeholders:

{PAGE_TITLE} {NAME}

So we now need to write code that will use FastTemplate in conjunction with our simple.tpl.html file to generate the final page with the correct values. Lets look at this file now: ft.php
<?php include("class.FastTemplate.php"); // Create a FastTemplate object... $tpl = new FastTemplate("./tpl"); // Define the template files... $tpl->define(array(simple => "simple.tpl.html")); // Assign the placeholder values... $tpl->assign("PAGE_TITLE", "FastTemplate Example Script"); $tpl->assign("NAME", "Andrew"); // Parse the template $tpl->parse("FINAL", "simple");

Chapter 3 / Creating Web-Based Server Interfaces

// Display the final page $tpl->FastPrint(); ?>

When we load this into the browser (http://127.0.0.1/testsite/ft.php), we should see the following (note that we have moved the class.FastTemplate.php file from the CD into the testsite directory also):
Figure 3-24 ft.php

As you can see from Figure 3-24 (and if you run the script), {PAGE_ TITLE} has been replaced with FastTemplate Example Script and {NAME} has been replaced with Andrew. Lets now look line by line at how this works. The first line simply includes the FastTemplate source code:
include("class.FastTemplate.php");

After this, an instance of the FastTemplate class is created by passing in the folder the template files is contained in as a parameter; the reference to this is then returned in the $tpl variable.
$tpl = new FastTemplate("./tpl");

Note that the folder can be easily changed to allow your web applications to be skinned with different designs without changing any of the code (apart from the folder the templates are contained in, of course). After the FastTemplate instance is created, the next step is to define the actual template files we will be using within the project and assign them to keywords so that we can refer to them easily later in the code. This can be seen here:
$tpl->define(array(simple => "simple.tpl.html"));

As you can see, we have defined the keyword simple to refer to our simple.tpl.html file. Later in this section we will look at using multiple template files; however, one is sufficient for this example. Now we can assign the values to our placeholders by making a call to the assign method of the FastTemplate class for each placeholder using the following two lines of code:
$tpl->assign("PAGE_TITLE", "FastTemplate Example Script"); $tpl->assign("NAME", "Andrew");

Chapter 3 / Creating Web-Based Server Interfaces

In the assign function, we first pass in the name of the placeholder we wish to assign a value to (without the curly brackets). The second parameter is simply the value we wish to assign it. Next, we call the parse function, passing in a placeholder to parse the data to, along with the keyword referring to the template file we wish to parse:
$tpl->parse("FINAL", "simple");

Note that the placeholder name, FINAL, is unimportant as this is the final parse (well see in the next example why more than one parse would be required). The final line of code then prints out the last placeholder to be parsed to the browser, in this case the FINAL placeholder:
$tpl->FastPrint();

Multiple Templates
Now that we have looked at a simple example of using FastTemplate, lets look at how multiple template files can be used to construct a small table. First, we are going to create a new template called mainbody.tpl.html, which will look as follows: mainbody.tpl.html
<html> <title>{PAGE_TITLE}</title> <body> This text is above the table. <br><br> <table border=1 cellpadding=4> {TABLE_DATA} </table> <br><br> This text is below the table. </body> </html>

As you can see, it is very similar to the simple template we created in the previous example; however, this time we have placed a table open tag, then a placeholder called TABLE_DATA, and a table close tag. That defines our main page; now we need to define what a row in our table will look like. Assuming we would like to display a name, age, and location in our table, we can create a template called tablerow.tpl.html as follows:

Chapter 3 / Creating Web-Based Server Interfaces

tablerow.tpl.html
<tr> <td>{NAME}</td> <td>{AGE}</td> <td>{LOCATION}</td> </tr>

Within our tablerow template, we have defined three placeholders to be replaced with the name, age, and location in each row of the table. Lets now look at the final script that will create the table with three rows in it: ft2.php
<?php include("class.FastTemplate.php"); // Create a FastTemplate object... $tpl = new FastTemplate("./tpl"); // Define the template files... $tpl->define(array(mainbody => "mainbody.tpl.html", tablerow => "tablerow.tpl.html")); // Assign the placeholder values... $tpl->assign("PAGE_TITLE", "FastTemplate Example Script"); $tpl->assign("NAME", "Andrew"); $tpl->assign("AGE", "21"); $tpl->assign("LOCATION", "Scotland"); $tpl->parse("TABLE_DATA", ".tablerow"); $tpl->assign("NAME", "Teijo"); $tpl->assign("AGE", "22"); $tpl->assign("LOCATION", "Finland"); $tpl->parse("TABLE_DATA", ".tablerow"); $tpl->assign("NAME", "Wes"); $tpl->assign("AGE", "unknown"); $tpl->assign("LOCATION", "USA"); $tpl->parse("TABLE_DATA", ".tablerow"); // Parse the template $tpl->parse("FINAL", "mainbody"); // Display the final page $tpl->FastPrint(); ?>

Chapter 3 / Creating Web-Based Server Interfaces

When this ft2.php script is run in the browser, it will look like the following:
Figure 3-25 ft2.php output

This script is very similar to ft.php, but there are some interesting changes. The first is that we have defined an additional template file (and of course changed the simple one to refer to the new mainbody one), as shown below:
$tpl->define(array(mainbody => "mainbody.tpl.html", tablerow => "tablerow.tpl.html"));

After this, the page title is assigned. Then we put the first row of the table in place by using the following four lines of code:
$tpl->assign("NAME", "Andrew"); $tpl->assign("AGE", "21"); $tpl->assign("LOCATION", "Scotland"); $tpl->parse("TABLE_DATA", ".tablerow");

Here we are assigning the NAME, AGE, and LOCATION placeholders, then parsing the tablerow template with these values into the TABLE_DATA placeholder within the mainbody template. Notice how we use the . before the tablerow keyword in the parse function. This simply means that we want the tablerow template to be appended to the TABLE_DATA keyword. (If we did not use the ., only the last row would be visible.)

Converting the Command Parser Example to FastTemplate

Here we are going to convert our slightly larger command parser example (where users can be added, edited, and deleted) to instead use FastTemplate. The first step is to create the templates. In the command parser example, we have two main screens (the list of users and the Add User form page). The table that displays the list of users also needs to have a

Chapter 3 / Creating Web-Based Server Interfaces

template. We can also create a generic look-and-feel template instead of using the header and footer idea. So we will be creating the following list of templates:
Template main.tpl.html userlist.tpl.html userlist_row.tpl.html adduser.tpl.html Purpose Main look-and-feel (previously displayed with the showHeader and showFooter functions) Contains the elements of the page that displays the list of users Defines how a row in the user list should look Contains the form required for the add user page

And here is what each of these template files should contain: main.tpl.html
<html> <head> <title>Command Processor (using FastTemplate)</title> </head> <body bgcolor="#0000AA"> <center> <table width="600" border=1 bgcolor="#FFFFFF"> <tr> <td> {MAIN_CONTENT} </td> </tr> </table> </center> </body> </html>

userlist.tpl.html
<center> <br><b>Registered Users</b><br><br> <a href="mysqlcore.php?cmd=adduser">Add User</a> <br><br> <table width="450" border="1" cellpadding="4"> <tr bgcolor="#BBBBBB"> <td width="25%"><b>Name</td> <td width="25%"><b>Age</td> <td width="25%"><b>Alias</td> <td width="25%"><b>Password</td> <td width="25%"><b>Edit</td> <td width="25%"><b>Delete</td> </tr> {TABLE_ROW}

Chapter 3 / Creating Web-Based Server Interfaces

</table> </center>

userlist_row.tpl.html
<tr> <td width="25%">{USER_NAME}</td> <td width="25%">{USER_AGE}</td> <td width="25%">{USER_ALIAS}</td> <td width="25%">{USER_PASSWORD}</td> <td width="25%"><input type="button" value="Edit" onClick= "window.location = 'mysqlcore.php?cmd=adduser&id={USER_ID}'"></td> <td width="25%"><input type="button" value="Delete" onClick= "window.location = 'mysqlcore.php?cmd=deleteuser&id={USER_ID}';"></td> </tr>

adduser.tpl.html
<center> <br> <b>{FORM_TYPE} User Form</b> <br><br> <form method="POST" action="mysqlcore.php"> <input type="hidden" name="cmd" value="doadduser"> <input type="hidden" name="id" value="{USER_ID}"> <table width="450" border="1" cellpadding="4"> <tr> <td>Name:</td> <td><input type="text" name="name" value= "{USER_NAME}"></td> </tr> <tr> <td>Age:</td> <td><input type="text" name="age" size=3 value= "{USER_AGE}"></td> </tr> <tr> <td>Alias:</td> <td><input type="text" name="alias" value= "{USER_ALIAS}"></td> </tr> <tr> <td>Password:</td> <td><input type="password" name="password" value= "{USER_PASSWORD}"></td> </tr> </table> <br><br> <input type="submit" value="{SUBMIT_TEXT}"> <input type="button" value="Cancel" onClick="window.location = 'mysqlcore.php'"> </form>

Chapter 3 / Creating Web-Based Server Interfaces

Next, we have edited mysqlcore.php and renamed it ftcore.php. Here is how it looks now: ftcore.php
<?php // include libraries here... include("class.FastTemplate.php"); // Create a FastTemplate object... $tpl = new FastTemplate("./tpl"); // Define the template files... $tpl->define(array(main => "main.tpl.html", userlist => "userlist.tpl.html", userlist_row => "userlist_row.tpl.html", adduser => "adduser.tpl.html")); include("ftusers.php"); // special commands (no header/footer) switch($cmd) { case "doadduser": doAddUser(); exit(); case "deleteuser": deleteUser(); exit(); } // command processing... switch($cmd) { case "adduser": addUser(); break; default: showUsers(); break; } // Display the final page $tpl->parse("FINAL", "main"); $tpl->FastPrint(); ?>

Chapter 3 / Creating Web-Based Server Interfaces

And our new ftusers.php (formally users.php): ftusers.php

<?php function showUsers() { global $tpl; // connect... $dbh = mysql_connect("localhost", "root", ""); // select the database... mysql_select_db("phptest", $dbh); // fetch all the users... $results = mysql_query("SELECT * FROM users ORDER BY name", $dbh); while($row = mysql_fetch_array($results)) { $tpl->assign("USER_ID", htmlspecialchars($row["id"])); $tpl->assign("USER_NAME", htmlspecialchars($row["name"])); $tpl->assign("USER_AGE", htmlspecialchars($row["age"])); $tpl->assign("USER_ALIAS", htmlspecialchars($row["alias"])); $tpl->assign("USER_PASSWORD", htmlspecialchars ($row["password"])); $tpl->parse("TABLE_ROW", ".userlist_row"); } mysql_close($dbh); $tpl->parse("MAIN_CONTENT", "userlist"); } function addUser() { $id = $_GET["id"]; if($id != "") { $dbh = mysql_connect("localhost", "root", ""); mysql_select_db("phptest", $dbh); $results = mysql_query("SELECT * FROM users WHERE id = '$id'", $dbh); if($row = mysql_fetch_array($results)) { $tpl->assign("USER_ID", htmlspecialchars($row["id"])); $tpl->assign("USER_NAME", htmlspecialchars($row["name"])); $tpl->assign("USER_AGE", htmlspecialchars($row["age"])); $tpl->assign("USER_ALIAS", htmlspecialchars ($row["alias"])); $tpl->assign("USER_PASSWORD", htmlspecialchars ($row["password"])); }

100

Chapter 3 / Creating Web-Based Server Interfaces

mysql_close($dbh); $tpl->assign("FORM_TYPE", "Edit"); $tpl->assign("SUBMIT_TEXT", "Save Changes"); } else { $tpl->assign("USER_ID", ""); $tpl->assign("USER_NAME", ""); $tpl->assign("USER_AGE", ""); $tpl->assign("USER_ALIAS", ""); $tpl->assign("USER_PASSWORD", ""); $tpl->assign("FORM_TYPE", "Add"); $tpl->assign("SUBMIT_TEXT", "Add User"); } $tpl->parse("MAIN_CONTENT", "adduser"); } function doAddUser() { $id = $_POST["id"]; // connect... $dbh = mysql_connect("localhost", "root", ""); // select the database... mysql_select_db("phptest", $dbh); // add the new user... if($id == "") mysql_query("INSERT INTO users VALUES (NULL, \"".$_POST["name"]."\", \"".$_POST["age"]."\", \"".$_POST["alias"]."\", \"".$_POST["password"]. "\")", $dbh); else { // update the details... mysql_query("UPDATE users SET name = \"".$_POST["name"]."\", age = \"".$_POST["age"]."\", alias = \"".$_POST["alias"]."\", password = \"".$_POST["password"]."\" WHERE id = '$id'", $dbh); } // close the connection... mysql_close($dbh); // Redirect the page... header("Location: mysqlcore.php"); } function deleteUser() {

Chapter 3 / Creating Web-Based Server Interfaces

101

$id = $_GET["id"]; // connect... $dbh = mysql_connect("localhost", "root", ""); // select the database... mysql_select_db("phptest", $dbh); // delete the user... mysql_query("DELETE FROM users WHERE id = '$id'", $dbh); // close the connection... mysql_close($dbh); // Redirect the page... header("Location: mysqlcore.php"); } ?>

There is little need for a screenshot for this example as it looks exactly the same as it did before from the users point of view. There is nothing new as far as FastTemplate goes in this final example; however, it should put you in the right direction for structuring your web-based interfaces.

Summary
In this chapter we learned how to set up an Apache server, install PHP4, and the basics of creating a web-based interface to a MySQL database. With this knowledge it is possible, with experimentation, to create complete administration tools for your online games so that they can be administered from a standard web site. Additionally, you can display player statistics and other such information on your site or even go as far as creating an online community site for your game. In the next chapter, we move away from databases and web programming and proceed by looking into TCP/IP .

This page intentionally left blank.

Chapter 4

Introduction to TCP/IP
Introduction
Modern computer games take advantage of the Internet, the worlds largest network, more and more every day. If we want to use the full potential of it in a game, we must understand how the Internet works. It doesnt help much if we can write flawless code but dont get the theory right. This chapter introduces the TCP/IP protocol and teaches how to write TCP/IP applications, or more accurately games, using the sockets application program interface (API).

What Is a Protocol?
To fully understand what TCP/IP is, we must understand what a network protocol is. Basically, a network protocol is the language that two or more computers use when they share information with each other. Naturally, all the computers must speak the same language to make the information flow between them. Imagine talking to someone who speaks a different language. You probably will not understand anything this person says to you. This would be the same as two computers using different protocols to communicate with each other. One computer simply will not understand what the other is trying to send to it. Hence, no information is shared between them.

103

104

Chapter 4 / Introduction to TCP/IP

A protocol defines the rules of communication and the format of the data. Protocols are standards that work on different kind of computers. This means that if two computers can be physically connected in a network, they can share information no matter what operating system they are running or even what type of computers they are if they use the same protocol.

OSI Model
The International Organization for Standardization (ISO) developed a model for data communications in the late 1970s called the Open Systems Interconnection (OSI) model. This model is the standard for all data communications, and it also defines the basics of other standards. The model is based on seven layers, all of which have their own unique area of the data communication process. These layers are connected to each other one by one. Between two layers is an interface that these layers use to work with each other. OSI defines the layers function and interface. It does not define how the layers are created. This makes it easy to create new data communication solutions based on the OSI model. But more importantly, it makes it easy to modify the existing ones. Because of this, we have the ability to choose from many different options to act on one layer. A good example is TCP and UDP , which are described later in this chapter. These two protocols belong to the transport layer, and we can choose which one to use in our application. We dont have to worry about anything extra, because the interface between the transport layer and the two other layers connected to it makes sure the data flows correctly. Figure 4-1 shows the OSI model layers and their order.

OSI Model Layers

Figure 4-1 OSI model layers

Chapter 4 / Introduction to TCP/IP

105

Layer 1: Physical: The first layer takes care of all the physical data transfers. It includes all the physical electronic devices such as circuits, cables, and connectors. Basically it defines the medium to be used in the data transfer. Layer 2: Data Link: The second layer consists of our computers network interface card and the device driver. The data is put into frames that are checked for errors and fixed if any errors exist. As game developers we should not be too concerned about this layer, because if the hardware can do normal networking, we have no problems. Layer 3: Network: The third layer is taken care of by the Internet Protocol (IP). It addresses the packets and frames and routes them to the correct address over subnets. More about IP later in this chapter. Layer 4: Transport: Layer four defines the method of data transfer. Depending on the method, the data is checked for errors, big packets are repackaged into smaller packets, and care is taken that the data is sent and received correctly. These features vary from method to method. This layer is the most interesting layer for the network programmer, as TCP and UDP , which are described later in this chapter, belong to this layer. Layer 5: Session: The fifth layer is fairly simple. It defines how two computers establish and end a session. It also takes care of the sessions synchronization, and defines when a computer can transfer data and when it can receive data. Layer 6: Presentation: The sixth layer takes care of possibly compressing and/or encrypting data. It defines what the data looks like when it is being transferred. Layer 7: Application: The seventh layer defines the network application, such as file transfers or e-mail.

A good example of the usage of the OSI model is a normal phone conversation. The telecommunication companies provide us layer 1 the cables and connectors through which the conversation is transferred from one place to another. Other companies take care of layer 2 they build phones for us. Telephone companies switches belong to layer 3. They direct the call to the correct place. Our phones use the method of transfer used by the switches and other phones to make our voice move from one place to another. This is layer 4. We are on layer 5 when we dial the number to call and the other ends phone rings. Modern phones modulate our speech using a method called pulse code modulation (PCM), which belongs to layer 6. Now we have a complete phone conversation (layer 7).

106

Chapter 4 / Introduction to TCP/IP

Internet Protocol
The protocol used in Internet communications is called TCP/IP . As the name implies, it is divided into two layers. These layers are part of the Open Systems Interconnection (OSI) model. See the section titled OSI Model for a description of the OSI model. Transmission Control Protocol (TCP) is part of the transport layer of the OSI model, and Internet Protocol (IP) is part of the network layer. TCP/IP is a protocol suite that has more members than the name suggests. One important member is another protocol from the transport layer User Datagram Protocol (UDP). Both TCP and UDP are connected to the IP protocol. Because the OSI model is based on layers, it is easy to develop better and better network solutions by just replacing one layer with a new, better one. We can change our network interface card (NIC), drivers, or network connection type, or we can develop a better protocol. All of these are happening all the time all around the world. Companies develop new NICs and people buy them. The same companies develop new drivers for their cards and people install them on their computers. For some people, this is not enough they want better Internet connections. Also, the Internet Protocol is improving. The current version is IPv4, but the next public version is IPv6. Requirements of the modern computer culture grow every day. As computers get faster and faster, people are buying more and more computers. But the reason people buy computers is not because they are fast. The reason is the Internet. Imagine what an average computer user did with his or her computer in the 1970s or 1980s. Word processing was likely one of the most common reasons for buying a computer then. Back then, average users had not even heard of the Internet. But even if they had, they didnt dream of ever using it at home. Nowadays people need computers for everything. Or at least they think they do. Getting an Internet connection is very easy now and even fairly cheap, so there really is no point in not getting one. This has led to problems. The number of free IP addresses is running out quickly as companies reserve addresses for themselves and individuals reserve their own IPs not necessarily for all the time, but at least for the time they are online. Demand for a good network system for universities was increasing in the U.S. in the 1960s. This led to the development of IPv4 during the 1970s. Its addresses consist of four decimal dotted bytes, for example, 192.168.0.1. But because the human mind is better with names than numbers, some IP addresses are given names, for example, www.huntedcow.com. A Domain Name Service (DNS) then looks up the name and the corresponding IP address number.

Chapter 4 / Introduction to TCP/IP

107

The addresses are divided into subnets class A, class B, and class C. Class A means that the first byte of the address is used to define the subnet. The very first bit is set to 0 to identify that it is a class A address. This bit is taken from the network byte, so only the remaining seven bits are used for identifying the network. The last three bytes are used to define the address itself. In class B, the first two bytes define the subnet and the other two the address. The first two bits of a class B address are 1 and 0, leaving 14 bits for the network part. In class C, the first three bytes define the subnet and the last byte the address. The first three bits of class C addresses are 1, 1, and 0, and the next 21 bits are used for the network part. There are about 16,000,000 class A IP addresses, all of which are in use already. Class B consists of about 65,000 addresses and class C has 254 addresses. The whole address space of IPv4 is about 4 billion addresses. While 4 billion is a big number, it is not big enough for IP addresses in the future. Class D addresses are multicast addresses. The first four bits are the identifier bits. On a class D address they are 1, 1, 1, and 0. Multicast defines a group of IP addresses. So when you send something to a multicast address, you are not sending to a single computer but a group of computers. Class E addresses are reserved for future use. Figure 4-2 shows the different address classes and how their data space is divided.
Figure 4-2 Address classes

The solution for the problem is IPv6. Among other improvements, it provides a 128-bit addressing system that increases the address space so much that we would need to colonize new worlds to use them all. At least thats the way it looks like right now. Only time will tell if it is true. IPv6 was developed in the 1990s and will most likely go into use during the next decade. Because of this, it is a good idea to prepare for

108

Chapter 4 / Introduction to TCP/IP

it now. Theres no point in making a network application that may not work tomorrow. Of course, IPv6 is compatible with IPv4 to a certain point, but to be absolutely sure we should develop our applications to be protocol independent from the beginning. This book teaches you how to do so.

Introduction to the Transport Layer

As we have already learned, the TCP/IP protocol suite consists of two layers of the OSI model network and transport. The transport layer is as important to TCP/IP as the network layer. For the network programmer, it is the most interesting part. When we write network applications, we write commands for the transport layer. In Internet communications using TCP/IP , the transport layer is either Transmission Control Protocol (TCP) or User Datagram Protocol (UDP). Sometimes people just say TCP when they really mean TCP/IP , so that may be confusing if you do not know the context.
NOTE We can also use raw sockets to get direct connection in our code to the Internet Protocol, bypassing TCP and UDP . However, this is beyond the scope of this book.

Transmission Control Protocol

Transmission Control Protocol is a connection-oriented protocol. This means that every time we want to communicate with a remote host, we must first establish a connection. Once we have established a connection, we do not have to worry about directing the messages we send to the correct place. When we are done with the connection, we must close it. TCP is also a reliable protocol. It makes sure the other end receives the messages we send, and it handles such things as duplicated packets. Sometimes when a packet is sent to the network, it may get lost for a while. This may happen, for example, if a router is experiencing some problems. The packet gets stuck on the router and so the receiving host never sends a confirmation that the packet has reached its destination. The sending host assumes that the packet is lost forever and retransmits it after a certain time. This new packet may find another route to the receiving host, while the original packet is still stuck on the router. If the router starts working normally again, it routes the packet to the correct place and the receiving host may receive the same packet again. TCP notices the duplicate packet and destroys it, as it is not needed anymore.

Chapter 4 / Introduction to TCP/IP

109

User Datagram Protocol

Unlike TCP , User Datagram Protocol is a connectionless protocol. No actual connection is established between the two communicating hosts. This means that you cannot say when a connection is open or closed; you just communicate with the other host by sending it data and waiting for data back. The sending host must define the target address every time it sends something. When a host receives a message, it knows where the message came from, and therefore it can transmit something back to the remote host. UDP is not reliable, as it does not ensure that the transmission is received. It also does not take care of duplicates. However, duplicates are not that common with UDP because the sending host wont retransmit anything if the message is not received simply because it does not care. UDP can be used in applications that need the best possible efficiency but very little reliability. Computer games fall into this category, and that is why UDP is so important to us. UDP can be made reliable, but this requires that we write the needed checking algorithms ourselves. Some checking is always good, but no one forces us to make any.

Ports
Taking into account that most network application servers today are very simple, it would be foolish to make a network server run only one service. A game server may require all of the servers resources, so there are exceptions too. But how can we identify the service we want to use if the server is running dozens of services? We cannot connect every service one by one and check to see if it is the one we want, as this may take a long time. Additionally, some services may seem like the one we are looking for but are not. Therefore, we need to give each service a number that we define when we are connecting to it. These numbers are called port numbers. The port number is a 16-bit value, so there are 65,535 possible ports available (there is no port 0 therefore, 216 1). Actually, available ports are not that straightforward. Ports from 1 to 1023 are so-called well-known ports, and thus we cannot use them for our servers. These ports are reserved for common network services such as FTP (port 21) and daytime server (port 13). Ports from 1024 to 65535 are generally free to be used, but it is a good idea to check that the port we are about pick is not used by any other known service. The Internet Assigned Numbers Authority (IANA) records used ports if they are well known enough.

110

Chapter 4 / Introduction to TCP/IP

TCP and UDP ports are unique. For example, TCP port 1024 is not the same as UDP port 1024. Usually, if a port number is registered by the IANA, it is registered for both protocols at the same time, even if the application does not use both protocols at that time. We need to define the port number only for the server. Clients usually use so-called ephemeral ports, which we have no control over. The kernel of the operating system chooses them for us. The range of ephemeral ports varies from platform to platform, as shown in Figure 4-3. We can also assign an ephemeral port for a server, but only in a special case, which is beyond the scope of this book.
Figure 4-3

Not all ports are always open to use, even if they are not well-known ports. Firewalls block most ports on servers to keep unwanted people out of the server, so the servers administrator must open a port on the firewall before it can be used. There are different kinds of firewalls available, so consult your servers administrator about the ports.

Introduction to Sockets
Sockets API is a programming interface that we use to write network applications. It is a multiplatform API, so we can make different operating systems communicate with each other using sockets. This book teaches you how to make your network code work on multiple platforms without any changes to the code. There is no clear definition for a socket. Different people have different opinions about them, but basically a socket is a pipe between two computers on a network through which data flows. It is not the physical cable or anything concrete like that. A socket exists only in the world of bits and bytes. The two computers each have their own unique socket, and these sockets can be identified as the two ends of the pipe. Request for Comment 147 (RFC147) defines what a socket is for the ARPA network. As the ARPA network is the predecessor of the Internet, this documentation applies to Internet sockets as well.

Chapter 4 / Introduction to TCP/IP

111

Figure 4-4

In computer memory, sockets are 32-bit numbers. So when a new socket is created, it is given a unique number that defines the socket on the local computer.

Socket Types
There are two kinds of sockets available: stream sockets and datagram sockets. A stream socket is a connection-oriented socket; thus, a connection has to be established before it can send or receive data. Stream sockets use the TCP transport protocol. A datagram socket is a connectionless socket. Datagram sockets use the UDP transport protocol. You may think that a connection must be established before usage. This is not as straightforward as it sounds, though. We can think of this in terms of multiple levels of connections. The last level is when we have established the connection between two stream sockets. Before that there is a level where there is no real connection, but the two hosts know of each other and their addresses. They can sort of throw things at each other and hope that they reach their destination. With an established connection, however, the two hosts could throw the things into a pipe (the connection between the two hosts), and the probability of receiving the data would be much higher than without a connection. In a stream socket, the data flows constantly in the socket, as the name implies. It works like a stream of water, except that it moves in both directions. Stream sockets are reliable in many ways. Both sides know if the other side disconnects or crashes and cannot receive anything anymore. Every packet is monitored to see if it reaches its destination. If a packet has not reached its destination in a certain time, it will be retransmitted. In a datagram socket, the data is transmitted in datagrams. This means that whenever there is something to send, the data is sent to the address defined. When there is nothing to send, no data flows between the two hosts. If one side crashes, the other side will not notice unless a system has been built that checks if the other side is alive. Datagram sockets are not reliable. The packets are not monitored by TCP/IP to

112

Chapter 4 / Introduction to TCP/IP

see if they reach their destination. If we want, we can create our own monitoring system, but if we use UDP for the monitoring system as well, the monitoring system is also not reliable. As this may get very confusing, some people use only stream sockets.

Address
Each socket has its own address information. We can create a socket on any port as long as the port is free and it is okay to choose that port. We can also define the IP address to connect to for every socket. Usually, a single socket is not enough on a server application. We need one socket to listen to the incoming clients and another socket to handle the client/server communication. At the same time, the listening socket is still listening for clients and creates a new socket for all other clients as well. This type of server is called a concurrent server. If the server uses only one socket to do all the communication, it can communicate with only one client at a time. This type of server is called an iterative server. When creating computer games, we need to process multiple clients at once, so a concurrent server is the only way to do it. This book covers both iterative and concurrent servers.

Platforms
The sockets API works on many platforms, and thus we should know how to take advantage of it. This section covers Unix (Linux) and Windows versions of the API; once we are done, we will know how to write code that works on both platforms. The sockets API was originally developed for the Unix operating system. The 4.2BSD (Berkeley Software Distribution) system had the first version of sockets in 1983. At the same time, TCP/IP was released widely to the public for the first time. The API has developed from these beginnings. Many Unix platforms use the same networking code as BSD does, but others, for example Linux, do not. Linuxs network code has been written from scratch, but this does not mean these different implementations would not be compatible. Linux is a free, very popular Unix-based operating system. All Unix code in this book has been developed and tested on a Linux system. Many non-Unix operating systems have the sockets API as well. For example, Microsoft Windows has its own sockets library called WinSock (Windows Sockets). WinSock is compatible with BSD sockets, but it also has many features that the BSD version does not have. Because of operating system differences, there are some noticeable differences in the APIs, but they do not interfere with compatibility when connecting two computers. The latest version of WinSock is WinSock2.

Chapter 4 / Introduction to TCP/IP

113

WinSock2 software development kit (SDK) is available on the Internet. This SDK is required for writing applications that use WinSock.

History of WinSock
The Windows sockets API was born October 10, 1991, at Interop 91 in San Jose, California. A committee was established to design a specification for a sockets library for the Windows operating system from a proposal by Martin Hall of JSB Corporation. There have been over 40 companies involved in the design of WinSock. WinSock is not the property of Microsoft, although it is an important part of Windows nowadays. It was developed by independent sources who were interested in taking part of this project. On January 20, 1993, the specification for WinSock 1.1 was published. This version had support for TCP/IP only. WinSock2 was published in 1996. It provided support for multiple transport protocols, such as Novell IPX/SPX and Digitals DECNet, and officially supported the OSI model. Version 2 also includes features like multicasting and Quality of Service (QoS), both of which are explained later in this book. WinSock is currently compatible with version 4.3 of the Berkeley Software Distribution sockets.

Summary
Because we are writing games, we consider WinSock essential. The fact is that Windows is the operating system for which most games nowadays are written, and we make no exception here. But why do we need Unix code too? We could write the game for Unix platforms, but the main reason is the server, as most network games use Unix as the platform for their servers. There are many reasons for this, but the two most important reasons are the great stability of all Unix systems and easy remote controlling. Every Unix system can be controlled remotely, so we do not actually have to sit down in front of the server every time we want to control the server. Using Windows 9x as the server platform is out of the question because of certain restrictions. Windows NT and Windows 2000 are also good options for the server platform, as they do not have the same restrictions as Windows 9x systems because they were designed for different kinds of usage. In the end, it is up to you which operating system you wish to use. But remember, the client and the server can be run on different platforms.

This page intentionally left blank.

Chapter 5

Basic Sockets Programming

Introduction
The Unix sockets API does not require any extra initialization before we can actually use the sockets. However, this is not the case for Windows sockets. What we need to initialize is the WinSock 2 dynamic-link library (DLL) and then enumerate the available protocols. As this book focuses only on the TCP/IP protocol, we only look for TCP and UDP protocols. In this chapter, we learn the basic initialization functions of WinSock. Then we see how to actually use them by creating an initialization function that does all the necessary initialization.

WinSock Initialization
This section explains the basics of WinSock initialization.

WSAStartup Function (Win32)

int WSAStartup(WORD wVersionRequested, LPWSADATA lpWSAData);

The WSAStartup function is used to initialize the WinSock API. As the first parameter, we must provide the version number we request. It is a word value, and so we must fill the variable using the MAKEWORD macro. MAKEWORD creates an unsigned 16-bit integer from the two 115

116

Chapter 5 / Basic Sockets Programming

unsigned characters we give it. The second parameter is a pointer used to get the data from the WinSock DLL. The data is stored in a WSADATA data type. The function returns 0 if everything went fine. If not, it returns one of the non-zero values listed in Table 5-1.
Table 5-1: WSAStartup function return values WSASYSNOTREADY WSAVERNOTSUPPORTED WSAEINPROGRESS WSAEPROCLIM WSAEFAULT The network subsystem is not ready. The version we requested is not supported. A blocking WinSock 1.1 operation is in progress, or the service providers callback function is in progress. The limit of WinSock tasks has been reached. WSAData is not a valid pointer.

WSACleanup Function (Win32)

int WSACleanup(void);

We use WSACleanup to uninitialize the WinSock API. This function will unregister the WinSock DLL used by our application. Windows keeps a record of all the DLLs used by applications and updates the reference count of each DLL. If the count is higher than zero, Windows knows that this DLL is used by an application and therefore it is kept open in the system memory. If the reference count is zero, the DLL is not open. It is our responsibility as programmers to keep Windows aware of all the registered DLLs. In other words, we must call WSACleanup at the end of every WinSock application if the WinSock API has been initialized. The return value is 0 if the operation was successful. On error, the return value is SOCKET_ERROR, but we can get a more accurate error value by using WSAGetLastError. Table 5-2 lists the return values of the WSACleanup function.
Table 5-2: WSACleanup function return values WSANOTINITIALIZED WSAENETDOWN WSAEINPROGRESS WinSock API is not initialized. A network subsystem error occurred. A blocking WinSock 1.1 operation is in progress, or the service providers callback function is in progress.

Chapter 5 / Basic Sockets Programming

117

WSAEnumProtocols Function (Win32)

int WSAEnumProtocols(LPINT lpiProtocols, LPWSAPROTOCOL_INFO lpProtocolBuffer, LPDWORD lpdwBufferLength);

This function is used to enumerate the available protocols on the local computer. The first parameter is a pointer to a list of protocols we wish to look for. We must create an integer list of the protocols. In this book we use only TCP and UDP , so we put only these two in the list: n IPPROTO_TCP Transmission Control Protocol n IPPROTO_UDP User Datagram Protocol The second parameter is a pointer to a WSAPROTOCOL_INFO buffer that is filled when the function is run. The third parameter is a value-result parameter. It is used to tell the function how big a buffer we need for the protocols, but the function may change it if it is not big enough. This is important, because when we start enumerating the protocols, we actually run this function twice. The first time, we do not fill the first two parameters because all we need to do is get the buffer size. We provide a pointer to a zero-size buffer so the function will increase the size for us. The next time we run the function, we give it all the needed parameter info, including the buffer size we got from the first call to WSAEnumProtocols.

WinSock Initialization Function

The following code shows us how to initialize WinSock API. It is a function from the network library we will create later. But as this function is essential for getting WinSock ready to be used, we introduce it here. After the code listing, we will go through the important parts to see what is really happening.
int dreamSock_InitializeWinSock(void) { WORD versionRequested; WSADATA wsaData; DWORD bufferSize = 0; LPWSAPROTOCOL_INFO SelectedProtocol; int NumProtocols; // Start WinSock2. If it fails, we do not need to call WSACleanup() versionRequested = MAKEWORD(2, 0); int error = WSAStartup(versionRequested, &wsaData); if(error) { return 1;

118

Chapter 5 / Basic Sockets Programming

} else { // Confirm that the WinSock2 DLL supports the exact version // we want. If not, call WSACleanup(). if(LOBYTE(wsaData.wVersion) != 2 || HIBYTE(wsaData.wVersion) != 0) { WSACleanup(); return 1; } } // Call WSAEnumProtocols to figure out how big of a buffer we need NumProtocols = WSAEnumProtocols(NULL, NULL, &bufferSize); if((NumProtocols != SOCKET_ERROR) && (WSAGetLastError() != WSAENOBUFS)) { WSACleanup(); return 1; } // Allocate a buffer; call WSAEnumProtocols to get an array of // WSAPROTOCOL_INFO structs SelectedProtocol = (LPWSAPROTOCOL_INFO) malloc(bufferSize); if(SelectedProtocol == NULL) { WSACleanup(); return 1; } // Allocate memory for protocol list and define what protocols to // look for int *protos = (int *) calloc(2, sizeof(int)); protos[0] = IPPROTO_TCP; protos[1] = IPPROTO_UDP; NumProtocols = WSAEnumProtocols(protos, SelectedProtocol, &bufferSize); free(protos); protos = NULL; free(SelectedProtocol); SelectedProtocol = NULL; if(NumProtocols == SOCKET_ERROR) { WSACleanup(); return 1; }

Chapter 5 / Basic Sockets Programming

119

return 0; }

First, we use the MAKEWORD macro to fill the WORD versionRequested with the information we want. In this case, we want to check that the version number of WinSock DLL is 2.0, so we fill the word with bytes representing 2 and 0. Then we run WSAStartup to start initializing the WinSock API.
versionRequested = MAKEWORD(2, 0); int error = WSAStartup(versionRequested, &wsaData);

The function fills wsaData for us. We check the wsaDatas wVersion member (WORD) to see what version the DLL supports. We use the LOBYTE and HIBYTE macros to check the two bytes of the WORD. If the bytes do not match the version number we want, we clean up WinSock and return 1 to indicate there was an error.
if(LOBYTE(wsaData.wVersion) != 2 || HIBYTE(wsaData.wVersion) != 0) { WSACleanup(); return 1; }

Next, we need to find out how big a buffer we need for the protocols. We use WSAEnumProtocols for this. During the initialization process, we run this function twice. The first time we run it, we do not provide any parameters for it other than the buffer size parameter. The function will increase the buffer size for us if it is too small. We start with a zero size buffer, so the first time we run the function it will definitely fail. We want this function to fail for now. This means that if the function does not return SOCKET_ERROR, something is wrong, probably because the buffer size is too small if it is 0. Then we check that the last error occurred because the buffer was too small. We do this by using the WSAGetLastError function. If the buffer was too small, the last error message was WSAENOBUFS. If it was something else, there is something else wrong and we should quit the initialization process.
NumProtocols = WSAEnumProtocols(NULL, NULL, &bufferSize); if((NumProtocols != SOCKET_ERROR) && (WSAGetLastError() != WSAENOBUFS)) { WSACleanup(); return 1; }

Here is how we define the protocols we are looking for. First we allocate memory for the protocol list quite normally, and then simply fill the list with protocols we want. Because we only want the two TCP/IP

120

Chapter 5 / Basic Sockets Programming

protocols, that is what we put in the list. The order of the list members is not important.
int *protos = (int *) calloc(2, sizeof(int)); protos[0] = IPPROTO_TCP; protos[1] = IPPROTO_UDP;

Finally, we get to enumerate the actual protocols. We now have all the information we need for the second call to WSAEnumProtocols. For the first parameter we provide the protocol list; for the second, the protocol info structure pointer; and for the last, the buffer size. The return value is the number of protocols it found. The return value is SOCKET_ ERROR if something is wrong. We do not need to check what actually is wrong. It is enough to know that something did go wrong, and we stop initializing if this happens.
NumProtocols = WSAEnumProtocols(protos, SelectedProtocol, &bufferSize);

NOTE WSAEnumProtocols looks for protocols installed on your Windows operating system. If you do not have TCP/IP protocols installed on your operating system, WSAEnumProtocols will not find them.

Error Handling
There is one WinSock function in the previous code that we have not discussed yet. This is the WSAGetLastError function, which is a WinSock-only function.

WSAGetLastError Function (Win32)

int WSAGetLastError(void);

This function retrieves the value of the last error that occurred in the last Windows sockets operation function. Some functions only indicate that an error occurred but do not give the actual error value. This function is used to get the error value. Unix does not have a function to receive the last error that occurred. It does, however, have the global variable errno that works just like the return value of WSAGetLastError, but you do not have to fetch it by running a function. The error values on different platforms may have equal integer values (they also may differ), but their constant names usually are not the same. For example, WinSock error values have the prefix WSA.

Chapter 5 / Basic Sockets Programming

121

Sockets Data Types

Sockets have more than one address data type because different protocols require unique information. We can still write protocolindependent code thanks to the generic address data types. These data types are explained here.

Platform-specific Data Types

As we will be writing our code for multiple platforms, we must be aware of different data types that correspond to the same thing on different platforms. We can easily define new data types and rename the required data types to match the other platforms types. This can be done because usually only the data type name is different; the actual data is the same. Table 5-3 shows two data types that have different names on Unix and Win32 but are really the same thing.
Table 5-3: Equivalent data types Data Type Address length data type Socket descriptor data type Unix socklen_t int Win32 int SOCKET

From now on we use the more informative data type name in the text and all source code examples. For socket address length, we use socklent_t instead of int, and for socket descriptors, we use SOCKET instead of int. We learn how to create these new definitions for our data types in Tutorial 2, Creating Your Network Library.

Address Structures
The address information of the sockets is stored in structures. Naturally, IPv4 and IPv6 have their own structures.

IPv4 Address Structure

The following code shows the structure for IPv4:
struct in_addr { in_addr_t s_addr; };

// 32-bit IP address

122

Chapter 5 / Basic Sockets Programming

struct sockaddr_in { uint8_t sin_len; sa_family_t sin_family; in_port_t sin_port; struct in_addr sin_addr; char sin_zero[8]; };

// // // // //

structure length: 16 bytes protocol family: AF_INET 16-bit port number 32-bit IP address Not used

Lets take a look at the structure members. The structure length variable is handled by the kernel, so we do not need to worry about it. The protocol family for IPv4 addresses is always AF_INET. The port number is stored in a 16-bit unsigned integer in network byte order. The IP address is stored in a 32-bit unsigned integer, also in network byte order. It is not stored in the normal dotted decimal format. We discuss how to get the dotted decimal format out of the 32-bit unsigned integer later.

IPv6 Address Structure

This code shows the IPv6 address structure:
struct in6_addr { uint8_t s6_addr[16]; }; #define SIN6_LEN struct sockaddr_in6 { uint8_t sin6_len; sa_family_t sin6_family; in_port_t sin6_port; uint32_t sin6_flowinfo; struct in6_addr sin6_addr; };

// 128-bit IP address

// required for the compiler

// // // // //

structure length: 24 bytes protocol family: AF_INET6 16-bit port number 32-bit flow label and priority 128-bit IP address

SIN6_LEN must be defined if the length member is supported by the system. The structure length of an IPv6 address is 24 bytes, but the kernel takes care of this. The protocol family for IPv6 addresses is AF_INET6. The port number is a 16-bit unsigned integer, just like the IPv4 ports. The IPv6 address structure has a member that is not included in the IPv4 address structure. It stores the flow label and priority values. The first 24 bits are used for the flow label, the next 4 bits are for the priority, and the remaining 4 bits are reserved for future use. The IP address is stored in 16 8-bit unsigned integers.

Chapter 5 / Basic Sockets Programming

123

Generic Address Structure

Having more than one protocol family that can be used creates a small problem. Because IPv4 and IPv6 have their own address structures, we cannot pass either of the structures as a pointer to any socket functions that use the address structure as a parameter, and still remain protocol independent. That is why we have a generic address structure, which is shown below.
struct sockaddr { uint8_t sa_len; sa_family_t sa_family; char sa_data[14]; };

// structure length // protocol family // the address (either IPv4 or IPv6)

By typecasting the protocol-specific address structures into this generic structure, we can use any version of address structures in any of the socket functions. Of course, the functions must have this generic structure as the parameter instead of the protocol-specific ones. Typecasting means that you are providing a mask for your pointer in the memory. This mask is used to divide the block of memory into the variables that are stored there. There is no need to typecast a pointer if the data type we want to use is the original data type. The beginning of the data types must match the generic one. But as we see in the protocol-specific address structures and the generic address structure, only the first two members of the structures match. The third member in the generic structure is chars only, and these chars are used to store the actual address information. The function itself must understand this. Figure 5-1 shows an example of how the typecast mask works. In this example, the first two members of the cast structure are 8-bit integers, or shorts (one X in the memory block means 8 bits), and after that there are only characters that hold the data in a format that can be transformed into any data type.
Figure 5-1 Typecasting

TIP

As you have probably already noticed, typecasting is a useful way to make code work with various data types. It is a good idea to take advantage of it whenever possible.

124

Chapter 5 / Basic Sockets Programming

Basic Sockets Functions

There are some functions in the sockets API that nearly all sockets applications use at some point. These functions are explained in this section.

socket Function (Unix, Win32)

SOCKET socket(int family, int type, int protocol);

This function creates a socket with the provided information. This function only creates the descriptor of the socket; it does not really start using any port or any IP address yet. If you are familiar with Unix programming, you may notice that the socket descriptors are just like file descriptors on Unix. The first parameter (int family) specifies the protocol family. It can be one of these: n n n AF_INET: IPv4 protocols AF_INET6: IPv6 protocols AF_ROUTE: Routing sockets

There are more options for this parameter on different platforms. We cover only the first two, AF_INET and AF_INET6, in this book. The second parameter (int type) defines the socket type, which can be one of the following: n SOCK_STREAM: Stream socket n SOCK_DGRAM: Datagram socket n SOCK_RAW: Raw socket This book covers only SOCK_STREAM and SOCK_DGRAM. SOCK_STREAM is used for TCP sockets and SOCK_DGRAM for UDP sockets. The third parameter (int protocol) is set to 0 when using either stream or datagram sockets. Win32 return values: n Success: A nonnegative descriptor (integer) n Failure: INVALID_SOCKET Unix return values: n Success: A nonnegative descriptor (integer) n Failure: 1

Chapter 5 / Basic Sockets Programming

125

bind Function (Unix, Win32)

int bind(SOCKET s, const struct sockaddr *addr, socklen_t addrlen);

The bind function makes your socket have its own address information, i.e., the IP address and the port number. The function binds the local address information to each socket. You cannot define a non-local IP address when calling bind, but you can define any port number (keeping in mind that not all ports are available ports). This function is usually used right after the call to the socket function. To listen to incoming events on a certain port, you must bind the port to the socket first. You cannot listen to incoming events if you have not bound the socket to the port to be listened to. Both stream and datagram sockets must be bound to an address before they can be used. You do not have to bind a clients local sockets, but it is okay to do so. The first parameter (SOCKET s) defines the unbound socket to be bound. The socket must have been created with the socket function. The second parameter (const struct sockaddr *addr) is a pointer to a sockaddr address information structure. The third parameter (socklen_t addrlen) is the size of the address information structure. The address you are about to bind may already be in use by another application or by your own application. If this is the case, bind returns the error value EADDRINUSE (Unix) or WSEADDRINUSE (Windows) (the value must be retrieved by using the WSAGetLastError function in Windows and the errno variable in Unix). With certain socket options we can still bind the address, even if it is already in use. If you set the port number to 0, the operating systems kernel will choose an emphemeral port. Usually this is the first free ephemeral port. Client applications use ephemeral ports because there really is no need for us to know the port the client uses. The remote host the client is sending data to can figure out the port itself. If we have multiple network interface cards on our local host, we can either choose the IP address for the socket ourself or we can make the kernel choose it for us. To do this, we use the constant INADDR_ANY.
NOTE A single host can have multiple IP addresses if it has multiple network interface cards. Usually only server machines have this kind of arrangement.

If an error is encountered with the bind function, we must use the WSAGetLastError function in Windows or the errno variable in Unix to get the actual error value.

126

Chapter 5 / Basic Sockets Programming

Win32 return values: n n Success: 0 Failure: SOCKET_ERROR

Unix return values: n Success: 0 n Failure: 1

connect Function (Unix, Win32)

int connect(SOCKET s, const struct sockaddr *addr, socklen_t addrlen);

This function is used to connect a TCP client with a TCP server. We do not need to know the clients address information, hence there is no need to call the bind function on the client before we call connect. The kernel will do all the dirty work for us in this case. It chooses the ephemeral port and retrieves the IP address information. The first parameter (SOCKET s) defines the socket to connect. The second parameter (const struct sockaddr *addr) must provide information about the servers address: the IP address and port number. If we fail to give the correct address information for the server, the connection will fail. The third parameter (socklen_t addrlen) is used to define the length of the address structure in bytes. This function does not return before the TCPs three-way handshake is complete. If an error occurs before that, the function will return. So when this function returns, we know that either the connection is established or it could not be established. The client will wait a total of 75 seconds for the remote server to respond to its SYN segment of the three-way handshake. If no response is received during this time, the function returns an error. If the client does receive a response to the SYN segment, but it is an RST (reset), it means that the server machine is running but not waiting for connections at the TCP port we specified. The function will return an error value. These and any other specific error values must be retrieved by using the WSAGetLastError function in Windows or the errno variable in Unix. Win32 return values: n Success: 0 n Failure: SOCKET_ERROR Unix return values: n Success: 0 n Failure: 1

Chapter 5 / Basic Sockets Programming

127

listen Function (Unix, Win32)

int listen(SOCKET s, int backlog);

The listen function is used for connection-oriented sockets to set the socket to listen for incoming connections. This means that it is not used for UDP sockets. This function is called after the calls to the socket and bind functions. This function must be called before the accept function is called. This is a server-side function. The first parameter (SOCKET s) must be a bound, unconnected socket. The socket must have been created with the socket function, and it must be bound with the bind function. The second and last parameter (int backlog) defines the maximum number of connections allowed. This sounds very simple but in fact is not. First of all, all operating systems have their own maximum for this. For example, Windows NT 4.0 Server has a maximum backlog value of 100. Normally, on operating systems that are not designed for server usage, the backlog value is about 4 to 5. The kernel keeps track of two connection queues: incomplete and complete connections. The latter one is clear; they are the connections that have been established all the way. This means that the TCP three-way handshake is complete with them. The incomplete connections are connections that have the first part of the TCP three-way handshake complete. On a packet level it means that the first SYN packet from the client has been received by the server. The sum of these two queues cannot exceed the backlog value. After this function successfully returns, the used socket is called a listening socket. If the backlog value is set to be larger than the maximum of the operating system, the value is silently set to the nearest valid value. Silently means that there is no way to know about it; no error value is returned. If the queue is full and there is an incoming connection, listen will return ECONNREFUSED on Unix and WSAECONNREFUSED on Windows. These error values must be retrieved by the WSAGetLastError function in Windows or the errno variable in Unix. Win32 return values: n Success: 0 n Failure: SOCKET_ERROR Unix return values: n Success: 0 n Failure: 1

128

Chapter 5 / Basic Sockets Programming

accept Function (Unix, Win32)

int accept(SOCKET s, struct sockaddr *addr, socklen_t *addrlen);

This function is used to accept an incoming TCP connection. This function is run after the listen function has filled the completed connections queue with at least one connection. When the queue is empty and this function is run, the process is either put to sleep or moves on to the next command, depending on the socket I/O option (blocking or non-blocking). This is a server-side function. The first parameter (SOCKET s) defines the socket that holds the connection to be accepted. This must be a listening socket. The second parameter (struct sockaddr *addr) is used to get the address information of the client whose connection we are just about to accept. This can be set to NULL if we are not interested in this information. The third parameter (socklen_t *addrlen) is a value-result parameter, which must be set to the size of the address structure before calling accept. When the function returns, this parameter holds the number of bytes allocated for the address structure by the function. This can be set to NULL if we are not interested in this information. After accept successfully returns, the used socket is called a connected socket. The return value is a new descriptor for the socket if everything went fine. The listening socket given to the function as a parameter stays untouched. It is not removed, and its descriptor is not changed. We can still listen for new incoming connections with the very same socket. What we get from accept is a brand new connected socket. If there was an error, accept returns an error value indicating it, but to get accurate error values we use the WSAGetLastError function in Windows or the errno variable in Unix. So if we have a TCP server that will serve multiple clients at once, we create one listening socket and then spawn new connected sockets for each connection. Win32 return values: n Success: A nonnegative descriptor (integer) n Failure: INVALID_SOCKET Unix return values: n Success: A nonnegative descriptor (integer) n Failure: 1

Chapter 5 / Basic Sockets Programming

129

close Function (Unix)/closesocket Function (Win32)

int close(SOCKET s); int closesocket(SOCKET s);

The Unix and Windows versions of this function have different names, but the function is the same on both operating systems. This function closes the TCP socket. Before actually terminating the connection, TCP will send all queued data. We cannot send or receive any more data even if the connection has not been terminated for good. When all queued data is sent, TCPs four-packet termination process initiates. We can make this function work differently by adjusting socket options. We discuss these options in Chapter 6, I/O Operations. The socket descriptor is returned to be reused. So after closing a socket, we may encounter a socket with the same descriptor as the one we closed. However, they have nothing do with each other. For example, if we close a socket and then immediately create a new one, the new socket must be initialized normally even if the descriptor is the same as the one we just closed. If an error occurs, we need to use the WSAGetLastError function in Windows or the errno variable in Unix to retrieve the actual error value. Win32 return values: n Success: 0 n Failure: SOCKET_ERROR Unix return values: n Success: 0 n Failure: 1

Input/Output Functions
There are four basic functions for sending/receiving data in the sockets API two for each operation. All four are explained in this section.

send Function (Unix, Win32)

int send(SOCKET s, const void *buf, size_t len, int flags);

This function is used to send data to a socket. The socket must be successfully connected before this function can be used. This function is usually used by stream (TCP) sockets only, but it is possible to use this

130

Chapter 5 / Basic Sockets Programming

with datagram (UDP) sockets too. The UDP socket must be connected using the sockets connect function, but that is beyond the scope of this book. The first parameter (SOCKET s) is the socket to which we want to send the data. It must be a connected socket. The second parameter (const void *buf) is the data itself. It is a pointer to a buffer containing the data we want to send. The third parameter (size_t len) defines the number of bytes to send. We do not need to send the whole data buffer (and it is not always even possible). The fourth and last parameter (int flags) is used to set different flags (options) with the current send process. These flags are temporary, so they must be set every time they are required by a send process. We can OR various flags together. Table 5-4 lists the possible flags.
HINT ORing is a bit-wise operation. Bit-wise operations are the lowest level of operations available as computers work using bits 1s and 0s. Because of this, they are also very fast and take very little memory. I recommend learning more about bit-wise operations and using them wherever possible. Table 5-4: send/sendto flags Flags MSG_DONTROUTE Description Target host is locally connected to the network. Do not look for the target host from a routing table. Set the current output non-blocking. Do not wait for the output of data. Send out-of-band data. Unix Yes Win32 Yes

MSG_DONTWAIT MSG_OOB

Yes Yes

No Yes

When send returns, it either returns the number of bytes sent or indicates that an error occurred. We cannot send zero size data with TCP protocol. If we get a 0 return value from send, it means that the other host has closed the connection or the connection is broken. If there was an error, we need to retrieve the actual error values by using the WSAGetLastError function in Windows or the errno variable in Unix. Even if send returns successfully, it does not mean that the other end successfully received the data. Win32 return values: n Success: Number of bytes sent n Failure: SOCKET_ERROR

Chapter 5 / Basic Sockets Programming

131

Unix return values: n n Success: Number of bytes sent Failure: 1

recv Function (Unix, Win32)

int recv(SOCKET s, void *buf, size_t len, int flags);

This function is very similar to the send function, but this time we are receiving data instead of sending it. The parameters are very similar. The first parameter (SOCKET s) is the socket we want to receive the data. It must be a connected socket. The second parameter (void *buf) is the data buffer. It is a pointer to a buffer where we want to store the data. The third parameter (size_t len) defines the number of bytes to receive. The fourth parameter (int flags) works the same way as in the send function. We can set temporary socket input options here by ORing various options together or simply set one on. Table 5-5 lists the input flags.
Table 5-5: recv/recvfrom flags Flags MSG_DONTWAIT MSG_OOB MSG_PEEK Description Set the current output non-blocking. Do not wait for the output of data. Receive out-of-band data. Peek at the incoming data. The data is copied to the receive buffer, but it is not removed from the incoming data queue. Unix Yes Yes Yes Win32 No Yes Yes

MSG_WAITALL

Wait for all the data to be received before Yes returning.

The return value of recv tells us how many bytes it received or, in case of an error, an error value. If recv returns 0, the other side has closed the connection or the connection is broken. If there was an error, we need to retrieve the actual error values by using the WSAGetLastError function in Windows or the errno variable in Unix. Win32 return values: n Success: Number of bytes received n Failure: SOCKET_ERROR

132

Chapter 5 / Basic Sockets Programming

Unix return values: n n Success: Number of bytes received Failure: 1

sendto Function (Unix, Win32)

int sendto(SOCKET s, const void *buf, size_t len, int flags, const struct sockaddr *to, socklen_t addrlen);

Like the send function, sendto is also used to send data to a socket, but with this function it is not necessary for the socket to be connected. This is the function that is normally used for datagram (UDP) socket output, but it can be used for stream (TCP) sockets too. The first parameter (SOCKET s) is the socket to which we want to send the data. The second parameter (const void *buf) is the data itself. It is a pointer to a buffer containing the data we want to send. The third parameter (size_t len) defines the number of bytes to send. We do not need to send the whole data buffer (and it is not always even possible). The fourth parameter (int flags) is used to set different options on with the current send process. These flags are temporary, and so they must be set every time they are required by a send process. We can OR various flags together. The flags listed in Table 5-4 work with sendto also. The fifth parameter (const struct sockaddr *to) is the address structure where we want to send the data. The address structure must contain the IP address and port. The sixth and last parameter (socklen_t addrlen) is used to define the size of the address structure. Similarly to the send function, sendto returns the number of bytes sent or indicates that an error occurred. It is okay to send zero size datagrams. When we send zero size datagrams, only the IP (v4 or v6) header and the UDP header are sent, so sendto can return 0 and still function normally. If there was an error, we need to retrieve the actual error values by using the WSAGetLastError function in Windows or the errno variable in Unix. Even if sendto returns successfully, it does not mean that the other end successfully received the data. Win32 return values: n Success: Number of bytes sent n Failure: SOCKET_ERROR Unix return values: n Success: Number of bytes sent n Failure: 1

Chapter 5 / Basic Sockets Programming

133

recvfrom Function (Unix, Win32)

int recvfrom(SOCKET s, void *buf, size_t len, int flags, struct sockaddr *from, socklen_t *addrlen);

This function is usually used for datagram (UDP) socket input. It is similar to the recv function, but we define the IP address and port to read from as well as the socket. The first parameter (SOCKET s) is the socket where we want to receive the data. The second parameter (void *buf) is the data buffer. Again, it is a pointer to a buffer where we want to store the data. The third parameter (size_t len) defines the number of bytes to receive. The fourth parameter (int flags) is used to set the temporary input flags. Table 5-5 lists the possible flags. The fifth parameter (struct sockaddr *from) is a pointer to the address structure that holds the address information of the host we want to receive data from. The sixth parameter (socklen_t *addrlen) is a value-result argument. It defines the size of the address structure and also returns the updated size when the function returns. The number of received bytes is returned if everything went fine. Unlike with the recv function, a 0 return value does not mean that the other host has closed the connection. That is obvious as there is no such thing as connection in UDP . This also means that we can write datagrams of zero size. If there was an error, we need to retrieve the actual error values by using the WSAGetLastError function in Windows or the errno variable in Unix. Win32 return values: n Success: Number of bytes received n Failure: SOCKET_ERROR Unix return values: n Success: Number of bytes received n Failure: 1

Address Data Conversion Functions

We humans prefer to have our information in written text or numbers instead of just ones and zeroes like computers. An Internet address is one of those things that may be displayed in many ways, depending on who, or better yet what, uses it. Humans can remember the addresses best when they are written as text, for example, www.huntedcow.co.uk. As computers do not have to worry about forgetting

134

Chapter 5 / Basic Sockets Programming

anything, they can be optimized to store the addresses in a format that takes the least amount of memory. We have three functions for IPv4 addresses and two functions for IPv6 addresses. These two new functions are protocol independent though, so it does not matter which protocol we use with them. We explain only one of these functions here because we need that function in our network library, and the others are easy to understand after you understand this one.

inet_aton Function (Unix, Win32)

int inet_aton(const char *string, struct in_addr *address);

This function converts the provided string address into a network byte ordered binary value. Note that depending on the DNS server configuration, the string can be an IP address in numeric or dotted decimal format or in written text as most web addresses are.

Client/Server Programming
Now that we have everything set up for the actual network programming, we can move on to write our server and client code. Servers and clients work very differently. Most of the time the server is waiting for connections to come in. Once one does come in, it serves the client that is connected.

Server Methods
Servers can be very passive applications. There can be times when they do not do anything at all. This can happen if there are no clients to be served. We must keep this in mind when creating the servers, because if we make our server loop when there are no connections, it can easily drain all the CPU time just by looping an empty loop. If our server is supposed to serve only one client at a time, it is called an iterative server. It does not listen for incoming connections once a connection is already open. When the connection terminates, the server starts to listen for a new connection. This kind of server is not very useful, because multiple clients may be on hold since the server can process only one connection at a time. Computer game servers simply cannot work this way, because it is against the principal idea of these servers. The idea is that multiple players connect to a server and then communicate via that server (i.e., play the game). That is why this kind of play is called multiplay.

Chapter 5 / Basic Sockets Programming

135

Servers that process multiple clients at once are called concurrent servers. When the server is started, it starts listening for incoming connections. When a connection comes in, it creates a child process (a thread) and the main program continues listening for incoming connections. The main program in this case is often also a thread, and not really the actual main program. The reason is simple: If it were the main program, it would not be much of a server. The main program is the very core, the part that initiates the listening functions and so on. Figure 5-2 shows how a concurrent server works. The listening socket is always there and, depending on the state of the application (i.e., whether it will allow any more connections), it will accept all the new connections coming in. Clients 1 to 3 have already established a connection and are being served normally by the server. Client 4 has just connected to the server, and is still in the process of the TCP three-way handshake. The listening socket is already free and has started to listen for more connections. Looking from the outside, a concurrent UDP server works the same way as TCP servers. Clients inform that they are there and the server takes care of them. There are lots of technical differences though, and these are discussed in the next section.
Figure 5-2 A concurrent server

Clients
Clients are active applications. They initiate the connection to the server, so they never wait for things to happen (other than wait for data from the server). The client application is the application that the normal user uses. In the world of gaming, the client is the game itself. The player should not know anything else about the server except the address and port

136

Chapter 5 / Basic Sockets Programming

number. Sometimes even this information is built into the game, so the player does not have to set the IP address and/or port manually. When working in a local area network (LAN), we can create a server search system, which is one way to eliminate the need for manually setting the address information. We discuss this system in Chapter 6, I/O Operations.

Byte Ordering
There is no standard way of ordering bytes in computer memory. Most PCs nowadays use Intels way of ordering the bytes, which is to store them in little-endian order. This means that if we have a two-byte variable, the last actual byte is stored first in the memory. Big-endian means that the bytes are stored in the correct order first byte first. Different processors store and access the bytes differently. As we said, Intels way is to store them in little-endian order. But because we are creating multiplatform applications, we cannot just use this one way. We must have a way to transform the bytes into a format that all computers understand. Therefore, we have to use network byte order. The network byte order is big-endian with the Internet protocols we use. There are functions to convert network byte ordered bytes into host byte ordered.
Figure 5-3 Byte ordering

Creating a Server
Now we are going to learn how to create a server on Unix and Windows, using both TCP and UDP . Lets go through the most important events when creating a server. Every sockets application (client or server) must first create the socket. Depending on the protocols we use, the parameters change accordingly.
// A stream (TCP) IPv4 socket SOCKET listeningSocket; listeningSocket = socket(AF_INET, SOCK_STREAM, 0);

Then, when the socket is successfully created, we usually fill in the address information of the server. We do not have to enter the local IP address of the server if we have only one network interface card on the server machine. If we have more than one, we would just enter the IP address we want to use. Here we assume that there is only one card, so

Chapter 5 / Basic Sockets Programming

137

we let the kernel automatically fill in the IP address. But we do have to enter the port number ourselves. We must remember the restrictions that exist when choosing a port number for our server. Chapter 4, Introduction to TCP/IP explains those things that limit port number availability.
struct sockaddr *servAddr; struct sockaddr_in *inetServAddr; int portNumber; // Allocate memory for the address structure and set it to zero. servAddr = (struct sockaddr *) malloc(sizeof(sockaddr)); memset((char *) servAddr, 0, sizeof(sockaddr)); // Fill the address structure. servAddr->sa_family = (u_short) AF_INET; inetServAddr = (struct sockaddr_in *) servAddr; inetServAddr->sin_port = htons((u_short) portNumber);

Of course, if we are developing a server for personal use only (LAN only), we can forget some of these restrictions. But even in this case we cannot use just any number, because the operating systems use some ports without having any external application installed. Because we are developing a game server for the public, we must choose a port that has no restrictions at all. Once we have filled in the required address information, we must bind this information to the socket we created in the beginning.
// Bind the address information to the socket. error = bind(listeningSocket, servAddr, sizeof(sockaddr));

NOTE Whatever we are programming, we should never forget to check every possible function for errors. A lot of crashes that people blame on the operating system are actually caused by an application that does not handle errors properly.

Now we need to discuss the TCP and UDP code separately because of their obvious differences.

TCP
As we have already learned, TCP is the easy transport protocol (for the programmer). This is also true when creating servers. All we need to do now is make the server listen for incoming connections and accept them. For every new connection, the server creates a child process and a new socket (if we are talking about a concurrent server).
// Listen for incoming connections. Queue max 5 connections. error = listen(listeningSocket, 5);

138

Chapter 5 / Basic Sockets Programming

... // Accept the connection. Accept is a blocking function. connectedSocket = accept(listeningSocket, NULL, NULL);

Figure 5-4 shows normal TCP client/server operation. You can see how much easier and faster it is to use UDP , as shown in Figure 5-5.
Figure 5-4 Normal TCP client/server operation

UDP
Because UDP is a connectionless protocol, it does not have a function for listening for incoming connections. The principal idea of UDP is that it simply does not have to listen to them. A UDP server just reads the incoming datagrams and acts how the programmer wants it to act.
Figure 5-5 Normal UDP client/server operation

But how do we keep the server organized with all the clients sending data to it? A game server needs to send data to a client pretty much whenever it is required, so we cannot let the client do all the active sending. The solution for this is a so-called knocking system. When the client wants to tell the server that it exists and wants to interact with the server, the client sends a knock datagram to the server. This datagram can be anything; it is up to you to decide what information is stored within it. When the server receives a datagram or message like this, it updates its list of clients and either creates a new child process to serve the client or simply acknowledges that there may be incoming

Chapter 5 / Basic Sockets Programming

139

messages from that client. In the latter case, the server uses the one and only socket to interact with all the clients. A UDP server that uses only one socket to interact with clients must always retrieve the address information when it is receiving data (assuming that it wants to send data back also). With UDP we cannot simply send data to a socket without providing the exact address. This kind of server is not necessarily an iterative server. It may process multiple clients at a time, as it responds only when the client sends data to the server.

Simple Echo TCP Server

It is time to create our very first sockets application: a simple echo TCP server. This server is an iterative server (serves only one client at a time), and its function is to return the character we send to it in uppercase. The protocol used is TCP , but later we will see how easy it is to make a UDP server from our existing TCP code. This example program is of no real use and is intended to show the basic idea of every server, which is the following: A request from the client is sent to the server, and the server sends feedback to the client. In this example, the request is simply the letter sent to the server, and the feedback is the uppercase letter that is sent back to the client. The following code shows the simple echo TCP server in its entirety. The code is written for the Windows operating system, but it is easy to change it so that it compiles and runs on Unix too. However, that is not the idea of this example, so lets skip it for now. The functional concept of this server is that the server listens for a connection to come in and when one does, it accepts it. Then the client is supposed to send single letters to the server, and the server will send them back in uppercase. If the user on the client end enters the letter q (lowercase), the server stops automatically. The whole application is stopped, so it will not accept any other connections. For this example, there is no need to write any extra checking if the user is exiting on the client end, because the recv function will notice if the connection is closed.
#ifndef _WINSOCKAPI_ #define _WINSOCKAPI_ #endif #include <windows.h> #include <winsock2.h> #include <stdio.h> // Declare the sockets we use.

140

Chapter 5 / Basic Sockets Programming

SOCKET listeningSocket; SOCKET connectedSocket; int InitSockets(void) { struct sockaddr *servAddr; struct sockaddr_in *inetServAddr; int error = 0; // Create the socket. listeningSocket = socket(AF_INET, SOCK_STREAM, 0); if(listeningSocket == INVALID_SOCKET) { printf("error: socket() failed"); return 1; } // Allocate memory for the address structure and set it to zero. servAddr = (struct sockaddr *) malloc(sizeof(sockaddr)); memset((char *) servAddr, 0, sizeof(sockaddr)); // Fill the address structure. servAddr->sa_family = (u_short) AF_INET; inetServAddr = (struct sockaddr_in *) servAddr; inetServAddr->sin_port = htons((u_short) 9009); // Bind the address information to the socket. error = bind(listeningSocket, servAddr, sizeof(sockaddr)); if(error == SOCKET_ERROR) { printf("error: bind() failed"); free(servAddr); return 1; } free(servAddr); servAddr = NULL; // Listen for incoming connections. Queue only one connection. error = listen(listeningSocket, 1); if(error == SOCKET_ERROR) { printf("error: listen() failed"); return 1; } // Accept the connection. Accept is a blocking function. connectedSocket = accept(listeningSocket, NULL, NULL); if(connectedSocket == INVALID_SOCKET)

Chapter 5 / Basic Sockets Programming

141

{ printf("error: socket() failed"); return 1; } return 0; } void ServerProcess(void) { int connectionOpen; char buf[2]; connectionOpen = 1; // Loop as long as connection is open. while(connectionOpen) { // Read the incoming data from the connected socket. if(recv(connectedSocket, buf, 2, 0)) { // Set the received letter to uppercase and // make sure the string ends after that by setting the next // byte to NULL. buf[0] = toupper(buf[0]); buf[1] = '\0'; printf("Got message from client: %s\n", buf); // Send the feedback. if(send(connectedSocket, buf, 2, 0) == SOCKET_ERROR) { connectionOpen = 0; } } else { closesocket(connectedSocket); connectionOpen = 0; } } } int main(void) { if(NET_WinSockInitialize() != 0) { printf("Critical error, quitting\n"); return 1; } if(InitSockets() != 0)

142

Chapter 5 / Basic Sockets Programming

{ printf("Critical error, quitting\n"); WSACleanup(); return 1; } ServerProcess(); WSACleanup(); return 0; }

Now lets see what is going on in the program. Lets start from the main function, as that is where the application always starts.

main Function
You may notice that there is not much in the main function, specifically no basic socket functions. But there are two function calls that are very important. Lets have a closer look at them:
NET_WinSockInitialize();

This is the function we introduced at the beginning of this chapter. It belongs to the network library that we will create later on. This function initializes the WinSock API for us. After a successful call, the WinSock API is ready to be used.
NOTE On Unix, we do not call the NET_WinSockInitialize function.
WSACleanup();

This function is very important also. It is used to uninitialize WinSock. We are not allowed to exit the application without calling this function if the WinSock API is initialized. As we have already learned, one of this functions tasks is to unregister the WinSock DLL from our application. Windows DLL registration system will lose track of the registered DLLs if we do not call this function at the end of each WinSock application.

InitSockets Function
The InitSockets function does all the initialization of sockets so they can be used and the server process itself can start. First we create the listening socket and check that it is successfully created. A listening socket is a socket that the server uses to listen for

Chapter 5 / Basic Sockets Programming

143

incoming connections. Once a connection comes in and is accepted, the server will start using a connected socket and leave the listening socket free for other clients. The socket we create here is an IPv4 stream socket. Because it is a stream socket, it uses the TCP transport protocol. The flags parameter is set to 0 as it is currently not used in the sockets API. Then we check whether the socket descriptor is invalid. If it is, we simply exit the program. If the socket is created as it is supposed to be, we move on.
listeningSocket = socket(AF_INET, SOCK_STREAM, 0); if(listeningSocket == INVALID_SOCKET) { printf("error: socket() failed"); return 1; }

Now we fill in the address information of the server. First we need to allocate memory for the structure and set it to 0. Then we fill the structure with the required information: protocol family and port number. There is no need to tell the program the IP address of the computer, as it can automatically retrieve it. If there is more than one network interface card installed, we can choose the one we want here. For this example program we have selected the port number 9009.
servAddr = (struct sockaddr *) malloc(sizeof(sockaddr)); memset((char *) servAddr, 0, sizeof(sockaddr)); // Fill the address structure. servAddr->sa_family = (u_short) AF_INET; inetServAddr = (struct sockaddr_in *) servAddr; inetServAddr->sin_port = htons((u_short) 9009);

Now that we have the address information ready in the structure, lets tell the socket to use that information. We bind the information to the listening socket with the bind function. Again, we must check whether an error occurred in the call to bind. If so, we need to free the allocated memory before we are allowed to exit. If everything went fine, we free the memory we allocated for the address structure and move on again.
// Bind the address information to the socket. error = bind(listeningSocket, servAddr, sizeof(sockaddr)); if(error == SOCKET_ERROR) { printf("error: bind() failed"); free(servAddr); return 1; }

144

Chapter 5 / Basic Sockets Programming

free(servAddr); servAddr = NULL;

Everything is now ready for listening for incoming connections. At first glance it may seem odd that it is just one call to a function without any loop as long as there are no connections coming in loop. Does the function just run once and then the process moves on to the next one? In this case, no. Some theory is required here to understand this. A socket can be blocking or non-blocking. With a blocking socket, some sockets functions will go to sleep when there is no action of any kind that needs to be processed. A non-blocking socket, on the other hand, will not put the functions to sleep. Once the functions are called, they check if there is an action to process; if there is not, they return and the next command (function) of the process is run. By default, all sockets are blocking. Blocking and non-blocking I/O is discussed more in Chapter 6, I/O Operations. So if we see a fragment of code similar to the following code, we cannot say how the application will perform, as we do not know if the socket is blocking or not. But as we do know that it is blocking in this example, we know that the application will stop at the listen function as long as there are no incoming connections. Now we put the listening socket to the use for which it was created by passing it as a parameter to the listen function. We set the backlog value to 1 because our server will process only one client at a time and, better yet, one client per instance. Again, we cannot forget checking for errors. If listen fails, we exit the application.
// Listen for incoming connections. Queue only one connection. error = listen(listeningSocket, 1); if(error == SOCKET_ERROR) { printf("error: listen() failed"); return 1; }

Now we have reached the accept function, so we know that someone wants to connect to our server. What we do now is create a new socket for the soon-to-be connected client. We pass the listening socket as a parameter to accept, as the connection we want to accept is on that socket. We have no interest in the address information of the client, so pass NULL as the two remaining parameters. Next we check for errors. If the connected socket is invalid after a call to accept, we exit the application. If it is valid, we are done initializing the server.

Chapter 5 / Basic Sockets Programming

145

// Accept the connection. connectedSocket = accept(listeningSocket, NULL, NULL); if(connectedSocket == INVALID_SOCKET) { printf("error: socket() failed"); return 1; }

ServerProcess Function
Now we get to the server process itself. This function has the main loop that every program has. It is looped as long as the connection is open. Lets have a closer look at the input/output functions we use here:
// Read the incoming data from the connected socket. if(recv(connectedSocket, buf, 2, 0)) { // Set the received letter to uppercase and // make sure the string ends after that by setting the next // byte to NULL. buf[0] = toupper(buf[0]); buf[1] = '\0'; printf("Got message from client: %s\n", buf); // Send the feedback. if(send(connectedSocket, buf, 2, 0) == SOCKET_ERROR) { connectionOpen = 0; } } else { closesocket(connectedSocket); connectionOpen = 0; }

At this point in the program, we can forget about the listening socket. This program is not going to use it anymore. The socket we use from now on is the connected socket that we got from the accept function. Lets get the data flowing! We call the recv function inside an if statement because we need to know if recv really read data from the socket or if it returned 0 to indicate that the connection has been closed or is lost. This is possible with a blocking socket, as recv will not return before it has data to read or before it notices that the connection is closed. A non-blocking socket would return 0 from a call to recv if there is no data to read but the connection is still alive. We read the data to a very small buffer in this example, only two bytes in size. When we notice that data has arrived, we process it.

146

Chapter 5 / Basic Sockets Programming

First, we simply set the received letter to uppercase, and then make sure the next letter is NULL; therefore, the string ends after the first letter. After this is done, we show the user what we received and send it back to the client. If we could not send the data (if send returns less than 0), we assume that the connection is broken and exit the program. We close the socket and exit the program if recv returns 0. Remember that in this example, because the sockets we use are blocking, the socket functions do not return if there is nothing happening (for example, if recv is not receiving data).

Simple Echo UDP Server

As we want to use UDP as well as TCP in our socket programs, we now modify the simple echo TCP server code to work with UDP . Modifying the TCP code is rather easy. The following shows all of the UDP server code.
#ifndef _WINSOCKAPI_ #define _WINSOCKAPI_ #endif #include <windows.h> #include <winsock2.h> #include <stdio.h> // Declare the sockets we use. SOCKET Socket; int InitSockets(void) { struct sockaddr *servAddr; struct sockaddr_in *inetServAddr; int error = 0; // Create the socket. Socket = socket(AF_INET, SOCK_DGRAM, 0); if(Socket < 0) { printf("error: socket() failed"); return 1; } // Allocate memory for the address structure and set it to zero. servAddr = (struct sockaddr *) malloc(sizeof(sockaddr)); memset((char *) servAddr, 0, sizeof(sockaddr)); // Fill the address structure. servAddr->sa_family = (u_short) AF_INET;

Chapter 5 / Basic Sockets Programming

147

inetServAddr inetServAddr->sin_port

= (struct sockaddr_in *) servAddr; = htons((u_short) 9009);

// Bind the address information to the socket. error = bind(Socket, servAddr, sizeof(sockaddr)); if(error == SOCKET_ERROR) { printf("error: bind() failed"); free(servAddr); return 1; } free(servAddr); servAddr = NULL; return 0; } void ServerProcess(void) { struct sockaddr_in inetClientAddr; int clientLen; int connectionOpen; char buf[2]; clientLen = sizeof(inetClientAddr); connectionOpen = 1; // Loop as long as connection is open. while(connectionOpen) { // Read the incoming data from the connected socket. if(recvfrom(Socket, buf, 2, 0, (struct sockaddr *) &inetClientAddr, &clientLen)) { // Set the received letter to uppercase and // make sure the string ends after that by setting the next // byte to NULL. buf[0] = toupper(buf[0]); buf[1] = '\0'; printf("Got message from client: %s\n", buf); // Send the feedback. if(sendto(Socket, buf, 2, 0, (struct sockaddr *) &inetClientAddr, clientLen) == SOCKET_ERROR) { connectionOpen = 0; }

148

Chapter 5 / Basic Sockets Programming

} else { connectionOpen = 0; } } } int main(void) { if(NET_WinSockInitialize() != 0) { printf("Critical error, quitting\n"); return 1; } if(InitSockets() != 0) { printf("Critical error, quitting\n"); WSACleanup(); return 1; } ServerProcess(); WSACleanup(); return 0; }

The biggest change in the UDP code in comparison to the TCP code is that we have only one socket. On the TCP server, we had a listening socket and a connected socket. On the UDP server, we have only one generic socket because there is no need to listen for incoming connections and connect them. This one UDP socket just reads the incoming data and sends data back. The following sections provide a closer look at the changes.

InitSockets Function
First, we change the calls to the socket and bind functions to match these two function calls. We have replaced the listening socket with the generic socket and we are now creating a datagram socket instead of a stream socket. After calling bind we are done. If we were using TCP , we would start the listening process now. But with UDP there is no need, so we can start the server process function right after we have bound the local address information to the socket.

Chapter 5 / Basic Sockets Programming

149

Socket = socket(AF_INET, SOCK_DGRAM, 0); ... error = bind(Socket, servAddr, sizeof(sockaddr));

ServerProcess Function
For the server processing function, we declare two variables that are not used in the TCP version. The first one is the Internet client address structure, and the second one is an integer holding the length of the structure. We need to set the length variable to the size of the structure before we pass it to any function, as shown in the following code.
struct sockaddr_in inetClientAddr; int clientLen; ... clientLen = sizeof(inetClientAddr);

The next thing we have changed is that we have replaced the recv and send functions with the recvfrom and sendto functions. They work almost like the ones in the TCP version, but there are two new parameters in recvfrom and sendto. We must pass the address information structure and its length as parameters in both functions. In recvfrom, the address structure is filled by the function. When the function is receiving data, it fills the structure with the corresponding address where the data is coming from. The function also updates the length variable, as it is a valueresult argument. Now that we have the address information for where the data came from, we use it to send data to the correct host. We pass the address structure and the length of the structure to sendto, updated by recvfrom. This way we are always sending data to the correct host, because in this example program we only send data when we have first received it. If we had a program that required sending data even when we have not received anything (we need the address information though, hence at least one datagram must have been received before), we would need to store all the addresses of the clients we want to send data to. That is why it is a good idea to have some kind of a system where there is a dedicated message for informing the server of the client.
if(recvfrom(Socket, buf, 2, 0, (struct sockaddr *) &inetClientAddr, &clientLen)) ... if(sendto(Socket, buf, 2, 0, (struct sockaddr *) &inetClientAddr, clientLen) == SOCKET_ERROR)

150

Chapter 5 / Basic Sockets Programming

These are all the changes we need to make the server work using the UDP protocol. The UDP version of the server works a little bit differently as it will not exit when the client is shut down. This is because there is no connection that is closed. This feature has its good and bad sides. One good thing is that the server can be used effectively because only one process is running all the time. A bad thing is the fact that we do not know when a client crashes or something else like that happens. But, as we have already said many times, UDP is unreliable, but it can be made reliable.

Creating a Client
What would we do with a server if we did not have a client application? Nothing. So lets make one. Remember that in computer games, the game itself is the client, so you must design your game so that sending and receiving data does not interfere with other parts of the game too much. Too much? It is almost impossible to make it work so that the communications library would not have any effect on the game flow. First, lets look at the functions we need for all clients.

TCP
We create the client-side socket exactly like the one on the server. We must set it to use the same protocols on both ends (IPv4 and TCP in this case).
SOCKET Socket; Socket = socket(AF_INET, SOCK_STREAM, 0);

Then we convert the Internet address from the server IP number to a form that the computer can use. After that we fill the address structure, but we do not bind this address to the socket ourselves, because the connect function will do it for us.
struct sockaddr_in inetServAddr; int portNumber; u_long inetAddr = inet_addr(IPaddress); memset((char *) &inetServAddr, inetServAddr.sin_family inetServAddr.sin_port inetServAddr.sin_addr.s_addr 0, sizeof(inetServAddr)); = AF_INET; = htons((u_short) portNumber); = inetAddr;

error = connect(Socket, (struct sockaddr *) &inetServAddr, sizeof(inetServAddr));

Chapter 5 / Basic Sockets Programming

151

That is all the basic initializing there is to do on the client side when using the TCP protocol.

UDP
Creating a UDP socket requires only one modification: We need to change the second parameter of the socket function to SOCK_ DGRAM. That is all. We fill the address structure in the same way as when using TCP .
SOCKET Socket; Socket = socket(AF_INET, SOCK_DGRAM, 0);

As the UDP protocol is a connectionless protocol, we do not need to run the the connect function at all when using UDP . This is the biggest difference in initializing client sockets. When sending data to a UDP server, we need to pass the address information of the server to the sending function every time we run it. Therefore we must store the address structure globally to be able to access it from all functions.

Simple Echo TCP Client

As we can see in the following code, the client source code is not very different from the server code.
#ifndef _WINSOCKAPI_ #define _WINSOCKAPI_ #endif #include <windows.h> #include <winsock2.h> #include <stdio.h> // Declare the only socket we need. SOCKET Socket; int InitSockets(char *IPaddress) { struct sockaddr_in inetServAddr; int error = 0; // Create a TCP socket. Socket = socket(AF_INET, SOCK_STREAM, 0); if(Socket < 0) { printf("error: socket() failed"); return 1;

152

Chapter 5 / Basic Sockets Programming

} // Create the Internet address from the IP number u_long inetAddr = inet_addr(IPaddress); memset((char *) &inetServAddr, inetServAddr.sin_family inetServAddr.sin_port inetServAddr.sin_addr.s_addr 0, sizeof(inetServAddr)); = AF_INET; = htons((u_short) 9009); = inetAddr;

// Try to connect the TCP server. error = connect(Socket, (struct sockaddr *) &inetServAddr, sizeof(inetServAddr)); if(error != 0) { printf("error: could not find server.\n"); return 1; } return 0; } void ClientProcess(void) { int connectionOpen; char transmitBuf[3]; char receiveBuf[3]; strcpy(transmitBuf, ""); strcpy(receiveBuf, ""); connectionOpen = 1; // Loop as long as connection is open. while(connectionOpen) { // Get the string to send. if(gets(transmitBuf)) { if(strcmp(transmitBuf, "q") == 0) { closesocket(Socket); connectionOpen = 0; break; } // Send the transmit buffer to the socket. if(send(Socket, transmitBuf, 2, 0) == SOCKET_ERROR) { connectionOpen = 0; } }

Chapter 5 / Basic Sockets Programming

153

// Read the incoming data from the connected socket. if(recv(Socket, receiveBuf, 2, 0)) { printf("Got reply from server: %s\n", receiveBuf); } else { connectionOpen = 0; } } } int main(int argc, char *argv[]) { if(argc < 2) { printf("Usage: SimpleEchoTCPClient.exe <Server IP>\n"); return 1; } NET_WinSockInitialize(); if(InitSockets(argv[1]) != 0) { printf("Critical error, quitting\n"); WSACleanup(); return 1; } ClientProcess(); WSACleanup(); return 0; }

Now lets see what is different.

main Function
The only thing that is different in this function in comparison to the server code is the following: the two parameters in the main function, the if statement, and the call to the InitSockets function. The if statement checks whether the user provided enough arguments when running the executable. The first argument is the executable name itself (in the form the user entered it), and the second argument in this case should be the IP address of the server. If there are not enough arguments, the program displays a usage message telling the user what arguments are needed. The call to the InitSockets function is

154

Chapter 5 / Basic Sockets Programming

a little bit different, because now we need to pass the IP address information to it. Other parts of the main function are similar to the server code just some simple function calls and the cleanup code.
int main(int argc, char *argv[]) { if(argc < 2) { printf("Usage: SimpleEchoTCPClient.exe <Server IP>\n"); return 1; } ... if(InitSockets(argv[1]) != 0) ...

InitSockets Function
The most obvious change in this function is that we now provide the server IP address within a parameter. We need to do this to make it possible to enter any IP address when running the client. We could hardcode an address to the code, but that would not be very wise. Then we should always have the server on the hardcoded IP address (which is not always even possible).
int InitSockets(char *IPaddress)

As we can see in the following code, the socket is created exactly like it is on the server end. There really is no way to make it different because we must use the very same protocols. Once more, we must check for errors.
// Create a TCP socket. Socket = socket(AF_INET, SOCK_STREAM, 0); if(Socket < 0) { printf("error: socket() failed"); return 1; }

Now that we have the server IP address in a string, we need to convert it to a form the computer understands. After that we reset the address structure memory to 0. Then we fill the address information structure with this address and the well-known port number, which is 9009 in this example. We must convert the integer value 9009 (host byte ordered value) to network byte ordered format. We also set the protocol family to AF_INET since we are using IPv4.

Chapter 5 / Basic Sockets Programming

155

// Create the Internet address from the IP number u_long inetAddr = inet_addr(IPaddress); memset((char *) &inetServAddr, inetServAddr.sin_family inetServAddr.sin_port inetServAddr.sin_addr.s_addr 0, sizeof(inetServAddr)); = AF_INET; = htons((u_short) 9009); = inetAddr;

Then all there is left to do in this function is to connect the server. To do this we use connect. You probably noticed that we do not call bind at all on the client. This is because we do not have to; connect does the address binding for us. Therefore, we must provide the address structure we just filled earlier for connect. We typecast the structure to the generic address format, because connect is designed to work on both IPv4 and IPv6, and so it accepts only generic format addresses. It is very important to check for errors here. If there is an error, it usually means that we could not find the server. Some other errors may occur too, but usually it is enough to inform the user that we could not connect the server.
// Try to connect the TCP server. error = connect(Socket, (struct sockaddr *) &inetServAddr, sizeof(inetServAddr)); if(error != 0) { printf("error: could not find server.\n"); return 1; }

connect also is a blocking function, but not like other blocking functions. If we set the socket to non-blocking mode, the connect function is not affected by this. It will still wait a certain amount of time for the connection to succeed, and after that time it will fail.

ClientProcess Function
This function matches the ServerProcess function on the server. Its purpose is to wait for the user to enter the letter to send and then send it. After that it will wait for a response from the server, then the loop starts all over again. Lets see what is going on in the function. We have separate buffers for transmitting and receiving to prevent mix-ups. The application could work with one buffer only, but it is much better for the programmer that we have two buffers. In bigger programs, it is sometimes a must to have different buffers.
char transmitBuf[3]; char receiveBuf[3];

156

Chapter 5 / Basic Sockets Programming

strcpy(transmitBuf, ""); strcpy(receiveBuf, "");

Here we get the letter to send into the transmit buffer using the gets function. Some problems arise if the user enters more than one letter, but in this example we do not worry about that because it really is beyond the scope of this example. For now, it is enough that we take care of this on the server by making sure the string ends after the first letter. Next we check whether the letter entered was q (lowercase). If it was, we close the socket and exit the loop, thus exiting the whole program. The server will notice that we have closed the socket and it will exit too. If the letter we entered was something other than q, we send it to the server. Because a string always contains at least two bytes, assuming that the string is not empty (one letter + NULL), and because we want to send only one letter, we send two bytes. If send returns SOCKET_ERROR, something went wrong and we exit the program without any extra checking.
// Get the string to send. if(gets(transmitBuf)) { if(strcmp(transmitBuf, "q") == 0) { closesocket(Socket); connectionOpen = 0; break; } // Send the transmit buffer to the socket. if(send(Socket, transmitBuf, 2, 0) == SOCKET_ERROR) { connectionOpen = 0; } }

After we have sent the data to the server, we immediately start to wait for the response by using the function recv. If the number of bytes received is more than zero (the return value of recv is more than zero), it means that data has successfully arrived. If not, we exit the loop and the whole program. After this, the loop starts all over again, assuming that no errors have been encountered and that the user did not enter the letter q.
// Read the incoming data from the connected socket. if(recv(Socket, receiveBuf, 2, 0)) { printf("Got reply from server: %s\n", receiveBuf); } else

Chapter 5 / Basic Sockets Programming

157

{ connectionOpen = 0; }

Simple Echo UDP Client

The following code shows the UDP version of the simple echo client.
#ifndef _WINSOCKAPI_ #define _WINSOCKAPI_ #endif #include <windows.h> #include <winsock2.h> #include <stdio.h> // Declare the only socket we need. SOCKET Socket; struct sockaddr_in inetServAddr; int servLen = sizeof(inetServAddr); int InitSockets(char *IPaddress) { int error = 0; // Create a TCP socket. Socket = socket(AF_INET, SOCK_DGRAM, 0); if(Socket < 0) { printf("error: socket() failed"); return 1; } // Create the Internet address from the IP number u_long inetAddr = inet_addr(IPaddress); memset((char *) &inetServAddr, inetServAddr.sin_family inetServAddr.sin_port inetServAddr.sin_addr.s_addr return 0; } void ClientProcess(void) { int connectionOpen; char transmitBuf[3]; char receiveBuf[3]; 0, sizeof(inetServAddr)); = AF_INET; = htons((u_short) 9009); = inetAddr;

158

Chapter 5 / Basic Sockets Programming

strcpy(transmitBuf, ""); strcpy(receiveBuf, ""); connectionOpen = 1; // Loop as long as connection is open. while(connectionOpen) { // Get the string to send. if(gets(transmitBuf)) { // Send the transmit buffer to the socket. if(sendto(Socket, transmitBuf, 2, 0, (struct sockaddr *) &inetServAddr, servLen) == SOCKET_ERROR) { connectionOpen = 0; } } // If the letter the user entered is "q", stop the application. if(strcmp(transmitBuf, "q") == 0) { connectionOpen = 0; break; } // Read the incoming data from the connected socket. if(recvfrom(Socket, receiveBuf, 2, 0, NULL, NULL)) { printf("Got reply from server: %s\n", receiveBuf); } else { connectionOpen = 0; } } } int main(int argc, char *argv[]) { if(argc < 2) { printf("Usage: SimpleEchoTCPClient.exe <Server IP>\n"); return 1; } NET_WinSockInitialize(); if(InitSockets(argv[1]) != 0) { printf("Critical error, quitting\n");

Chapter 5 / Basic Sockets Programming

159

WSACleanup(); return 1; } ClientProcess(); WSACleanup(); return 0; }

Note that it is not much different from the TCP version. The following sections look at what has changed.

InitSockets Function
We create the socket exactly like in the TCP version but with one change. Instead of creating a stream socket, we create a datagram socket. To do this, we set the second parameter to SOCK_DGRAM. Then we fill the address information structure that holds the servers address. Again, we do it exactly like in the TCP version, only this time the structure is a global variable as we need it elsewhere in the code. Once the structure is filled completely, we are done. There is no need to call bind or connect and we can move on to the client process function.
Socket = socket(AF_INET, SOCK_DGRAM, 0);

ClientProcess Function
In this function, we replace send and recv with the following functions and remove the socket closing functions. Unlike on the server-side code, we do not need to receive any data before we know the address of the server. This is obvious, is it not? If we do not know the address of the server, we hit the wall. We cannot do anything without that piece of information. Fortunately in computer games today, the servers can be found automatically using a built-in or external application that uses a server (we do not need to know the address of this server, as it is built in) to retrieve the IP addresses and ports of the servers. Lets assume that we know the address of the simple echo server and that we have entered the correct address when running this client. Now we use the address information structure that we filled in for the InitSockets function to send the data to the correct host. Then we assume that the server address does not change and we do not care about checking the address structure returned by recvfrom.
if(sendto(Socket, transmitBuf, 2, 0, (struct sockaddr *) &inetServAddr, servLen) == SOCKET_ERROR)

160

Chapter 5 / Basic Sockets Programming

... if(recvfrom(Socket, receiveBuf, 2, 0, NULL, NULL))

NOTE It is possible that the server address changes between a sendto and recvfrom. For example, a concurrent UDP server spawns the connections on new ephemeral ports because it must use the well-known port for listening for incoming connections. So when a client sends data to the server, the server spawns a child process and a new port for that connection. Then the next time the client receives data from the server it must update its address information for the server because the port has changed.

Running the Simple Echo Application

To try out the simple echo application, we must first start the server program and then start the client program. We must start the programs in this order because the client application will exit if it cannot find anything on the server port. To run the server, from the command prompt type:
> SimpleEchoTCPServer.exe

Then we run the client, passing the server IP address as the first and only argument:
> SimpleEchoTCPClient.exe 127.0.0.1

In this example, we are running both programs on the same host, and therefore we can use the IP address 127.0.0.1, which is a local host address. When you use this address, you are pointing to your own computer. Now that both of the programs are running, we can start sending data back and forth. On the client side, enter any letter except q and then press Enter. You will see on the serverside how the letter is received, and almost immediately on the client side how the letter is sent back in uppercase. To stop the application, enter the letter q on the client program. The TCP version will end both client and server, but the UDP version ends only the client. Figure 5-6 and 5-7 show a normal run of the simple echo application.

Chapter 5 / Basic Sockets Programming

161

Figure 5-6 Simple echo server running

Figure 5-7 Simple echo client running

Summary
In this chapter, we encountered the basic socket functions for the first time. We learned about the parameters we pass to them and the values they return. We learned some of the technical differences between TCP and UDP . More importantly, we learned how to use the functions by creating our first sockets application, the simple echo application. We can now move on to more advanced technology, as we now know the basics.

This page intentionally left blank.

Chapter 6

I/O Operations
Introduction
A network input/output operation requires more than just knowing how to send and receive data. We need to know when to start receiving the data and how to set it up. This chapter discusses the input/output operations of socket network events. We will also learn a new way to send data and how to modify the way our sockets act.

Detecting Network Events

To be able to start reading the socket for incoming data at the correct time, we need to know when there is something to read. We could poll the read function on the socket all the time, but that is not always a good way to do it. We need to be able to wake up the read function whenever there is any incoming data. The following functions are used to accomplish this: n select Unix, Win32 n WSAAsyncSelect/WSAEventSelect Win32

select (Unix, Win32)

int select(int maxfdp1, fd_set *readset, fd_set *writeset, fd_set *exceptset, const struct timeval *timeout);

The select function is used to set the kernel to wake up the process when a network event occurs. We can set the function to wake up the 163

164

Chapter 6 / I/O Operations

process on any type of network event (writing, reading, and exception condition pending) at the same time. Plus, we can set a timeout value to make the process wake up after a certain amount of time. The first parameter (int maxfdp1) defines the number of socket descriptors to test for network events. Its value is set to the highest socket descriptor to test plus one, because it is the number of descriptors (and the descriptor values begin from 0). So if we want to test our socket for network events, we should set this value to (at least) the socket descriptor plus one. The second, third, and fourth parameters (fd_set *readset, fd_set *writeset, and fd_set *exceptset) define the network events we want to test for. We discuss only the reading and writing events in this book. These parameters are pointers to fd_set type of data, which is used to store the notification of a possible network event. After the select function is run, we check if the socket is a member of a set by using the macro FD_SET. The macros are explained below. The fifth and last parameter (const struct timeval *timeout) defines the timeout value for the function. This is a pointer to a timeval structure, which holds two members: seconds and microseconds. To make the function wait forever, we set this parameter to NULL. We can set the timeout value to 0 seconds and 0 microseconds. In that case, the function returns immediately after first checking for the network events. Win32 return values: n Success: Positive number of ready descriptors, 0 on timeout n Failure: SOCKET_ERROR Unix return values: n Success: Positive number of ready descriptors; 0 on timeout n Failure: 1

Macros
The following four macros are used to modify and check the socket descriptor sets: n FD_ZERO n FD_SET n n FD_CLR FD_ISSET

FD_ZERO resets a set so that no socket descriptors belong to it. It is a good idea to do this before any other macro is used.

Chapter 6 / I/O Operations

165

FD_SET adds the socket descriptor to the set. FD_CLR removes the socket from the set. FD_ISSET checks if the socket descriptor is a member of the set.

WSAAsyncSelect (Win32)
int WSAAsyncSelect(SOCKET s, HWND hWnd, unsigned int wMsg, long lEvent);

This function is used to set Windows kernel to send a message to notify of a network event on a socket. This system can provide information about numerous network events, such as the following: n n n n FD_READ Ready to read data FD_WRITE Ready to write data FD_ACCEPT Incoming connection FD_CLOSE Socket closing

The first parameter (SOCKET s) defines the socket to monitor. This does not have to be a connected socket, because we can also monitor for incoming connections with this function. Note that datagram sockets do not tell about incoming connections only stream sockets do. The second parameter (HWND hWnd) defines the window handle to which the message is sent. The third parameter (unsigned int wMsg) defines the message to send when the event defined by the next parameter occurs. We can (and should) create our own message for this. The fourth parameter (long lEvent) defines the network event(s) to monitor. We can define multiple events at once by ORing them together here, such as FD_READ | FD_WRITE. Win32 return values: n Success: 0 n Failure: SOCKET_ERROR

WSAEventSelect (Win32)
int WSAEventSelect(SOCKET s, WSAEVENT hEventObject, long lNetworkEvents);

This function sets an event object to receive a notification of the specified network events. The event object can then be used to see which network event happened, if any. The network events we can specify are exactly the same as in the WSAAsyncSelect function. This function gives us some more breathing room because we do not have to tie the network events to a window. We can specify any number of event objects and then use them wherever we wish. The first parameter (SOCKET s) defines the socket to monitor. The second parameter (WSAEVENT hEventObject) defines the event

166

Chapter 6 / I/O Operations

object handle that will receive the notification of the network events. The third parameter (long lNetworkEvents) defines the network events to monitor. As with the WSAAsyncSelect function, we can OR multiple events together. Win32 return values: n Success: 0 n Failure: SOCKET_ERROR

WSAWaitForMultipleEvents (Win32)
DWORD WSAWaitForMultipleEvents(DWORD cEvents, const WSAEVENT far *lphEvents, BOOL fWaitAll, DWORD dwTimeOUT, BOOL fAlertable);

This function polls for an event to happen and actually tells us if there is a network event that we should process. This function returns either when there is a network event happening (one we are waiting for) or when the timeout value has been reached. We can set this value to infinite so the function returns only when an event occurs. The first parameter (DWORD cEvents) is the number of events to wait for. This is the number of members in the lphEvents array (parameter two). At least one must be specified, but we cannot specify more than what WSA_MAXIMUM_WAIT_EVENTS specifies. The second parameter (const WSAEVENT far *lphEvents) is a pointer to the array of network event objects. The third parameter (BOOL fWaitAll) defines whether the function should wait for all the events to occur before it returns. Possible values are TRUE and FALSE (1 and 0). If set to FALSE, this function returns when at least one event occurs. The fourth parameter (DWORD dwTimeOUT) defines the timeout value in milliseconds. If this value is reached, the function returns no matter what, even if the fWaitAll flag is set to TRUE and the timeout value is reached. The fifth and last parameter (BOOL fAlertable) defines whether the function should return if there is an I/O completion routine queued by the system for execution. Possible values are TRUE and FALSE (1 and 0). Win32 return values: n Success: The event object that caused the function to return

Event Object
An event object is a normal Windows handle that is used to store the state of a network event (or any other event). For example, if we want to know when there is data to read on a socket, we create an event

Chapter 6 / I/O Operations

167

object, set it to inform us of incoming network events, and check its state. Lets take a look at a more detailed example. First we create the event object handle and a socket:
HANDLE readEvent; SOCKET s;

We presume the socket is initialized properly somewhere. Then we set the event object to receive notification of incoming data:
WSAEventSelect(s, readEvent, FD_READ);

And finally, when we are ready to read data from a socket, we check if there is anything to read:
WSAEVENT EventArray[1]; EventArray[0] = SocketInputEvent; int waitStatus = WSAWaitForMultipleEvents(1, EventArray, FALSE, WSA_INFINITE, FALSE);

Multithreading
It is safe to say that multithreading is a must in a non-iterative network application. It is the only way to keep all the clients handled properly by the server since there can be multiple clients connected to one server at a time. If each client waited for the server to handle the other clients first, we could call our system Wait Wait Wait.
NOTE A UDP server does not need multithreading as much as a TCP server. This is because a UDP client can just throw in a message to the server, then wait for a response from the server and we are done. The server could handle all the clients on one socket, thus removing one reason for multithreading. It depends on the design of the UDP network application whether we should use multithreading or not.

What Is Multithreading?
Before we tell you how to make our application multithreaded, lets take a moment to think about what multithreading really is. Multithreading means that there is more than one process running in one application. These processes run constantly and at the same time. All the threads and the main application share the same memory. This means that if you have a global variable, it can be accessed and modified by each thread. Each process usually has its own loop to keep it running. A normal single-threaded application has only one main loop,

168

Chapter 6 / I/O Operations

which is the backbone of the application. When this loop breaks, the application terminates. This is the same for every extra thread. When a thread reaches the end of the thread function, it terminates and the thread is destroyed. When the main application ends, so do the threads. Although the threads seem to be running all at the same time, this is not actually true on a single-processor system. One processor can process only one thing at a time, so the threads are really run one by one, but only a little bit at a time. An example from real life explains this best. Imagine you have three papers to write. You could write one completely, then write the next one, and so on. But if you wanted to write all the papers at the same time (for some weird reason), you would have to write one word at a time for each paper. First, you write one word on the first paper, then write one on the next paper, and lastly write one on the third paper. You are not really writing them at the same time, but if you do it really fast (I mean really fast), it seems like you actually are writing them simultaneously, because the words seem to appear on the papers at the same time. This is exactly the same for multithreading. It is all about speed.

CreateThread (Win32)
HANDLE CreateThread(LPSECURITY_ATTRIBUTES lpThreadAttributes, DWORD dwStackSize, LPTHREAD_START_ROUTINE lpStartAddress, LPVOID lpParameter, DWORD dwCreationFlags, LPWORD lpThreadId);

The CreateThread function creates a new thread on Windows. The first parameter (LPSECURITY_ATTRIBUTES lpThreadAttributes) is a pointer to a security attributes structure. In this book, we always set this to NULL. The second parameter (DWORD dwStackSize) defines the stack size for the thread. If this is set to 0, the default value is used. This value is the size of the calling threads stack. The third parameter (LPTHREAD_START_ROUTINE lpStartAddress) specifies the threads routine function. The fourth parameter (LPVOID lpParameter) is a pointer to the parameter that will be passed to the threads routine function. The fifth parameter (DWORD dwCreationFlags) defines the flags for how to create the thread. We can only set this to 0 or CREATE_SUSPENDED. If the latter option is used, the thread starts in suspended mode and will not start before the ResumeThread function is called. The sixth parameter (LPWORD lpThreadId) is a pointer to a variable that will receive the thread identifier.

Chapter 6 / I/O Operations

169

Win32 return values: n n Success: Handle to the thread Failure: NULL

pthread_create (Unix)
pthread_create(pthread_t *tid, const pthread_attr_t *attr, void *(*func) (void *), void *arg);

The pthread_create function creates a new thread on Unix. This is similar to Windows CreateThread function. The first parameter (pthread_t *tid) is a pointer to the variable that will receive the thread identifier. The second parameter (const pthread_attr_t *attr) is a pointer to the thread attributes variable. Usually we set this parameter to NULL to use the defaults. The third parameter (void *(*func)(void *)) defines the thread routine function. This is similar to the Windows CreateThread functions lpStartAddress parameter. The last parameter (void *arg) specifies the parameter to pass to the thread routine function. This is similar to the Windows CreateThread functions lpParameter parameter. Unix return values: n Success: 0 n Failure: Non-zero

I/O Strategy
Each network application has its own input/output strategy for how data flow is controlled within the application. For example, will the server open multiple sockets for the clients (UDP) or will it use multithreading? Lets take a look at some of the possible strategies: n Blocking I/O n Non-blocking I/O n Signal-driven I/O n Multiplexing I/O

Blocking I/O
Blocking I/O is the simplest form of I/O strategies. As we have already learned, a socket can be blocking or non-blocking. A blocking socket means that some of the socket functions we run, such as passing a blocking socket as a parameter, wait for the action to be accomplished

170

Chapter 6 / I/O Operations

before they return. For example, if we try to read data off a blocking socket, the read function we use will not return before the data is read. If there is no one sending anything to that socket, the function will block until there is someone sending data to it. Each socket is blocking by default.
Figure 6-1 Blocking I/O

Non-blocking I/O
We can set a socket to non-blocking if we want to. Then the socket functions will return even if the action cannot be accomplished immediately. Usually this means that we need to loop the function to create our own blocking effect. This is called polling. If we do not poll the incoming data, we would most likely miss it because the data must have reached the local host before we call the function to read the data.
Figure 6-2 Non-blocking I/O

Signal-driven I/O
We do not always have to use the data reading function to monitor the socket for incoming data. We can set up our operating system to signal us when incoming data is available for us to read. Then all we have to do is call the function to read the data from the socket. The obvious advantage in this strategy compared to the two previous ones is that we

Chapter 6 / I/O Operations

171

can do other things while we wait for the data to come in. The application that waits for data input can run the rest of the application when there is no data to be read. Then when data input exists, everything else is stopped for the time it takes to read the data and possibly process it.
Figure 6-3 Signal-driven I/O

Multiplexing I/O
Another way to avoid using the actual data reading function to tell if there is incoming data is to use the select function. It is used similarly to a blocking socket read function call, as the select function will stop and wait for the data to come in. Multiplexing means that we have multiple sources, but we use only one at a time. We need to be able to choose the source that has something to offer us. We use the select function for that. This is the advantage of multiplexing I/O; we can wait for more than one socket to have data to read with one function call.
Figure 6-4 Multiplexing I/O

By combining these I/O strategies with multithreading, we can unleash all the power of our network applications.

172

Chapter 6 / I/O Operations

I/O Control
Now that we know some of the possible I/O strategies, we need to know how to get our sockets to work using the strategy we choose. Some do not need any extra setting up, but some do. Also, we can create some nice features for our network application by controlling the I/O mode of our sockets. To control the input/output mode of our sockets, we use the following two functions: n ioctl/ioctlsocket n setsockopt

ioctl (Unix)/ioctlsocket (Win32)

int ioctl(int fd, int request, /* void *arg */); int ioctlsocket(SOCKET s, long cmd, u_long FAR *argp);

These two functions work alike; ioctl is for Unix and ioctlsocket is for Windows. The functions are used to control the I/O mode of the socket. The most common use for these functions is setting the blocking/non-blocking mode of a socket. The first parameter (int fd/SOCKET s) defines the socket to control. The second parameter (int request/long cmd) is the command to give the socket. This is used with the third parameter (void *arg/u_long FAR *argp), which tells the function how to give out the command. The Windows version (ioctlsocket) does not have all the commands that the Unix version (ioctl) has. The only command we really use within this book is FIONBIO, which sets/clears the non-blocking flag of the socket. Win32 return values: n Success: 0 n Failure: SOCKET_ERROR Unix return values: n Success: 0 n Failure: 1 An example best explains the usage of this function. To set a socket into non-blocking mode, we call the I/O control function like this:
u_long on = 1; ioctl(mysocket, FIONBIO, &on); ioctlsocket(mysocket, FIONBIO, &on); // Unix // Windows

Chapter 6 / I/O Operations

173

To set the non-blocking mode off:

u_long off = 0; ioctl(mysocket, FIONBIO, &off); ioctlsocket(mysocket, FIONBIO, &off); // Unix // Windows

setsockopt/getsockopt (Unix, Win32)

int setsockopt(int sockfd, int level, int optname, const void *optval, socklen_t optlen); int getsockopt(int sockfd, int level, int optname, void *optval, socklen_t *optlen);

These functions are used to set and get the socket options. There are numerous socket options available, but we discuss only a few in this book. The first parameter of both functions (int sockfd) defines the socket to set the option for. The second parameter of both functions (int level) specifies the level at which the options are defined. Possible values (for compatibility issues) are SOL_SOCKET and IPPROTO_TCP . The third parameter is also the same for both functions (int optname); it defines the actual option to set. The fourth parameter of setsockopt (const void *optval) is a pointer to the buffer where the value for the option is stored. The fourth parameter of getsockopt (void *optval) is a pointer to the buffer where the value of the option will be stored. The fifth parameter of both functions (socklen_t optlen/ socklen_t *optlen) defines the size of the buffer used in parameter four. In this book we use only the SOL_SOCKET level, as the other levels are beyond the scope of this book. Here are some options from the SOL_SOCKET level that we should be aware of: n SO_BROADCAST Set/get broadcasting on/off n n n n SO_LINGER Set/get lingering on/off SO_RCVBUF Set/get receive buffer size SO_SNDBUF Set/get send buffer size SO_TYPE Get socket type

Win32 return values: n Success: 0 n Failure: SOCKET_ERROR

174

Chapter 6 / I/O Operations

Unix return values: n n Success: 0 Failure: 1

Lets take a look at an example of how to make our socket linger on a call to close/closesocket to make sure all the data is sent before the socket is closed:
struct linger Ling; Ling.l_onoff = 1; Ling.l_linger = 0; setsockopt(mysocket, SOL_SOCKET, SO_LINGER, (const char *) &Ling, sizeof(struct linger));

When we want to close the socket, we should put the close/ closesocket function call in a loop that loops as long as the function returns succesfully:
shutdown(mysocket, SD_BOTH); int ret = WSAEWOULDBLOCK; while(ret == WSAEWOULDBLOCK) ret = closesocket(mysocket); -shutdown(mysocket, SD_RDWR); int ret = WSAEWOULDBLOCK; while(ret == WSAEWOULDBLOCK) ret = close(mysocket); // Unix // Windows

// Windows

// Unix

shutdown (Unix, Win32)

int shutdown(int sockfd, int howto);

This function is used to disable sending and receiving on a socket. It does not close the socket, but depending on the parameters we pass to it, the socket will not be able to send or receive (or both) any more data. The first parameter (int sockfd) defines the socket to shut down. The second parameter (int howto) specifies how to shut down the socket. Possible values are listed in Table 6-1.

Chapter 6 / I/O Operations

175

Table 6-1: howto parameter values OS Unix Windows Shut Reading SHUT_RD SD_RECEIVE Shut Writing SHUT_WR SD_SEND Shut Both SHUT_RDWR SD_BOTH

Win32 return values: n n n n Success: 0 Failure: SOCKET_ERROR Success: 0 Failure: 1

Unix return values:

Broadcasting
Broadcasting is a very useful way of sending data. When you broadcast, you send to everybody on the network with only one call to the sending function. There are some restrictions however, which make broadcasting much less interesting than it first sounds. First of all, it can be used only with datagram sockets. This means that we need to use UDP if we want to broadcast. The second, and much more limiting, issue is that it can be used only on a local area network. We cannot broadcast messages through the Internet, because broadcasting works using a unique broadcast IP address that every LAN has. Usually it is of the form subnet.255. For example if our subnet is 192.168.0, the broadcast address is 192.168.0.255. We can also use a global broadcast address, 255.255.255.255, that works on all LANs.
NOTE IPv6 does not support broadcasting. IPv6 uses multicasting instead, but that is beyond the scope of this book.

Searching for Servers

One very good use for broadcasting is to search for available servers on a LAN. Anyone who has played a network game on a LAN knows that it is annoying to have to continually enter the IP address of a certain server to be able to connect to it. And if you do not know what servers there are to join, you have to find that out beforehand. Broadcasting is a solution for this. We assume that all the servers use the same port and that the client application (the game) knows this port number. Now all we need to do is broadcast to that port number and wait for the servers

176

Chapter 6 / I/O Operations

to send their info to us. All the servers that are up and running in the LAN will receive the broadcast message. They can find out the clients address from the datagram and send a reply to that address. The client will then build a list of servers that replied in a certain amount of time.

Broadcast Function
To make our socket broadcast, we need to set the socket option of it as shown in the following code. We also need to create a datagram socket and fill in the address information so that the data is sent to the broadcast address. In this example, we use the address 255.255.255.255, which works on all LANs as a broadcast address.
void Broadcast(char *buf, size_t count) { // Define on const int on = 1; // Create a datagram socket SOCKET sock; sock = socket(AF_INET, SOCK_DGRAM, 0); struct sockaddr_in servaddr; socklen_t len; // Use the example broadcast address u_long inetAddr = inet_addr("255.255.255.255"); // Fill address information structure memset(&servaddr, 0, sizeof(struct sockaddr_in)); servaddr.sin_family = AF_INET; servaddr.sin_port = htons(9009); servaddr.sin_addr.s_addr = inetAddr; len = sizeof(servaddr); // Set socket broadcasting option on setsockopt(sock, SOL_SOCKET, SO_BROADCAST, (const char *) &on, sizeof(on)); // Broadcast! sendto(sock, buf, count, 0, (struct sockaddr *) &servaddr, len); }

Now lets discuss the code. First, we must define the socket option switch in a variable because the setsockopt function always retrieves the switch from a variable.
// Define on const int on = 1;

Chapter 6 / I/O Operations

177

Then we create a normal datagram socket to be used in the broadcasting operation.

// Create a datagram socket SOCKET sock; sock = socket(AF_INET, SOCK_DGRAM, 0);

Next, the address information structure must be filled with the correct information. The IP address is set to the example broadcast address 255.255.255.255. The port number is set to 9009, assuming that is the port that all the recipients use.
// Use the example broadcast address u_long inetAddr = inet_addr("255.255.255.255"); // Fill address information structure memset(&servaddr, 0, sizeof(struct sockaddr_in)); servaddr.sin_family = AF_INET; servaddr.sin_port = htons(9009); servaddr.sin_addr.s_addr = inetAddr;

And finally, the socket option SO_BROADCAST is set on and we broadcast the message using the normal sendto function.
// Set socket broadcasting option on setsockopt(sock, SOL_SOCKET, SO_BROADCAST, (const char *) &on, sizeof(on)); // Broadcast! sendto(sock, buf, count, 0, (struct sockaddr *) &servaddr, len);

Summary
In this chapter we learned what I/O strategies we can use in our network applications and how to implement them. We also learned the concept of multithreading and the reasons to use it. Along with the I/O methods, we learned a new way to send data efficiently in a LAN by broadcasting. We are now ready to create our own network library.

This page intentionally left blank.

Part II

Tutorials
The tutorial section leads you step by step through how to create a working online game. Since we wish to focus on the network programming side, we have supplied you with a simple OpenGL-based 2D graphics library. The first tutorial consists of an introduction to our graphics library. It explains the basics of the library and also gives you code examples to experiment with. Then we move on to developing our network library, which is reusable so that it can be used in your own online titles as well as in our sample online game. Once we have the foundations, we then move on to how to implement a login and lobby system that is ready for implementing our online game. In the final tutorial, we develop our final working online game that you can both learn from and expand upon. In creating these tutorials, we have aimed to give you a practical hands-on approach to learning how to create an online game. We have found that other books tell you more about how things work rather than explain how to actually get it working so you can experiment with it. Remember that all the source code for the tutorials can be found on the companion CD.

179

This page intentionally left blank.

Tutorial 1

Using 2DLIB
Introduction
In this first tutorial, we cover the use of our simple 2D OpenGL library that we will be using in the following tutorials to create our online network game. The reason we are covering this is to keep confusing graphics code from getting in the way of the core network code, making the network code easier for us to understand. You will quickly be able to understand and use our 2D library, but if you prefer, you can use your own DirectX or OpenGL routines.

Configuring Visual Studio

The OpenGL library and our 2D library are located on the companion CD. First, you must add the 2D librarys API directory into the Visual Studio environment. To do this select Tools, Options from the main Visual Studio menu. Now select the Directories tab on the dialog box that is visible in the center of the screen.

181

182

Tutorial 1 / Using 2DLIB

Figure 1

Next, add the librarys API directory to both the Include files and Library files. This will allow Visual Studio to find the 2D library when you try to compile a program that uses it. Finally, you need to add the OpenGL library into Visual Studio. You do this in the same way except you need to add the Include folder into the Include directories and the Library folder into the Library directories. If you would like additional information on the creation and use of static link libraries, see Chapter 1.

Creating a Skeleton Project

It is generally a good idea to create basic code that never changes so you can reuse it in future programs and projects. Here we create a simple windowed program using the 2D graphics library that will create a window on the screen. This skeleton can be developed further into any Windows application, as you will see throughout this online game tutorial. First, we must create a workspace for our project. This entails creating a project, adding our 2D library and OpenGL into the workspace, and finally adding a main source file to code in.

Creating the Workspace

To create our workspace, select File, New from the main Visual Studio menu. Next, click the Projects tab and select Win32 Application (not a console application). Set where you wish the project to be created (i.e., c:\projects\) and give the project a name (i.e., OnlineGame). Once you click OK, the next window will ask which type of application you would like to create. For this project, leave it as an empty project and

Tutorial 1 / Using 2DLIB

183

click Finish. Simply click OK on the next screen, as it is simply a summary of what you have just done. Now that we have our workspace created, we need to add the 2D library into the workspace so it is compiled with our code (so the functions are available to us).

Adding the Static Libraries

To add the static libraries, select Project, Settings from the main Visual Studio menu. On the dialog box that appears, change the drop-down box at the top left from Win32 Debug to All Configurations. This means that when we add in the library it will be included in both the debug and release versions of our game code. Next select the Link tab on the right of the dialog and add the following before kernel32.lib:
opengl32.lib glu32.lib glaux.lib 2dlib.lib

This will add the required OpenGL libraries and our 2D library into the workspace when we compile the program.
Figure 2

Adding the Source File

Finally, we need to add our main source file into the project. To do this, select File, New from the main Visual Studio menu and select the Files tab on the dialog that appears. Now select C++ Source File and enter main in the File name edit box. Click OK and the main source file will be added into our project.

184

Tutorial 1 / Using 2DLIB

Creating a Basic Windowed Application with 2DLIB

Our workspace is now ready to have code added. For the 2D library to function, we need to create a Windows message loop and a callback function for the window, which is passed into the 2D library as a parameter. Now we will cover the main parts of the main program. The complete code can be found at the end of this section.

The WinMain Function

The following code segment is the main Windows message loop, but it also contains our initialization code and main drawing loop.
int WINAPI WinMain(HINSTANCE hThisInst, HINSTANCE hPrevInst, LPSTR lpszArgs, int nWinMode) { MSG msg; GFX_Init("App Title", 640, 480, 16, 0, WndProc); while(msg.message != WM_QUIT) { if(PeekMessage(&msg, NULL, 0, 0, PM_REMOVE)) { TranslateMessage(&msg); DispatchMessage(&msg); } else { GFX_Begin(); { } GFX_End(); } } GFX_Shutdown(); return msg.wParam; }

The GFX_Init() function sets up our window for use by the 2D graphics library. The first parameter is the title of the window, although this is not relevant if you wish to create a full-screen application. The next two parameters determine the width and height of the window in pixels. The fourth specifies the color depth you wish to use; usually this is set to 16 bits per pixel, but other values such as 24 or 32 can be used too. Next is a flag to determine whether you wish to run the

Tutorial 1 / Using 2DLIB

185

application full screen or not; set this to 0 for windowed or 1 for fullscreen mode. The final parameter is a pointer to the Windows procedure function that is explained in the next section.
GFX_Init("App Title", 640, 480, 16, 0, WndProc);

Next is the message loop for Windows; this controls all the Windows messages that are required to allow the program to run correctly in the Windows environment.
while(msg.message != WM_QUIT) { if(PeekMessage(&msg, NULL, 0, 0, PM_REMOVE)) { TranslateMessage(&msg); DispatchMessage(&msg); } else { GFX_Begin(); { } GFX_End(); } }

As you can see in the preceding code, the Windows messaging loop is a simple while loop that continues until a quit message is received. The PeekMessage function is a faster version of the GetMessage function and is the optimal way to get Windows messages. This is not important for us, but the else part of the if statement is, however. This is where we place all our 2D drawing commands; it is, if you like, our game loop. Once the 2D library is initialized using the GFX_Init() function, you must call GFX_Begin() before you start drawing and then GFX_End() once everything is drawn. The first function is used to clear the drawing buffer and ready it for the next frame of graphics to be drawn to it. The ending function is used to swap the buffer onto the visible screen so the user can see it without any shearing or other nasty graphical glitches. Finally, once the user quits the program, the graphics library must be shut down. To do this, we simply call GFX_Shutdown() after the while loop. This closes the graphics library and frees any memory that was allocated internally by the 2D graphics library.

The Windows Procedure

Every time our window receives a Windows message, such as a key being pressed on the keyboard, this function is called to process the message correctly. The code that follows is the entire Windows procedure.

186

Tutorial 1 / Using 2DLIB

LRESULT CALLBACK WndProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { switch (uMsg) { case WM_CLOSE: { PostQuitMessage(0); return 0; } case WM_KEYDOWN: { keys[wParam] = TRUE; return 0; } case WM_KEYUP: { keys[wParam] = FALSE; return 0; } case WM_SIZE: { GFX_Resize(LOWORD(lParam),HIWORD(lParam)); return 0; } } // Pass all unhandled messages to DefWindowProc return DefWindowProc(hWnd,uMsg,wParam,lParam); }

In basic terms, the function is simply a switch statement that reacts correctly to different events (cases) in Windows. If the function has no handling routine for an event, it simply passes it back to Windows to deal with accordingly. It is possible to add other event handles to this procedure, but the four we have included in this skeleton application are all that are required for now. Lets take a look at the events in a little more detail.

The WM_CLOSE Event

case WM_CLOSE: { PostQuitMessage(0); return 0; }

This event is triggered when the user clicks on the X button at the top right of a window or when a close message is manually sent to the window. When a WM_CLOSE message is sent, this routine sends a WM_ QUIT message that, if you remember from earlier, is the condition for

Tutorial 1 / Using 2DLIB

187

our main while loop (i.e., when a WM_QUIT message is received, our program will shut down and exit).

The WM_KEYDOWN Event

case WM_KEYDOWN: { keys[wParam] = TRUE; return 0; }

In the 2D library, we have defined an array that allows for easy use of the keyboard without having to learn DirectInput or another similar input library. When a key has been pressed on the keyboard, this routine sets the correct key in the array to TRUE, meaning that the key is currently down. Later in this tutorial we will cover how to use the keyboard for input.

The WM_KEYUP Event

case WM_KEYUP: { keys[wParam] = FALSE; return 0; }

This routine works in the same way as the WM_KEYDOWN routine except that it handles the event of a key being released on the keyboard. When the key has been released, it sets the correct value in the keys array to FALSE, meaning the key is not being pressed.

The WM_SIZE Event

case WM_SIZE: { GFX_Resize(LOWORD(lParam),HIWORD(lParam)); return 0; }

This event handles the resizing of a windowed mode application. It is not relevant for full-screen applications, but it does no harm to leave it in. It tells the 2D library the new width and height of the window so that it can react accordingly. (The width is the low word of the lParam and the height is the high word of the lParam.)

The Complete Code

The final part we require is what files to include. We need to include the Windows header file (since we are creating a Windows application), the three OpenGL header files for the 2D library, and the 2D librarys header files. Therefore, our complete code listing for the skeleton application resembles the following:

188

Tutorial 1 / Using 2DLIB

#include #include #include #include #include

<windows.h> <gl/gl.h> <gl/glu.h> <gl/glaux.h> <2dlib.h>

// WINDOWS PROCEDURE LRESULT CALLBACK WndProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { switch (uMsg) { case WM_CLOSE: { PostQuitMessage(0); return 0; } case WM_KEYDOWN: { keys[wParam] = TRUE; return 0; } case WM_KEYUP: { keys[wParam] = FALSE; return 0; } case WM_SIZE: { GFX_Resize(LOWORD(lParam),HIWORD(lParam)); return 0; } } // Pass all unhandled messages to DefWindowProc return DefWindowProc(hWnd,uMsg,wParam,lParam); }

// WINDOWS MESSAGE LOOP AND APPLICATION ENTRY POINT int WINAPI WinMain(HINSTANCE hThisInst, HINSTANCE hPrevInst, LPSTR lpszArgs, int nWinMode) { MSG msg; GFX_Init("Skeleton App", 640, 480, 16, 0, WndProc); while(msg.message != WM_QUIT) {

Tutorial 1 / Using 2DLIB

189

if(PeekMessage(&msg, NULL, 0, 0, PM_REMOVE)) { TranslateMessage(&msg); DispatchMessage(&msg); } else { GFX_Begin(); { } GFX_End(); } } GFX_Shutdown(); return msg.wParam; }

If everything has gone according to plan, when you compile and execute the preceding code you should be able to see the following screen:
Figure 3

Not very exciting, is it? However, it is very useful as we now have the foundation for our 2D graphics engine, meaning we can easily use the 2D library to create primitives and bitmapped graphics on the screen.

190

Tutorial 1 / Using 2DLIB

Using the 2DLIB Graphics Routines

Now we will cover all the graphical functions in the 2D library with an explanation of how to use them. The final sections of this tutorial demonstrate a couple of examples of how to use these functions with the skeleton application. Remember that all graphics functions must be placed between the GFX_Begin() and GFX_End() functions or the graphics will not be visible on the screen.

2D Positions on the Screen

For the x coordinates, the values start from 0 at the left-hand side of the application and end at the width of your application minus 1. The y coordinates start from 0 at the top of your application and end at the height of your application minus 1. Figure 4 displays what the screen coordinates would be like for a 640 x 480 application in the format (x, y).
Figure 4

Use of Colors
Each primitive drawing function in the library has three parameters at the end that specify the red, green, and blue values for what you wish to draw. These values range as integers from 0 to 255, where 0 is black and 255 is full brightness. Therefore, if you wanted your color to be bright red, you would use 255 red, 0 green, and 0 blue.

Tutorial 1 / Using 2DLIB

191

Plotting a Single Pixel

void GFX_Pixel(int x, int y, int r, int g, int b);

The GFX_Pixel() function will display a single pixel on the screen at the specified (x, y) position. The position is specified by the first two parameters of the function, the first being x and the second being the y position. The final three parameters determine the color of the pixel in RGB format. (See the information about colors in the Use of Colors section.)

Drawing a Line
void GFX_Line(int x1, int y1, int x2, int y2, int r, int g, int b);

The GFX_Line() function draws a line between two points. The first two parameters are the (x, y) position of the first point. The next two parameters are the (x, y) position of the point to connect the first point to. The final three are to specify the color in RGB format.

Drawing a Rectangle/Filled Rectangle

void GFX_Rect(int x1, int y1, int x2, int y2, int r, int g, int b); void GFX_RectFill(int x1, int y1, int x2, int y2, int r, int g, int b);

The GFX_Rect() and GFX_RectFill() functions draw a rectangle based on the two points you specify. The first two parameters determine the top-left corner for the rectangle and the next two specify the bottom-right corner for the rectangle. Again, the final three determine the color in RGB format. The first function draws an outlined rectangle in the specified color, whereas the second draws a filled rectangle in the specified color.

Drawing a Triangle/Filled Triangle

void GFX_Tri(int x1, int y1, int x2, int y2, int x3, int y3, int r, int g, int b); void GFX_TriFill(int x1, int y1, int x2, int y2, int x3, int y3, int r, int g, int b);

The GFX_Tri() and GFX_TriFill() functions allow you to draw triangles and filled triangles on the screen. The first six parameters determine the three two-dimensional coordinates that are required to construct a triangle. The final three then determine the red, green, and blue values the same way as the other primitive functions.

192

Tutorial 1 / Using 2DLIB

Graphic Loading Functions

Our 2D library has support for Windows bitmaps and 24-bit and 32-bit uncompressed Truevision Targa format. First, we need to declare a variable that will hold our graphic data. You do this as follows:
GFX_IMAGE2D my_image;

This creates a variable called my_image that you can use with the graphics loading commands in the library to load pictures from your hard drive in the application. Next, you want to actually load an image. Lets say we have a Windows bitmap image called brick.bmp. We would call the following function to load it into our my_image variable. Notice you pass a pointer to the GFX_IMAGE2D variable, followed by the filename of the graphic you wish to load.
GFX_LoadBitmap(&my_image, brick.bmp);

The Truevision Targa loading function works in the same manner. If you have an image called phone.tga, you would load it like this:
GFX_LoadTarga(&my_image, phone.tga);

That is all there is to loading graphics. Here are the prototypes for the graphics loading functions for your reference:
void GFX_LoadBitmap(GFX_IMAGE2D *pImage, char *filename); void GFX_LoadTarga(GFX_IMAGE2D *pImage, char *filename);

Graphics Display (Blitting) Function

Once you have successfully loaded in your graphics, you use the following function to display them on the screen:
void GFX_Blit(GFX_IMAGE2D *pImage, int x, int y, int w, int h, float rotate);

The first parameter is a pointer to a GFX_IMAGE2D. You simply pass a pointer to a graphic you have loaded using one of the loading functions explained previously. Next you specify the (x, y) position using the second and third parameters of the function. The function also handles image scaling so you can state the width and height at which the image is to be drawn on the screen. These are set in the forth and fifth parameters, respectively. Finally, the function handles image rotation in degrees. Therefore, you can specify a floating-point value between 0.0 and 360.0 for the image to be drawn at (the image rotates around its center point).

Tutorial 1 / Using 2DLIB

193

For example, if you wanted to display an image you loaded into a my_image variable at point (25, 50) with a width of 100, a height of 150, and no rotation, you would call the following function:
GFX_Blit(&my_image, 25, 50, 100, 150, 0.0);

If you then decided you wanted it in the same position and with the same dimensions but at an angle of 30 degrees, the function would be as follows:
GFX_Blit(&my_image, 25, 50, 100, 150, 30.0);

That is all there is to the graphical side of the library. As you can see, there is nothing very complicated in it, and it simplifies many of the initialization procedures and other such things.

Keyboard Input Method

This is a very simple method of getting key input. We would really recommend using a library such as DirectInput, but this makes the process simple and it does what we require it to do for now. To check a key on the keyboard, simply place an if statement after the GFX_End() function as follows:
if(keys[VkKeyScan('a')]) { // place what is to be done here }

The preceding if statement would check if the a key had been pressed. The a can be replaced with any other character, symbol, or number from the keyboard. For example, if you wished to check if the ] key had been pressed you would use this statement:
if(keys[VkKeyScan(']')]) { // place what is to be done here }

There are special cases such as the function keys (F1, F2, etc.) and other keys like the Spacebar and Return. Table 1 lists most of these special cases and an if statement follows that shows how to use a special key.

194

Tutorial 1 / Using 2DLIB

Table 1: Special keys Statement VK_F1 VK_F2 VK_F3 VK_F4 VK_F5 VK_F6 VK_F7 VK_F8 VK_F9 VK_F10 VK_F11 VK_F12 VK_INSERT VK_HOME VK_PAGE_UP VK_DELETE VK_END VK_PAGE_DOWN VK_UP VK_DOWN VK_LEFT VK_RIGHT VK_ESCAPE VK_SPACE VK_CONTROL VK_ALT VK_ADD VK_SUBTRACT VK_MULTIPLY VK_DIVIDE Description Function key F1 Function key F2 Function key F3 Function key F4 Function key F5 Function key F6 Function key F7 Function key F8 Function key F9 Function key F10 Function key F11 Function key F12 Insert key Home key Page Up key Delete key End key Page Down key Arrow key up Arrow key down Arrow key left Arrow key right Escape key Spacebar Control key Alt key Numeric keypad + key Numeric keypad key Numeric keypad * key Numeric keypad / key

Tutorial 1 / Using 2DLIB

195

Statement VK_EQUALS VK_TAB

Description = key Tab key

Unlike the other if statements where you use a function (VkKeyScan) to get the correct value for the character, you simply place the VK_ value as follows:
if(keys[VK_SPACE]) { // place what is to be done here }

The preceding example would check to see if the Spacebar had been pressed; if so, it would execute the code within the if statement.

2DLIB Example 1 Moving Primitives with the Cursor Keys

In this first example application, we create a program that allows the user to move a filled rectangle around the screen using the cursor keys, but not outside the borders of the application. In addition, the user will be able to select whether the rectangle is red, green, or blue by pressing r, g, or b, respectively. The complete code listing for this example is at the end of this section. First, we must set up a project file for this application. This is done in exactly the same manner as the skeleton application that was explained earlier in this tutorial. Once we have the skeleton application running, the first thing we must do is create variables to store the position, dimensions, and color of the rectangle. We declare these globally as follows:
int rect_x, rect_y; int rect_w, rect_h; float rect_r; float rect_g; float rect_b; // (x, y) position // width and height // red color // green color // blue color

Next, after the initialization code for 2DLIB, we want to set these variables for the rectangle. This is done as follows:
rect_x rect_y rect_w rect_h = = = = 10; 10; 100; 100; // // // // x position is 10 y position is 10 width is 100 pixels height is 100 pixels

196

Tutorial 1 / Using 2DLIB

rect_r = 255.0; rect_g = 0.0; rect_b = 0.0;

// red is full intensity // no blue // no green

Now, when we draw our rectangle it will appear at position (10, 10) and be 100 pixels in both height and width. In addition, we have set the red value to the maximum and the green and blue to the minimum, making the rectangle appear a bright red color. In the drawing loop, we will then place the following function to draw it:
GFX_RectFill(rect_x, rect_y, rect_x+rect_w, rect_y+rect_h, rect_r, rect_g, rect_b);

This will draw the filled rectangle at the position we specified, using the values from the variables we previously declared. For more information on this function, see the Using the 2DLIB Graphics Routines section. Finally, we wish to be able to move the rectangle around the screen, so we need to have the correct keyboard code to do this. Here is what the movement code should look like:
if(keys[VK_UP]) { if(rect_y > 0) rect_y--; } if(keys[VK_DOWN]) { if(rect_y+rect_h < 480) rect_y++; } if(keys[VK_RIGHT]) { if(rect_x+rect_w < 640) rect_x++; } if(keys[VK_LEFT]) { if(rect_x > 0) rect_x--; }

This simply adjusts the variables for the rectangle based on which arrow key the user presses on the keyboard. The colors work in a similar fashion, as shown below:
if(keys[VkKeyScan('r')]) { rect_r = 255.0; rect_g = 0.0; rect_b = 0.0; } if(keys[VkKeyScan('g')])

Tutorial 1 / Using 2DLIB

197

{ rect_r = 0.0; rect_g = 255.0; rect_b = 0.0; } if(keys[VkKeyScan('b')]) { rect_r = 0.0; rect_g = 0.0; rect_b = 255.0; }

This code checks whether r, g, or b has been pressed, and if so, changes the rectangle variables to set the specified color. That covers everything for our first example. When you compile the following code, you should be able to move the rectangle around with the arrow keys and change the color with the r, g, and b keys.

Complete Code Listing for Example 1

#include #include #include #include #include <windows.h> <gl/gl.h> <gl/glu.h> <gl/glaux.h> <2dlib.h>

// APPLICATION SPECIFIC int rect_x, rect_y; int rect_w, rect_h; float rect_r; float rect_g; float rect_b; // (x, y) position // width and height // red color // green color // blue color

198

Tutorial 1 / Using 2DLIB

case WM_KEYUP: { keys[wParam] = FALSE; return 0; } case WM_SIZE: { GFX_Resize(LOWORD(lParam),HIWORD(lParam)); return 0; } } // Pass all unhandled messages to DefWindowProc return DefWindowProc(hWnd,uMsg,wParam,lParam); }

// WINDOWS MESSAGE LOOP AND APPLICATION ENTRY POINT int WINAPI WinMain(HINSTANCE hThisInst, HINSTANCE hPrevInst, LPSTR lpszArgs, int nWinMode) { MSG msg; // Initialize 2DLIB GFX_Init("2DLIB - Example 1", 640, 480, 16, 0, WndProc); // Set rect_x rect_y rect_w rect_h up the rectangle = 10; // x position is 10 = 10; // y position is 10 = 100; // width is 100 pixels = 100; // height is 100 pixels // red is full intensity // no blue // no green

rect_r = 255.0; rect_g = 0.0; rect_b = 0.0;

// Start the Windows message loop while(msg.message != WM_QUIT) { if(PeekMessage(&msg, NULL, 0, 0, PM_REMOVE)) { TranslateMessage(&msg); DispatchMessage(&msg); } else { GFX_Begin(); { // Draw the rectangle

Tutorial 1 / Using 2DLIB

199

GFX_RectFill(rect_x, rect_y, rect_x+rect_w, rect_y+rect_h, rect_r, rect_g, rect_b); } GFX_End(); } // Check for keyboard input // -> Movement if(keys[VK_UP]) { if(rect_y > 0) rect_y--; } if(keys[VK_DOWN]) { if(rect_y+rect_h < 480) rect_y++; } if(keys[VK_RIGHT]) { if(rect_x+rect_w < 640) rect_x++; } if(keys[VK_LEFT]) { if(rect_x > 0) rect_x--; } // -> Colors if(keys[VkKeyScan('r')]) { rect_r = 255.0; rect_g = 0.0; rect_b = 0.0; } if(keys[VkKeyScan('g')]) { rect_r = 0.0; rect_g = 255.0; rect_b = 0.0; } if(keys[VkKeyScan('b')]) { rect_r = 0.0; rect_g = 0.0; rect_b = 255.0; } } GFX_Shutdown(); return msg.wParam; }

200

Tutorial 1 / Using 2DLIB

2DLIB Example 2 Loading and Rotating Graphics

In this example we load a single bitmap file, then display it twice on the screen and make the two copies rotate in opposite directions. The first thing we must do is create a skeleton project as described earlier in this tutorial. Once we have this in place, we can begin by creating one variable that will hold the image and another to hold the current angle in degrees.
GFX_IMAGE2D cdrom; float rotation;

The preceding declarations are global. The first is used to store the image in a variable called cdrom, and the second is used to store the angle in degrees at which the image is to be rotated. To spin the second image in the opposite direction, we simply subtract the rotation value from 360 degrees. Next, we need to load our graphic file into the application. In this case, the file is an image of a cdrom and is called cdrom.bmp. Therefore, to load it we would want to call the GFX_LoadBitmap() function as follows:
GFX_LoadBitmap(&cdrom, "cdrom.bmp");

This would load the cdrom.bmp into the cdrom global variable. This is done after we initialize the 2D library. Once this is done we can display the image on the screen using the following technique:
GFX_Blit(&cdrom, 32, 100, 256, 256, rotation); GFX_Blit(&cdrom, 352, 100, 256, 256, 360-rotation);

These two functions are placed between the GFX_Begin() and GFX_End() functions as they display the image onto the screen. The first image is displayed at (32, 100) and the second is displayed at (352, 100). Both images are 256 x 256 pixels. Additionally, the first is rotated clockwise and the latter is rotated counterclockwise. The only thing left to do now is adjust the rotation variable every time through the loop. This is simply done as follows:
rotation++; if(rotation>360) { rotation = 360; }

When the rotation value becomes greater than 360 degrees, 360 is subtracted from the value.

Tutorial 1 / Using 2DLIB

201

That completes the second example. When you compile and execute the following code listing, you should see two images of CD-ROMs displayed on the screen spinning in opposite directions.

Complete Code Listing for Example 2

#include #include #include #include #include <windows.h> <gl/gl.h> <gl/glu.h> <gl/glaux.h> <2dlib.h>

// APPLICATION SPECIFIC GFX_IMAGE2D cdrom; float rotation; // variable to hold 'cdrom' graphic // variable to hold current rotation value

202

Tutorial 1 / Using 2DLIB

// WINDOWS MESSAGE LOOP AND APPLICATION ENTRY POINT int WINAPI WinMain(HINSTANCE hThisInst, HINSTANCE hPrevInst, LPSTR lpszArgs, int nWinMode) { MSG msg; // Initialize 2DLIB GFX_Init("2DLIB - Example 1", 640, 480, 16, 0, WndProc); // Load in the image GFX_LoadBitmap(&cdrom, "cdrom.bmp"); // Start the Windows message loop while(msg.message != WM_QUIT) { if(PeekMessage(&msg, NULL, 0, 0, PM_REMOVE)) { TranslateMessage(&msg); DispatchMessage(&msg); } else { GFX_Begin(); { GFX_Blit(&cdrom, 32, 100, 256, 256, rotation); GFX_Blit(&cdrom, 352, 100, 256, 256, 360-rotation); } GFX_End(); rotation++; if(rotation>360) { rotation = 360; } } // Check for keyboard input } GFX_Shutdown(); return msg.wParam; }

Tutorial 1 / Using 2DLIB

203

Summary
In this tutorial, you learned how to create a simple 2D graphics application using our 2D OpenGL library. This knowledge is sufficient for you to understand the network game programming tutorials that are to follow, although we would recommend learning more about OpenGL and Direct3D, as they are very useful pieces of knowledge. On a final note, it is also possible to add OpenGL commands directly. The library creates an OpenGL window with a two-dimensional orthographic view; therefore you could, for example, change it to a perspective view and create a 3D application, but this is beyond the scope of these tutorials. The network code, however, is not tied to any one graphics library, so you can implement any style of game using the same network library whether it is 2D or 3D.

This page intentionally left blank.

Tutorial 2

Creating Your Network Library

Introduction
In this tutorial we learn how to create the most essential part of our tutorial game: the network library. We learn how to send and receive data, how to ping hosts, and so on. We make a library of all the network code in order to gain access to it in every application we want. In the case of our tutorial game, we use the library in the game itself (client software) and in the server. Also, if we want to make any extra servers with functionality different from the normal game server, we would most likely want to use the same library for that too. We create the library so that it can be used under both Windows and Unix (Linux). The basic idea is that the game (client) is run on Windows and the server on Unix. But because of our platform-independent code, we do not have to worry about the network code if we want to change platforms. We are going to create our network library using only UDP protocol, as it is more suitable for computer games than TCP . You could use both, but to make things simpler, we only concentrate on UDP . You may wonder if this is what other games on the market do. The anwer is yes; all the games you play on the Internet use UDP protocol. Well, most of them anyway.

205

206

Tutorial 2 / Creating Your Network Library

The name for the library we are creating here is dreamSock. The name comes from a game development group to which the authors of this book belonged. The group is no longer active, but dreamSock lives on!

Why Create a Network Library of Our Own?

One big question arises when you start to think about creating your own network library: Why? Why would anyone want to make one when there are perfectly good network libraries out there already? For example, why not use Microsofts DirectPlay? DirectPlay is an excellent library, and because it is designed solely for gaming, it is an excellent choice for a game programmer. But it has one major drawback: You do not have Unix connectivity with it. Now you may think that no one really needs that. Well, it is really up to you if you want Unix connectivity, but because Unix is an absolutely stable system, it is the best choice for the server platform. There are some examples of bad experiences stemming from not using any Unix operating system as the server platform in recent gaming history, so why repeat those mistakes? If you create your own network library using sockets, you can have full Unix connectivity if you want. Plus, you will have a library that fits your needs exactly. And you can always improve upon it.

Planning the Structure

Before doing anything else we should take a moment to think about the structure of the library. We need to know all the facts about the library before we start to code it. If we started coding the library without any planning beforehand, we would most likely need to modify the structure many times before it met our requirements. So lets look at some of the facts. Fact number one is that the library must be platform independent. This leads to the conclusion that it is best to use platform-independent functions as much as we can, and also divide the code into separate files to store platform-dependent code. We can easily add support for more platforms this way. We need one main source file that wraps up all the code and selects the correct functions to be run. Fact number two is that in a network connection, we have a server, the clients, and the connection between them (or in the case of UDP , only the packets we send between them). So it is a good idea to make

Tutorial 2 / Creating Your Network Library

207

separate C++ classes a class for the server, a class for the client, and a class for the messages between them. We also want an easy-to-use interface for the library so the C++ classes internally take care of everything for the end user. But it would be nice to have access to the socket functions too, so we will make global functions for them first. Because our game needs to have only Windows and Unix support, we create three source files and their header files: dreamSock.cpp, dreamSock.h, dreamSockGlobal.cpp, dreamWinSock.cpp, and dreamLinuxSock.cpp. We also create dreamSockLog.cpp and dreamSockLog.h for our logging system. Some people want to put each class into a separate file, and that is fine, but we will put all the classes into dreamSock.cpp and dreamSock.h. A network library may vary in size, depending on the functionality we want, so it is a good idea to make the code as modular as possible. Therefore, we use the C++ class as the base of our library, with each object being one connection on a client and one service on a server (with multiple connections). Other factors do not affect the structure of the library much.

Planning the Functionality

Knowing the basic functionality before starting to code is also a must. Of course, we could just write functions for sending and receiving data without anything extra, but what would that accomplish? Nothing. But we are going to do that anyway. Yes, it sounds stupid, but read on. We write some global functions before we write the actual library functionality, and then we use these global functions in the C++ class we are going to write. We want to create the library to make it as easy to use as possible. We do this by moving all the complex network technology from the end programmer to the background, so that the end programmer can merely point and tell the library what to do. The end programmer should not worry about binding address information to a socket; that is a job for the network library. The end programmer should not have to understand the complex technical functionality of the network code. And most of all, the network library must take care of sending the data to the correct location.

Identifying Hosts
Being able to tell one host from another is not really required when sending because whenever we send data, we send it to all the clients at once (or only to the server). We do send unique data to a host when it

208

Tutorial 2 / Creating Your Network Library

connects to the server, but we only reply to the address from which the connection request came. A client, on the other hand, can only send to a server. The clients that join a server are stored in a client list. This is a linked list, so when a client disconnects, the original order of the list is lost. Clients are stored in the order in which they join the server. It is possible to send to a certain client by browsing that list and comparing the addresses to see which one is the one we want, but normally there is no need for that. But we do need to identify a host when we receive data from it. How else would we know what client to process? When we receive data, we also receive the address it came from, and we simply compare this address to the ones we have stored.

Sending Data to Hosts

Sending data to hosts is not as straightforward as it may seem at first. This is because we need to decide if we want to let the clients send data between each other directly without having the server as a router of any kind. To be able to send data directly between clients, all the clients need to know the address information of all the other clients. If there are a lot of clients in the network, it would consume vast amounts of memory and time (as the address information must be sent via the network). Also, the server cannot process data that is sent between two clients. This is why we decided not to allow clients to send data between each other directly. In fact, we will not provide functionality to send data between clients indirectly either. That is the job for the end user if it is required. Normally, a client sends its information to the server, and the server then sends that information to every client (even to the one that originally sent it to the server). We will create a message sequence system that allows us to see if a packet is dropped. Each packet is given a sequence number, and the remote host knows what packet number to expect next. If those numbers do not match, one more packet was lost somewhere. We will continue processing the packets normally, but the number of dropped packets is noted. It is up to the network application whether it cares about dropped packets. What about multithreading, then? Do we need to open a socket for each client on the server? With UDP , the answer is no. All we need is one socket on the server side that takes care of sending and receiving data for every client. That is possible because we store the clients addresses in the client list, and we can then pick the address from there when we want to send data or when we need to identify a host. Do not

Tutorial 2 / Creating Your Network Library

209

worry about the server missing some packets even though it uses only one socket, as the operating system buffers all the incoming messages for us. So to summarize: A client sends data only to the server, and the server sends data to all clients at once. Each host opens only one socket.

Pinging Calculating Network Latency

Pinging is a useful way to see how fast the network connection is between two hosts. Pinging is very simple: Have you ever seen a submarine movie where a sub pings another sub to see if it is there (or how far away it is)? The sound is deflected from the other subs hull, and those on the first sub will hear the ping sound come back. Pinging in network programming is exactly the same. We send a ping message to the host we want, and then we listen for the ping to come back. Then we calculate the time difference between sending and receiving the ping, and there it is, network latency between those two hosts. Remember that this value is the time it takes for a client to send a message and receive a response, so it takes roughly half of that time for one message to reach its destination. We create pinging directly in our network library. Only the server calculates the values because we want to be able to tell all the clients pings to everyone.

Timing Out
Some clients may crash without being able to disconnect properly from the server, so we need a way to drop those clients from the server. One way to do that is to keep track of when the client sent us something. If the client did not send us anything in a certain amount of time, we assume it has crashed and should be dropped from the server. We implement this in our network library with the timeout value set to 30 seconds.

Building the Library

Now we will learn how to build the library file. We will actually learn to set up the compiler so that it builds a library file for us, which can then be linked to other applications.

210

Tutorial 2 / Creating Your Network Library

Windows
Building and compiling the library on Windows is very easy with Microsoft Visual C++ 6.0. Because we are building a library, there is no need (and no way) to link any extra libraries with it. All the linking of other libraries is done when we build the application that uses our library. But there is a way to make our library link other libraries automatically. Because this network library requires the WinSock 2 library on Windows, we need to link it with the application. We could link it normally among the other libraries in Microsoft Visual C++ 6.0 settings, as shown in the following figure.
Figure 1

But we can also add the following line in our network librarys header file to make the WinSock 2 library be linked automatically whenever our network library is linked.
#pragma comment (lib, "ws2_32.lib")

We must also define the constant WIN32, as shown in the following piece of code, to make the compiler compile the correct version of the source code.
#define Win32

Or we can add it to the preprocessor list as shown in the following figure.

Tutorial 2 / Creating Your Network Library

211

Figure 2

This is our own constant, so we must also know where and how to use it. It is used to define parts of code that belong to Windows only or Unix only. It is used as follows:
#ifdef Win32 // place Windows code here #else // place other platform code here (in our case: Unix/Linux code) #endif

Unix/Linux
Building the library on Unix is a little bit more complicated, as is pretty much everything on Unix systems. We need to use the following makefile and a command-line compiler. This should work on most Unix systems, but depending on your setup, you may have to change some things. Note also that C_ARGS is commented out, but you can uncomment it to see more warnings during compilation.
#C_ARGS = Wall CC = gcc all: dreamSock.o dreamSockLog.o dreamLinuxSock.o dreamSockGlobal.o ar q libdreamSock.a dreamSock.o dreamSockLog.o dreamLinuxSock.o dreamSockGlobal.o cp libdreamSock.a ../server/ cp dreamSock.h ../server/ cp dreamSockLog.h ../server/

clean: rm f *.o rm libdreamSock.a

212

Tutorial 2 / Creating Your Network Library

dreamSock.o: dreamSock.cpp $(CC) $(C_ARGS) c dreamSock.cpp dreamSockLog.o: dreamSockLog.cpp $(CC) $(C_ARGS) c dreamSockLog.cpp dreamLinuxSock.o: dreamLinuxSock.cpp $(CC) $(C_ARGS) c dreamLinuxSock.cpp dreamSockGlobal.o: dreamSockGlobal.cpp $(CC) $(C_ARGS) c dreamSockGlobal.cpp

Now all we need to do is run make on the command line and the library is built. The above makefile will copy the required header files and library file to a directory called server. That is where you should put your server code.
NOTE Make sure you have enough rights to make the library on the Unix machine you use.

Creating Independent Code

In this section we discuss how we can make our code platform independent. We use the same function calls, but the insides differ. To the end programmer who will be using the library, it looks as if the code is exactly the same, regardless of the platform we are using.

Creating Definitions for Data Types

To create the new definitions we must know which platform is used for the build. This is very easy: You can either add a definition for the current platform at the beginning of your code or use the preprocessor of the compiler. Both do the same thing really. If you want to use #define, just add this line when you are compiling under Windows:
#define Win32

You do not have to add a define for Unix builds because you can simply comment out the definition of WIN32. This kind of system assumes that we are going to use the code only on Windows and some other (unspecified) operating system. In our case, the other operating system is Unix (or Linux). We could provide definitions for all the operating systems we are going to support, but it is easier this way when we use only two known systems.

Tutorial 2 / Creating Your Network Library

213

Using the preprocessor is even easier. Just add the string Win32 in the Windows version makefile (or workspace settings) preprocessors list of definitions. Now you do not have to make any modifications to your code when you move from one platform to another. The makefile you use defines the platform for the code. But how do we check if the definition of WIN32 is there? Again, this is very simple.
#ifdef Win32 typedef int socklen_t; #else typedef int SOCKET; #endif // will be run if we are on Win32 // will be run if we are not on Win32

This small piece of code checks the platform used during the build. The compiler ignores the non-true statement, and it will not be added to the executable. Hence, it does not slow down the run-time process. Now we will create definitions for some data types, so we can use one data type identifier on all platforms. These are the types to define: n n n n socklen_t = int SOCKET = int TRUE = 1 FALSE = 0

Here is how we do that in our code:

// Define SOCKET data type for Unix (defined in WinSock for Win32) // And socklen_t for Win32 #ifdef Win32 typedef int socklen_t; #else typedef int SOCKET; #ifndef #define #endif #ifndef #define #endif #endif TRUE TRUE 1 FALSE FALSE 0

Log System
It is a good idea to keep a log of what is happening so we can see exactly what happened when. For example, if something does go wrong we do not have to shut down the server and start investigating. We can first check the log file to see if that helps.

214

Tutorial 2 / Creating Your Network Library

The log system goes into the dreamSockLog.cpp and dreamSockLog.h files. Lets first take a look at the header file:
#ifndef __DREAMSOCKLOG_H__ #define __DREAMSOCKLOG_H__ #ifdef Win32 class dreamConsole { public: dreamConsole(char *title); ~dreamConsole(); void println(char *string, int type, ...); }; #define CONSOLE_NOTIFY #define CONSOLE_ERROR #define CONSOLE_FATAL 0 1 2

// Function prototypes void StartLogConsole(void); #endif // Function prototypes void StartLog(void); void StopLog(void); void LogString(char *string, ...); #endif

As you see, the log systems header file is rather simple. You should notice that the dreamConsole class is defined only for Windows. This console class is used to display run-time log information. Unix systems do not need that because they are nothing but big consoles themselves.

StartLogConsole Function
The StartLogConsole function simply creates a dreamConsole object.
#ifdef Win32 void StartLogConsole(void) { console = new dreamConsole("dreamSock application"); } #endif

Tutorial 2 / Creating Your Network Library

215

dreamConsole Constructor
The constructor function for dreamConsole is listed here:
dreamConsole::dreamConsole(char *title) { AllocConsole(); SetConsoleTitle(title); }

dreamConsole Destructor
The destructor function for dreamConsole is as follows:
dreamConsole::~dreamConsole() { FreeConsole(); }

The constructor and destructor functions allocate and deallocate the Win32 console system.

println Function
The println function is used to print a line on the console screen.
void dreamConsole::println(char *string, int type, ...) { char buf[1024]; char buf2[1024]; va_list ap; va_start(ap, string); vsprintf(buf, string, ap); va_end(ap); sprintf(buf2, "-> %s\n", buf); HANDLE console = GetStdHandle(STD_OUTPUT_HANDLE); switch(type) { case 0: SetConsoleTextAttribute(console, FOREGROUND_GREEN | FOREGROUND_INTENSITY); break; case 1: SetConsoleTextAttribute(console, FOREGROUND_RED | FOREGROUND_INTENSITY); break; }

216

Tutorial 2 / Creating Your Network Library

WriteConsole(console, buf2, strlen(buf2), NULL, NULL); }

Here we first get a handle to a standard console output and then set the attributes for the text color and background. Finally the text is written on the console screen.
HANDLE console = GetStdHandle(STD_OUTPUT_HANDLE); switch(type) { case 0: SetConsoleTextAttribute(console, FOREGROUND_GREEN | FOREGROUND_INTENSITY); break; case 1: SetConsoleTextAttribute(console, FOREGROUND_RED | FOREGROUND_INTENSITY); break; } WriteConsole(console, buf2, strlen(buf2), NULL, NULL);

StartLog Function
StartLog is used to start the log system.
void StartLog(void) { time_t current = time(NULL); if((LogFile = fopen("dreamSock.log", "w")) != NULL) { fprintf(LogFile, "Log file started %s", ctime(&current)); fclose(LogFile); } if((LogFile = fopen("dreamSock.log", "a")) != NULL) { } }

The following piece of code is used to retrieve the current time. This information is written every time something is logged so we know when it happened.
time_t current = time(NULL);

We open the file for writing and write the current time at the beginning of the file. Then we close the log file and open it for adding so we can start adding text to it. We leave the file handle open.

Tutorial 2 / Creating Your Network Library

217

if((LogFile = fopen("dreamSock.log", "w")) != NULL) { fprintf(LogFile, "Log file started %s", ctime(&current)); fclose(LogFile); } if((LogFile = fopen("dreamSock.log", "a")) != NULL) { }

LogString Function
LogString is used to add a string to the log file.
void LogString(char *string, ...) { char buf[1024]; va_list ap; va_start(ap, string); vsprintf(buf, string, ap); va_end(ap); // Get current time and date time_t current = time(NULL); char timedate[64]; sprintf(timedate, ctime(&current)); // Remove linefeed from time/date string int i = 0; while(timedate[i] != '\n') { i++; } timedate[i] = '\0'; // Output log string #ifdef Win32 fprintf(LogFile, "%s: %s\n", timedate, buf); if(console) console->println(buf, 0); #else // Linux outputs to screen and to the open file when running as daemon printf("%s: %s\n", timedate, buf); #endif }

218

Tutorial 2 / Creating Your Network Library

First we will format the text so that we can add any variable values to it. The final string will be stored in string buf.
char buf[1024]; va_list ap; va_start(ap, string); vsprintf(buf, string, ap); va_end(ap);

Again we retrieve the current time to add a time stamp with each log string. We remove the almost annoying linefeed character from the end of the timedate string so we can write stuff on the same line as the time stamp.
time_t current = time(NULL); char timedate[64]; sprintf(timedate, ctime(&current)); // Remove linefeed from time/date string int i = 0; while(timedate[i] != '\n') { i++; } timedate[i] = '\0';

We write the time stamp and the string and follow it by a linefeed, so the user does not have to add the linefeed to the string every time.
// Output log string #ifdef Win32 fprintf(LogFile, "%s: %s\n", timedate, buf); if(console) console->println(buf, 0); #else // Linux outputs to screen and to the open file when running as daemon printf("%s: %s\n", timedate, buf); #endif

StopLog Function
After we are done logging, we need to stop the log system. That is done with the following StopLog function:
void StopLog(void) { fclose(LogFile); #ifdef Win32 if(console != NULL)

Tutorial 2 / Creating Your Network Library

219

delete console; #endif }

The log file handle is closed. On Windows systems, the console object is removed if it exists.

Getting Started
Lets get started. It is a good idea to start writing the code so that we can test it as soon as possible. It is very frustrating to write tons of code and not know if it works at all. It is also much easier to debug the code if we notice when a bug appears (of course, we are aiming not to meet any critters). So we will start from scratch and move on to the final library step by step.

Setting Up Source and Header Files

Before we actually start coding the library, we need to set up some things in the source files and header files. In the main source file, dreamSock.cpp, we need to have the following build-time definition check in the Windows version to make sure that WinSock 2 API is used instead of an older version:
#ifndef _WINSOCKAPI_ #define _WINSOCKAPI_ #endif

If we do not define _WINSOCKAPI_ before including windows.h, an old version of the WinSock header file will be included and the build will fail. So the preceding piece of code is required for all WinSock 2 applications that include the windows.h header. The beginning of dreamSock.cpp looks like this:
#ifdef Win32 // Windows-specific headers #ifndef _WINSOCKAPI_ #define _WINSOCKAPI_ #endif #include <windows.h> #include <winsock2.h> #else // Unix-specific headers #include <memory.h> #include <errno.h> #include <sys/ioctl.h> #include <sys/socket.h> #include <sys/time.h> #include <netinet/in.h>

220

Tutorial 2 / Creating Your Network Library

#include <arpa/inet.h> #endif // Common headers #include <stdio.h> #include <stdarg.h> #include <stdlib.h> #include <ctype.h> #include <time.h> #include "dreamSock.h" #include "dreamSockLog.h"

As we can see in the preceding code, we have separate header files for Windows and Unix, along with some common header files that both operating systems require. As we learned in the section titled Creating Independent Code, we use build-time definition checks to see what piece of code is to be linked and what is not, depending on the platform we are on.

dreamSock.h File
#ifndef __DREAMSOCK_H #define __DREAMSOCK_H #include "dreamSockLog.h" #ifdef Win32 #pragma comment (lib,"ws2_32.lib") #pragma message ("Auto linking WinSock2 library") #include <winsock2.h> #else #include <string.h> #include <netinet/in.h> #endif #include <stdio.h> #include <stddef.h> // Define SOCKET data type for Unix (defined in WinSock for Win32) // And socklen_t for Win32 #ifdef Win32 typedef int socklen_t; #else typedef int SOCKET; #ifndef #define #endif #ifndef #define #endif TRUE TRUE 1 FALSE FALSE 0

Tutorial 2 / Creating Your Network Library

221

#endif // Host types #define DREAMSERVER #define DREAMCLIENT // Connection protocols #define DREAMSOCK_TCP #define DREAMSOCK_UDP // Connection states #define DREAMSOCK_CONNECTING #define DREAMSOCK_CONNECTED #define DREAMSOCK_DISCONNECTING #define DREAMSOCK_DISCONNECTED // Error codes #define DREAMSOCK_SERVER_ERROR #define DREAMSOCK_CLIENT_ERROR

1 0

0 1

0 1 2 4

1 2

#ifdef Win32 #define DREAMSOCK_INVALID_SOCKET #else #define DREAMSOCK_INVALID_SOCKET #endif

INVALID_SOCKET 1

// System messages // Note (for all messages system and user): // positive = sequenced message // negative = unsequenced message #define DREAMSOCK_MES_CONNECT 101 #define DREAMSOCK_MES_DISCONNECT 102 #define DREAMSOCK_MES_ADDCLIENT 103 #define DREAMSOCK_MES_REMOVECLIENT 104 #define DREAMSOCK_MES_PING 105 // Introduce classes class dreamMessage; class dreamClient; class dreamServer; class dreamSock; class dreamMessage { private: bool int int int char public: void

overFlow; maxSize; size; readCount; *GetNewPoint(int length);

Init(char *d, int length);

222

Tutorial 2 / Creating Your Network Library

void void void void void void void void void void char char short long float char bool int void char char }; class dreamClient { private: void void int

Clear(void); Write(void *d, int length); AddSequences(dreamClient *client); WriteByte(char c); WriteShort(short c); WriteLong(long c); WriteFloat(float c); WriteString(char *s); BeginReading(void); BeginReading(int s); *Read(int s); ReadByte(void); ReadShort(void); ReadLong(void); ReadFloat(void); *ReadString(void); GetOverFlow(void) GetSize(void) SetSize(int s) *data; outgoingData[1400]; {return overFlow;} {return size;} {size = s;}

DumpBuffer(void); ParsePacket(dreamMessage *mes); connectionState; // Connecting, connected, // disconnecting, disconnected // // // // // // // // // // Outgoing packet sequence Incoming packet sequence Last packet acknowledged by other end Dropped packets Port IP address Client index (starts from 1, running number) Client name

unsigned short unsigned short unsigned short unsigned short int char int char SOCKET struct sockaddr int int

outgoingSequence; incomingSequence; incomingAcknowledged; droppedPackets; serverPort; serverIP[32]; index; name[32]; socket; myaddress; pingSent; ping;

// Socket // Socket address // When did we send ping? // Network latency

Tutorial 2 / Creating Your Network Library

223

int bool public:

lastMessageTime; init;

dreamClient(); ~dreamClient(); int void void void void void void int int void void unsigned short void void unsigned short void unsigned short void unsigned short void Initialize(char *localIP, char *remoteIP, int port); Uninitialize(void); Reset(void); SendConnect(char *name); SendDisconnect(void); SendPing(void); SetConnectionState(int con) {connectionState = con;} GetConnectionState(void) {return connectionState;} GetPacket(char *data, struct sockaddr *from); SendPacket(void); SendPacket(dreamMessage *message); GetOutgoingSequence(void) {return outgoingSequence;} SetOutgoingSequence(unsigned short seq) {outgoingSequence = seq;} IncreaseOutgoingSequence(void) {outgoingSequence++;} GetIncomingSequence(void) {return incomingSequence;} SetIncomingSequence(unsigned short seq) {incomingSequence = seq;} GetIncomingAcknowledged(void) {return incomingAcknowledged;} SetIncomingAcknowledged(unsigned short seq) {incomingAcknowledged = seq;} GetDroppedPackets(void) {return droppedPackets;} SetDroppedPackets(unsigned short drop) {droppedPackets = drop;} GetInit(void) {return init;} GetIndex(void) {return index;} SetIndex(int ind) {index = ind;} *GetName(void) {return name;} SetName(char *n) {strcpy(name, n);} GetSocket(void) {return socket;} SetSocket(SOCKET sock) {socket = sock;} *GetSocketAddress(void) {return &myaddress;} SetSocketAddress(struct sockaddr *address) {memcpy(&myaddress, address, sizeof(struct sockaddr));} GetPingSent(void) {return pingSent;}

bool int void char void SOCKET void struct sockaddr void

int

224

Tutorial 2 / Creating Your Network Library

void int void dreamMessage dreamClient }; class dreamServer { private: void void void void void int dreamClient int SOCKET int

SetPing(int p)

{ping = p;}

GetLastMessageTime(void) {return lastMessageTime;} SetLastMessageTime(int t) {lastMessageTime = t;} message; *next;

SendAddClient(dreamClient *newClient); SendRemoveClient(dreamClient *client); AddClient(struct sockaddr *address, char *name); RemoveClient(dreamClient *client); ParsePacket(dreamMessage *mes, struct sockaddr *address); CheckForTimeout(char *data, struct sockaddr *from); *clientList; port; socket; runningIndex; // // // // Port Socket Running index numbers for new clients

bool public:

init;

dreamServer(); ~dreamServer(); int void void int void bool dreamClient int }; /*************************************** dreamSock global functions ***************************************/ // Function prototypes int dreamSock_Initialize(void); int dreamSock_InitializeWinSock(void); Initialize(char *localIP, int serverPort); Uninitialize(void); SendPing(void); GetPacket(char *data, struct sockaddr *from); SendPackets(void); GetInit(void) {return init;} *GetClientList(void) {return clientList;} GetPort(void) {return port;}

Tutorial 2 / Creating Your Network Library

225

void dreamSock_Shutdown(void); SOCKET dreamSock_Socket(int protocol); int dreamSock_SetNonBlocking(SOCKET sock, u_long setMode); int dreamSock_SetBroadcasting(SOCKET sock, int mode); int dreamSock_StringToSockaddr(char *addressString, struct sockaddr *sadr); SOCKET dreamSock_OpenUDPSocket(char netInterface[32], int port); void dreamSock_CloseSocket(SOCKET sock); int dreamSock_GetPacket(SOCKET sock, char *data, struct sockaddr *from); void dreamSock_SendPacket(SOCKET sock, int length, char *data, struct sockaddr addr); void dreamSock_Broadcast(SOCKET sock, int length, char *data, int port); #ifndef Win32 int dreamSock_Linux_GetCurrentSystemTime(void); #else int dreamSock_Win_GetCurrentSystemTime(void); #endif int dreamSock_GetCurrentSystemTime(void); #endif

dreamMessage Class
Now lets start creating our library by first creating the header files. Here we list the dreamMessage class, which is typically used to build outgoing messages and parse incoming messages.
class dreamMessage { private: bool int int int char public: void void void void void void void void void void void char

overFlow; maxSize; size; readCount; *GetNewPoint(int length);

Init(char *d, int length); Clear(void); Write(void *d, int length); AddSequences(dreamClient *client); WriteByte(char c); WriteShort(short c); WriteLong(long c); WriteFloat(float c); WriteString(char *s); BeginReading(void); BeginReading(int s); *Read(int s);

226

Tutorial 2 / Creating Your Network Library

char short long float char bool int void char char };

ReadByte(void); ReadShort(void); ReadLong(void); ReadFloat(void); *ReadString(void); GetOverFlow(void) {return overFlow;} GetSize(void) {return size;} SetSize(int s) {size = s;} *data; outgoingData[1400];

We take a closer look at the methods and variables later in this tutorial, but as you can probably see, this class is rather simple. It contains many write and read methods, as well as size control methods.

dreamClient Class
Next comes the dreamClient class, which is a bit more complicated than dreamMessage. Do not worry too much about that now, everything will be explained in detail later.
class dreamClient { private: void void int

DumpBuffer(void); ParsePacket(dreamMessage *mes); connectionState; // Connecting, connected, // disconnecting, disconnected

unsigned short unsigned short unsigned short unsigned short int char int char

outgoingSequence; // Outgoing packet sequence incomingSequence; // Incoming packet sequence incomingAcknowledged; // Last packet acknowledged // by other end droppedPackets; // Dropped packets serverPort; serverIP[32]; index; name[32]; // // // // // Port IP address Client index (starts from 1, running number) Client name

SOCKET socket; struct sockaddr myaddress; int int int pingSent; ping; lastMessageTime;

// Socket // Socket address // When did we send ping? // Network latency

Tutorial 2 / Creating Your Network Library

227

bool public:

init;

bool int void char void SOCKET void struct sockaddr void

int void

228

Tutorial 2 / Creating Your Network Library

int void dreamMessage dreamClient };

GetLastMessageTime(void) {return lastMessageTime;} SetLastMessageTime(int t) {lastMessageTime = t;} message; *next;

This class is used on both the server and client applications. On the server side, this class works as a linked list to store all the clients that have connected to that server. When the server wants to send data to a client, it uses an object in this list to do so. On the client side, this class is used to achieve and maintain the connection with the server.

dreamServer Class
Last but not least is the server class dreamServer. This is actually simpler than you might think, because all the magic happens in the dreamClient class.
class dreamServer { private: void void void void void int dreamClient int SOCKET int

bool public:

init;

dreamServer(); ~dreamServer(); int void void int void bool dreamClient Initialize(char *localIP, int serverPort); Uninitialize(void); SendPing(void); GetPacket(char *data, struct sockaddr *from); SendPackets(void); GetInit(void) *GetClientList(void) {return init;} {return clientList;}

Tutorial 2 / Creating Your Network Library

229

int };

GetPort(void)

{return port;}

This class is used only on the server side. Its function is to listen for incoming clients and communicate with them as long as they are connected. This class contains a linked list of all connected clients, and that is where this class handles network communication.

Global Setup Functions

There are some global setup functions that are not in any class but are very important for the library. These functions are: n n n int dreamSock_Initialize(void); int dreamSock_InitializeWinSock(void); void dreamSock_Shutdown(void);

We have one global variable in our library that is not in any class. This is a boolean variable (bool dreamSock_init) to store information about dreamSocks current state whether it has been initialized or not. This is important to know because many functions may call dreamSock_Initialize or dreamSock_Shutdown in our application. With the dreamSock_init variable, we can avoid running initialization multiple times.

dreamSock_Initialize
This function is used to to initialize the library, and it is a platformindependent function. This means that whatever the platform is, this function must be run before any other dreamSock function can be used. As we can see in the following code listing, this function currently has no use on Unix (other than starting the log system). But this function must exist to keep the source code platform independent. Also, we can add Unix code later if we need to. On Windows, this runs a specific function to initialize the WinSock API. The following code listing is the first version of the dreamSock_Initialize function. We will update this function later on.
int dreamSock_Initialize(void) { if(dreamSock_init == true) return 0; dreamSock_init = true; StartLog(); #ifdef Win32 return dreamSock_InitializeWinSock();

230

Tutorial 2 / Creating Your Network Library

#else return 0; #endif }

dreamSock_InitializeWinSock
This function, in contrast to the previous function, is used to initialize only WinSock API. The programmer using our library does not have to worry about this function, as dreamSock_Initialize will automatically run this if we are on Windows. More accurately speaking, the compiler will compile the source code so that this function will be run. We introduced and examined this function in detail in Chapter 5.
int dreamSock_InitializeWinSock(void) { WORD versionRequested; WSADATA wsaData; DWORD bufferSize = 0; LPWSAPROTOCOL_INFO SelectedProtocol; int NumProtocols; // Start WinSock2. If it fails, we do not need to call WSACleanup() versionRequested = MAKEWORD(2, 0); int error = WSAStartup(versionRequested, &wsaData); if(error) { LogString("FATAL ERROR: WSAStartup failed (error = %d)", error); return 1; } else { LogString("WSAStartup OK"); // Confirm that the WinSock2 DLL supports the exact version // we want. If not, call WSACleanup(). if(LOBYTE(wsaData.wVersion) != 2 || HIBYTE(wsaData.wVersion) != 0) { LogString("FATAL ERROR: WinSock2 DLL does not support the correct version (%d.%d)", LOBYTE(wsaData.wVersion), HIBYTE(wsaData.wVersion)); WSACleanup(); return 1; } } // Call WSAEnumProtocols to figure out how big of a buffer we need NumProtocols = WSAEnumProtocols(NULL, NULL, &bufferSize);

Tutorial 2 / Creating Your Network Library

231

if((NumProtocols != SOCKET_ERROR) && (WSAGetLastError() != WSAENOBUFS)) { WSACleanup(); return 1; } // Allocate a buffer and call WSAEnumProtocols to get an array of // WSAPROTOCOL_INFO structs SelectedProtocol = (LPWSAPROTOCOL_INFO) malloc(bufferSize); if(SelectedProtocol == NULL) { WSACleanup(); return 1; } // Allocate memory for protocol list and define the protocols to // look for int *protos = (int *) calloc(2, sizeof(int)); protos[0] = IPPROTO_TCP; protos[1] = IPPROTO_UDP; NumProtocols = WSAEnumProtocols(protos, SelectedProtocol, &bufferSize); free(protos); protos = NULL; free(SelectedProtocol); SelectedProtocol = NULL; if(NumProtocols == SOCKET_ERROR) { LogString("FATAL ERROR: Didn't find any required protocols"); WSACleanup(); return 1; } return 0; }

dreamSock_Shutdown
This function, naturally, is used to shut down our network library. Like dreamSock_Initialize, this function also currently has no real use on Unix. But again, it must exist and must be run on both operating systems code to achieve full platform independency. On Windows this function will shut down the WinSock API. It must be run to shut down the API; otherwise, we may be interfering the WinSock DLL registration process. The following code listing shows the first version of the dreamSock_Shutdown function (we will update this function later).

232

Tutorial 2 / Creating Your Network Library

void dreamSock_Shutdown(void) { if(dreamSock_init == false) return; LogString("Shutting down dreamSock"); dreamSock_init = false; StopLog(); #ifdef Win32 WSACleanup(); #endif }

Global Socket Functions

Before we go into detail about all the methods and variables in the classes, we should take a look at the global socket functions that are used throughout the classes. The purpose of these functions is to wrap some common socket functions into more compact ones, so it is easier to start programming the actual network application. Here are the functions:
SOCKET dreamSock_Socket(int protocol) { int type; int proto; SOCKET sock; // Check which protocol to use if(protocol == DREAMSOCK_TCP) { type = SOCK_STREAM; proto = IPPROTO_TCP; } else { type = SOCK_DGRAM; proto = IPPROTO_UDP; } // Create the socket if((sock = socket(AF_INET, type, proto)) == 1) { LogString("dreamSock_Socket - socket() failed"); #ifdef Win32 errno = WSAGetLastError(); LogString("Error: socket() code %d : %s", errno, strerror(errno));

Tutorial 2 / Creating Your Network Library

233

#else LogString("Error: socket() : %s", strerror(errno)); #endif return DREAMSOCK_INVALID_SOCKET; } return sock; }

int dreamSock_SetNonBlocking(SOCKET sock, u_long setMode) { u_long set = setMode; // Set the socket option #ifdef Win32 return ioctlsocket(sock, FIONBIO, &set); #else return ioctl(sock, FIONBIO, &set); #endif }

int dreamSock_SetBroadcasting(SOCKET sock, int mode) { // make it broadcast capable if(setsockopt(sock, SOL_SOCKET, SO_BROADCAST, (char *) &mode, sizeof(int)) == 1) { LogString("DreamSock_SetBroadcasting failed"); #ifdef Win32 int err = WSAGetLastError(); LogString("Error code %d: setsockopt() : %s", err, strerror(err)); #else LogString("Error code %d: setsockopt() : %s", errno, strerror(errno)); #endif return DREAMSOCK_INVALID_SOCKET; } return 0; }

int dreamSock_StringToSockaddr(char *addressString, struct sockaddr *sadr) { char copy[128]; memset(sadr, 0, sizeof(struct sockaddr));

234

Tutorial 2 / Creating Your Network Library

struct sockaddr_in *addressPtr = (struct sockaddr_in *) sadr; addressPtr->sin_family = AF_INET; addressPtr->sin_port = htons(0); strcpy(copy, addressString); // If the address string begins with a number, assume an IP address if(copy[0] >= '0' && copy[0] <= '9') { *(int *) &addressPtr->sin_addr = inet_addr(copy); return 0; } else return 1; }

SOCKET dreamSock_OpenUDPSocket(char *netInterface, int port) { SOCKET sock; struct sockaddr_in address; sock = dreamSock_Socket(DREAMSOCK_UDP); if(sock == DREAMSOCK_INVALID_SOCKET) return sock; dreamSock_SetNonBlocking(sock, 1); dreamSock_SetBroadcasting(sock, 1); // If no address string provided, use any interface available if(!netInterface || !netInterface[0] || !strcmp(netInterface, "localhost")) { LogString("No net interface given, using any interface available"); address.sin_addr.s_addr = htonl(INADDR_ANY); } else { LogString("Using net interface = '%s'", netInterface); dreamSock_StringToSockaddr(netInterface, (struct sockaddr *) &address); } // If no port number provided, use any port number available if(port == 0) { LogString("No port defined, picking one for you"); address.sin_port = 0; } else {

Tutorial 2 / Creating Your Network Library

235

address.sin_port = htons((short) port); } address.sin_family = AF_INET; // Bind the address to the socket if(bind(sock, (struct sockaddr *) &address, sizeof(address)) == 1) { #ifdef Win32 errno = WSAGetLastError(); LogString("Error code %d: bind() : %s", errno, strerror(errno)); #else LogString("Error code %d: bind() : %s", errno, strerror(errno)); #endif return DREAMSOCK_INVALID_SOCKET; } // Get the port number (if we did not define one, // we get the assigned port number here) socklen_t len = sizeof(address); getsockname(sock, (struct sockaddr *) &address, &len); LogString("Opening UDP port = %d", ntohs(address.sin_port)); return sock; }

void dreamSock_CloseSocket(SOCKET sock) { #ifdef Win32 closesocket(sock); #else close(sock); #endif }

int dreamSock_GetPacket(SOCKET sock, char *data, struct sockaddr *from) { int ret; struct sockaddr tempFrom; socklen_t fromlen; fromlen = sizeof(tempFrom); ret = recvfrom(sock, data, 1400, 0, (struct sockaddr *) &tempFrom, &fromlen); // Copy the address to the from pointer if(from != NULL) *(struct sockaddr *) from = tempFrom;

236

Tutorial 2 / Creating Your Network Library

if(ret == 1) { #ifdef Win32 errno = WSAGetLastError(); // Silently handle wouldblock if(errno == WSAEWOULDBLOCK) return ret; if(errno == WSAEMSGSIZE) { // ERROR: Oversize packet return ret; } LogString("Error code %d: recvfrom() : %s", errno, strerror(errno)); #else // Silently handle wouldblock if(errno == EWOULDBLOCK || errno == ECONNREFUSED) return ret; LogString("Error code %d: recvfrom() : %s", errno, strerror(errno)); #endif return ret; } return ret; }

void dreamSock_SendPacket(SOCKET sock, int length, char *data, struct sockaddr addr) { int ret; ret = sendto(sock, data, length, 0, &addr, sizeof(addr)); if(ret == 1) { #ifdef Win32 errno = WSAGetLastError(); // Silently handle wouldblock if(errno == WSAEWOULDBLOCK) return; LogString("Error code %d: sendto() : %s", errno, strerror(errno)); #else // Silently handle wouldblock

Tutorial 2 / Creating Your Network Library

237

if(errno == EWOULDBLOCK) return; LogString("Error code %d: sendto() : %s", errno, strerror(errno)); #endif } }

void dreamSock_Broadcast(SOCKET sock, int length, char *data, int port) { struct sockaddr_in servaddr; socklen_t len; // Use broadcast address u_long inetAddr = inet_addr("255.255.255.255"); // Fill address information structure memset(&servaddr, 0, sizeof(struct sockaddr_in)); servaddr.sin_family = AF_INET; servaddr.sin_port = htons(port); servaddr.sin_addr.s_addr = inetAddr; len = sizeof(servaddr); // Broadcast! int ret = sendto(sock, data, length, 0, (struct sockaddr *) &servaddr, len); if(ret == 1) { #ifdef Win32 errno = WSAGetLastError(); // Silently handle wouldblock if(errno == WSAEWOULDBLOCK) return; LogString("Error code %d: sendto() : %s", errno, strerror(errno)); #else // Silently handle wouldblock if(errno == EWOULDBLOCK) return; LogString("Error code %d: sendto() : %s", errno, strerror(errno)); #endif } }

int dreamSock_GetCurrentSystemTime(void)

238

Tutorial 2 / Creating Your Network Library

{ #ifndef Win32 return dreamSock_Linux_GetCurrentSystemTime(); #else return dreamSock_Win_GetCurrentSystemTime(); #endif }

dreamSock_Socket Function
This function creates a UDP or TCP socket, based on the given parameter.
SOCKET dreamSock_Socket(int protocol) { int type; int proto; SOCKET sock; // Check which protocol to use if(protocol == DREAMSOCK_TCP) { type = SOCK_STREAM; proto = IPPROTO_TCP; } else { type = SOCK_DGRAM; proto = IPPROTO_UDP; } // Create the socket if((sock = socket(AF_INET, type, proto)) == 1) { LogString("dreamSock_Socket - socket() failed"); #ifdef Win32 errno = WSAGetLastError(); LogString("Error: socket() code %d : %s", errno, strerror(errno)); #else LogString("Error: socket() : %s", strerror(errno)); #endif return DREAMSOCK_INVALID_SOCKET; } return sock; }

This function takes one parameter (int protocol) and uses it to define what kind of socket to create. The actual library functionality does not include TCP networking, so DREAMSOCK_TCP is there for later use only. Of course, you can still create a TCP socket with this

Tutorial 2 / Creating Your Network Library

239

function, and then use that socket with basic socket functions. So all this function really does is run the socket function and do some error checking for it.

dreamSock_SetNonBlocking Function
This function sets a socket blocking or non-blocking, based on the given parameters. Most of the time it is used to set a socket non-blocking, hence the name.
int dreamSock_SetNonBlocking(SOCKET sock, u_long setMode) { u_long set = setMode; // Set the socket option #ifdef Win32 return ioctlsocket(sock, FIONBIO, &set); #else return ioctl(sock, FIONBIO, &set); #endif }

The function takes two parameters (SOCKET sock and u_long setMode). The first one defines the socket to modify and the latter sets the mode. If setMode is 1, the socket is set non-blocking.

dreamSock_SetBroadcasting Function
This is used to set a socket to broadcast over a local area network.
int dreamSock_SetBroadcasting(SOCKET sock, int mode) { // make it broadcast capable if(setsockopt(sock, SOL_SOCKET, SO_BROADCAST, (char *) &mode, sizeof(int)) == 1) { LogString("DreamSock_SetBroadcasting failed"); #ifdef Win32 int err = WSAGetLastError(); LogString("Error code %d: setsockopt() : %s", err, strerror(err)); #else LogString("Error code %d: setsockopt() : %s", errno, strerror(errno)); #endif return DREAMSOCK_INVALID_SOCKET; } return 0; }

240

Tutorial 2 / Creating Your Network Library

Like dreamSock_SetNonBlocking, this function also has two parameters (SOCKET sock and int mode) that define the socket to modify and the mode to set. If mode is 1, the socket is set to broadcast.
Note A socket does not broadcast automatically even if it is set to broadcast. To broadcast over a local area network, you must use the broadcast address 255.255.255.255.

dreamSock_StringToSockaddr Function
This function converts the address string (for example, 192.168.0.1) to socket address format.
int dreamSock_StringToSockaddr(char *addressString, struct sockaddr *sadr) { char copy[128]; memset(sadr, 0, sizeof(struct sockaddr)); struct sockaddr_in *addressPtr = (struct sockaddr_in *) sadr; addressPtr->sin_family = AF_INET; addressPtr->sin_port = htons(0); strcpy(copy, addressString); // If the address string begins with a number, assume an IP address if(copy[0] >= '0' && copy[0] <= '9') { *(int *) &addressPtr->sin_addr = inet_addr(copy); return 0; } else return 1; }

This function takes two parameters (char *addressString and struct sockaddr *sadr). The first one is used to give the function the IP address string, which we want to convert to a socket address. The latter one is the socket address itself.
if(copy[0] >= '0' && copy[0] <= '9') { *(int *) &addressPtr->sin_addr = inet_addr(copy); return 0; } else return 1;

The function checks whether the given address string actually is an IP string by checking if the first letter is a number. Otherwise, the function fails and returns 1 to indicate this.

Tutorial 2 / Creating Your Network Library

241

dreamSock_OpenUDPSocket Function
Here is a function that does the magic to open a UDP socket and even binds a port to it while doing so.
SOCKET dreamSock_OpenUDPSocket(char *netInterface, int port) { SOCKET sock; struct sockaddr_in address; sock = dreamSock_Socket(DREAMSOCK_UDP); if(sock == DREAMSOCK_INVALID_SOCKET) return sock; dreamSock_SetNonBlocking(sock, 1); dreamSock_SetBroadcasting(sock, 1); // If no address string provided, use any interface available if(!netInterface || !netInterface[0] || !strcmp(netInterface, "localhost")) { LogString("No net interface given, using any interface available"); address.sin_addr.s_addr = htonl(INADDR_ANY); } else { LogString("Using net interface = '%s'", netInterface); dreamSock_StringToSockaddr(netInterface, (struct sockaddr *) &address); } // If no port number provided, use any port number available if(port == 0) { LogString("No port defined, picking one for you"); address.sin_port = 0; } else { address.sin_port = htons((short) port); } address.sin_family = AF_INET; // Bind the address to the socket if(bind(sock, (struct sockaddr *) &address, sizeof(address)) == 1) { #ifdef Win32 errno = WSAGetLastError(); LogString("Error code %d: bind() : %s", errno, strerror(errno)); #else

242

Tutorial 2 / Creating Your Network Library

LogString("Error code %d: bind() : %s", errno, strerror(errno)); #endif return DREAMSOCK_INVALID_SOCKET; } // Get the port number (if we did not define one, // we get the assigned port number here) socklen_t len = sizeof(address); getsockname(sock, (struct sockaddr *) &address, &len); LogString("Opening UDP port = %d", ntohs(address.sin_port)); return sock; }

The function takes two parameters (char *netInterface, int port). The first one defines which local IP address we want to use for this socket. Remember that a computer can indeed have more than one IP address (for example, if it has more than one network interface card). If this parameter is NULL, the default network interface is used.
if(!netInterface || !netInterface[0] || !strcmp(netInterface, "localhost")) { LogString("No net interface given, using any interface available"); address.sin_addr.s_addr = htonl(INADDR_ANY); } else { LogString("Using net interface = '%s'", netInterface); dreamSock_StringToSockaddr(netInterface, (struct sockaddr *) &address); }

The same happens for the port number. If no port is defined, the function lets the kernel pick one for you.
if(port == 0) { LogString("No port defined, picking one for you"); address.sin_port = 0; } else { address.sin_port = htons((short) port); }

We do not always know what port number the socket received, so we should get the address information. This is done in the following piece of code:
socklen_t len = sizeof(address); getsockname(sock, (struct sockaddr *) &address, &len); LogString("Opening UDP port = %d", ntohs(address.sin_port));

Tutorial 2 / Creating Your Network Library

243

As you can see, this function uses many of the global functions we have introduced here. This makes the functionality simpler now that we know how the small parts of this function works. Nice, eh?

dreamSock_CloseSocket Function
At some point we most likely want to close a socket. Here is a function for that.
void dreamSock_CloseSocket(SOCKET sock) { #ifdef Win32 closesocket(sock); #else close(sock); #endif }

This is a simple function, but we still have to use two different basic socket functions in it. We must use different function calls on different platforms, but the functionality is the same; only the names of the functions are different.

dreamSock_GetPacket Function
We need a function for receiving data, and this is it.
int dreamSock_GetPacket(SOCKET sock, char *data, struct sockaddr *from) { int ret; struct sockaddr tempFrom; socklen_t fromlen; fromlen = sizeof(tempFrom); ret = recvfrom(sock, data, 1400, 0, (struct sockaddr *) &tempFrom, &fromlen); // Copy the address to the from pointer if(from != NULL) *(struct sockaddr *) from = tempFrom; if(ret == 1) { #ifdef Win32 errno = WSAGetLastError(); // Silently handle wouldblock if(errno == WSAEWOULDBLOCK) return ret; if(errno == WSAEMSGSIZE) {

244

Tutorial 2 / Creating Your Network Library

// ERROR: Oversize packet return ret; } LogString("Error code %d: recvfrom() : %s", errno, strerror(errno)); #else // Silently handle wouldblock if(errno == EWOULDBLOCK || errno == ECONNREFUSED) return ret; LogString("Error code %d: recvfrom() : %s", errno, strerror(errno)); #endif return ret; } return ret; }

This function retrieves the incoming data. It has three parameters (SOCKET sock, char *data, struct sockaddr *from). The first one is obviously used to define the socket to read from, the second one stores the incoming data, and the last one stores the senders address information. So every time we receive data, we know the IP address and port of the sender. On a successful call to recvfrom, this function returns the amount of data read in bytes. If the socket is non-blocking and there is no incoming data, this function returns 1. If the socket is blocking, this function never returns until there is incoming data.

dreamSock_SendPacket Function
We probably want to send some data too, and this is the function we would use.
void dreamSock_SendPacket(SOCKET sock, int length, char *data, struct sockaddr addr) { int ret; ret = sendto(sock, data, length, 0, &addr, sizeof(addr)); if(ret == 1) { #ifdef Win32 errno = WSAGetLastError(); // Silently handle wouldblock if(errno == WSAEWOULDBLOCK)

Tutorial 2 / Creating Your Network Library

245

return; LogString("Error code %d: sendto() : %s", errno, strerror(errno)); #else // Silently handle wouldblock if(errno == EWOULDBLOCK) return; LogString("Error code %d: sendto() : %s", errno, strerror(errno)); #endif } }

This function works pretty much the same way as dreamSock_GetPacket. The difference is the direction of the data. This time we send it. The function has four parameters (SOCKET sock, int length, char *data, struct sockaddr addr). The first defines the socket, the second defines the length of the data to send (in bytes), the third is the data itself, and the fourth is the address of the recipient. Again, this function returns the amount of data sent in bytes. If something went wrong, the function returns 1.

dreamSock_Broadcast Function
This function is not used in the dreamSock library, but you can use it for broadcasting.
void dreamSock_Broadcast(SOCKET sock, int length, char *data, int port) { struct sockaddr_in servaddr; socklen_t len; // Use broadcast address u_long inetAddr = inet_addr("255.255.255.255"); // Fill address information structure memset(&servaddr, 0, sizeof(struct sockaddr_in)); servaddr.sin_family = AF_INET; servaddr.sin_port = htons(port); servaddr.sin_addr.s_addr = inetAddr; len = sizeof(servaddr); // Broadcast! int ret = sendto(sock, data, length, 0, (struct sockaddr *) &servaddr, len); if(ret == 1) {

246

Tutorial 2 / Creating Your Network Library

#ifdef Win32 errno = WSAGetLastError(); // Silently handle wouldblock if(errno == WSAEWOULDBLOCK) return; LogString("Error code %d: sendto() : %s", errno, strerror(errno)); #else // Silently handle wouldblock if(errno == EWOULDBLOCK) return; LogString("Error code %d: sendto() : %s", errno, strerror(errno)); #endif } }

This function works the same way as dreamSock_SendPacket, but this time we do not define the entire address of the recipient, only the port. This is because we have to send to a broadcast address instead of a known remote address.

dreamSock_GetCurrentSystemTime Function
We need to be aware of time, and this is the function for that.
int dreamSock_GetCurrentSystemTime(void) { #ifndef Win32 return dreamSock_Linux_GetCurrentSystemTime(); #else return dreamSock_Win_GetCurrentSystemTime(); #endif }

This function wraps the platform-specific functions, which are introduced next.

dreamSock_Linux_GetCurrentSystemTime Function
This is the Unix/Linux version of dreamSock_GetCurrentSystemTime.
int dreamSock_Linux_GetCurrentSystemTime(void) { struct timeval tp; struct timezone tzp; static int basetime;

Tutorial 2 / Creating Your Network Library

247

gettimeofday(&tp, &tzp); if(!basetime) { basetime = tp.tv_sec; return tp.tv_usec / 1000; } return (tp.tv_sec basetime) * 1000 + tp.tv_usec / 1000; }

The function returns the current system time (starting from 0).
if(!basetime) { basetime = tp.tv_sec; return tp.tv_usec / 1000; }

In this piece of code we retrieve the base time, which is the current absolute time. This happens only once. After we have the base time value, we use it to calculate the application run time by subtracting the base time from the current time.

dreamSock_Win_GetCurrentSystemTime Function
This is the Windows version of dreamSock_GetCurrentSystemTime.
int dreamSock_Win_GetCurrentSystemTime(void) { int curtime; static int base; static bool initialized = false; if(!initialized) { base = timeGetTime() & 0xffff0000; initialized = true; } curtime = timeGetTime() base; return curtime; }

This function returns the current system time (starting from 0).
if(!initialized) { base = timeGetTime() & 0xffff0000; initialized = true; }

248

Tutorial 2 / Creating Your Network Library

In this piece of code we retrieve the base time, which is the current absolute time. This happens only once, and after we have the base-time value, we use it to calculate the application run time by subtracting the base time from the current time.

Retrieving Error Values

You most likely noticed that in most of the functions we check a variable called errno, which is obviously short for error number. This variable holds the error value number if something goes wrong. On Windows, we need to get the value ourselves, by using the following piece of code.
errno = WSAGetLastError();

On Unix systems, this is not required as the socket and other functions fill in errno for us. We can then try to retrieve a text string concerning that error value since a plain number does not tell us much. This is done with the following piece of code:
strerror(errno)

This function does not always give us anything useful, but it is worth a try.

Summary of Global Functions

Those are all the global socket functions we need and which form the basis of our dreamSock network library. Remember that these functions can be used directly if you need to do something that is not done in the library.

Creating dreamSock Network Library

Now we will begin creating our network library by creating the three C++ classes that make the library come alive. We will start off by creating the message buffer handling system, dreamMessage class. Then we move on to the most complex system the dreamClient class. The last one is the dreamServer class. Once these classes are ready, we can start writing our network application. So lets get kicking!

Tutorial 2 / Creating Your Network Library

249

dreamMessage Class Member Variables

Each class has its own member variables. The following are the variables for dreamMessage and their usage:
private: bool int int int public: char char overFlow; maxSize; size; readCount;

*data; outgoingData[1400];

The boolean variable overFlow is used to store message size overflow information. Normally this variable is set to FALSE, but whenever we try to write more bytes into the message than the maximum size is, this variable is set to TRUE. When this happens, you cannot send the message. The integer maxSize defines the maximum size of the message in bytes. We can set this manually. The integer size tells us the current size of the message. The integer readCount stores the number of bytes read from the message so far. The char data is a pointer to the messages data buffer. We must set this to point to an actual char buffer. The char outgoingData is a helper data buffer for outgoing data. We may make the data pointer point to this buffer, but it is not required.

dreamMessage Class Functionality

The dreamMessage class is used to create the messages that we send over the network. This class does not have the functionality to send the messages, only to build and read them. Lets see how everything works. The following code listing shows all the methods of the dreamMessage class:
void dreamMessage::Init(char *d, int length) { data = d; maxSize = length; size = 0; readCount = 0; overFlow = false; } void dreamMessage::Clear(void) { size = 0; readCount = 0; overFlow = false;

250

Tutorial 2 / Creating Your Network Library

} char *dreamMessage::GetNewPoint(int length) { char *tempData; // Check for overflow if(size + length > maxSize) { Clear(); overFlow = true; } tempData = data + size; size += length; return tempData; } void dreamMessage::AddSequences(dreamClient *client) { WriteShort(client->GetOutgoingSequence()); WriteShort(client->GetIncomingSequence()); } void dreamMessage::Write(void *d, int length) { memcpy(GetNewPoint(length), d, length); } void dreamMessage::WriteByte(char c) { char *buf; buf = GetNewPoint(1); memcpy(buf, &c, 1); } void dreamMessage::WriteShort(short c) { char *buf; buf = GetNewPoint(2); memcpy(buf, &c, 2); } void dreamMessage::WriteLong(long c) { char *buf; buf = GetNewPoint(4);

Tutorial 2 / Creating Your Network Library

251

memcpy(buf, &c, 4); } void dreamMessage::WriteFloat(float c) { char *buf; buf = GetNewPoint(4); memcpy(buf, &c, 4); } void dreamMessage::WriteString(char *s) { if(!s) { return; } else Write(s, strlen(s) + 1); } void dreamMessage::BeginReading(void) { readCount = 0; } void dreamMessage::BeginReading(int s) { size = s; readCount = 0; } char *dreamMessage::Read(int s) { static char c[2048]; if(readCount+s > size) return NULL; else memcpy(&c, &data[readCount], s); readCount += s; return c; } char dreamMessage::ReadByte(void) { char c; if(readCount+1 > size) c = 1; else

252

Tutorial 2 / Creating Your Network Library

memcpy(&c, &data[readCount], 1); readCount++; return c; } short dreamMessage::ReadShort(void) { short c; if(readCount+2 > size) c = 1; else memcpy(&c, &data[readCount], 2); readCount += 2; return c; } long dreamMessage::ReadLong(void) { long c; if(readCount+4 > size) c = 1; else memcpy(&c, &data[readCount], 4); readCount += 4; return c; } float dreamMessage::ReadFloat(void) { float c; if(readCount+4 > size) c = 1; else memcpy(&c, &data[readCount], 4); readCount += 4; return c; } char *dreamMessage::ReadString(void) { static char string[2048]; int l, c;

Tutorial 2 / Creating Your Network Library

253

l = 0; do { c = ReadByte(); if (c == 1 || c == 0) break; string[l] = c; l++; } while(l < sizeof(string)1); string[l] = 0; return string; }

Init Function
The Init function is pretty straightforward. All it does is initialize the classs member variables.
void dreamMessage::Init(char *d, int length) { data = d; maxSize = length; size = 0; readCount = 0; overFlow = false; }

This function takes two parameters (char *d, int length). The first sets the buffer for the messages actual data. Whenever we write something to a message, we really write it into this buffer. Notice that dreamMessages data member variable is only a pointer. So we must have the buffer elsewhere, and then using Init function, we make the data pointer point to that buffer. The second parameter defines the maximum size of a message in bytes. The function also resets some other variables.

Clear Function
The Clear function resets the messages member variables without modifying the data buffer.
void dreamMessage::Clear(void) { size = 0; readCount = 0; overFlow = false; }

254

Tutorial 2 / Creating Your Network Library

GetNewPoint Function
This function gets a new point from the data buffer based on the given parameter.
char *dreamMessage::GetNewPoint(int length) { char *tempData; // Check for overflow if(size + length > maxSize) { Clear(); overFlow = true; } tempData = data + size; size += length; return tempData; }

This function takes only one parameter (int length), which defines the point in the buffer to retrieve. The function then returns a pointer that points to that position. With this function we can move around in the data buffer and write data to its correct location. The function checks if we are trying to write more than the maximum size of the message. If yes, the message is cleared and the overflow flag is raised.
if(size + length > maxSize) { Clear(); overFlow = true; }

The following piece of code is essential to this function. The temporary pointer tempData is set to point to the messages data pointer and this pointer is moved by the current size of the message. Now we have a completely new, untouched position in the data buffer. Well, that may not always be true because the data buffer itself may hold some old data at that position. But because we have the variable that holds the current size of the message, we do not need to worry about the old data. We will never read over to the old data part, and we will always write on top of the old one.
tempData = data + size; size += length;

Tutorial 2 / Creating Your Network Library

255

AddSequences Function
Understanding what this function does requires us to understand the dreamClient class first, so do not worry about this function yet. All you need to know now is that this function writes two shorts into the messages data buffer. These two shorts are the clients message sequence numbers. There is more about this later in this tutorial.
void dreamMessage::AddSequences(dreamClient *client) { WriteShort(client->GetOutgoingSequence()); WriteShort(client->GetIncomingSequence()); }

This function takes one parameter (dreamClient *client). This parameter is a pointer to a dreamClient object. Each client uses this function to add its own message sequences to the message.

Write Function
Here we have the first writing function. This is a generic write function that lets us write any type of data to the message.
void dreamMessage::Write(void *d, int length) { memcpy(GetNewPoint(length), d, length); }

This function takes two parameters (void *d, int length), which define the data to write and the length of the data. Normally when we write data, we use the other write functions that are discussed in this section. When we use those functions, we do not need to define the size of the data ourselves because the size is known and is always the same. But when we use the generic Write function, we have to define the size, since it can be anything. The function gets a new point in the data buffer and copies the given data to that point.

WriteByte Function
The WriteByte function writes a byte into the message.
void dreamMessage::WriteByte(char c) { char *buf; buf = GetNewPoint(1); memcpy(buf, &c, 1); }

256

Tutorial 2 / Creating Your Network Library

This function takes the data to write as a parameter and then writes it to the buffer.

WriteShort Function
The WriteShort function writes a short into the message.
void dreamMessage::WriteShort(short c) { char *buf; buf = GetNewPoint(2); memcpy(buf, &c, 2); }

This function takes the data to write as a parameter and writes it to the buffer. A short takes 2 bytes.

WriteLong Function
The WriteLong function writes a long into the message.
void dreamMessage::WriteLong(long c) { char *buf; buf = GetNewPoint(4); memcpy(buf, &c, 4); }

This function takes the data to write as a parameter and writes it to the buffer. A long takes 4 bytes.

WriteFloat Function
The WriteFloat function writes a float into the message.
void dreamMessage::WriteFloat(float c) { char *buf; buf = GetNewPoint(4); memcpy(buf, &c, 4); }

This function takes the data to write as a parameter and then writes it to the buffer. A float takes 4 bytes.

Tutorial 2 / Creating Your Network Library

257

WriteString Function
The WriteString function writes a string into the message.
void dreamMessage::WriteString(char *s) { if(!s) { return; } else Write(s, strlen(s) + 1); }

This function takes the data to write as a parameter and writes it to the buffer. If the string is NULL, nothing is written. We do not need to define the length of the string, because we can use the strlen function to get that.

BeginReading Function
There are two versions of the BeginReading function. They both set the current size of the message.
void dreamMessage::BeginReading(void) { readCount = 0; } void dreamMessage::BeginReading(int s) { size = s; readCount = 0; }

The purpose of this function is to reset the readCount member variable so reading begins from the beginning. The second version takes one parameter (int s) that defines the current size of the message.

Read Function
This function reads a defined amount of data.
char *dreamMessage::Read(int s) { static char c[2048]; if(readCount+s > size) return NULL; else memcpy(&c, &data[readCount], s); readCount += s;

258

Tutorial 2 / Creating Your Network Library

return c; }

This function has one parameter (int s) that defines the amount of data to read. Member variable readCount keeps track of the current position in the data buffer. The function checks for buffer overflow, as shown in the following piece of code.
if(readCount+s > size) return NULL; else memcpy(&c, &data[readCount], s);

If we try to read more data than the message holds, nothing is read.

ReadByte Function
This function reads a byte from the data buffer.
char dreamMessage::ReadByte(void) { char c; if(readCount+1 > size) c = 1; else memcpy(&c, &data[readCount], 1); readCount++; return c; }

Once the reading is done, the function returns a byte that holds the data it just read. The function checks for buffer overflow.

ReadShort Function
This function reads a short from the data buffer.
short dreamMessage::ReadShort(void) { short c; if(readCount+2 > size) c = 1; else memcpy(&c, &data[readCount], 2); readCount += 2; return c; }

Tutorial 2 / Creating Your Network Library

259

Once the reading is done, the function returns a short that holds the data it just read. The function checks for buffer overflow.

ReadLong Function
This function reads a long from the data buffer.
long dreamMessage::ReadLong(void) { long c; if(readCount+4 > size) c = 1; else memcpy(&c, &data[readCount], 4); readCount += 4; return c; }

Once the reading is done, the function returns a long that holds the data it just read. The function checks for buffer overflow.

ReadFloat Function
This function reads a float from the data buffer.
float dreamMessage::ReadFloat(void) { float c; if(readCount+4 > size) c = 1; else memcpy(&c, &data[readCount], 4); readCount += 4; return c; }

Once the reading is done, the function returns a float that holds the data it just read. The function checks for buffer overflow.

ReadString Function
This function reads a string from the data buffer.
char *dreamMessage::ReadString(void) { static char string[2048]; int l, c;

260

Tutorial 2 / Creating Your Network Library

l = 0; do { c = ReadByte(); if (c == 1 || c == 0) break; string[l] = c; l++; } while(l < sizeof(string)1); string[l] = 0; return string; }

Once the reading is done, the function returns a pointer to a string that holds the data it just read. The function checks for buffer overflow. The function reads the string byte by byte. If the byte is 0 or 1, reading is stopped. That means the string ends. This is shown here.
do { c = ReadByte(); if (c == 1 || c == 0) break; string[l] = c; l++; } while(l < sizeof(string)1);

dreamMessage Summary
That is all of the dreamMessage class. We now have the functionality to write messages but no way to send them yet. So next we need to create dreamClient and dreamServer.

dreamClient Class Member Variables

Here are the member variables for dreamClient and their purposes.
private: int connectionState; // Connecting, connected, // disconnecting, disconnected

unsigned short unsigned short unsigned short

outgoingSequence; // Outgoing packet sequence incomingSequence; // Incoming packet sequence incomingAcknowledged; // Last packet acknowledged // by other end

Tutorial 2 / Creating Your Network Library

261

unsigned short int char int char SOCKET struct sockaddr int int int bool public: dreamMessage dreamClient

droppedPackets; serverPort; serverIP[32]; index; name[32]; socket; myaddress; pingSent; ping; lastMessageTime; init;

// Dropped packets // // // // // Port IP address Client index (starts from 1, running number) Client name

// Socket // Socket address // When did we send ping? // Network latency

message; *next;

The integer connectionState tells us the current state of the connection. It can be one of these: connecting, connected, disconnecting, or disconnected. The unsigned short outgoingSequence keeps track of the message numbers that we have sent. The first one has number or sequence 1, the next one 2, and so on. The unsigned short incomingSequence keeps track of received messages sequence numbers. This number is set every time we receive a message from a remote host. The sequence is set to match the remote hosts outgoingSequence. This may be old information though, because the sequence is picked from the last received message, and before we got that message, the remote host may have sent another packet. But that is not usually a problem so do not worry about it. The unsigned short incomingAcknowledged keeps track of the last packet the remote host acknowledged. This is actually the remote hosts incomingSequence. Now we have a nice little loop so we dont have to worry about dropped packets. If a packet is dropped, it is lost, but the sequence numbers will not get mixed up. The unsigned short droppedPackets stores information about the amount of dropped packets this frame. The integer serverPort stores the servers port number. The char serverIP stores the servers IP address. The integer index is unique for every client. The server gives this number to us. The number is incremented starting from 1. Every time a client connects the server, this number increases and the new client gets that new number. This number is unique to each client since the index number is never decreased on the server. The char name stores the name of this client.

262

Tutorial 2 / Creating Your Network Library

The SOCKET socket variable holds the socket for each client. This is not used on the server side. The struct sockaddr myaddress is the sockets socket address. The integer pingSent tells us the time we sent our ping message. The integer ping tell us the network latency. The integer lastMessageTime tells us when the server received the last message from this client. The boolean Init is a flag that tells us whether the client has been initialized. The dreamMessage message is the clients internal message. This is more like a helper variable than a requirement. We can use an external message to send data, but we also can use this one. The dreamClient next is a pointer to the next client in the client list. This is used on the server only, as clients do not have client lists.

dreamClient Class Functionality

The dreamClient class is used on both the server side and the client side. On the server side, there is a list of clients that controls all the communication with the real clients. On the client side, this class is used to connect to the server. Here are all the methods of dreamClient:
dreamClient::dreamClient() { connectionState outgoingSequence incomingSequence incomingAcknowledged droppedPackets init serverPort pingSent ping lastMessageTime next = NULL; } dreamClient::~dreamClient() { dreamSock_CloseSocket(socket); } int dreamClient::Initialize(char *localIP, char *remoteIP, int port) { // Initialize dreamSock if it is not already initialized

= DREAMSOCK_DISCONNECTED; = = = = 1; 0; 0; 0;

= false; = 0; = 0; = 0; = 0;

Tutorial 2 / Creating Your Network Library

263

dreamSock_Initialize(); // Save server's address information for later use serverPort = port; strcpy(serverIP, remoteIP); LogString("Server's information: IP address: %s, port: %d", serverIP, serverPort); // Create client socket socket = dreamSock_OpenUDPSocket(localIP, 0); // Check that the address is not empty u_long inetAddr = inet_addr(serverIP); if(inetAddr == INADDR_NONE) { return DREAMSOCK_CLIENT_ERROR; } if(socket == DREAMSOCK_INVALID_SOCKET) { return DREAMSOCK_CLIENT_ERROR; } init = true; return 0; } void dreamClient::Uninitialize(void) { dreamSock_CloseSocket(socket); Reset(); init = false; } void dreamClient::Reset(void) { connectionState outgoingSequence incomingSequence incomingAcknowledged droppedPackets pingSent ping lastMessageTime next = NULL;

= DREAMSOCK_DISCONNECTED; = = = = 1; 0; 0; 0;

= 0; = 0; = 0;

264

Tutorial 2 / Creating Your Network Library

} void dreamClient::DumpBuffer(void) { char data[1400]; int ret; while((ret = dreamSock_GetPacket(socket, data, NULL)) > 0) { } } void dreamClient::SendConnect(char *name) { // Dump buffer so there won't be any old packets to process DumpBuffer(); connectionState = DREAMSOCK_CONNECTING; message.Init(message.outgoingData, sizeof(message.outgoingData)); message.WriteByte(DREAMSOCK_MES_CONNECT); message.WriteString(name); SendPacket(&message); } void dreamClient::SendDisconnect(void) { message.Init(message.outgoingData, sizeof(message.outgoingData)); message.WriteByte(DREAMSOCK_MES_DISCONNECT); SendPacket(&message); Reset(); connectionState = DREAMSOCK_DISCONNECTING; } void dreamClient::SendPing(void) { pingSent = dreamSock_GetCurrentSystemTime(); message.Init(message.outgoingData, sizeof(message.outgoingData)); message.WriteByte(DREAMSOCK_MES_PING); SendPacket(&message); } void dreamClient::ParsePacket(dreamMessage *mes) { mes->BeginReading(); int type = mes->ReadByte(); // Check if the type is a positive number // = is the packet sequenced

Tutorial 2 / Creating Your Network Library

265

if(type > 0) { unsigned short sequence unsigned short sequenceAck

= mes->ReadShort(); = mes->ReadShort();

if(sequence <= incomingSequence) { LogString("Client: (sequence: %d <= incoming seq: %d)", sequence, incomingSequence); LogString("Client: Sequence mismatch"); } droppedPackets = sequence incomingSequence + 1; incomingSequence = sequence; incomingAcknowledged = sequenceAck; } // Parse through the system messages switch(type) { case DREAMSOCK_MES_CONNECT: connectionState = DREAMSOCK_CONNECTED; LogString("LIBRARY: Client: got connect confirmation"); break; case DREAMSOCK_MES_DISCONNECT: connectionState = DREAMSOCK_DISCONNECTED; LogString("LIBRARY: Client: got disconnect confirmation"); break; case DREAMSOCK_MES_ADDCLIENT: LogString("LIBRARY: Client: adding a client"); break; case DREAMSOCK_MES_REMOVECLIENT: LogString("LIBRARY: Client: removing a client"); break; case DREAMSOCK_MES_PING: SendPing(); break; } } int dreamClient::GetPacket(char *data, struct sockaddr *from) { // Check if the client is set up or if it is disconnecting if(!socket) return 0;

266

Tutorial 2 / Creating Your Network Library

int ret; dreamMessage mes; mes.Init(data, sizeof(data)); ret = dreamSock_GetPacket(socket, mes.data, from); if(ret <= 0) return 0; mes.SetSize(ret); // Parse system messages ParsePacket(&mes); return ret; } void dreamClient::SendPacket(void) { // Check that everything is set up if(!socket || connectionState == DREAMSOCK_DISCONNECTED) { LogString("SendPacket error: Could not send because the client is disconnected"); return; } // If the message overflowed, do not send it if(message.GetOverFlow()) { LogString("SendPacket error: Could not send because the buffer overflowed"); return; } // Check if serverPort is set. If it is, we are a client sending to // the server. Otherwise we are a server sending to a client. if(serverPort) { struct sockaddr_in sendToAddress; memset((char *) &sendToAddress, 0, sizeof(sendToAddress)); u_long inetAddr = inet_addr(serverIP); sendToAddress.sin_port = htons((u_short) serverPort); sendToAddress.sin_family = AF_INET; sendToAddress.sin_addr.s_addr = inetAddr; dreamSock_SendPacket(socket, message.GetSize(), message.data, *(struct sockaddr *) &sendToAddress); } else { dreamSock_SendPacket(socket, message.GetSize(), message.data,

Tutorial 2 / Creating Your Network Library

267

myaddress); } // Check if the packet is sequenced message.BeginReading(); int type = message.ReadByte(); if(type > 0) { outgoingSequence++; } } void dreamClient::SendPacket(dreamMessage *theMes) { // Check that everything is set up if(!socket || connectionState == DREAMSOCK_DISCONNECTED) { LogString("SendPacket error: Could not send because the client is disconnected"); return; } // If the message overflowed, do not send it if(theMes->GetOverFlow()) { LogString("SendPacket error: Could not send because the buffer overflowed"); return; } // Check if serverPort is set. If it is, we are a client sending to // the server. Otherwise we are a server sending to a client. if(serverPort) { struct sockaddr_in sendToAddress; memset((char *) &sendToAddress, 0, sizeof(sendToAddress)); u_long inetAddr = inet_addr(serverIP); sendToAddress.sin_port = htons((u_short) serverPort); sendToAddress.sin_family = AF_INET; sendToAddress.sin_addr.s_addr = inetAddr; dreamSock_SendPacket(socket, theMes->GetSize(), theMes->data, *(struct sockaddr *) &sendToAddress); } else { dreamSock_SendPacket(socket, theMes->GetSize(), theMes->data, myaddress); } // Check if the packet is sequenced theMes->BeginReading();

268

Tutorial 2 / Creating Your Network Library

int type = theMes->ReadByte(); if(type > 0) { outgoingSequence++; } }

dreamClient Constructor
The dreamClient constructor sets up everything for network connection.
dreamClient::dreamClient() { connectionState outgoingSequence incomingSequence incomingAcknowledged droppedPackets init serverPort pingSent ping lastMessageTime next = NULL; }

= DREAMSOCK_DISCONNECTED; = = = = 1; 0; 0; 0;

= false; = = = = 0; 0; 0; 0;

dreamClient Destructor
The destructor does nothing but try to close the socket. It does not matter if the socket is already closed.
dreamClient::~dreamClient() { dreamSock_CloseSocket(socket); }

Initialize Function
This function initializes the client and readies it for a network connection with the server. This function does not send anything to the server, as it only sets up the client. This is a client-side function only.
int dreamClient::Initialize(char *localIP, char *remoteIP, int port) { // Initialize dreamSock if it is not already initialized dreamSock_Initialize(); // Save server's address information for later use

Tutorial 2 / Creating Your Network Library

269

serverPort = port; strcpy(serverIP, remoteIP); LogString("Server's information: IP address: %s, port: %d", serverIP, serverPort); // Create client socket socket = dreamSock_OpenUDPSocket(localIP, 0); // Check that the address is not empty u_long inetAddr = inet_addr(serverIP); if(inetAddr == INADDR_NONE) { return DREAMSOCK_CLIENT_ERROR; } if(socket == DREAMSOCK_INVALID_SOCKET) { return DREAMSOCK_CLIENT_ERROR; } init = true; return 0; }

This function takes three parameters (char *localIP, char *remoteIP, and int port). The first parameter defines the local IP address to use (if we have multiple network interface cards). If this is NULL, the default is used. The second parameter is the servers IP address and the third one is the servers port number. The function then opens a UDP socket using the given information.
socket = dreamSock_OpenUDPSocket(localIP, 0);

After this function successfully returns, the client is ready to start sending data to the server.

Uninitialize Function
This function uninitializes the clients network connection ability.
void dreamClient::Uninitialize(void) { dreamSock_CloseSocket(socket); Reset(); init = false; }

The socket is closed and all the member variables are reset with the Reset function.

270

Tutorial 2 / Creating Your Network Library

Reset Function
The Reset function does what the name tells you it resets the client.
void dreamClient::Reset(void) { connectionState outgoingSequence incomingSequence incomingAcknowledged droppedPackets pingSent ping lastMessageTime next = NULL; }

= DREAMSOCK_DISCONNECTED; = = = = 1; 0; 0; 0;

= 0; = 0; = 0;

All the important member variables are set to their initial state so a connection process can be started all over again if required.

DumpBuffer Function
This functions purpose is to dump the incoming data buffer by reading all the incoming packets and just dumping them (not processing them at all).
void dreamClient::DumpBuffer(void) { char data[1400]; int ret; while((ret = dreamSock_GetPacket(socket, data, NULL)) > 0) { } }

The buffer is dumped by calling dreamSock_GetPacket as long there is anything to read. This way a new connection will not get the packets from any old connections.

System Messages vs. User Messages

Now lets take a moment to talk about system messages and user messages. A system message is a message that is built into the system (in this case, into our network library). A user message, on the other hand, is the users own message, and all the processing for that happens in the final network application, not in the library.

Tutorial 2 / Creating Your Network Library

271

Every messages first byte must contain its type. A message cannot be identified without having a type value attached to it. This is the first thing that we read from a packet, and once we have identified the message type we can move on to the correct direction. We will create five system messages for dreamSock one for connecting to the server, one for disconnecting from the server, one for adding a client, and one for removing a client. The fifth is for pinging clients to calculate the network latency. They are defined as follows:
#define #define #define #define #define DREAMSOCK_MES_CONNECT DREAMSOCK_MES_DISCONNECT DREAMSOCK_MES_ADDCLIENT DREAMSOCK_MES_REMOVECLIENT DREAMSOCK_MES_PING 101 102 103 104 105

Did you notice that the messages type value is actually negative? This is used to tell dreamSock that the message should not be sequenced. And that means that no sequence numbers are attached to the messages. This works for user messages also; just make the type value negative. We will see how all this works later. dreamSock will then process these messages and act accordingly. All the system messages can be reprocessed in the final application. This is very useful as we can then use the same messages for adding a client, for example. You may wonder why we would need to add a client there also, if that is done in the library already. Well, the answer is the games data for each client. Since dreamClient only holds the information required for the network connection, we probably need another structure to hold the game data. But enough about that for now. We will now create some functions for dreamSock that process the system messages. So when we want to connect a server, we just call one function. Simple, but effective.

SendConnect Function
This function sends a connection request to the server.
void dreamClient::SendConnect(char *name) { // Dump buffer so there won't be any old packets to process DumpBuffer(); connectionState = DREAMSOCK_CONNECTING; message.Init(message.outgoingData, sizeof(message.outgoingData)); message.WriteByte(DREAMSOCK_MES_CONNECT); message.WriteString(name); SendPacket(&message); }

272

Tutorial 2 / Creating Your Network Library

The function takes one parameter (char *name), which is used to define the name of the client. The function will add this name to the message and then send it to the server. When the server receives this message, it will add a client to its client list and send an add client message to each connected client. As you can see, we use the dreamMessage class to create the message. First we initialize it by setting the data buffer and its maximum size. Then we write one byte to it, which tells the system the type of the message. Finally we add the name string to the message and send the packet to the server. The SendPacket function is introduced later in this tutorial.
message.Init(message.outgoingData, sizeof(message.outgoingData)); message.WriteByte(DREAMSOCK_MES_CONNECT); message.WriteString(name); SendPacket(&message);

SendDisconnect Function
This function sends a disconnect message to the server.
void dreamClient::SendDisconnect(void) { message.Init(message.outgoingData, sizeof(message.outgoingData)); message.WriteByte(DREAMSOCK_MES_DISCONNECT); SendPacket(&message); Reset(); connectionState = DREAMSOCK_DISCONNECTING; }

This function works the same way as SendConnect does. It sends a system message to the server, telling it that we want to disconnect. After the message is sent, the client resets itself.

SendPing Function
This function sends a ping message to the server.
void dreamClient::SendPing(void) { pingSent = dreamSock_GetCurrentSystemTime(); message.Init(message.outgoingData, sizeof(message.outgoingData)); message.WriteByte(DREAMSOCK_MES_PING); SendPacket(&message); }

Tutorial 2 / Creating Your Network Library

273

This function is run when the server pings the client first. The function responds to the server by pinging back.

ParsePacket Function
This function parses incoming system messages and handles sequenced messages.
void dreamClient::ParsePacket(dreamMessage *mes) { mes->BeginReading(); int type = mes->ReadByte(); // Check if the type is a positive number // = is the packet sequenced if(type > 0) { unsigned short sequence = mes->ReadShort(); unsigned short sequenceAck = mes->ReadShort(); if(sequence <= incomingSequence) { LogString("Client: (sequence: %d <= incoming seq: %d)", sequence, incomingSequence); LogString("Client: Sequence mismatch"); } droppedPackets = sequence incomingSequence + 1; incomingSequence = sequence; incomingAcknowledged = sequenceAck; } // Parse through the system messages switch(type) { case DREAMSOCK_MES_CONNECT: connectionState = DREAMSOCK_CONNECTED; LogString("LIBRARY: Client: got connect confirmation"); break; case DREAMSOCK_MES_DISCONNECT: connectionState = DREAMSOCK_DISCONNECTED; LogString("LIBRARY: Client: got disconnect confirmation"); break; case DREAMSOCK_MES_ADDCLIENT: LogString("LIBRARY: Client: adding a client"); break;

274

Tutorial 2 / Creating Your Network Library

case DREAMSOCK_MES_REMOVECLIENT: LogString("LIBRARY: Client: removing a client"); break; case DREAMSOCK_MES_PING: SendPing(); break; } }

This function takes one parameter (dreamMessage *mes). It is a pointer to the message to parse. Parsing begins by reading the type of the message from the packet:
mes->BeginReading(); int type = mes->ReadByte();

Then we check if the type value is a positive number in other words, whether the message is sequenced. If it is, we read the sequence numbers from the packet. First is the remote hosts outgoing sequence number and then its incoming sequence number. We call them incoming sequence and acknowledged sequence. Sound confusing? Well, at first it may be so, but if you really think about it for a while, it is quite simple.
if(type > 0) { unsigned short sequence unsigned short sequenceAck

= mes->ReadShort(); = mes->ReadShort();

Then we parse the system messages. On the client side, there is nothing important going on here, if you do not count the connection state changes and ping response. These are mainly confirmations only.
// Parse through the system messages switch(type) { case DREAMSOCK_MES_CONNECT: connectionState = DREAMSOCK_CONNECTED;

Tutorial 2 / Creating Your Network Library

275

LogString("LIBRARY: Client: got connect confirmation"); break; case DREAMSOCK_MES_DISCONNECT: connectionState = DREAMSOCK_DISCONNECTED; LogString("LIBRARY: Client: got disconnect confirmation"); break; case DREAMSOCK_MES_ADDCLIENT: LogString("LIBRARY: Client: adding a client"); break; case DREAMSOCK_MES_REMOVECLIENT: LogString("LIBRARY: Client: removing a client"); break; case DREAMSOCK_MES_PING: SendPing(); break; }

GetPacket Function
This function reads packets from the server and then moves them on to parsing.
int dreamClient::GetPacket(char *data, struct sockaddr *from) { // Check if the client is set up or if it is disconnecting if(!socket) return 0; int ret; dreamMessage mes; mes.Init(data, sizeof(data)); ret = dreamSock_GetPacket(socket, mes.data, from); if(ret <= 0) return 0; mes.SetSize(ret); // Parse system messages ParsePacket(&mes); return ret; }

This function takes two parameters (char *data, struct sockaddr *from). The first one is a pointer to the data buffer, which will be filled by this function. The other is a socket address pointer,

276

Tutorial 2 / Creating Your Network Library

which will also be filled by this function. So when we receive a packet, we also get the address from where it came. This information can be ignored, but it is useful to know. We read the data into a dreamMessage so we can then parse it in the ParsePacket function. If dreamSock_GetPacket returns a negative number or 0, we do not process the packet any further, because there is no packet to process.
dreamMessage mes; mes.Init(data, sizeof(data)); ret = dreamSock_GetPacket(socket, mes.data, from); if(ret <= 0) return 0; mes.SetSize(ret); // Parse system messages ParsePacket(&mes);

SendPacket Function (Internal Message)

There are two versions of the SendPacket function. This one sends the internal message of dreamClient (message member variable). This function can be run on both the server side and the client side. In fact, the server sends packets to clients with this very function.
void dreamClient::SendPacket(void) { // Check that everything is set up if(!socket || connectionState == DREAMSOCK_DISCONNECTED) { LogString("SendPacket error: Could not send because the client is disconnected"); return; } // If the message overflowed, do not send it if(message.GetOverFlow()) { LogString("SendPacket error: Could not send because the buffer overflowed"); return; } // Check if serverPort is set. If it is, we are a client sending to // the server. Otherwise we are a server sending to a client. if(serverPort) { struct sockaddr_in sendToAddress;

Tutorial 2 / Creating Your Network Library

277

memset((char *) &sendToAddress, 0, sizeof(sendToAddress)); u_long inetAddr = inet_addr(serverIP); sendToAddress.sin_port = htons((u_short) serverPort); sendToAddress.sin_family = AF_INET; sendToAddress.sin_addr.s_addr = inetAddr; dreamSock_SendPacket(socket, message.GetSize(), message.data, *(struct sockaddr *) &sendToAddress); } else { dreamSock_SendPacket(socket, message.GetSize(), message.data, myaddress); } // Check if the packet is sequenced message.BeginReading(); int type = message.ReadByte(); if(type > 0) { outgoingSequence++; } }

First we do some checking that it is okay to send data. We check that we have a socket and that we are not disconnected. If the message overflowed, we will not send it because some parts of the message are missing.
// Check that everything is set up if(!socket || connectionState == DREAMSOCK_DISCONNECTED) { LogString("SendPacket error: Could not send because the client is disconnected"); return; } // If the message overflowed, do not send it if(message.GetOverFlow()) { LogString("SendPacket error: Could not send because the buffer overflowed"); return; }

Then we check if we are a server trying to send to a client or vice versa. An easy way to check this is to see if the server port number is set. If it is, we are a client sending to the server, because only clients set the serverPort variable. So we fill in the servers address information and send the packet to that address. On the server side we just

278

Tutorial 2 / Creating Your Network Library

use the clients address, which is set when the client connects to the server. Notice that the data we send comes from the dreamClients message member variable.
if(serverPort) { struct sockaddr_in sendToAddress; memset((char *) &sendToAddress, 0, sizeof(sendToAddress)); u_long inetAddr = inet_addr(serverIP); sendToAddress.sin_port = htons((u_short) serverPort); sendToAddress.sin_family = AF_INET; sendToAddress.sin_addr.s_addr = inetAddr; dreamSock_SendPacket(socket, message.GetSize(), message.data, *(struct sockaddr *) &sendToAddress); } else { dreamSock_SendPacket(socket, message.GetSize(), message.data, myaddress); }

Finally we increase the outgoing sequence number, if the message is sequenced.

// Check if the packet is sequenced message.BeginReading(); int type = message.ReadByte(); if(type > 0) { outgoingSequence++; }

SendPacket Function (External Message)

This function sends an external message given as a parameter. It can be run on both the server side and the client side.
void dreamClient::SendPacket(dreamMessage *theMes) { // Check that everything is set up if(!socket || connectionState == DREAMSOCK_DISCONNECTED) { LogString("SendPacket error: Could not send because the client is disconnected"); return; } // If the message overflowed, do not send it if(theMes->GetOverFlow())

Tutorial 2 / Creating Your Network Library

279

{ LogString("SendPacket error: Could not send because the buffer overflowed"); return; } // Check if serverPort is set. If it is, we are a client sending to // the server. Otherwise we are a server sending to a client. if(serverPort) { struct sockaddr_in sendToAddress; memset((char *) &sendToAddress, 0, sizeof(sendToAddress)); u_long inetAddr = inet_addr(serverIP); sendToAddress.sin_port = htons((u_short) serverPort); sendToAddress.sin_family = AF_INET; sendToAddress.sin_addr.s_addr = inetAddr; dreamSock_SendPacket(socket, theMes->GetSize(), theMes->data, *(struct sockaddr *) &sendToAddress); } else { dreamSock_SendPacket(socket, theMes->GetSize(), theMes->data, myaddress); } // Check if the packet is sequenced theMes->BeginReading(); int type = theMes->ReadByte(); if(type > 0) { outgoingSequence++; } }

This function works exactly the same way as the internal message version. The only difference is that this function takes one parameter (dreamMessage *theMes), which is the message to send.
dreamSock_SendPacket(socket, theMes->GetSize(), theMes->data, myaddress);

dreamClient Summary
That concludes the dreamClient methods. Now we have the client side fully ready, but we cannot do anything with it yet, because the server side is not ready. So what are we waiting for? Lets go finish our library!

280

Tutorial 2 / Creating Your Network Library

dreamServer Class Member Variables

Here are the member variables for dreamServer and their purposes.
private: dreamClient *clientList; int SOCKET int bool port; socket; runningIndex; init; // Port // Socket // Running index numbers for new clients

The dreamClient clientList is a linked list of all the connected clients. If it is NULL, no clients are connected. The integer port stores the port number of the server. The SOCKET socket holds the socket of the server. All network communication happens through this socket. The integer runningIndex is an index number for incoming clients. Every time a client connects to the server, this number is increased and given to the client as its index number. The number is unique for every client. The boolean init tells us whether the server has been initialized.

dreamServer Class Functionality

The dreamServer class is used only on the server side. This class handles the incoming clients and their network connections. Here are all the methods of dreamServer:
dreamServer::dreamServer() { init = false; port runningIndex socket clientList } dreamServer::~dreamServer() { dreamClient *list = clientList; dreamClient *next; while(list != NULL) { next = list->next; if(list) { free(list); = = = = 0; 1; 0; NULL;

Tutorial 2 / Creating Your Network Library

281

} list = next; } clientList = NULL; dreamSock_CloseSocket(socket); } int dreamServer::Initialize(char *localIP, int serverPort) { // Initialize dreamSock if it is not already initialized dreamSock_Initialize(); // Store the server IP and port for later use port = serverPort; // Create server socket socket = dreamSock_OpenUDPSocket(localIP, port); if(socket == DREAMSOCK_INVALID_SOCKET) { return DREAMSOCK_SERVER_ERROR; } init = true; return 0; } void dreamServer::Uninitialize(void) { dreamSock_CloseSocket(socket); init = false; } void dreamServer::SendAddClient(dreamClient *newClient) { // Send connection confirmation newClient->message.Init(newClient->message.outgoingData, sizeof(newClient->message.outgoingData)); newClient->message.WriteByte(DREAMSOCK_MES_CONNECT); newClient->SendPacket(); // Send 'Add client' message to every client dreamClient *client = clientList; // First inform the new client of the other clients for(; client != NULL; client = client->next) { // type

282

Tutorial 2 / Creating Your Network Library

newClient->message.Init(newClient->message.outgoingData, sizeof(newClient->message.outgoingData)); newClient->message.WriteByte(DREAMSOCK_MES_ADDCLIENT); // type if(client == newClient) { newClient->message.WriteByte(1); // local client newClient->message.WriteByte(client->GetIndex()); newClient->message.WriteString(client->GetName()); } else { newClient->message.WriteByte(0); // not local client newClient->message.WriteByte(client->GetIndex()); newClient->message.WriteString(client->GetName()); } newClient->SendPacket(); } // Then tell the others about the new client for(client = clientList; client != NULL; client = client->next) { if(client == newClient) continue; client->message.Init(client->message.outgoingData, sizeof(client->message.outgoingData)); client->message.WriteByte(DREAMSOCK_MES_ADDCLIENT); // type client->message.WriteByte(0); client->message.WriteByte(newClient->GetIndex()); client->message.WriteString(newClient->GetName()); client->SendPacket(); } } void dreamServer::SendRemoveClient(dreamClient *client) { int index = client->GetIndex(); // Send 'Remove client' message to every client dreamClient *list = clientList; for(; list != NULL; list = list->next) { list->message.Init(list->message.outgoingData, sizeof(list->message.outgoingData)); list->message.WriteByte(DREAMSOCK_MES_REMOVECLIENT); list->message.WriteByte(index); // type // index

Tutorial 2 / Creating Your Network Library

283

} SendPackets(); // Send disconnection confirmation client->message.Init(client->message.outgoingData, sizeof(client->message.outgoingData)); client->message.WriteByte(DREAMSOCK_MES_DISCONNECT); client->SendPacket(); } void dreamServer::SendPing(void) { // Send ping message to every client dreamClient *list = clientList; for(; list != NULL; list = list->next) { list->SendPing(); } } void dreamServer::AddClient(struct sockaddr *address, char *name) { // First get a pointer to the beginning of client list dreamClient *list = clientList; dreamClient *prev; dreamClient *newClient; LogString("LIB: Adding client, index %d", runningIndex); // No clients yet, adding the first one if(clientList == NULL) { LogString("LIB: Server: Adding first client"); clientList = (dreamClient *) calloc(1, sizeof(dreamClient)); clientList->SetSocket(socket); clientList->SetSocketAddress(address); clientList->SetConnectionState(DREAMSOCK_CONNECTING); clientList->SetOutgoingSequence(1); clientList->SetIncomingSequence(0); clientList->SetIncomingAcknowledged(0); clientList->SetIndex(runningIndex); clientList->SetName(name); clientList->next = NULL; newClient = clientList; } else {

284

Tutorial 2 / Creating Your Network Library

LogString("LIB: Server: Adding another client"); prev = list; list = clientList->next; while(list != NULL) { prev = list; list = list->next; } list = (dreamClient *) calloc(1, sizeof(dreamClient)); list->SetSocket(socket); list->SetSocketAddress(address); list->SetConnectionState(DREAMSOCK_CONNECTING); list->SetOutgoingSequence(1); list->SetIncomingSequence(0); list->SetIncomingAcknowledged(0); list->SetIndex(runningIndex); list->SetName(name); list->next = NULL; prev->next = list; newClient = list; } runningIndex++; SendAddClient(newClient); } void dreamServer::RemoveClient(dreamClient *client) { dreamClient *list = NULL; dreamClient *prev = NULL; dreamClient *next = NULL; int index = client->GetIndex(); LogString("LIB: Removing client with index %d", index); SendRemoveClient(client); for(list = clientList; list != NULL; list = list->next) { if(client == list) { if(prev != NULL) { prev->next = client->next;

Tutorial 2 / Creating Your Network Library

285

} break; } prev = list; } if(client == clientList) { LogString("LIB: Server: removing first client in list"); if(list) next = list->next; if(client) free(client); client = NULL; clientList = next; } else { LogString("LIB: Server: removing a client"); if(list) next = list->next; if(client) free(client); client = next; } } void dreamServer::ParsePacket(dreamMessage *mes, struct sockaddr *address) { mes->BeginReading(); int type = mes->ReadByte(); // Find the correct client by comparing addresses dreamClient *clList = clientList; // If we do not have clients yet, skip to message type checking if(clList != NULL) { for(; clList != NULL; clList = clList->next) { if(memcmp(clList->GetSocketAddress(), address, sizeof(address)) == 0) { break; } } if(clList != NULL) { clList->SetLastMessageTime(dreamSock_ GetCurrentSystemTime());

286

Tutorial 2 / Creating Your Network Library

// Check if the type is a positive number // -> is the packet sequenced if(type > 0) { unsigned short sequence = mes->ReadShort(); unsigned short sequenceAck = mes->ReadShort(); if(sequence <= clList->GetIncomingSequence()) { LogString("LIB: Server: Sequence mismatch (sequence: %ld <= incoming seq: %ld)", sequence, clList->GetIncomingSequence()); } clList->SetDroppedPackets(sequence clList->GetIncomingSequence() 1); clList->SetIncomingSequence(sequence); clList->SetIncomingAcknowledged(sequenceAck); } // Wait for one message before setting state to connected if(clList->GetConnectionState() == DREAMSOCK_CONNECTING) clList->SetConnectionState(DREAMSOCK_CONNECTED); } } // Parse through the system messages switch(type) { case DREAMSOCK_MES_CONNECT: AddClient(address, mes->ReadString()); LogString("LIBRARY: Server: a client connected successfully"); break; case DREAMSOCK_MES_DISCONNECT: if(clList == NULL) break; } } RemoveClient(clList); LogString("LIBRARY: Server: a client disconnected"); break; case DREAMSOCK_MES_PING: clList->SetPing(dreamSock_GetCurrentSystemTime() clList->GetPingSent()); break; int dreamServer::CheckForTimeout(char *data, struct sockaddr *from) { int currentTime = dreamSock_GetCurrentSystemTime();

Tutorial 2 / Creating Your Network Library

287

dreamClient *clList = clientList; dreamClient *next; for(; clList != NULL;) { next = clList->next; // Don't timeout when connecting if(clList->GetConnectionState() == DREAMSOCK_CONNECTING) { clList = next; continue; } // Check if the client has been silent for 30 seconds // If yes, assume crashed and remove the client if(currentTime clList->GetLastMessageTime() > 30000) { LogString("Client timeout, disconnecting (%d %d = %d)", currentTime, clList->GetLastMessageTime(), currentTime clList->GetLastMessageTime()); // Build a 'fake' message so the application will also // receive notification of a client disconnecting dreamMessage mes; mes.Init(data, sizeof(data)); mes.WriteByte(DREAMSOCK_MES_DISCONNECT); *(struct sockaddr *) from = *clList->GetSocketAddress(); RemoveClient(clList); return mes.GetSize(); } clList = next; } return 0; } int dreamServer::GetPacket(char *data, struct sockaddr *from) { // Check if the server is set up if(!socket) return 0; // Check for timeout int timeout = CheckForTimeout(data, from); if(timeout) return timeout;

288

Tutorial 2 / Creating Your Network Library

// Wait for a while or incoming data int maxfd = socket; fd_set allset; struct timeval waittime; waittime.tv_sec = 10 / 1000; waittime.tv_usec = (10 % 1000) * 1000; FD_ZERO(&allset); FD_SET(socket, &allset); fd_set reading = allset; int nready = select(maxfd + 1, &reading, NULL, NULL, &waittime); if(!nready) return 0; // Read data of the socket int ret = 0; dreamMessage mes; mes.Init(data, sizeof(data)); ret = dreamSock_GetPacket(socket, mes.data, from); if(ret <= 0) return 0; mes.SetSize(ret); // Parse system messages ParsePacket(&mes, from); return ret; } void dreamServer::SendPackets(void) { // Check if the server is set up if(!socket) return; dreamClient *clList = clientList; for(; clList != NULL; clList = clList->next) { if(clList->message.GetSize() == 0) continue; clList->SendPacket(); } }

Tutorial 2 / Creating Your Network Library

289

dreamServer Constructor
The dreamServer constructor makes everything ready for clients to join the server.
dreamServer::dreamServer() { init = false; port runningIndex socket clientList } = = = = 0; 1; 0; NULL;

dreamServer Destructor
The dreamServer destructor frees all the allocated memory, if any.
dreamServer::~dreamServer() { dreamClient *list = clientList; dreamClient *next; while(list != NULL) { next = list->next; if(list) { free(list); } list = next; } clientList = NULL; dreamSock_CloseSocket(socket); }

The client list is parsed through and all the allocated memory is freed. The server socket is also closed. Note that before we free the memory of a client in the list, we must store its next pointer to an external pointer, because the original is lost once the memory is freed. Then we start it all over again.
next = list->next; if(list) { free(list);

290

Tutorial 2 / Creating Your Network Library

} list = next;

Initialize Function
This function initializes dreamServer by creating the socket so clients start connecting to it.
int dreamServer::Initialize(char *localIP, int serverPort) { // Initialize dreamSock if it is not already initialized dreamSock_Initialize(); // Store the server IP and port for later use port = serverPort; // Create server socket socket = dreamSock_OpenUDPSocket(localIP, port); if(socket == DREAMSOCK_INVALID_SOCKET) { return DREAMSOCK_SERVER_ERROR; } init = true; return 0; }

The function takes two parameters (char *localIP and int serverPort). The first one defines the local IP address we want to use. If this is NULL, the default is used. The latter parameter sets the server port.

Uninitialize Function
Here we uninitialize dreamServer by closing the socket.
void dreamServer::Uninitialize(void) { dreamSock_CloseSocket(socket); init = false; }

SendAddClient Function
This function sends an add client message to each connected client and a connection confirmation to the new client.
void dreamServer::SendAddClient(dreamClient *newClient) {

Tutorial 2 / Creating Your Network Library

291

// Send connection confirmation newClient->message.Init(newClient->message.outgoingData, sizeof(newClient->message.outgoingData)); newClient->message.WriteByte(DREAMSOCK_MES_CONNECT); newClient->SendPacket(); // Send 'Add client' message to every client dreamClient *client = clientList; // First inform the new client of the other clients for(; client != NULL; client = client->next) { newClient->message.Init(newClient->message.outgoingData, sizeof(newClient->message.outgoingData)); newClient->message.WriteByte(DREAMSOCK_MES_ADDCLIENT); // type if(client == newClient) { newClient->message.WriteByte(1); // local client newClient->message.WriteByte(client->GetIndex()); newClient->message.WriteString(client->GetName()); } else { newClient->message.WriteByte(0); // not local client newClient->message.WriteByte(client->GetIndex()); newClient->message.WriteString(client->GetName()); } newClient->SendPacket(); } // Then tell the others about the new client for(client = clientList; client != NULL; client = client->next) { if(client == newClient) continue; client->message.Init(client->message.outgoingData, sizeof(client->message.outgoingData)); client->message.WriteByte(DREAMSOCK_MES_ADDCLIENT); // type client->message.WriteByte(0); client->message.WriteByte(newClient->GetIndex()); client->message.WriteString(newClient->GetName()); client->SendPacket(); } } // type

292

Tutorial 2 / Creating Your Network Library

This function takes one parameter (dreamClient *newClient). This is a pointer to the client that connected the server and triggered this function to be run. First we send a connect confirmation to the new client:
newClient->message.Init(newClient->message.outgoingData, sizeof(newClient->message.outgoingData)); newClient->message.WriteByte(DREAMSOCK_MES_CONNECT); newClient->SendPacket(); // type

Then we send an add client message to all the connected clients (including the new client). This tells the clients to add a client to the game. With this message we tell the clients their index numbers and send a local client flag. This means that the new client knows to assign a local client pointer to its own client list.
if(client == newClient) { newClient->message.WriteByte(1); // local client newClient->message.WriteByte(client->GetIndex()); newClient->message.WriteString(client->GetName()); } else { newClient->message.WriteByte(0); // not local client newClient->message.WriteByte(client->GetIndex()); newClient->message.WriteString(client->GetName()); }

SendRemoveClient Function
Here we send a remove client message to each client.
void dreamServer::SendRemoveClient(dreamClient *client) { int index = client->GetIndex(); // Send 'Remove client' message to every client dreamClient *list = clientList; for(; list != NULL; list = list->next) { list->message.Init(list->message.outgoingData, sizeof(list->message.outgoingData)); list->message.WriteByte(DREAMSOCK_MES_REMOVECLIENT); list->message.WriteByte(index); } SendPackets(); // Send disconnection confirmation // type // index

Tutorial 2 / Creating Your Network Library

293

client->message.Init(client->message.outgoingData, sizeof(client->message.outgoingData)); client->message.WriteByte(DREAMSOCK_MES_DISCONNECT); client->SendPacket(); }

This function takes one parameter (dreamClient *client), which is a pointer to the client to remove. First we tell all the clients to remove the client and then we send the disconnecting client a confirmation of disconnection.

SendPing Function
This function pings the clients to calculate the network latency.
void dreamServer::SendPing(void) { // Send ping message to every client dreamClient *list = clientList; for(; list != NULL; list = list->next) { list->SendPing(); } }

What this function actually does is call the SendPing function of each client. The server then sends the ping message to each client.

AddClient Function
This function adds a client to the servers client list. The function is run when a client connects to the server.
void dreamServer::AddClient(struct sockaddr *address, char *name) { // First get a pointer to the beginning of client list dreamClient *list = clientList; dreamClient *prev; dreamClient *newClient; LogString("LIB: Adding client, index %d", runningIndex); // No clients yet, adding the first one if(clientList == NULL) { LogString("LIB: Server: Adding first client"); clientList = (dreamClient *) calloc(1, sizeof(dreamClient)); clientList->SetSocket(socket); clientList->SetSocketAddress(address);

294

Tutorial 2 / Creating Your Network Library

clientList->SetConnectionState(DREAMSOCK_CONNECTING); clientList->SetOutgoingSequence(1); clientList->SetIncomingSequence(0); clientList->SetIncomingAcknowledged(0); clientList->SetIndex(runningIndex); clientList->SetName(name); clientList->next = NULL; newClient = clientList; } else { LogString("LIB: Server: Adding another client"); prev = list; list = clientList->next; while(list != NULL) { prev = list; list = list->next; } list = (dreamClient *) calloc(1, sizeof(dreamClient)); list->SetSocket(socket); list->SetSocketAddress(address); list->SetConnectionState(DREAMSOCK_CONNECTING); list->SetOutgoingSequence(1); list->SetIncomingSequence(0); list->SetIncomingAcknowledged(0); list->SetIndex(runningIndex); list->SetName(name); list->next = NULL; prev->next = list; newClient = list; } runningIndex++; SendAddClient(newClient); }

The function takes two parameters (struct sockaddr *address and char *name). The first one is the socket address of the new client, and the second one is the clients name. This information is then added to the clients own structure. After the server has added the information to its own list, it sends the add client message to all the connected clients.

Tutorial 2 / Creating Your Network Library

295

The function checks if the client list is empty; if it is, it creates the first entry for it. All the same information is stored for each client. The following piece of code shows the first entry being added.
// No clients yet, adding the first one if(clientList == NULL) { LogString("LIB: Server: Adding first client"); clientList = (dreamClient *) calloc(1, sizeof(dreamClient)); clientList->SetSocket(socket); clientList->SetSocketAddress(address); clientList->SetConnectionState(DREAMSOCK_CONNECTING); clientList->SetOutgoingSequence(1); clientList->SetIncomingSequence(0); clientList->SetIncomingAcknowledged(0); clientList->SetIndex(runningIndex); clientList->SetName(name); clientList->next = NULL; newClient = clientList; }

RemoveClient Function
This function removes a client from the client list.
void dreamServer::RemoveClient(dreamClient *client) { dreamClient *list = NULL; dreamClient *prev = NULL; dreamClient *next = NULL; int index = client->GetIndex(); LogString("LIB: Removing client with index %d", index); SendRemoveClient(client); for(list = clientList; list != NULL; list = list->next) { if(client == list) { if(prev != NULL) { prev->next = client->next; } break; } prev = list;

296

Tutorial 2 / Creating Your Network Library

} if(client == clientList) { LogString("LIB: Server: removing first client in list"); if(list) next = list->next; if(client) free(client); client = NULL; clientList = next; } else { LogString("LIB: Server: removing a client"); if(list) next = list->next; if(client) free(client); client = next; } }

The function takes one parameter (dreamClient *client). This is a pointer to the client to remove. The server first sends each client a message to remove the client and then removes the client itself. We must update the client list so that the previous clients next pointer does not point to the removed client anymore. We want it to point to the next client in the list.
for(list = clientList; list != NULL; list = list->next) { if(client == list) { if(prev != NULL) { prev->next = client->next; } break; } prev = list; }

We also need to make sure that if the first entry on the list is removed, the next one in the list becomes the new first entry.
if(client == clientList) { LogString("LIB: Server: removing first client in list"); if(list) next = list->next;

Tutorial 2 / Creating Your Network Library

297

if(client) free(client); client = NULL; clientList = next; } else { LogString("LIB: Server: removing a client"); if(list) next = list->next; if(client) free(client); client = next; }

ParsePacket Function
This function parses the servers system messages and handles sequenced messages.
void dreamServer::ParsePacket(dreamMessage *mes, struct sockaddr *address) { mes->BeginReading(); int type = mes->ReadByte(); // Find the correct client by comparing addresses dreamClient *clList = clientList; // If we do not have clients yet, skip to message type checking if(clList != NULL) { for(; clList != NULL; clList = clList->next) { if(memcmp(clList->GetSocketAddress(), address, sizeof(address)) == 0) { break; } } if(clList != NULL) { clList->SetLastMessageTime(dreamSock_ GetCurrentSystemTime()); // Check if the type is a positive number // -> is the packet sequenced if(type > 0) { unsigned short sequence = mes->ReadShort(); unsigned short sequenceAck = mes->ReadShort(); if(sequence <= clList->GetIncomingSequence()) {

298

Tutorial 2 / Creating Your Network Library

LogString("LIB: Server: Sequence mismatch (sequence: %ld <= incoming seq: %ld)", sequence, clList->GetIncomingSequence()); } clList->SetDroppedPackets(sequence clList->GetIncomingSequence() 1); clList->SetIncomingSequence(sequence); clList->SetIncomingAcknowledged(sequenceAck); } // Wait for one message before setting state to connected if(clList->GetConnectionState() == DREAMSOCK_CONNECTING) clList->SetConnectionState(DREAMSOCK_CONNECTED); } } // Parse through the system messages switch(type) { case DREAMSOCK_MES_CONNECT: AddClient(address, mes->ReadString()); LogString("LIBRARY: Server: a client connected successfully"); break; case DREAMSOCK_MES_DISCONNECT: if(clList == NULL) break; RemoveClient(clList); LogString("LIBRARY: Server: a client disconnected"); break; case DREAMSOCK_MES_PING: clList->SetPing(dreamSock_GetCurrentSystemTime() clList->GetPingSent()); break; } }

The function takes two parameters (dreamMessage *mes and struct sockaddr *address). The first one is a pointer to the message to parse. The second one is the socket address of the client from which we got the message. We compare this address to the ones in the client list, and if it matches one there, the message is from an old client. If it does not find a match, the client is new. Note that we just break the search loop if we find a match, and then we use the last pointer as our client pointer.

Tutorial 2 / Creating Your Network Library

299

for(; clList != NULL; clList = clList->next) { if(memcmp(clList->GetSocketAddress(), address, sizeof(address)) == 0) { break; } }

The sequences are processed just like on the client side. We also control the connection state of a client here. When a client connects, its connection state is set to connecting. When we get a message from this client again, we set the connection state to connected. So we wait for one message from the client before we give it a connected state. This is to make sure we do not time out the client before it has even had time to connect properly.
// Wait for one message before setting state to connected if(clList->GetConnectionState() == DREAMSOCK_CONNECTING) clList->SetConnectionState(DREAMSOCK_CONNECTED);

Finally, we parse the system messages. The ping value is calculated by subtracting the ping sent time from the current time.
// Parse through the system messages switch(type) { case DREAMSOCK_MES_CONNECT: AddClient(address, mes->ReadString()); LogString("LIBRARY: Server: a client connected successfully"); break; case DREAMSOCK_MES_DISCONNECT: if(clList == NULL) break; RemoveClient(clList); LogString("LIBRARY: Server: a client disconnected"); break; case DREAMSOCK_MES_PING: clList->SetPing(dreamSock_GetCurrentSystemTime() clList->GetPingSent()); break; }

CheckForTimeout Function
Here we check if a client times out. This means that the server has not received any message from the client in a certain amount of time.
int dreamServer::CheckForTimeout(char *data, struct sockaddr *from) {

300

Tutorial 2 / Creating Your Network Library

int currentTime = dreamSock_GetCurrentSystemTime(); dreamClient *clList = clientList; dreamClient *next; for(; clList != NULL;) { next = clList->next; // Don't timeout when connecting if(clList->GetConnectionState() == DREAMSOCK_CONNECTING) { clList = next; continue; } // Check if the client has been silent for 30 seconds // If yes, assume crashed and remove the client if(currentTime clList->GetLastMessageTime() > 30000) { LogString("Client timeout, disconnecting (%d %d = %d)", currentTime, clList->GetLastMessageTime(), currentTime clList->GetLastMessageTime()); // Build a 'fake' message so the application will also // receive notification of a client disconnecting dreamMessage mes; mes.Init(data, sizeof(data)); mes.WriteByte(DREAMSOCK_MES_DISCONNECT); *(struct sockaddr *) from = *clList->GetSocketAddress(); RemoveClient(clList); return mes.GetSize(); } clList = next; } return 0; }

This function takes two parameters (char *data and struct sockaddr *from). These parameters are used to create a fake message to tell everybody a client timed out. Actually, all they will know is that the client disconnected, because this fake message makes the server believe it got a disconnect message from the client that timed out. We should not time out while the client is connecting.
// Don't timeout when connecting if(clList->GetConnectionState() == DREAMSOCK_CONNECTING)

Tutorial 2 / Creating Your Network Library

301

{ clList = next; continue; }

Here we check if the client has been silent for 30 seconds; if it has, we create the fake message to make the server remove the client.
// Check if the client has been silent for 30 seconds // If yes, assume crashed and remove the client if(currentTime clList->GetLastMessageTime() > 30000) { LogString("Client timeout, disconnecting (%d %d = %d)", currentTime, clList->GetLastMessageTime(), currentTime clList->GetLastMessageTime()); // Build a 'fake' message so the application will also // receive notification of a client disconnecting dreamMessage mes; mes.Init(data, sizeof(data)); mes.WriteByte(DREAMSOCK_MES_DISCONNECT); *(struct sockaddr *) from = *clList->GetSocketAddress(); RemoveClient(clList); return mes.GetSize(); }

GetPacket Function
This function receives packets from the clients.
int dreamServer::GetPacket(char *data, struct sockaddr *from) { // Check if the server is set up if(!socket) return 0; // Check for timeout int timeout = CheckForTimeout(data, from); if(timeout) return timeout; // Wait for a while or incoming data int maxfd = socket; fd_set allset; struct timeval waittime; waittime.tv_sec = 10 / 1000; waittime.tv_usec = (10 % 1000) * 1000; FD_ZERO(&allset);

302

Tutorial 2 / Creating Your Network Library

FD_SET(socket, &allset); fd_set reading = allset; int nready = select(maxfd + 1, &reading, NULL, NULL, &waittime); if(!nready) return 0; // Read data of the socket int ret = 0; dreamMessage mes; mes.Init(data, sizeof(data)); ret = dreamSock_GetPacket(socket, mes.data, from); if(ret <= 0) return 0; mes.SetSize(ret); // Parse system messages ParsePacket(&mes, from); return ret; }

The function takes two parameters (char *data and struct sockaddr *from). The first one is a pointer to the incoming data buffer, which holds the actual data that comes from the client. The second one is the socket address of the client from which we got the data. First we check for timeouts, and then we start to listen to the servers socket. We use the select function to save processor time by waiting for incoming data. If no data is received in 10 milliseconds, the function returns. If we did not do this, the server would consume all the processor time, and the server machine could not run anything else efficiently.
// Wait for a while or incoming data int maxfd = socket; fd_set allset; struct timeval waittime; waittime.tv_sec = 10 / 1000; waittime.tv_usec = (10 % 1000) * 1000; FD_ZERO(&allset); FD_SET(socket, &allset); fd_set reading = allset;

Tutorial 2 / Creating Your Network Library

303

int nready = select(maxfd + 1, &reading, NULL, NULL, &waittime); if(!nready) return 0;

After that we read the sockets data and parse for system messages.
// Read data of the socket int ret = 0; dreamMessage mes; mes.Init(data, sizeof(data)); ret = dreamSock_GetPacket(socket, mes.data, from); if(ret <= 0) return 0; mes.SetSize(ret); // Parse system messages ParsePacket(&mes, from); return ret;

SendPackets Function
This function sends the internal messages of all the clients in the client list. The server should write the messages into these internal messages inside dreamClient and then send them all at once with this function.
void dreamServer::SendPackets(void) { // Check if the server is set up if(!socket) return; dreamClient *clList = clientList; for(; clList != NULL; clList = clList->next) { if(clList->message.GetSize() == 0) continue; clList->SendPacket(); } }

This function goes through all of the clients in the servers client list and sends off all of their internal messages by using dreamClients SendPacket function.

304

Tutorial 2 / Creating Your Network Library

dreamServer Summary
We have now created dreamServer, which processes all the clients on the server side, and we know how to send to clients and how to receive messages from them.

Summary
That concludes our network library tutorial. We created the dreamMessage class to build and read network messages, the dreamClient class to handle client-side and server-side network connections, and the dreamServer class to handle the clients on the server side. We now have a fully working library ready, and we can start writing our multiplayer game.

Tutorial 3

Creating a Basic Network Application with dreamSock

Introduction
In this tutorial we learn how to create a basic network application using dreamSock. This application will be the sign-in part of our tutorial game. We do not talk about the actual sign-in process that happens with MySQL but instead learn how to create a server application and a client application that work together over a network. We have our network library dreamSock ready, so we will be using that for the network communications.

305

306

Tutorial 3 / Creating a Basic Network Application with dreamSock

Planning the Functionality

We are now going to create a basic dreamSock network application in the form of the sign-in server of our tutorial game. The final game will have three such applications: sign-in, lobby, and the game itself. So when the game is ready, the server will actually be running more than one service. Sign-in and lobby services are running all the time, and the game services are run when needed (when games are created). All these parts of the application work exactly the same way; only the messages are different. On both the server and client we have a loop that reads the socket every frame. Messages are sent as needed.

Catching Exceptions
Catching exceptions from the loops we run is a very, very good idea, especially on the server. Imagine a client connecting to the server and then somehow crashing and confusing the server code (hopefully our code is good enough not to do that though). It is possible that the server does something that it should not do, such as trying to use some unallocated memory. Without catching exceptions, the server would crash completely, leaving the clients hanging. But if we use try and catch statements to catch this unplanned exception, the server code will ignore the malfunction and move on. The parts of the code that depended on what the server was trying to do will also fail, but the server will keep running and serving all the other clients. Also, when a client catches an exception, it can safely disconnect from the server before it is shut down.

Creating a Basic Client Application

We will now create a basic client application, which is a normal Windows dialog-based program. We have the dialogs ready, so all we need to do now is link our network library with the program we are creating. To do this, select Project, Settings from the main Visual Studio menu. Then select the Link tab in the dialog that appears in the middle of the screen. Ensure that the top-left drop-down box is set to All Configurations, then add the following library before or after the other libraries in the Object/library modules edit box:
dreamSock.lib

Tutorial 3 / Creating a Basic Network Application with dreamSock

307

When you enter these libraries, the dialog should look like Figure 1.
Figure 1

Once this is done, we can make new source code files called main.cpp, signin.cpp, main.h, common.h, network.h, and signin.h. We will write our client code into these files.

signin.h File
It is a good idea to start with header files, because that is where our data structures are, and we need those before we can do anything with the application. So lets take a look at the signin.h file that contains class CSignin.
#ifndef __SIGNIN_H__ #define __SIGNIN_H__ #include "network.h" #include "main.h" #define #define #define #define SIGNIN_RESULT_ACCEPTED SIGNIN_RESULT_USERNAMEBAD SIGNIN_RESULT_PASSWORDBAD SIGNIN_RESULT_MYSQLERROR 200 201 202 203

class CSignin { private: dreamClient clientLoginData clientLoginData

networkClient; clientList; *localClient;

// Pointer to the local client // in the client list

public: CSignin();

308

Tutorial 3 / Creating a Basic Network Application with dreamSock

dreamClient clientLoginData

GetNetworkClient(void) {return networkClient;} GetLocalClient(void) {return localClient;}

void ReadPackets(void); void AddClient(int local, int index, char *name); void RemoveClient(int index); void RemoveClients(void); void SendSignIn(char *nickname, char *firstname, char *surname, int age, char *gender, char *password, char *email); void SendKeepAlive(void); void Connect(char *name); void Disconnect(void); void RunNetwork(int msec); }; extern CSignin Signin; #endif

CSignin Class
Every dreamSock network application should have a class similar to this one (or many such classes as you learn in the other tutorials). This is the network clients base class, because it holds the dreamClient object and other members that are closely connected to networking. We look at these members later in this tutorial.
class CSignin { private: dreamClient clientLoginData clientLoginData

networkClient; clientList; *localClient;

// Pointer to the local client // in the client list

public: CSignin(); dreamClient clientLoginData *GetNetworkClient(void) {return networkClient;} *GetLocalClient(void) {return localClient;}

void ReadPackets(void); void AddClient(int local, int index, char *name); void RemoveClient(int index); void RemoveClients(void);

Tutorial 3 / Creating a Basic Network Application with dreamSock

309

void SendSignIn(char *nickname, char *firstname, char *surname, int age, char *gender, char *password, char *email); void SendKeepAlive(void); void Connect(char *name); void Disconnect(void); void RunNetwork(int msec); };

network.h File
This header file holds some common data structures and definitions that are used throughout the application. The importance of this file grows in the later tutorials, as more message types are created. network.h holds the clientLoginData structure that is used to store each clients data. As you have probably noticed, the CSignin class has a linked list of this data type.
#ifndef NETWORK_H #define NETWORK_H #define USER_MES_SERVEREXIT #define USER_MES_SIGNIN #define USER_MES_KEEPALIVE 1 2 3

typedef struct clientLoginData { int index; char nickname[30]; clientLoginData *next; } clientLoginData; #endif

main.h File
This is the main header and contains only some externs (one for now).
#ifndef __TUTMAIN_H__ #define __TUTMAIN_H__ extern HWND hWnd_LoginDialog; #endif

310

Tutorial 3 / Creating a Basic Network Application with dreamSock

common.h File
The common.h file is used to include the header files all at once. This makes things simpler, as all you need to do is include one header file.
#ifndef __COMMON_H__ #define __COMMON_H__ #include "dreamSock.h" #include "network.h" #include "signin.h" #include "main.h" #endif

main.cpp File
Here is the main application source file:
/******************************************/ /* Programming Multiplayer Games */ /* Tutorial game client */ /* Programming: */ /* Teijo Hakala */ /******************************************/ #include "common.h" #include "resource.h" // Some global stuff CSignin Signin; char serverIP[32]; HINSTANCE hInst; HWND hWnd_Application; HWND hWnd_CreateAccountDialog; HWND hWnd_LoginDialog; //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------LRESULT CALLBACK ApplicationProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { switch (uMsg) { case WM_CLOSE: { PostQuitMessage(0);

Tutorial 3 / Creating a Basic Network Application with dreamSock

311

break; } default: break; } // Pass all unhandled messages to DefWindowProc return DefWindowProc(hWnd,uMsg,wParam,lParam); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------LRESULT CALLBACK CreateAccountDialogProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { char nickname[30]; char firstname[50]; char surname[50]; int age; char gender[10]; char password[50]; char password2[50]; char email[150]; int ret; switch (uMsg) { case WM_COMMAND: { switch(LOWORD(wParam)) { case IDC_CREATEACCOUNT_CANCEL: DestroyWindow(hWnd_CreateAccountDialog); hWnd_CreateAccountDialog = NULL; break; case IDC_CREATEACCOUNT_CONTINUE: // -> First get the IP address of the server // from the dialog GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_IPADDRESS, serverIP, 20); // -> Store the player data in local variables GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_NICKNAME, nickname, 30); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_FIRSTNAME, firstname, 50); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_SURNAME, surname, 50);

312

Tutorial 3 / Creating a Basic Network Application with dreamSock

age = GetDlgItemInt(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_AGE, NULL, FALSE); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_GENDER, gender, 10); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_PASSWORD1, password, 50); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_PASSWORD2, password2, 50); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_EMAIL, email, 150); // -> Check that all fields have been filled in if(!strcmp(nickname,"") || !strcmp (firstname,"") || !strcmp(surname,"") || !strcmp(gender,"") || !strcmp(password,"") || !strcmp(email,"") || age < 1) { MessageBox(hWnd_CreateAccountDialog, "Not all fields have been filled in!\n\nPlease check and try again...", "Information Error", MB_OK); break; } // -> Check to see if passwords match if(strcmp(password,password2)) { MessageBox(hWnd_CreateAccountDialog, "The two passwords you entered do not match!\n\nPlease check and try again...", "Password Error", MB_OK); break; } ret = Signin.GetNetworkClient()->Initialize ("", serverIP, 30002); if(ret == DREAMSOCK_CLIENT_ERROR) { char text[64]; sprintf(text, "Could not open client socket"); MessageBox(NULL, text, "Error", MB_OK); } Signin.Connect(nickname); DestroyWindow(hWnd_CreateAccountDialog); hWnd_CreateAccountDialog = NULL;

Tutorial 3 / Creating a Basic Network Application with dreamSock

313

Signin.SendSignIn(nickname, firstname, surname, age, gender, password, email); break; default: break; } return 0; } case WM_CLOSE: break; case WM_DESTROY: break; } return 0; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------LRESULT CALLBACK LoginDialogProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { switch (uMsg) { case WM_COMMAND: { switch(LOWORD(wParam)) { case IDC_LOGIN_QUIT: PostQuitMessage(0); break; case IDC_LOGIN_CREATEACCOUNT: if(!hWnd_CreateAccountDialog) { hWnd_CreateAccountDialog = CreateDialog(hInst, MAKEINTRESOURCE (IDD_CREATEACCOUNT), hWnd_Application, (DLGPROC) CreateAccountDialogProc); } break; case IDC_DOLOGIN: break; default: break; } return 0;

314

Tutorial 3 / Creating a Basic Network Application with dreamSock

} case WM_CLOSE: { PostQuitMessage(0); break; } } return 0; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------int PASCAL WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, TCHAR *pCmdLine, int nCmdShow) { int time, oldTime, newTime; WNDCLASSEX wcl; // Create our main window wcl.cbSize = sizeof(WNDCLASSEX); wcl.hInstance wcl.lpszClassName wcl.lpfnWndProc wcl.style wcl.hIcon wcl.hIconSm wcl.hCursor wcl.lpszMenuName wcl.cbClsExtra wcl.cbWndExtra = = = = hInstance; "ArmyWar"; ApplicationProc; 0;

= LoadIcon(NULL, IDI_APPLICATION); = LoadIcon(NULL, IDI_WINLOGO); = LoadCursor(NULL, IDC_ARROW); = NULL; = 0; = 0;

Tutorial 3 / Creating a Basic Network Application with dreamSock

315

); // Initialize the network library if(dreamSock_Initialize() != 0) { MessageBox(NULL, "Error initializing Communication Library!", "Fatal Error", MB_OK); return 1; } ShowWindow(hWnd_Application, nCmdShow); UpdateWindow(hWnd_Application); // Set global instance variable hInst = hInstance; // Display the LoginDialog hWnd_LoginDialog = CreateDialog(hInst, MAKEINTRESOURCE(IDD_LOGINDIALOG), hWnd_Application, (DLGPROC)LoginDialogProc); MSG msg; BOOL bMsg = FALSE; PeekMessage(&msg, NULL, 0, 0, PM_NOREMOVE); oldTime = dreamSock_GetCurrentSystemTime(); bool done = false; try { while(!done) { while(PeekMessage(&msg, NULL, 0, 0, PM_NOREMOVE)) { if(!GetMessage(&msg, NULL, 0, 0)) { Signin.Disconnect(); done = true; } TranslateMessage(&msg); DispatchMessage(&msg); } do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); // Run sign-in network

316

Tutorial 3 / Creating a Basic Network Application with dreamSock

Signin.RunNetwork(time); oldTime = newTime; } } catch(...) { Signin.Disconnect(); LogString("Unknown Exception caught in main loop"); MessageBox(NULL, "Unknown Exception caught in main loop", "Error", MB_OK | MB_TASKMODAL); return 1; } return msg.wParam; }

Global Variables
There are some global variables that we need to create also. Global variables should generally be avoided, but sometimes they make life simpler!
// Some global stuff CSignin Signin; char serverIP[32]; HINSTANCE hInst; HWND hWnd_Application; HWND hWnd_CreateAccountDialog; HWND hWnd_LoginDialog;

The CSignin Signin is an object that we use for the clients network part. The char serverIP holds the servers IP address string once it is entered in the main login window. The other global variables are required for dialogs and such.

CreateAccountDialogProc Function
This is a dialog procedure function for creating a new account. It checks that the required information has been entered and connects the sign-in server.
LRESULT CALLBACK CreateAccountDialogProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { char nickname[30];

Tutorial 3 / Creating a Basic Network Application with dreamSock

317

char firstname[50]; char surname[50]; int age; char gender[10]; char password[50]; char password2[50]; char email[150]; int ret; switch (uMsg) { case WM_COMMAND: { switch(LOWORD(wParam)) { case IDC_CREATEACCOUNT_CANCEL: DestroyWindow(hWnd_CreateAccountDialog); hWnd_CreateAccountDialog = NULL; break; case IDC_CREATEACCOUNT_CONTINUE: // -> First get the IP address of the server // from the dialog GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_IPADDRESS, serverIP, 20); // -> Store the player data in local variables GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_NICKNAME, nickname, 30); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_FIRSTNAME, firstname, 50); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_SURNAME, surname, 50); age = GetDlgItemInt(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_AGE, NULL, FALSE); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_GENDER, gender, 10); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_PASSWORD1, password, 50); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_PASSWORD2, password2, 50); GetDlgItemText(hWnd_CreateAccountDialog, IDC_CREATEACCOUNT_EMAIL, email, 150); // -> Check that all fields have been filled in if(!strcmp(nickname,"") || !strcmp(firstname,"") || !strcmp(surname,"") || !strcmp(gender,"") || !strcmp(password,"") || !strcmp(email,"") || age < 1) { MessageBox(hWnd_CreateAccountDialog, "Not all fields have been filled

318

Tutorial 3 / Creating a Basic Network Application with dreamSock

in!\n\nPlease check and try again...", "Information Error", MB_OK); break; } // -> Check to see if passwords match if(strcmp(password,password2)) { MessageBox(hWnd_CreateAccountDialog, "The two passwords you entered do not match!\n\nPlease check and try again...", "Password Error", MB_OK); break; } ret = Signin.GetNetworkClient()->Initialize("", serverIP, 30002); if(ret == DREAMSOCK_CLIENT_ERROR) { char text[64]; sprintf(text, "Could not open client socket"); MessageBox(NULL, text, "Error", MB_OK); } Signin.Connect(nickname); DestroyWindow(hWnd_CreateAccountDialog); hWnd_CreateAccountDialog = NULL; Signin.SendSignIn(nickname, firstname, surname, age, gender, password, email); break; default: break; } return 0; } case WM_CLOSE: break; case WM_DESTROY: break; } return 0; }

Tutorial 3 / Creating a Basic Network Application with dreamSock

319

Here we see how to connect the server. As you can see, we use the CSignin class object Signin. First we initialize our network client with dreamClients Initialize function. We want to use the default local IP and the server IP we entered in the text box. The sign-in server uses UDP port 30002. If everything went fine, we connect to the server and send our sign-in information. The functions we use here will be introduced later in this tutorial.
ret = Signin.GetNetworkClient()->Initialize("", serverIP, 30002); if(ret == DREAMSOCK_CLIENT_ERROR) { char text[64]; sprintf(text, "Could not open client socket"); MessageBox(NULL, text, "Error", MB_OK); } Signin.Connect(nickname); DestroyWindow(hWnd_CreateAccountDialog); hWnd_CreateAccountDialog = NULL; Signin.SendSignIn(nickname, firstname, surname, age, gender, password, email);

WinMain Function
The second function with new code here is the WinMain function. As you know, this is the first function that runs in a Windows environment, so we put our initialization code here (among other things).
int PASCAL WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, TCHAR *pCmdLine, int nCmdShow) { int time, oldTime, newTime; WNDCLASSEX wcl; // Create our main window wcl.cbSize = sizeof(WNDCLASSEX); wcl.hInstance wcl.lpszClassName wcl.lpfnWndProc wcl.style wcl.hIcon wcl.hIconSm wcl.hCursor wcl.lpszMenuName wcl.cbClsExtra = = = = hInstance; "ArmyWar"; ApplicationProc; 0;

= LoadIcon(NULL, IDI_APPLICATION); = LoadIcon(NULL, IDI_WINLOGO); = LoadCursor(NULL, IDC_ARROW); = NULL; = 0;

320

Tutorial 3 / Creating a Basic Network Application with dreamSock

wcl.cbWndExtra

= 0;

wcl.hbrBackground = (HBRUSH) GetStockObject(LTGRAY_BRUSH); if(!RegisterClassEx(&wcl)) return 0; hWnd_Application = CreateWindow( "ArmyWar", "ARMY WAR Online 2.0", WS_OVERLAPPEDWINDOW, CW_USEDEFAULT, CW_USEDEFAULT, 640, 480, HWND_DESKTOP, NULL, hInstance, NULL ); // Initialize the network library if(dreamSock_Initialize() != 0) { MessageBox(NULL, "Error initializing Communication Library!", "Fatal Error", MB_OK); return 1; } ShowWindow(hWnd_Application, nCmdShow); UpdateWindow(hWnd_Application); // Set global instance variable hInst = hInstance; // Display the LoginDialog hWnd_LoginDialog = CreateDialog(hInst, MAKEINTRESOURCE(IDD_LOGINDIALOG), hWnd_Application, (DLGPROC)LoginDialogProc); MSG msg; BOOL bMsg = FALSE; PeekMessage(&msg, NULL, 0, 0, PM_NOREMOVE); oldTime = dreamSock_GetCurrentSystemTime(); bool done = false; try { while(!done) { while(PeekMessage(&msg, NULL, 0, 0, PM_NOREMOVE)) {

Tutorial 3 / Creating a Basic Network Application with dreamSock

321

if(!GetMessage(&msg, NULL, 0, 0)) { Signin.Disconnect(); done = true; } TranslateMessage(&msg); DispatchMessage(&msg); } do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); // Run sign-in network Signin.RunNetwork(time); oldTime = newTime; } } catch(...) { Signin.Disconnect(); LogString("Unknown Exception caught in main loop"); MessageBox(NULL, "Unknown Exception caught in main loop", "Error", MB_OK | MB_TASKMODAL); return 1; } return msg.wParam; }

The main window here is created the same way it is in the next tutorial. Because it is shown there, we skip it here. Initializing dreamSock interests us more, and it is done as follows. All we do is call the dreamSock_Initialize function and see if it succeeded or not. If it returns non-zero, the function failed to initialize dreamSock for some reason.
// Initialize the network library if(dreamSock_Initialize() != 0) { MessageBox(NULL, "Error initializing Communication Library!", "Fatal Error", MB_OK); return 1; }

322

Tutorial 3 / Creating a Basic Network Application with dreamSock

The next thing that interests us is the main application loop. We use try and catch statements to try to catch some undesired exceptions. If it does catch something, the client tries to disconnect from the server. The main loop calculates each frames time in milliseconds. This is done by first getting the current time when the loop ends, and then taking the current time during the loop and subtracting it from the old loop end time. This way we see how long it takes to run the loop one time. This is called frame time and it is used as a parameter for Signins RunNetwork function.
MSG msg; BOOL bMsg = FALSE; PeekMessage(&msg, NULL, 0, 0, PM_NOREMOVE); oldTime = dreamSock_GetCurrentSystemTime(); bool done = false; try { while(!done) { while(PeekMessage(&msg, NULL, 0, 0, PM_NOREMOVE)) { if(!GetMessage(&msg, NULL, 0, 0)) { Signin.Disconnect(); done = true; } TranslateMessage(&msg); DispatchMessage(&msg); } do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); // Run sign-in network Signin.RunNetwork(time); oldTime = newTime; } } catch(...) { Signin.Disconnect();

Tutorial 3 / Creating a Basic Network Application with dreamSock

323

LogString("Unknown Exception caught in main loop"); MessageBox(NULL, "Unknown Exception caught in main loop", "Error", MB_OK | MB_TASKMODAL); return 1; }

signin.cpp File CSignin Class Methods

This file contains all the methods required for the sign-in network client to operate. We read and write data to the network in this source file.
/******************************************/ /* Programming Multiplayer Games */ /* Tutorial game client */ /* Programming: */ /* Teijo Hakala */ /******************************************/ #include "common.h" //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CSignin::CSignin() { networkClient = new dreamClient; clientList = NULL; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CSignin::~CSignin() { delete networkClient; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSignin::ReadPackets(void) { char data[1400]; struct sockaddr address; int type; int ind;

324

Tutorial 3 / Creating a Basic Network Application with dreamSock

int local; int ret; char name[30]; dreamMessage mes; mes.Init(data, sizeof(data)); while(ret = networkClient->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading(); type = mes.ReadByte(); switch(type) { case DREAMSOCK_MES_ADDCLIENT: local = mes.ReadByte(); ind = mes.ReadByte(); strcpy(name, mes.ReadString()); AddClient(local, ind, name); break; case DREAMSOCK_MES_REMOVECLIENT: ind = mes.ReadByte(); RemoveClient(ind); break; case USER_MES_SERVEREXIT: MessageBox(NULL, "Server disconnected", "Info", MB_OK); Disconnect(); break; case USER_MES_SIGNIN: // Skip sequences mes.ReadShort(); mes.ReadShort(); ret = mes.ReadShort(); LogString("Got lobby signin respond %d", ret); if(ret != SIGNIN_RESULT_ACCEPTED) { MessageBox(hWnd_LoginDialog, "Sign-in did not succeed. Try again.", "Error", MB_OK); } else { MessageBox(hWnd_LoginDialog, "Sign-in successful. You can now login.",

Tutorial 3 / Creating a Basic Network Application with dreamSock

325

"Info", MB_OK); } Disconnect(); break; } } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSignin::AddClient(int local, int ind, char *name) { // First get a pointer to the beginning of client list clientLoginData *list = clientList; clientLoginData *prev; LogString("App: Client: Adding client with index %d", ind); // No clients yet, adding the first one if(clientList == NULL) { LogString("App: Client: Adding first client"); clientList = (clientLoginData *) calloc(1, sizeof(clientLoginData)); if(local) { LogString("App: Client: This one is local"); localClient = clientList; } clientList->index = ind; strcpy(clientList->nickname, name); clientList->next = NULL; } else { LogString("App: Client: Adding another client"); prev = list; list = clientList->next; while(list != NULL) { prev = list; list = list->next; }

326

Tutorial 3 / Creating a Basic Network Application with dreamSock

list = (clientLoginData *) calloc(1, sizeof(clientLoginData)); if(local) { LogString("App: Client: This one is local"); localClient = list; } list->index = ind; strcpy(list->nickname, name); list->next = NULL; prev->next = list; } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSignin::RemoveClient(int ind) { clientLoginData *list = clientList; clientLoginData *prev = NULL; clientLoginData *next = NULL; for(; list != NULL; list = list->next) { if(list->index == ind) { if(prev != NULL) { prev->next = list->next; } break; } prev = list; } if(list == clientList) { if(list) { next = list->next; free(list); } list = NULL; clientList = next; } else

Tutorial 3 / Creating a Basic Network Application with dreamSock

327

{ if(list) { next = list->next; free(list); } list = next; } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSignin::RemoveClients(void) { clientLoginData *list = clientList; clientLoginData *next; while(list != NULL) { if(list) { next = list->next; free(list); } list = next; } clientList = NULL; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSignin::SendSignIn(char *nickname, char *firstname, char *surname, int age, char *gender, char *password, char *email) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_SIGNIN); message.AddSequences(networkClient); message.WriteString(nickname); message.WriteString(firstname); message.WriteString(surname); message.WriteByte(age); message.WriteString(gender); message.WriteString(password);

328

Tutorial 3 / Creating a Basic Network Application with dreamSock

message.WriteString(email); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSignin::SendKeepAlive(void) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_KEEPALIVE); message.AddSequences(networkClient); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSignin::Connect(char *name) { LogString("CSignin::Connect"); networkClient->SendConnect(name); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSignin::Disconnect(void) { LogString("CSignin::Disconnect"); localClient = NULL; RemoveClients(); networkClient->SendDisconnect(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSignin::RunNetwork(int msec) { if(networkClient->GetConnectionState() == DREAMSOCK_DISCONNECTED)

Tutorial 3 / Creating a Basic Network Application with dreamSock

329

return; static int keepalive = 0; keepalive += msec; if(keepalive > 20000) { SendKeepAlive(); keepalive = 0; } ReadPackets(); }

CSignin Constructor
Here is the constructor for CSignin. All it does is create the dreamClient object and set clientList to NULL, which is very important. Otherwise, the client would not know that the list is empty.
CSignin::CSignin() { networkClient clientList }

= new dreamClient; = NULL;

CSignin Destructor
This is the destructor for CSignin. The dreamClient object is deleted.
CSignin::~CSignin() { delete networkClient; }

ReadPackets Function
Finally we get to read some packets from the server! This function listens to the sign-in server and parses through system and user messages.
void CSignin::ReadPackets(void) { char data[1400]; struct sockaddr address; int type; int ind; int local; int ret; char name[30];

330

Tutorial 3 / Creating a Basic Network Application with dreamSock

dreamMessage mes; mes.Init(data, sizeof(data)); while(ret = networkClient->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading(); type = mes.ReadByte(); switch(type) { case DREAMSOCK_MES_ADDCLIENT: local = mes.ReadByte(); ind = mes.ReadByte(); strcpy(name, mes.ReadString()); AddClient(local, ind, name); break; case DREAMSOCK_MES_REMOVECLIENT: ind = mes.ReadByte(); RemoveClient(ind); break; case USER_MES_SERVEREXIT: MessageBox(NULL, "Server disconnected", "Info", MB_OK); Disconnect(); break; case USER_MES_SIGNIN: // Skip sequences mes.ReadShort(); mes.ReadShort(); ret = mes.ReadShort(); LogString("Got lobby signin respond %d", ret); if(ret != SIGNIN_RESULT_ACCEPTED) { MessageBox(hWnd_LoginDialog, "Sign-in did not succeed. Try again.", "Error", MB_OK); } else { MessageBox(hWnd_LoginDialog, "Sign-in successful. You can now login.", "Info", MB_OK); } Disconnect();

Tutorial 3 / Creating a Basic Network Application with dreamSock

331

break; } } }

When the server sends us any data, this is the method that reads the sockets data. If no data is coming, the function returns immediately. Here we have two variables that are filled by dreamClients GetPacket function. One is the data buffer where all the incoming data goes (we will set this up soon), and the other is a socket address structure that will hold the remote hosts address after we have first received something from it. Remember that a client receives data only from the server, so this address should always be the same (and we normally have no use for it at all).
char data[1400]; struct sockaddr address;

There are some variables here that are used to store parts of the incoming data. These are not required, but they make life a bit easier.
int type; int ind; int local; int ret; char name[30];

Next we initialize a dreamMessage object to hold the incoming message. See how simple it is to set the data buffer for a message?
dreamMessage mes; mes.Init(data, sizeof(data));

Now we are ready to read the packet of the socket, if there is any. When there is, we read the incoming data to our dreamMessage object and set the correct size for it. Then we begin reading the message, message type first.
while(ret = networkClient->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading(); type = mes.ReadByte(); ...

Now that we know the messages type, we can parse the rest of the message. As you can see, there are dreamSock system messages and user messages here. We use the same add client and remove client messages as dreamSock to add and remove clients. If the client receives a server exit message, we know the server has shut down and we should unitialize the client. This is a user

332

Tutorial 3 / Creating a Basic Network Application with dreamSock

message; we will see how to send those in the server-side part of this tutorial. If we receive a sign-in message, we know that the server has processed our sign-in request and it has a response for us. So we read it from the message. Before we can read the response part of the message, we must read the message sequence part. We cannot just read the response because the message sequence numbers come first in the message array. In this case, we simply read and ignore the sequence numbers and then move on to the actual message. Then the client disconnects from the server, because there is nothing else the sign-in server has to offer to us.
switch(type) { case DREAMSOCK_MES_ADDCLIENT: local = mes.ReadByte(); ind = mes.ReadByte(); strcpy(name, mes.ReadString()); AddClient(local, ind, name); break; case DREAMSOCK_MES_REMOVECLIENT: ind = mes.ReadByte(); RemoveClient(ind); break; case USER_MES_SERVEREXIT: MessageBox(NULL, "Server disconnected", "Info", MB_OK); Disconnect(); break; case USER_MES_SIGNIN: // Skip sequences mes.ReadShort(); mes.ReadShort(); ret = mes.ReadShort(); LogString("Got lobby signin respond %d", ret); if(ret != SIGNIN_RESULT_ACCEPTED) { MessageBox(hWnd_LoginDialog, "Sign-in did not succeed. Try again.", "Error", MB_OK); } else { MessageBox(hWnd_LoginDialog, "Sign-in successful. You can now login.", "Info", MB_OK);

Tutorial 3 / Creating a Basic Network Application with dreamSock

333

} Disconnect(); break; }

AddClient Function
This function adds a client to the clients own client list. This list holds the data essential to the network application itself.
void CSignin::AddClient(int local, int ind, char *name) { // First get a pointer to the beginning of client list clientLoginData *list = clientList; clientLoginData *prev; LogString("App: Client: Adding client with index %d", ind); // No clients yet, adding the first one if(clientList == NULL) { LogString("App: Client: Adding first client"); clientList = (clientLoginData *) calloc(1, sizeof(clientLoginData)); if(local) { LogString("App: Client: This one is local"); localClient = clientList; } clientList->index = ind; strcpy(clientList->nickname, name); clientList->next = NULL; } else { LogString("App: Client: Adding another client"); prev = list; list = clientList->next; while(list != NULL) { prev = list; list = list->next; } list = (clientLoginData *) calloc(1, sizeof(clientLoginData)); if(local)

334

Tutorial 3 / Creating a Basic Network Application with dreamSock

{ LogString("App: Client: This one is local"); localClient = list; } list->index = ind; strcpy(list->nickname, name); list->next = NULL; prev->next = list; } }

This function works the same way as the AddClient function of dreamServer. The only difference is that we give more data to store as a parameter (int local, int ind, char *name). If the first parameter is 1, the client we are adding to the list is the local client. So in this case, set the localClient pointer to point to the correct client in the list. The other parameters are data to put in the client list.

RemoveClient Function
This function removes a client from the clients own client list.
void CSignin::RemoveClient(int ind) { clientLoginData *list = clientList; clientLoginData *prev = NULL; clientLoginData *next = NULL; for(; list != NULL; list = list->next) { if(list->index == ind) { if(prev != NULL) { prev->next = list->next; } break; } prev = list; } if(list == clientList) { if(list) { next = list->next; free(list);

Tutorial 3 / Creating a Basic Network Application with dreamSock

335

} list = NULL; clientList = next; } else { if(list) { next = list->next; free(list); } list = next; } }

The function takes one parameter (int ind) that is used to identify the client to remove. In the following loop, we locate the correct client in the list and update the list. After that, the client can be safely removed from memory.
for(; list != NULL; list = list->next) { if(list->index == ind) { if(prev != NULL) { prev->next = list->next; } break; } prev = list; }

RemoveClients Function
This function removes all clients from the clients own client list.
void CSignin::RemoveClients(void) { clientLoginData *list = clientList; clientLoginData *next; while(list != NULL) { if(list) { next = list->next; free(list); }

336

Tutorial 3 / Creating a Basic Network Application with dreamSock

list = next; } clientList = NULL; }

SendSignIn Function
This function sends the sign-in message to the server.
void CSignin::SendSignIn(char *nickname, char *firstname, char *surname, int age, char *gender, char *password, char *email) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_SIGNIN); message.AddSequences(networkClient); message.WriteString(nickname); message.WriteString(firstname); message.WriteString(surname); message.WriteByte(age); message.WriteString(gender); message.WriteString(password); message.WriteString(email); networkClient->SendPacket(&message); }

This function takes several parameters, which are the data to send to the server. The only notable thing here is that we add the sequence numbers right after the message type.

SendKeepAlive Function
This function is used to keep the client alive on the server if necessary.
void CSignin::SendKeepAlive(void) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_KEEPALIVE); message.AddSequences(networkClient); networkClient->SendPacket(&message); }

Tutorial 3 / Creating a Basic Network Application with dreamSock

337

Connect Function
This function runs dreamClients SendConnect function to connect to the server.
void CSignin::Connect(char *name) { LogString("CSignin::Connect"); networkClient->SendConnect(name); }

Disconnect Function
This function runs dreamClients SendDisconnect function to disconnect from the server. We also remove all the clients from the local client list and set the local client pointer to NULL since they depend on the connection.
void CSignin::Disconnect(void) { LogString("CSignin::Disconnect"); localClient = NULL; RemoveClients(); networkClient->SendDisconnect(); }

RunNetwork Function
This function is run every frame to keep the network system running.
void CSignin::RunNetwork(int msec) { if(networkClient->GetConnectionState() == DREAMSOCK_DISCONNECTED) return; static int keepalive = 0; keepalive += msec; if(keepalive > 20000) { SendKeepAlive(); keepalive = 0; } ReadPackets(); }

The function takes one parameter (int msec), the frame time. We use it to calculate time. Here, we use it to send a keep alive message

338

Tutorial 3 / Creating a Basic Network Application with dreamSock

in 20-second intervals. The functions most important task is to read the packets of the socket.

Creating a Basic Server Application

Now all we have to do is make the server application. The server works basically the same way as the client application, but with some differences of course. This server can be built and run on Windows and Unix systems. Below is the makefile for Unix.
#C_ARGS = Wall CC = gcc all: dreamSockServer clean: rm f *.o dreamSockServer: main.o signin.o $(CC) o dreamSockServer main.o signin.o L. ldreamSock lm lstdc++ main.o: main.cpp $(CC) $(C_ARGS) c main.cpp L. ldreamSock signin.o: signin.cpp $(CC) $(C_ARGS) c signin.cpp L. ldreamSock

We need to link dreamSock.lib into the server program. That is done the same way as in the client application. We will create the following source code files: main.cpp, signin.cpp, common.h, network.h, and signin.h.

signin.h File
Like in the client application, signin.h contains the most important class of the server application. Class CSigninServer is used for all the network operations we need to process.
#ifndef __SIGNIN_H__ #define __SIGNIN_H__ #include "network.h" #define #define #define #define SIGNIN_RESULT_ACCEPTED SIGNIN_RESULT_USERNAMEBAD SIGNIN_RESULT_PASSWORDBAD SIGNIN_RESULT_MYSQLERROR 200 201 202 203

class CSigninServer {

Tutorial 3 / Creating a Basic Network Application with dreamSock

339

private: dreamServer clientLoginData public: CSigninServer(); ~CSigninServer();

*networkServer; *clientList;

CSigninServer Class
Here is the CSigninServer classs member methods and variables. As you can see, it contains a dreamServer object similar to how the client application contained a dreamClient object. It also contains another client list on the server side (remember that dreamServer also has one). This is used to store the application-specific data, and both lists have the same clients in the same order. We will take a closer look at the clientLoginData structure next.
class CSigninServer { private: dreamServer clientLoginData public: CSigninServer(); ~CSigninServer(); int InitNetwork(void); void ShutdownNetwork(void); void ReadPackets(void); void SendExitNotification(void); void AddClient(void); void RemoveClient(struct sockaddr *address);

*networkServer; *clientList;

340

Tutorial 3 / Creating a Basic Network Application with dreamSock

void RemoveClients(void); void Frame(int msec); };

network.h File
This file contains some common data types and definitions, just like on the client side. There is an important structure in this file: clientLoginData. This structure is used to hold the clients application-specific data. In this basic network application, there is nothing application specific, but later on in the tutorial game we will see some real data here (like the players position on the map). Notice that there is a pointer to a dreamClient object here. It is set to point to a correct client when we are adding a client. After that we can use this one structure to handle the clients because we have their application-specific data here, and we can also access their network-specific functions through the netClient pointer.
#ifndef NETWORK_H #define NETWORK_H #include "dreamSock.h" #define USER_MES_SERVEREXIT #define USER_MES_SIGNIN 1 2

typedef struct clientLoginData { dreamClient netClient; clientLoginData next; } clientLoginData; #endif

common.h File
This file is used to include common header files all at once.
#ifndef __COMMON_H__ #define __COMMON_H__ #include "dreamSock.h" #include "network.h" #include "signin.h" #endif

Tutorial 3 / Creating a Basic Network Application with dreamSock

341

main.cpp File
This file is the main source file for the server. As you can see, it contains a bit more code than the client version because we have code for both Unix and Windows systems here.
/******************************************/ /* Programming Multiplayer Games */ /* Tutorial game server */ /* Programming: */ /* Teijo Hakala */ /******************************************/ #ifdef Win32 #ifndef _WINSOCKAPI_ #define _WINSOCKAPI_ #endif #include <windows.h> #endif #ifdef Win32 #include <shellapi.h> #else #include #include #include #include #include #include #include #include #include #endif

<signal.h> <syslog.h> <errno.h> <unistd.h> <sys/time.h> <sys/stat.h> <sys/types.h> <sys/wait.h> <fcntl.h>

#include "common.h" #include <string.h> #include <stdlib.h> #include <stdio.h> // Win32 only #ifdef Win32 // Unix only #else int runningDaemon; #endif CSigninServer Signin;

342

Tutorial 3 / Creating a Basic Network Application with dreamSock

#ifdef Win32 //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------LRESULT CALLBACK WindowProc(HWND WindowhWnd, UINT Message, WPARAM wParam, LPARAM lParam) { // Process messages switch(Message) { case WM_CREATE: break; case WM_DESTROY: PostQuitMessage(0); break; default: break; } return DefWindowProc(WindowhWnd, Message, wParam, lParam); } //--------------------------------------------------------------------------// Name: WinMain() // Desc: Windows app start position //--------------------------------------------------------------------------int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, LPSTR lpCmdLine, int nCmdShow) { WNDCLASS WinClass; WinClass.style WinClass.lpfnWndProc WinClass.cbClsExtra WinClass.cbWndExtra WinClass.hInstance WinClass.hIcon WinClass.hCursor WinClass.hbrBackground WinClass.lpszMenuName WinClass.lpszClassName = = = = = = = = = = 0; WindowProc; 0; 0; hInstance; LoadIcon(NULL, IDI_APPLICATION); LoadCursor(NULL, IDC_ARROW); (HBRUSH) (COLOR_MENU); 0; "WINCLASS1";

if(!RegisterClass(&WinClass)) return 0; HWND hwnd = CreateWindow(WinClass.lpszClassName, "dreamSock server", WS_OVERLAPPEDWINDOW | WS_VISIBLE, 320, 240,

Tutorial 3 / Creating a Basic Network Application with dreamSock

343

320, 240, NULL, NULL, hInstance, NULL); ShowWindow(hwnd, SW_HIDE); StartLogConsole(); if(Signin.InitNetwork() == 1) { PostQuitMessage(0); } MSG WinMsg; bool done = false; int time, oldTime, newTime; // first peek at the message without removing it PeekMessage(&WinMsg, hwnd, 0, 0, PM_NOREMOVE); oldTime = dreamSock_GetCurrentSystemTime(); try { while(!done) { while (PeekMessage(&WinMsg, NULL, 0, 0, PM_NOREMOVE)) { if(!GetMessage(&WinMsg, NULL, 0, 0)) { Signin.ShutdownNetwork(); dreamSock_Shutdown(); done = true; } TranslateMessage(&WinMsg); DispatchMessage(&WinMsg); } do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); Signin.Frame(time); oldTime = newTime; } } catch(...)

344

Tutorial 3 / Creating a Basic Network Application with dreamSock

{ LogString("Unknown Exception caught in main loop"); Signin.ShutdownNetwork(); dreamSock_Shutdown(); MessageBox(NULL, "Unknown Exception caught in main loop", "Error", MB_OK | MB_TASKMODAL); return 1; } return WinMsg.wParam; } #else //--------------------------------------------------------------------------// Name: daemonInit() // Desc: Initialize Unix daemon //--------------------------------------------------------------------------static int daemonInit(void) { printf("Running daemon...\n\n"); runningDaemon = 1; pid_t pid; if((pid = fork()) < 0) { return 1; } else if(pid != 0) { exit(0); } setsid(); umask(0); close(1); close(2); close(3); return 0; } //--------------------------------------------------------------------------// Name: keyPress() // Desc: Check for a keypress //--------------------------------------------------------------------------int keyPress(void)

Tutorial 3 / Creating a Basic Network Application with dreamSock

345

{ static char keypressed; struct timeval waittime; int num_chars_read; fd_set mask; FD_SET(0, &mask); waittime.tv_sec = 0; waittime.tv_usec = 0; if(select(1, &mask, 0, 0, &waittime)) { num_chars_read = read(0, &keypressed, 1); if(num_chars_read == 1) return ((int) keypressed); } return (1); } //--------------------------------------------------------------------------// Name: main() // Desc: Unix app start position //--------------------------------------------------------------------------int main(int argc, char **argv) { LogString("Welcome to Army War Server v2.0"); LogString("-------------------------------\n"); if(argc > 1) { if(strcmp(argv[1], "daemon") == 0) { daemonInit(); } } // Ignore the SIGPIPE signal, so the program does not terminate if the // pipe gets broken signal(SIGPIPE, SIG_IGN); if(Signin.InitNetwork() == 1) { exit(0); } LogString("Init successful"); int time, oldTime, newTime; oldTime = dreamSock_GetCurrentSystemTime();

346

Tutorial 3 / Creating a Basic Network Application with dreamSock

// App main loop try { if(runningDaemon) { // Keep server alive while(1) { do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); Signin.Frame(time); oldTime = newTime; } } else { // Keep server alive (wait for keypress to kill it) while(keyPress() == 1) { do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); Signin.Frame(time); oldTime = newTime; } } } catch(...) { Signin.ShutdownNetwork(); dreamSock_Shutdown(); LogString("Unknown Exception caught in main loop"); return 1; } LogString("Shutting down everything"); Signin.ShutdownNetwork(); dreamSock_Shutdown(); return 0;

Tutorial 3 / Creating a Basic Network Application with dreamSock

347

} #endif

Global Variables
There are some global variables that we should take a look at first:
int runningDaemon; CSigninServer Signin;

The integer runningDaemon is a Unix-only variable that tells us if the program is set to run as a daemon program (on the background). The CSigninServer Signin is the object that controls the servers network communications.

WindowProc Function
Lets begin with Windows functions. This first one is the window procedure function that handles the window commands. There is nothing really interesting here as we will be using our log console system as the main window.
LRESULT CALLBACK WindowProc(HWND WindowhWnd, UINT Message, WPARAM wParam, LPARAM lParam) { // Process messages switch(Message) { case WM_CREATE: break; case WM_DESTROY: PostQuitMessage(0); break; default: break; } return DefWindowProc(WindowhWnd, Message, wParam, lParam); }

WinMain Function
This is the application entry point in Windows.
int APIENTRY WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, LPSTR lpCmdLine, int nCmdShow) {

348

Tutorial 3 / Creating a Basic Network Application with dreamSock

WNDCLASS WinClass; WinClass.style WinClass.lpfnWndProc WinClass.cbClsExtra WinClass.cbWndExtra WinClass.hInstance WinClass.hIcon WinClass.hCursor WinClass.hbrBackground WinClass.lpszMenuName WinClass.lpszClassName = = = = = = = = = = 0; WindowProc; 0; 0; hInstance; LoadIcon(NULL, IDI_APPLICATION); LoadCursor(NULL, IDC_ARROW); (HBRUSH) (COLOR_MENU); 0; "WINCLASS1";

if(!RegisterClass(&WinClass)) return 0; HWND hwnd = CreateWindow(WinClass.lpszClassName, "dreamSock server", WS_OVERLAPPEDWINDOW | WS_VISIBLE, 320, 240, 320, 240, NULL, NULL, hInstance, NULL); ShowWindow(hwnd, SW_HIDE); StartLogConsole(); if(Signin.InitNetwork() == 1) { PostQuitMessage(0); } MSG WinMsg; bool done = false; int time, oldTime, newTime; // first peek at the message without removing it PeekMessage(&WinMsg, hwnd, 0, 0, PM_NOREMOVE); oldTime = dreamSock_GetCurrentSystemTime(); try { while(!done) { while (PeekMessage(&WinMsg, NULL, 0, 0, PM_NOREMOVE)) { if(!GetMessage(&WinMsg, NULL, 0, 0)) {

Tutorial 3 / Creating a Basic Network Application with dreamSock

349

Signin.ShutdownNetwork(); dreamSock_Shutdown(); done = true; } TranslateMessage(&WinMsg); DispatchMessage(&WinMsg); } do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); Signin.Frame(time); oldTime = newTime; } } catch(...) { LogString("Unknown Exception caught in main loop"); Signin.ShutdownNetwork(); dreamSock_Shutdown(); MessageBox(NULL, "Unknown Exception caught in main loop", "Error", MB_OK | MB_TASKMODAL); return 1; } return WinMsg.wParam; }

We want to see what is happening on the server, so instead of a normal window we are going to open a console window. First though, we need to open a real window and hide it. It runs the window procedure for us and keeps the program alive. Then we simply start up our log console (implemented into dreamSock).
HWND hwnd = CreateWindow(WinClass.lpszClassName, "dreamSock server", WS_OVERLAPPEDWINDOW | WS_VISIBLE, 320, 240, 320, 240, NULL, NULL, hInstance, NULL);

350

Tutorial 3 / Creating a Basic Network Application with dreamSock

ShowWindow(hwnd, SW_HIDE); StartLogConsole();

Then we can initialize the sign-in server using the CSigninServer method InitNetwork. If something goes wrong here, we just bail out from the program by posting a quit message to the main loop.
if(Signin.InitNetwork() == 1) { PostQuitMessage(0); }

The main loop works exactly the same as in the client application. We try to catch exceptions here too, and that is a very useful thing to do it can keep the server running in cases when it normally would crash. Just like in the client application, we calculate the frame time and use it in the RunNetwork method of CSigninServer.
MSG WinMsg; bool done = false; int time, oldTime, newTime; // first peek at the message without removing it PeekMessage(&WinMsg, hwnd, 0, 0, PM_NOREMOVE); oldTime = dreamSock_GetCurrentSystemTime(); try { while(!done) { while (PeekMessage(&WinMsg, NULL, 0, 0, PM_NOREMOVE)) { if(!GetMessage(&WinMsg, NULL, 0, 0)) { Signin.ShutdownNetwork(); dreamSock_Shutdown(); done = true; } TranslateMessage(&WinMsg); DispatchMessage(&WinMsg); } do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); Signin.Frame(time);

Tutorial 3 / Creating a Basic Network Application with dreamSock

351

oldTime = newTime; } } catch(...) { LogString("Unknown Exception caught in main loop"); Signin.ShutdownNetwork(); dreamSock_Shutdown(); MessageBox(NULL, "Unknown Exception caught in main loop", "Error", MB_OK | MB_TASKMODAL); return 1; }

daemonInit Function
This function makes the program run as a daemon on Unix systems.
static int daemonInit(void) { printf("Running daemon...\n\n"); runningDaemon = 1; pid_t pid; if((pid = fork()) < 0) { return 1; } else if(pid != 0) { exit(0); } setsid(); umask(0); close(1); close(2); close(3); return 0; }

First we create a copy of our process with the fork function. Then we can close the parent process with the exit function.
pid_t pid; if((pid = fork()) < 0)

352

Tutorial 3 / Creating a Basic Network Application with dreamSock

{ return 1; } else if(pid != 0) { exit(0); }

Next we set a new process group for the process, detaching it from the terminal we used:
setsid();

Then we set the file creation mask to 0, so we can create files and so on:
umask(0);

Finally, we close the standard I/O descriptors (stdin, stdout, and stderr) because they are not required.
close(1); close(2); close(3);

keyPress Function
This function tells us if a key has been pressed down. It is used on non-daemon programs only.
int keyPress(void) { static char keypressed; struct timeval waittime; int num_chars_read; fd_set mask; FD_SET(0, &mask); waittime.tv_sec = 0; waittime.tv_usec = 0; if(select(1, &mask, 0, 0, &waittime)) { num_chars_read = read(0, &keypressed, 1); if(num_chars_read == 1) return ((int) keypressed); } return (1); }

Tutorial 3 / Creating a Basic Network Application with dreamSock

353

main Function
The last function here is the application entry point for Unix systems.
int main(int argc, char **argv) { LogString("Welcome to Army War Server v2.0"); LogString("-------------------------------\n"); if(argc > 1) { if(strcmp(argv[1], "daemon") == 0) { daemonInit(); } } // Ignore the SIGPIPE signal, so the program does not terminate if the // pipe gets broken signal(SIGPIPE, SIG_IGN); if(Signin.InitNetwork() == 1) { exit(0); } LogString("Init successful"); int time, oldTime, newTime; oldTime = dreamSock_GetCurrentSystemTime(); // App main loop try { if(runningDaemon) { // Keep server alive while(1) { do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); Signin.Frame(time); oldTime = newTime; } }

354

Tutorial 3 / Creating a Basic Network Application with dreamSock

else { // Keep server alive (wait for keypress to kill it) while(keyPress() == 1) { do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); Signin.Frame(time); oldTime = newTime; } } } catch(...) { Signin.ShutdownNetwork(); dreamSock_Shutdown(); LogString("Unknown Exception caught in main loop"); return 1; } LogString("Shutting down everything"); Signin.ShutdownNetwork(); dreamSock_Shutdown(); return 0; }

If we give a daemon parameter for the program when we start it, the daemon is initialized:
if(argc > 1) { if(strcmp(argv[1], "daemon") == 0) { daemonInit(); } }

There are two version of the main loop. One is for the daemon and the other is for the foreground application. If the program is not running as a daemon, we can terminate the program by pressing any key. Otherwise, the main loop works the same way as in Windows.
// Keep server alive (wait for keypress to kill it) while(keyPress() == 1) { do

Tutorial 3 / Creating a Basic Network Application with dreamSock

355

{ newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); Signin.Frame(time); oldTime = newTime; }

signin.cpp File CSigninServer Class Methods

This is the signin.cpp file that contains the methods for CSigninServer.
/******************************************/ /* Programming Multiplayer Games */ /* Tutorial game server */ /* Programming: */ /* Teijo Hakala */ /******************************************/ #include "common.h" #include <malloc.h> //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CSigninServer::CSigninServer() { networkServer = new dreamServer; clientList = NULL; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CSigninServer::~CSigninServer() { delete networkServer; } //--------------------------------------------------------------------------// Name: InitNetwork() // Desc: Initialize network //--------------------------------------------------------------------------int CSigninServer::InitNetwork(void) { // Initialize dreamSock and the server if(dreamSock_Initialize() != 0) { LogString("Error initializing Communication Library!");

356

Tutorial 3 / Creating a Basic Network Application with dreamSock

return 1; } int ret = networkServer->Initialize("", 30002); if(ret == DREAMSOCK_SERVER_ERROR) { #ifdef Win32 char text[64]; sprintf(text, "Could not open server on port %d", networkServer->GetPort()); MessageBox(NULL, text, "Error", MB_OK); #else LogString("Could not open server on port %d", networkServer->GetPort()); #endif return 1; } return 0; } //--------------------------------------------------------------------------// Name: ShutdownNetwork() // Desc: Shut down network //--------------------------------------------------------------------------void CSigninServer::ShutdownNetwork(void) { LogString("Shutting down sign-in server..."); SendExitNotification(); RemoveClients(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSigninServer::ReadPackets(void) { char data[1400]; int type; int ret; // Some incoming data char password[50]; int respond; char char char char nickname[30]; surname[30]; firstname[30]; gender[10];

Tutorial 3 / Creating a Basic Network Application with dreamSock

357

char email[30]; int age; struct sockaddr address; clientLoginData *clList; dreamMessage mes; mes.Init(data, sizeof(data)); // Get the packet from the socket try { while(ret = networkServer->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading(); type = mes.ReadByte(); // Check the type of the message switch(type) { case DREAMSOCK_MES_CONNECT: AddClient(); break; case DREAMSOCK_MES_DISCONNECT: RemoveClient(&address); break; case USER_MES_SIGNIN: // Skip sequences mes.ReadShort(); mes.ReadShort(); strcpy(nickname, mes.ReadString()); strcpy(firstname, mes.ReadString()); strcpy(surname, mes.ReadString()); age = mes.ReadByte(); strcpy(gender, mes.ReadString()); strcpy(password, mes.ReadString()); strcpy(email, mes.ReadString()); LogString("Signin: Player %s signed in", nickname); // MySQL connection comes here respond = SIGNIN_RESULT_ACCEPTED; // Find the correct client by comparing addresses clList = clientList; for(; clList != NULL; clList = clList->next)

358

Tutorial 3 / Creating a Basic Network Application with dreamSock

{ if(memcmp(clList->netClient-> GetSocketAddress(), &address, sizeof(address)) == 0) { clList->netClient->message.Init (clList->netClient-> message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte (USER_MES_SIGNIN); // type clList->netClient->message.AddSequences (clList->netClient); // sequences clList->netClient->message.WriteShort (respond); // respond clList->netClient->SendPacket(); LogString("Sending signin respond"); break; } } break; } } } catch(...) { LogString("Unknown Exception caught in Signin ReadPackets loop"); #ifdef Win32 MessageBox(NULL, "Unknown Exception caught in Signin ReadPackets loop", "Error", MB_OK | MB_TASKMODAL); #endif } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSigninServer::SendExitNotification(void) { clientLoginData *toClient = clientList; for(; toClient != NULL; toClient = toClient->next) { toClient->netClient->message.Init(toClient->netClient-> message.outgoingData, sizeof(toClient->netClient-> message.outgoingData));

Tutorial 3 / Creating a Basic Network Application with dreamSock

359

toClient->netClient->message.WriteByte(USER_MES_SERVEREXIT); // type toClient->netClient->message.AddSequences(toClient->netClient); // sequences } networkServer->SendPackets(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSigninServer::AddClient(void) { // First get a pointer to the beginning of client list clientLoginData *list = clientList; clientLoginData *prev; dreamClient *netList = networkServer->GetClientList(); // No clients yet, adding the first one if(clientList == NULL) { LogString("App: Server: Adding first client"); clientList = (clientLoginData *) calloc(1, sizeof(clientLoginData)); clientList->netClient = netList; clientList->next = NULL; } else { LogString("App: Server: Adding another client"); prev = list; list = clientList->next; netList = netList->next; while(list != NULL) { prev = list; list = list->next; netList = netList->next; } list = (clientLoginData *) calloc(1, sizeof(clientLoginData)); list->netClient = netList; list->next = NULL; prev->next = list;

360

Tutorial 3 / Creating a Basic Network Application with dreamSock

361

//--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSigninServer::RemoveClients(void) { clientLoginData *list = clientList; clientLoginData *next; while(list != NULL) { if(list) { next = list->next; free(list); } list = next; } clientList = NULL; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CSigninServer::RunNetwork(int msec) { ReadPackets(); }

CSigninServer Constructor
Here is the constructor for CSigninServer. The dreamServer object is created and clientList is set to NULL.
CSigninServer::CSigninServer() { networkServer = new dreamServer; clientList = NULL; }

CSigninServer Destructor
Here is the destructor for CSigninServer. The dreamServer object is deleted.
CSigninServer::~CSigninServer() { delete networkServer; }

362

Tutorial 3 / Creating a Basic Network Application with dreamSock

InitNetwork Function
This function initializes the servers socket so we can start serving our clients. The sign-in server uses UDP port 30002.
int CSigninServer::InitNetwork(void) { // Initialize dreamSock and the server if(dreamSock_Initialize() != 0) { LogString("Error initializing Communication Library!"); return 1; } int ret = networkServer->Initialize("", 30002); if(ret == DREAMSOCK_SERVER_ERROR) { #ifdef Win32 char text[64]; sprintf(text, "Could not open server on port %d", networkServer->GetPort()); MessageBox(NULL, text, "Error", MB_OK); #else LogString("Could not open server on port %d", networkServer->GetPort()); #endif return 1; } return 0; }

ShutdownNetwork Function
This function uninitializes the server and tells everyone we are going down.
void CSigninServer::ShutdownNetwork(void) { LogString("Shutting down sign-in server..."); SendExitNotification(); RemoveClients(); }

ReadPackets Function
This function reads the packets from the clients. It functions the same way as it does on the client side. Messages are read into a dreamMessage, and they are parsed based on the message type.

Tutorial 3 / Creating a Basic Network Application with dreamSock

363

void CSigninServer::ReadPackets(void) { char data[1400]; int type; int ret; // Some incoming data char password[50]; int respond; char nickname[30]; char surname[30]; char firstname[30]; char gender[10]; char email[30]; int age; struct sockaddr address; clientLoginData *clList; dreamMessage mes; mes.Init(data, sizeof(data)); // Get the packet from the socket try { while(ret = networkServer->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading(); type = mes.ReadByte(); // Check the type of the message switch(type) { case DREAMSOCK_MES_CONNECT: AddClient(); break; case DREAMSOCK_MES_DISCONNECT: RemoveClient(&address); break; case USER_MES_SIGNIN: // Skip sequences mes.ReadShort(); mes.ReadShort(); strcpy(nickname, mes.ReadString()); strcpy(firstname, mes.ReadString()); strcpy(surname, mes.ReadString());

364

Tutorial 3 / Creating a Basic Network Application with dreamSock

age = mes.ReadByte(); strcpy(gender, mes.ReadString()); strcpy(password, mes.ReadString()); strcpy(email, mes.ReadString()); LogString("Signin: Player %s signed in", nickname); // MySQL connection comes here respond = SIGNIN_RESULT_ACCEPTED; // Find the correct client by comparing addresses clList = clientList; for(; clList != NULL; clList = clList->next) { if(memcmp(clList->netClient-> GetSocketAddress(), &address, sizeof(address)) == 0) { clList->netClient->message.Init (clList->netClient-> message.outgoingData, sizeof(clList-> netClient->message.outgoingData)); clList->netClient->message.WriteByte (USER_MES_SIGNIN); // type clList->netClient->message.AddSequences (clList->netClient); // sequences clList->netClient->message.WriteShort (respond); // respond clList->netClient->SendPacket(); LogString("Sending signin respond"); break; } } break; } } } catch(...) { LogString("Unknown Exception caught in Signin ReadPackets loop"); #ifdef Win32 MessageBox(NULL, "Unknown Exception caught in Signin ReadPackets loop", "Error", MB_OK | MB_TASKMODAL); #endif } }

Tutorial 3 / Creating a Basic Network Application with dreamSock

365

Here we process the sign-in message that comes from a client. We do not use MySQL in this tutorial, but you can see where the MySQL connection would be placed. Obviously when we have the information to put into the database, we would put it there.
case USER_MES_SIGNIN: // Skip sequences mes.ReadShort(); mes.ReadShort(); strcpy(nickname, mes.ReadString()); strcpy(firstname, mes.ReadString()); strcpy(surname, mes.ReadString()); age = mes.ReadByte(); strcpy(gender, mes.ReadString()); strcpy(password, mes.ReadString()); strcpy(email, mes.ReadString()); LogString("Signin: Player %s signed in", nickname); // MySQL connection comes here respond = SIGNIN_RESULT_ACCEPTED; ...

We need to find out which host sent us this message. We compare the addresses of the old clients (in the client list) to the address the message came from. When we find the correct one, we send the sign-in response.
// Find the correct client by comparing addresses clList = clientList; for(; clList != NULL; clList = clList->next) { if(memcmp(clList->netClient->GetSocketAddress(), &address, sizeof(address)) == 0) { clList->netClient->message.Init(clList->netClient-> message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_MES_SIGNIN); // type clList->netClient->message.AddSequences(clList->netClient); // sequences clList->netClient->message.WriteShort(respond); // respond clList->netClient->SendPacket(); LogString("Sending signin respond");

366

Tutorial 3 / Creating a Basic Network Application with dreamSock

break; } }

SendExitNotification Function
This function sends a packet to each client to tell them that the server is going down. We use the dreamClients internal message, so we can use dreamServers SendPackets function to send the packets all at once.
void CSigninServer::SendExitNotification(void) { clientLoginData *toClient = clientList; for(; toClient != NULL; toClient = toClient->next) { toClient->netClient->message.Init(toClient->netClient-> message.outgoingData, sizeof(toClient->netClient-> message.outgoingData)); toClient->netClient->message.WriteByte(USER_MES_SERVEREXIT); // type toClient->netClient->message.AddSequences(toClient->netClient); // sequences } networkServer->SendPackets(); }

AddClient Function
The AddClient function adds a client to the server applications own client list (filling the application-specific data). Notice how we set the netClient pointer to point to the matching client in dreamSocks client list.
void CSigninServer::AddClient(void) { // First get a pointer to the beginning of client list clientLoginData *list = clientList; clientLoginData *prev; dreamClient *netList = networkServer->GetClientList(); // No clients yet, adding the first one if(clientList == NULL) { LogString("App: Server: Adding first client"); clientList = (clientLoginData *) calloc(1, sizeof(clientLoginData));

Tutorial 3 / Creating a Basic Network Application with dreamSock

367

clientList->netClient = netList; clientList->next = NULL; } else { LogString("App: Server: Adding another client"); prev = list; list = clientList->next; netList = netList->next; while(list != NULL) { prev = list; list = list->next; netList = netList->next; } list = (clientLoginData *) calloc(1, sizeof(clientLoginData)); list->netClient = netList; list->next = NULL; prev->next = list; } }

RemoveClient Function
This function removes a client from the server applications client list.
void CSigninServer::RemoveClient(struct sockaddr *address) { clientLoginData *list = clientList; clientLoginData *prev = NULL; clientLoginData *next = NULL; for(; list != NULL; list = list->next) { if(memcmp(list->netClient->GetSocketAddress(), address, sizeof(address)) == 0) { if(prev != NULL) { prev->next = list->next; } break; }

368

Tutorial 3 / Creating a Basic Network Application with dreamSock

prev = list; } if(list == clientList) { if(list) { next = list->next; free(list); } list = NULL; clientList = next; } else { if(list) { next = list->next; free(list); } list = next; } }

This function takes one parameter (struct sockaddr *address), which is the socket address of the client to remove. The address is looked up from the client list, and the matching client is removed.

RemoveClients Function
This function removes all the clients from the server applications client list.
void CSigninServer::RemoveClients(void) { clientLoginData *list = clientList; clientLoginData *next; while(list != NULL) { if(list) { next = list->next; free(list); } list = next; } clientList = NULL; }

Tutorial 3 / Creating a Basic Network Application with dreamSock

369

RunNetwork Function
This function is run every frame. Its purpose is to keep the network running.
void CSigninServer::RunNetwork(int msec) { ReadPackets(); }

Summary
In this tutorial we learned how to create a basic network application with dreamSock. We learned how to create both a client application and a server application. Together they make a complete network application. We now know how to send and receive messages, so we can move on to more complicated things in our game tutorial. But before we do, we create the game lobby in the next tutorial.

This page intentionally left blank.

Tutorial 4

Creating the Game Lobby

Introduction
In this tutorial we learn how to create the lobby for our game, where the players can chat and create games to play. We will see how to use the network library to make the lobby work, but we will also learn how to create the lobby dialog itself. The lobby dialog is the screen where you see all the players, their chat messages, and the existing games. After this tutorial we can actually create and join the games in the lobby, but we cannot start the game yet. Starting and making the game work is explained in Tutorial 5, Creating Your Online Game.

Creating the Lobby Client Application

We will first create the client application for the lobby, which is an updated version of the previous tutorial. This time when you log in, the lobby dialog is opened.

371

372

Tutorial 4 / Creating the Game Lobby

Creating the Dialogs

It is best to start off with creating the dialogs that we will use in the lobby. When we have the dialogs ready, we know what buttons there are, what dialog opens when you press the buttons, and so on.

Lobby Dialog
The lobby dialog is the main screen of the lobby. There you see all the players who have logged in, their chat messages, and the existing games. You can also see the buttons for creating a new game and joining an existing one. This dialog should also have a button for logging out from the server. We create the lobby dialog exactly the same way we created the earlier dialogs for login and signup. As we recall from the earlier tutorial, the dialogs are created from the Insert menu in Visual Studio. Select Resource, Dialog to choose what to create, and then click the New button to create a new dialog. Now you see a new dialog (like the one when you created the login dialogs) in front of you, but it does not look at all like we want it to. So lets modify it. Before we do anything else, we rename the dialog to IDD_LOBBYDIALOG. Click the right mouse button on the name of the new dialog in the list on the left side. Click on Properties and type the new name in. Now you can close the properties window.
Figure 1

Once the name is set up properly, we can change the style of the dialog to what we want. Right-click anywhere and click on Properties in the pop-up menu (or double-click on the empty space in the dialog preview window) to open the Dialog Properties window. The General tab does not have anything interesting for us in this case, so move on to the next tab the Styles tab. In the Styles tab you see many check boxes and two combo boxes. Uncheck all the check boxes, choose Child as the dialog style from the upper combo box, and choose None for the border in the lower combo box. Now our dialog is a child window and has no border of its own. It also does not have a title bar showing its name. We do all of this because this dialog fits into the

Tutorial 4 / Creating the Game Lobby

373

game main window, and therefore the lobby dialog actually is not a window of its own.
Figure 2

The other tabs in the Dialog Properties window do not interest us, so it can be closed now. Next, remove the OK and Cancel buttons that were added automatically to the dialog when it was created.
Figure 3

Now we have an empty dialog that is ready for adding new buttons, edit boxes, and lists. Figure 4 shows what the complete lobby dialog should look like.
Figure 4

Now that the dialog is all laid out, we need to name the controls so we can use them in the code as well. They do already have names, but those are the default names based on the control type, so to make

374

Tutorial 4 / Creating the Game Lobby

everything as clear as possible, we name the controls to correspond to their usage. Here are the three list control names starting from left to right:
IDC_CHATLIST IDC_PLAYERLIST IDC_GAMELIST

We do not wish to sort any of the list items in any list in the lobby, so we disable the sorting option from each one. Double-click on the list (or right-click and choose Properties from the pop-up menu) to open the List Box Properties window. Uncheck the Sort check box on the Styles tab and close the properties window. Repeat this for all the list boxes.
Figure 5

The chat edit box and the Send button are named as follows:
IDC_CHATMESSAGE IDC_SENDCHATMESSAGE

The Create New Game, Join Selected Game, and Log Out buttons are named as follows:
IDC_CREATEGAME IDC_JOINGAME IDC_LOGOUT

Note that the Join Selected Game button is disabled by default because you can join a game only when you have selected one and when the selected game is not in progress. Disabling the button by default is done from the Push Button Properties window, which is opened by double-clicking on the button (or right-clicking on it and choosing Properties from the pop-up menu). Check the Disabled check box and close the properties window.
Figure 6

Tutorial 4 / Creating the Game Lobby

375

The lobby dialog is complete. All we need to do now is create the other dialogs that open when you press the buttons in the lobby dialog.

Create Game Dialog

We create the Create Game dialog the same way as the lobby dialog. Lets start by naming the dialog IDD_CREATEGAME in the Dialog Properties window. Next, open another Dialog Properties window by double-clicking on the empty space in the preview window.
Figure 7

On the Styles tab, set the dialog style to Popup and the border to Dialog Frame. Make sure the Title bar check box is checked, and close the properties window. Lay out the dialog as shown in the following figure.
Figure 8

Then name the controls as shown in the following list, putting the edit box name first, then the Cancel and OK buttons.
IDC_GAMENAME IDC_CANCELCREATEGAME IDC_DOCREATEGAME

When the user presses the OK button, a dialog will open that shows the players who have joined that game and gives the game host the ability to start or cancel the game.

376

Tutorial 4 / Creating the Game Lobby

Create View Players Dialog

Again, we create another dialog as we have done before, this time naming it IDD_CREATEVIEWPLAYERS. This dialog has the same properties as the Create Game dialog, so we will not go through them here. Lay out the dialog as shown in the following figure.
Figure 9

Then name the controls as listed here. The first name in the list is for the player list box, the second for the Cancel button, and the last one for the Start Game button.
IDC_PLAYERSINGAME IDC_CANCELGAME IDC_STARTGAME

The Start Game button starts the actual game, but in this tutorial we stop at this dialog. Pressing the Start Game button at this point does nothing.

Join Game Dialog

Once more we create a dialog as we have created all the other dialogs. This is the very last dialog to create for this project. Name it IDD_JOINVIEWPLAYERS, as it is used only to show the players who have joined the game. We can also cancel the join if we do not wish to be part of the game after all. Lay out the dialog as shown here.

Figure 10

Tutorial 4 / Creating the Game Lobby

377

Then name the two controls as follows:

IDC_JOINPLAYERSINGAME IDC_JOINCANCEL

Now all the dialogs are done, and we can move on to program the lobby system.

Lobby System Code

The lobbys network part works the same way as the sign-in systems network part, so we will not focus on that here. If you have not read Tutorial 3 yet, you should do so now. This tutorial shares a lot of similar code with that one, but here we focus more on how to make the lobby work and build it on top of our network system.

Lobby Client Code

Lets begin with the client-side application. First we will take a look at the source code files: lobby.cpp, main.cpp, lobby.h, main.h, and network.h.

lobby.h File
This is the file for the lobbys CLobby class. It is similar to the CSignin class in the previous tutorial.
#ifndef __LOBBY_H__ #define __LOBBY_H__ #include "network.h" #include "main.h" #define #define #define #define LOBBYLOGIN_RESULT_ACCEPTED LOBBYLOGIN_RESULT_USERNAMEBAD LOBBYLOGIN_RESULT_PASSWORDBAD LOBBYLOGIN_RESULT_MYSQLERROR 200 201 202 203

class CLobby { private: dreamClient clientLoginData clientLoginData

networkClient; clientList; *localClient;

// Pointer to the local client in // the client list

378

Tutorial 4 / Creating the Game Lobby

int int

gameAmount; timeConnecting;

// How long we have tried // to connect

public: CLobby(); ~CLobby(); dreamClient clientLoginData void void void void void void void void void void void void void void void void void void void int }; extern CLobby Lobby; #endif *GetNetworkClient(void) {return networkClient;} *GetLocalClient(void) {return localClient;}

RefreshPlayerList(void); RefreshGameList(void); RefreshJoinedPlayersList(void); ReadPackets(void); AddClient(int local, int index, char *name); RemoveClient(int index); RemoveClients(void); AddGame(char *name, int index, bool inProgress); RemoveGame(char *name); RemoveGames(void); RequestGameData(void); SendChat(char *text); SendCreateGame(char *gamename); SendRemoveGame(char *gamename); SendStartGame(int ind); SendKeepAlive(void); Connect(char *name, char *password); Disconnect(void); RunNetwork(int msec); GetGameAmount(void) {return gameAmount;}

There are a couple of member variables that we should take a look at:
int int gameAmount; timeConnecting;

The integer gameAmount tells us how many games exist at the moment. The integer timeConnecting tells us how long we have tried to connect to the lobby server (in milliseconds).

Tutorial 4 / Creating the Game Lobby

379

network.h File
This file is almost the same as network.h in the sign-in tutorial, but now we have more message types.
#ifndef NETWORK_H #define NETWORK_H #define #define #define #define #define #define #define #define #define #define USER_MES_SERVEREXIT USER_MES_LOGIN USER_MES_SIGNIN USER_MES_CHAT USER_MES_CREATEGAME USER_MES_REMOVEGAME USER_MES_GAMEDATA USER_MES_STARTGAME USER_MES_MAPDATA USER_MES_KEEPALIVE 1 2 3 4 5 6 7 8 9 10

typedef struct clientLoginData { int index; char nickname[30]; clientLoginData *next; } clientLoginData; #endif

main.h File
This is the main application header file, which sounds more important than it is. We have only some externs and function prototypes here.
#ifndef __TUTMAIN_H__ #define __TUTMAIN_H__ LRESULT CALLBACK ApplicationProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam); extern char serverIP[32]; extern extern extern extern extern HINSTANCE hInst; HWND hWnd_Application; HWND hWnd_CreateAccountDialog; HWND hWnd_LoginDialog; HWND hWnd_LobbyDialog;

extern HWND hWnd_CreateGameDialog; extern HWND hWnd_JoinGameDialog; extern HWND hWnd_CreateViewPlayersDialog; #endif

380

Tutorial 4 / Creating the Game Lobby

main.cpp File
This time we will not look at the entire file as it is basically the same as before. Instead, we will look at the new functions directly.

CreateViewPlayersDialogProc Function
This function handles the dialog for viewing players who join the created game. The dialog has Start Game and Cancel buttons, and they are processed here.
LRESULT CALLBACK CreateViewPlayersDialogProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { switch (uMsg) { case WM_COMMAND: switch(LOWORD(wParam)) { case IDC_STARTGAME: DestroyWindow(hWnd_CreateViewPlayersDialog); break; case IDC_CANCELGAME: DestroyWindow(hWnd_CreateViewPlayersDialog); break; default: break; } break; } return 0; }

CreateGameDialogProc Function
This dialog procedure takes in the name for a new game and sends a create game message to the server if the player wants to create a game. It also makes sure that you cannot press the Create Game button unless you have entered a name for the game.
LRESULT CALLBACK CreateGameDialogProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { char gamename[32]; switch (uMsg) {

Tutorial 4 / Creating the Game Lobby

381

case WM_COMMAND: switch(LOWORD(wParam)) { case IDC_DOCREATEGAME: GetWindowText(GetDlgItem(hWnd_CreateGameDialog, IDC_GAMENAME), gamename, 32); DestroyWindow(hWnd_CreateGameDialog); hWnd_CreateViewPlayersDialog = CreateDialog(hInst, MAKEINTRESOURCE(IDD_CREATEVIEWPLAYERS), hWnd_Application, (DLGPROC) CreateViewPlayersDialogProc); ShowWindow(hWnd_CreateViewPlayersDialog, SW_SHOW); Lobby.SendCreateGame(gamename); break; case IDC_CANCELCREATEGAME: DestroyWindow(hWnd_CreateGameDialog); break; default: if(SendMessage(GetDlgItem(hWnd_CreateGameDialog, IDC_GAMENAME), EM_GETMODIFY, 0, 0)) { GetWindowText(GetDlgItem(hWnd_CreateGameDialog, IDC_GAMENAME), gamename, 32); if(strcmp(gamename, "") == 0) { EnableWindow(GetDlgItem(hWnd_CreateGameDialog, IDC_DOCREATEGAME), FALSE); } else { EnableWindow(GetDlgItem(hWnd_CreateGameDialog, IDC_DOCREATEGAME), TRUE); } } break; } break; } return 0; }

382

Tutorial 4 / Creating the Game Lobby

JoinGameDialogProc Function
This dialog procedure runs in the background of the dialog that shows us the players who have joined the game.
LRESULT CALLBACK JoinGameDialogProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { switch (uMsg) { case WM_COMMAND: switch(LOWORD(wParam)) { case IDC_JOINCANCEL: DestroyWindow(hWnd_JoinGameDialog); break; default: break; } break; } return 0; }

LoginDialogProc Function
This function takes in the login information and tries to connect the server. The lobby server runs on UDP port 30003.
LRESULT CALLBACK LoginDialogProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { char nickname[30]; char password[50]; int ret; switch (uMsg) { case WM_COMMAND: { switch(LOWORD(wParam)) { case IDC_LOGIN_QUIT: PostQuitMessage(0); break; case IDC_LOGIN_CREATEACCOUNT: if(!hWnd_CreateAccountDialog) { hWnd_CreateAccountDialog = CreateDialog(hInst,

Tutorial 4 / Creating the Game Lobby

383

MAKEINTRESOURCE(IDD_CREATEACCOUNT), hWnd_Application, (DLGPROC) CreateAccountDialogProc); } break; case IDC_DOLOGIN: // -> First get the IP address of the server // from the dialog GetDlgItemText(hWnd_LoginDialog, IDC_LOGIN_IPADDRESS, serverIP, 16); // -> Store the player data in local variables GetDlgItemText(hWnd_LoginDialog, IDC_LOGIN_NICKNAME, nickname, 30); GetDlgItemText(hWnd_LoginDialog, IDC_LOGIN_PASSWORD, password, 50); // -> Hide the login window ShowWindow(hWnd_LoginDialog, SW_HIDE); ret = Lobby.GetNetworkClient()->Initialize("", serverIP, 30003); if(ret == DREAMSOCK_CLIENT_ERROR) { char text[64]; sprintf(text, "Could not open client socket"); MessageBox(NULL, text, "Error", MB_OK); } Lobby.Connect(nickname, password); break; default: break; } return 0; } case WM_CLOSE: { PostQuitMessage(0); break; } } return 0; }

384

Tutorial 4 / Creating the Game Lobby

LobbyDialogProc Function
Now here is the main lobby dialog procedure function. It controls all the user commands in the lobby screen, such as chat messaging.
LRESULT CALLBACK LobbyDialogProc(HWND hWnd, UINT uMsg, WPARAM wParam, LPARAM lParam) { char chatMessage[256]; char temp[256]; int selectedGame; switch (uMsg) { case WM_COMMAND: switch(LOWORD(wParam)) { case IDC_LOGOUT: Lobby.Disconnect(); // Hide the lobby ShowWindow(hWnd_LobbyDialog, SW_HIDE); // Recreate the login dialog hWnd_LoginDialog = CreateDialog(hInst,MAKEINTRESOURCE (IDD_LOGINDIALOG),hWnd_Application, (DLGPROC)LoginDialogProc); break; case IDC_SENDCHATMESSAGE: GetWindowText(GetDlgItem(hWnd_LobbyDialog, IDC_CHATMESSAGE), temp, 255); sprintf(chatMessage, "%s: ", Lobby.GetLocalClient()-> nickname); strcat(chatMessage, temp); Lobby.SendChat(chatMessage); SetWindowText(GetDlgItem(hWnd_LobbyDialog, IDC_CHATMESSAGE), ""); break; case IDC_CREATEGAME: hWnd_CreateGameDialog = CreateDialog(hInst, MAKEINTRESOURCE(IDD_CREATEGAME), hWnd_Application, (DLGPROC) CreateGameDialogProc); ShowWindow(hWnd_CreateGameDialog, SW_SHOW); break; case IDC_JOINGAME: selectedGame = SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_GAMELIST), LB_GETCURSEL, 0, 0);

Tutorial 4 / Creating the Game Lobby

385

hWnd_JoinGameDialog = CreateDialog(hInst, MAKEINTRESOURCE(IDD_JOINVIEWPLAYERS), hWnd_Application, (DLGPROC) JoinGameDialogProc); ShowWindow(hWnd_JoinGameDialog, SW_SHOW); break; default: int count = SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_GAMELIST), LB_GETSELCOUNT, 0, 0); if(count) { int sel = SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_GAMELIST), LB_GETCURSEL, 0, 0); } else { EnableWindow(GetDlgItem(hWnd_LobbyDialog, IDC_JOINGAME), FALSE); } break; } break; case WM_CLOSE: Lobby.Disconnect(); PostQuitMessage(0); break; } return 0; }

If the player presses the Log Out button, the client disconnects from the server and hides the lobby window, bringing the login dialog in front.
case IDC_LOGOUT: Lobby.Disconnect(); // Hide the lobby ShowWindow(hWnd_LobbyDialog, SW_HIDE); // Recreate the login dialog hWnd_LoginDialog = CreateDialog(hInst,MAKEINTRESOURCE(IDD_LOGINDIALOG), hWnd_Application,(DLGPROC)LoginDialogProc); break;

When the player presses the Send Chat Message button, the text entered in the chat text box is sent to the server in a chat message. The players name is also included in the text.

386

Tutorial 4 / Creating the Game Lobby

case IDC_SENDCHATMESSAGE: GetWindowText(GetDlgItem(hWnd_LobbyDialog, IDC_CHATMESSAGE), temp, 255); sprintf(chatMessage, "%s: ", Lobby.GetLocalClient()->nickname); strcat(chatMessage, temp); Lobby.SendChat(chatMessage); SetWindowText(GetDlgItem(hWnd_LobbyDialog, IDC_CHATMESSAGE), ""); break;

When the player presses the Create Game button, the create game dialog opens.
case IDC_CREATEGAME: hWnd_CreateGameDialog = CreateDialog(hInst, MAKEINTRESOURCE (IDD_CREATEGAME), hWnd_Application, (DLGPROC) CreateGameDialogProc); ShowWindow(hWnd_CreateGameDialog, SW_SHOW); break;

When the player presses the Join Game button, the join game dialog opens. The selected game is retrieved with a Windows system function.
case IDC_JOINGAME: selectedGame = SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_GAMELIST), LB_GETCURSEL, 0, 0); hWnd_JoinGameDialog = CreateDialog(hInst, MAKEINTRESOURCE (IDD_JOINVIEWPLAYERS), hWnd_Application, (DLGPROC) JoinGameDialogProc); ShowWindow(hWnd_JoinGameDialog, SW_SHOW); break;

WinMain Function
This function is almost the same as before but with a small difference. This time we run the frame for both sign-in and lobby systems.
int PASCAL WinMain(HINSTANCE hInstance, HINSTANCE hPrevInstance, TCHAR *pCmdLine, int nCmdShow) { int time, oldTime, newTime; WNDCLASSEX wcl; // Create our main window wcl.cbSize = sizeof(WNDCLASSEX); wcl.hInstance = hInstance; wcl.lpszClassName = "ArmyWar";

Tutorial 4 / Creating the Game Lobby

387

wcl.lpfnWndProc wcl.style wcl.hIcon wcl.hIconSm wcl.hCursor wcl.lpszMenuName wcl.cbClsExtra wcl.cbWndExtra

= ApplicationProc; = 0; = LoadIcon(NULL, IDI_APPLICATION); = LoadIcon(NULL, IDI_WINLOGO); = LoadCursor(NULL, IDC_ARROW); = NULL; = 0; = 0;

wcl.hbrBackground = (HBRUSH) GetStockObject(LTGRAY_BRUSH); if(!RegisterClassEx(&wcl)) return 0; hWnd_Application = CreateWindow( "ArmyWar", "ARMY WAR Online 2.0", WS_OVERLAPPEDWINDOW, CW_USEDEFAULT, CW_USEDEFAULT, 640, 480, HWND_DESKTOP, NULL, hInstance, NULL ); // Initialize the network library if(dreamSock_Initialize() != 0) { MessageBox(NULL, "Error initializing Communication Library!", "Fatal Error", MB_OK); return 1; } ShowWindow(hWnd_Application, nCmdShow); UpdateWindow(hWnd_Application); // Set global instance variable hInst = hInstance; // Display the login dialog hWnd_LoginDialog = CreateDialog(hInst, MAKEINTRESOURCE(IDD_LOGINDIALOG), hWnd_Application, (DLGPROC)LoginDialogProc); // Create the lobby hWnd_LobbyDialog = CreateDialog(hInst, MAKEINTRESOURCE(IDD_LOBBYDIALOG), hWnd_Application, (DLGPROC)LobbyDialogProc); MSG msg;

388

Tutorial 4 / Creating the Game Lobby

BOOL bMsg = FALSE; PeekMessage(&msg, NULL, 0, 0, PM_NOREMOVE); oldTime = dreamSock_GetCurrentSystemTime(); bool done = false; try { while(!done) { while(PeekMessage(&msg, NULL, 0, 0, PM_NOREMOVE)) { if(!GetMessage(&msg, NULL, 0, 0)) { Lobby.Disconnect(); Signin.Disconnect(); done = true; } TranslateMessage(&msg); DispatchMessage(&msg); } do { newTime = dreamSock_GetCurrentSystemTime(); time = newTime oldTime; } while (time < 1); // Run lobby and sign-in network Lobby.RunNetwork(time); Signin.RunNetwork(time); // Run the game frame here oldTime = newTime; } } catch(...) { Lobby.Disconnect(); Signin.Disconnect(); LogString("Unknown Exception caught in main loop"); MessageBox(NULL, "Unknown Exception caught in main loop", "Error", MB_OK | MB_TASKMODAL); return 1; }

Tutorial 4 / Creating the Game Lobby

389

return msg.wParam; }

Here we run the frames for sign-in and lobby:

// Run lobby and sign-in network Lobby.RunNetwork(time); Signin.RunNetwork(time);

lobby.cpp File CLobby Class Methods

Here is the lobby.cpp file. As you can see, it is almost the same as signin.cpp. This time we will focus only on new and modified functions.
/******************************************/ /* Programming Multiplayer Games */ /* Tutorial game client */ /* Programming: */ /* Teijo Hakala */ /******************************************/ #include "common.h" #include "resource.h" //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CLobby::CLobby() { networkClient = new dreamClient; clientList = NULL; gameAmount = 0; timeConnecting = 0; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CLobby::~CLobby() { delete networkClient; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::RefreshPlayerList(void) {

390

Tutorial 4 / Creating the Game Lobby

SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_PLAYERLIST), LB_RESETCONTENT, 0, 0); clientLoginData *list = clientList; for(; list != NULL; list = list->next) { SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_PLAYERLIST), LB_ADDSTRING, 0, (LPARAM) list->nickname); } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::RefreshGameList(void) { } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::RefreshJoinedPlayersList(void) { } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::ReadPackets(void) { char data[1400]; struct sockaddr address; int type; int ind; int local; int ret; char name[30]; dreamMessage mes; mes.Init(data, sizeof(data)); while(ret = networkClient->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading(); type = mes.ReadByte(); switch(type)

Tutorial 4 / Creating the Game Lobby

391

{ case DREAMSOCK_MES_ADDCLIENT: local = mes.ReadByte(); ind = mes.ReadByte(); strcpy(name, mes.ReadString()); AddClient(local, ind, name); break; case DREAMSOCK_MES_REMOVECLIENT: ind = mes.ReadByte(); LogString("Got removeclient %d message", ind); RemoveClient(ind); break; case USER_MES_SERVEREXIT: MessageBox(NULL, "Server disconnected", "Info", MB_OK); Disconnect(); break; case USER_MES_LOGIN: // Skip sequences mes.ReadShort(); mes.ReadShort(); ret = mes.ReadShort(); LogString("Got lobby login respond %d", ret); if(ret != LOBBYLOGIN_RESULT_ACCEPTED) { MessageBox(NULL, "Nickname or password is not valid", "Error", MB_OK); Disconnect(); return; } SetWindowText(hWnd_Application, "ARMY WAR Online 2.0 connected"); timeConnecting = 1; ShowWindow(hWnd_LobbyDialog, SW_SHOW); break; case USER_MES_CHAT: // Skip sequences mes.ReadShort(); mes.ReadShort();

392

Tutorial 4 / Creating the Game Lobby

SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_CHATLIST), LB_ADDSTRING, 0, (LPARAM) mes.ReadString()); break; } } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::AddClient(int local, int ind, char *name) { // First get a pointer to the beginning of client list clientLoginData *list = clientList; clientLoginData *prev; LogString("App: Client: Adding client with index %d", ind); // No clients yet, adding the first one if(clientList == NULL) { LogString("App: Client: Adding first client"); clientList = (clientLoginData *) calloc(1, sizeof(clientLoginData)); if(local) { LogString("App: Client: This one is local"); localClient = clientList; } clientList->index = ind; strcpy(clientList->nickname, name); strcpy(clientList->nickname, name); clientList->next = NULL; } else { LogString("App: Client: Adding another client"); prev = list; list = clientList->next; while(list != NULL) { prev = list; list = list->next; }

Tutorial 4 / Creating the Game Lobby

393

list = (clientLoginData *) calloc(1, sizeof(clientLoginData)); if(local) { LogString("App: Client: This one is local"); localClient = list; } list->index = ind; strcpy(list->nickname, name); list->next = NULL; prev->next = list; } RefreshPlayerList(); // If we just joined the game, request all the game data if(local) RequestGameData(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::RemoveClient(int ind) { clientLoginData *list = clientList; clientLoginData *prev = NULL; clientLoginData *next = NULL; for(; list != NULL; list = list->next) { if(list->index == ind) { if(prev != NULL) { prev->next = list->next; } break; } prev = list; } if(list == clientList) { if(list) { next = list->next; free(list);

394

Tutorial 4 / Creating the Game Lobby

} list = NULL; clientList = next; } else { if(list) { next = list->next; free(list); } list = next; } RefreshPlayerList(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::RemoveClients(void) { clientLoginData *list = clientList; clientLoginData *next; while(list != NULL) { if(list) { next = list->next; free(list); } list = next; } clientList = NULL; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::AddGame(char *name, int ind, bool inProgress) { } //--------------------------------------------------------------------------// Name: empty() // Desc: //---------------------------------------------------------------------------

Tutorial 4 / Creating the Game Lobby

395

void CLobby::RemoveGame(char *name) { } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::RemoveGames(void) { } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::RequestGameData(void) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_GAMEDATA); message.AddSequences(networkClient); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::SendChat(char *text) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_CHAT); message.AddSequences(networkClient); message.WriteString(text); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::SendCreateGame(char *gamename) { char data[1400]; dreamMessage message;

396

Tutorial 4 / Creating the Game Lobby

message.Init(data, sizeof(data)); message.WriteByte(USER_MES_CREATEGAME); message.AddSequences(networkClient); message.WriteString(gamename); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::SendRemoveGame(char *gamename) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_REMOVEGAME); message.AddSequences(networkClient); message.WriteString(gamename); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::SendStartGame(int ind) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_STARTGAME); message.AddSequences(networkClient); message.WriteByte(ind); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::SendKeepAlive(void) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_KEEPALIVE);

Tutorial 4 / Creating the Game Lobby

397

message.AddSequences(networkClient); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::Connect(char *name, char *password) { LogString("CLobby::Connect"); timeConnecting = 0; SetWindowText(hWnd_Application, "ARMY WAR Online 2.0 - connecting ..."); networkClient->SendConnect(name); dreamMessage message; char data[1400]; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_LOGIN); message.AddSequences(networkClient); message.WriteString(name); message.WriteString(password); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::Disconnect(void) { LogString("CLobby::Disconnect"); timeConnecting = 1; SetWindowText(hWnd_Application, "ARMY WAR Online 2.0"); localClient = NULL; RemoveClients(); RemoveGames(); networkClient->SendDisconnect(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobby::RunNetwork(int msec) {

// // // //

type sequences name password

398

Tutorial 4 / Creating the Game Lobby

if(networkClient->GetConnectionState() == DREAMSOCK_DISCONNECTED) return; static int time = 0; static int keepalive = 0; time += msec; keepalive += msec; if(keepalive > 20000) { SendKeepAlive(); keepalive = 0; } // If timeconnecting is negative, we are connected or connection failed if(timeConnecting > 1) timeConnecting += msec; if(timeConnecting > 3000) { SetWindowText(hWnd_Application, "ARMY WAR Online 2.0"); MessageBox(NULL, "Could not connect", "Connection error", MB_OK); timeConnecting = 1; ShowWindow(hWnd_LoginDialog, SW_SHOW); } // framerate is too high if(time < (1000 / 30)) return; time = 0; ReadPackets(); }

RefreshPlayerList Function
This function updates the player list whenever called. First, the whole list is reset, and then it is filled with all the connected players.
void CLobby::RefreshPlayerList(void) { SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_PLAYERLIST), LB_RESETCONTENT, 0, 0); clientLoginData *list = clientList; for(; list != NULL; list = list->next) {

Tutorial 4 / Creating the Game Lobby

399

SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_PLAYERLIST), LB_ADDSTRING, 0, (LPARAM) list->nickname); } }

ReadPackets Function
Here is the ReadPackets function once again. The lobby system has a few of its own messages that we will look at now.
void CLobby::ReadPackets(void) { char data[1400]; struct sockaddr address; int type; int ind; int local; int ret; char name[30]; dreamMessage mes; mes.Init(data, sizeof(data)); while(ret = networkClient->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading(); type = mes.ReadByte(); switch(type) { case DREAMSOCK_MES_ADDCLIENT: local = mes.ReadByte(); ind = mes.ReadByte(); strcpy(name, mes.ReadString()); AddClient(local, ind, name); break; case DREAMSOCK_MES_REMOVECLIENT: ind = mes.ReadByte(); LogString("Got removeclient %d message", ind); RemoveClient(ind); break; case USER_MES_SERVEREXIT: MessageBox(NULL, "Server disconnected", "Info", MB_OK); Disconnect();

400

Tutorial 4 / Creating the Game Lobby

break; case USER_MES_LOGIN: // Skip sequences mes.ReadShort(); mes.ReadShort(); ret = mes.ReadShort(); LogString("Got lobby login respond %d", ret); if(ret != LOBBYLOGIN_RESULT_ACCEPTED) { MessageBox(NULL, "Nickname or password is not valid", "Error", MB_OK); Disconnect(); return; } SetWindowText(hWnd_Application, "ARMY WAR Online 2.0 connected"); timeConnecting = 1; ShowWindow(hWnd_LobbyDialog, SW_SHOW); break; case USER_MES_CHAT: // Skip sequences mes.ReadShort(); mes.ReadShort(); SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_CHATLIST), LB_ADDSTRING, 0, (LPARAM) mes.ReadString()); break; } } }

If the client receives a login message, we know that the server has processed our login request and now we get the result. If it was not accepted, we disconnect from the server. Otherwise, we show the lobby dialog and set a new window title that has the text connected in it. The variable timeConnecting is set to 1 to indicate that it should not be counted anymore. More on this variable later.
case USER_MES_LOGIN: // Skip sequences mes.ReadShort(); mes.ReadShort();

Tutorial 4 / Creating the Game Lobby

401

ret = mes.ReadShort(); LogString("Got lobby login respond %d", ret); if(ret != LOBBYLOGIN_RESULT_ACCEPTED) { MessageBox(NULL, "Nickname or password is not valid", "Error", MB_OK); Disconnect(); return; } SetWindowText(hWnd_Application, "ARMY WAR Online 2.0 - connected"); timeConnecting = 1; ShowWindow(hWnd_LobbyDialog, SW_SHOW); break;

Lets say someone has written something to the chat system, and the server has sent us the chat message. Here is how we put it up on the chat window.
case USER_MES_CHAT: // Skip sequences mes.ReadShort(); mes.ReadShort(); SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_CHATLIST), LB_ADDSTRING, 0, (LPARAM) mes.ReadString()); break;

RequestGameData Function
This function sends a request game data message to the server.
void CLobby::RequestGameData(void) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_GAMEDATA); message.AddSequences(networkClient); networkClient->SendPacket(&message); }

402

Tutorial 4 / Creating the Game Lobby

SendChat Function
This function sends a chat message to the server.
void CLobby::SendChat(char *text) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_CHAT); message.AddSequences(networkClient); message.WriteString(text); networkClient->SendPacket(&message); }

SendCreateGame Function
This function sends a create game message to the server.
void CLobby::SendCreateGame(char *gamename) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_CREATEGAME); message.AddSequences(networkClient); message.WriteString(gamename); networkClient->SendPacket(&message); }

SendRemoveGame Function
This function sends a remove game message to the server.
void CLobby::SendRemoveGame(char *gamename) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_REMOVEGAME); message.AddSequences(networkClient); message.WriteString(gamename); networkClient->SendPacket(&message); }

Tutorial 4 / Creating the Game Lobby

403

SendStartGame Function
This function sends a start game message to the server.
void CLobby::SendStartGame(int ind) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_STARTGAME); message.AddSequences(networkClient); message.WriteByte(ind); networkClient->SendPacket(&message); }

Connect Function
This function connects the lobby server and then sends the login request right after that. The variable timeConnecting is set to 0, so the system will start counting how long we have tried to connect.
void CLobby::Connect(char *name, char *password) { LogString("CLobby::Connect"); timeConnecting = 0; SetWindowText(hWnd_Application, "ARMY WAR Online 2.0 - connecting ..."); networkClient->SendConnect(name); dreamMessage message; char data[1400]; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_LOGIN); message.AddSequences(networkClient); message.WriteString(name); message.WriteString(password); networkClient->SendPacket(&message); }

// // // //

type sequences name password

Disconnect Function
This function disconnects from the lobby server. Everything is uninitialized.
void CLobby::Disconnect(void) { LogString("CLobby::Disconnect");

404

Tutorial 4 / Creating the Game Lobby

timeConnecting = 1; SetWindowText(hWnd_Application, "ARMY WAR Online 2.0"); localClient = NULL; RemoveClients(); RemoveGames(); networkClient->SendDisconnect(); }

RunNetwork Function
This function is run every frame to keep the lobby network running.
void CLobby::RunNetwork(int msec) { if(networkClient->GetConnectionState() == DREAMSOCK_DISCONNECTED) return; static int time = 0; static int keepalive = 0; time += msec; keepalive += msec; if(keepalive > 20000) { SendKeepAlive(); keepalive = 0; } // If timeConnecting is negative, we are connected or connection failed if(timeConnecting > 1) timeConnecting += msec; if(timeConnecting > 3000) { SetWindowText(hWnd_Application, "ARMY WAR Online 2.0"); MessageBox(NULL, "Could not connect", "Connection error", MB_OK); timeConnecting = 1; ShowWindow(hWnd_LoginDialog, SW_SHOW); } // framerate is too high if(time < (1000 / 30)) return; time = 0; ReadPackets(); }

Tutorial 4 / Creating the Game Lobby

405

Here we count how long we have tried to connect to the server. If it takes longer than it normally should, it probably means that the server is down or we have the wrong address. In this case, we tell the user that the client could not connect to the server, and we bring the login window back up.
// If timeConnecting is negative, we are connected or connection failed if(timeConnecting > 1) timeConnecting += msec; if(timeConnecting > 3000) { SetWindowText(hWnd_Application, "ARMY WAR Online 2.0"); MessageBox(NULL, "Could not connect", "Connection error", MB_OK); timeConnecting = 1; ShowWindow(hWnd_LoginDialog, SW_SHOW); }

Unimplemented Functions
There are some functions that cannot be implemented until we have the game data structure ready. We will implement these functions in the next tutorial, Creating Your Online Game.
void CLobby::RefreshGameList(void) { } void CLobby::RefreshJoinedPlayersList(void) { } void CLobby::AddGame(char *name, int ind, bool inProgress) { } void CLobby::RemoveGame(char *name) { } void CLobby::RemoveGames(void) { }

406

Tutorial 4 / Creating the Game Lobby

Lobby Server Code

We now learn how to create the lobby server. It is very similar to the sign-in server, and so we will only take a look at the new and modified parts of the code.

lobby.h File
Here we have the lobby.h file, which contains CLobbyServer class. It works just like CSigninServer.
#ifndef __LOBBY_H__ #define __LOBBY_H__ #define #define #define #define LOBBY_RESULT_ACCEPTED LOBBY_RESULT_USERNAMEBAD LOBBY_RESULT_PASSWORDBAD LOBBY_RESULT_MYSQLERROR 200 201 202 203

class CLobbyServer { private: dreamServer clientLoginData int gameAmount; public: CLobbyServer(); ~CLobbyServer();

*networkServer; *clientList;

int InitNetwork(void); void ShutdownNetwork(void); void ReadPackets(void); void SendExitNotification(void); void AddClient(void); void RemoveClient(struct sockaddr *address); void RemoveClients(void); void AddGame(char *name); void RemoveGame(char *name); void RemoveGames(void); void Frame(int msec); int }; #endif GetGameAmount(void) {return gameAmount;}

Tutorial 4 / Creating the Game Lobby

407

There is one new variable that tells us how many games exist at the moment: int gameAmount;.

network.h File
This file is almost the same as the one in the sign-in tutorial, but now we have some more message types.
#ifndef NETWORK_H #define NETWORK_H #include "dreamSock.h" #define #define #define #define #define #define USER_MES_SERVEREXIT USER_MES_LOGIN USER_MES_SIGNIN USER_MES_CHAT USER_MES_CREATEGAME USER_MES_REMOVEGAME 1 2 3 4 5 6

typedef struct clientLoginData { dreamClient netClient; clientLoginData next; } clientLoginData; #endif

main.cpp File
This file has so few changes that we will skip right to them. The application entry functions (WinMain and main) now initialize both the sign-in and lobby servers, and run both servers frames.
CLobbyServer Lobby; CSigninServer Signin; ... if(Lobby.InitNetwork() == 1) { PostQuitMessage(0); } if(Signin.InitNetwork() == 1) { PostQuitMessage(0); } ...

408

Tutorial 4 / Creating the Game Lobby

Lobby.Frame(time); Signin.Frame(time);

lobby.cpp File CLobbyServer Class Methods

Here is the lobby.cpp file, which contains the code for CLobbyServer. Note that the lobby server runs on UDP port 30003. We will now take a look at the new and modified functions.
/******************************************/ /* Programming Multiplayer Games */ /* Tutorial game server */ /* Programming: */ /* Teijo Hakala */ /******************************************/ #include "common.h" #include <malloc.h> #include <mysql++> //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CLobbyServer::CLobbyServer() { networkServer = new dreamServer; clientList gameAmount } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CLobbyServer::~CLobbyServer() { delete networkServer; } //--------------------------------------------------------------------------// Name: InitNetwork() // Desc: Initialize network //--------------------------------------------------------------------------int CLobbyServer::InitNetwork(void) { // Initialize dreamSock and the server if(dreamSock_Initialize() != 0) { LogString("Error initializing Communication Library!"); return 1; = NULL; = 0;

Tutorial 4 / Creating the Game Lobby

409

} int ret = networkServer->Initialize("", 30003); if(ret == DREAMSOCK_SERVER_ERROR) { #ifdef Win32 char text[64]; sprintf(text, "Could not open server on port %d", networkServer->GetPort()); MessageBox(NULL, text, "Error", MB_OK); #else LogString("Could not open server on port %d", networkServer->GetPort()); #endif return 1; } return 0; } //--------------------------------------------------------------------------// Name: ShutdownNetwork() // Desc: Shut down network //--------------------------------------------------------------------------void CLobbyServer::ShutdownNetwork(void) { LogString("Shutting down lobby server..."); SendExitNotification(); RemoveClients(); RemoveGames(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobbyServer::ReadPackets(void) { char data[1400]; int type; int ret; // Some incoming data char name[30]; char password[50]; char chatter[50]; int respond; int ind; struct sockaddr address;

410

Tutorial 4 / Creating the Game Lobby

clientLoginData *clList; dreamMessage mes; mes.Init(data, sizeof(data)); // Get the packet from the socket try { while(ret = networkServer->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading(); type = mes.ReadByte(); // Check the type of the message switch(type) { case DREAMSOCK_MES_CONNECT: AddClient(); break; case DREAMSOCK_MES_DISCONNECT: RemoveClient(&address); if(clientList == NULL) RemoveGames(); break; case USER_MES_LOGIN: // Skip sequences mes.ReadShort(); mes.ReadShort(); strcpy(name, mes.ReadString()); strcpy(password, mes.ReadString()); LogString("Lobby: Player %s logged in", name); try { // -> Create a connection to the database Connection con("onlinedata", "127.0.0.1"); // -> Create a query object that is bound // to our connection Query query = con.query(); // -> Assign the query to that object query << "SELECT id,firstname,password FROM playerdata WHERE nickname = \"" << name << "\"";

Tutorial 4 / Creating the Game Lobby

411

// -> Store the results from the query Result res = query.store(); Result::iterator i; Row row; i = res.begin(); if(i!=res.end()) { row = *i; if(!strcmp(password, row["password"])) { // -> Update the 'lastlogin' // field to current date and time query << "UPDATE playerdata SET lastlogin = NULL WHERE id = " << row["id"]; query.execute(); // -> Set the player to 'online' query << "UPDATE playerdata SET online = 1 WHERE id = " << row["id"]; query.execute(); // -> Player login successful! respond = LOBBYLOGIN_RESULT_ ACCEPTED; } else { // -> Password did not match respond = LOBBYLOGIN_RESULT_ PASSWORDBAD; } } else { // -> Nickname could not be found respond = LOBBYLOGIN_RESULT_USERNAMEBAD; } } catch (BadQuery er) // handle any connection errors { // -> MySQL server not running? respond = LOBBYLOGIN_RESULT_MYSQLERROR; } respond = LOBBY_RESULT_ACCEPTED; // Find the correct client by comparing addresses clList = clientList; for(; clList != NULL; clList = clList->next)

412

Tutorial 4 / Creating the Game Lobby

{ if(memcmp(clList->netClient-> GetSocketAddress(), &address, sizeof(address)) == 0) { clList->netClient->message.Init (clList->netClient-> message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte (USER_MES_LOGIN); // type clList->netClient->message.AddSequences (clList->netClient); // sequences clList->netClient->message.WriteShort (respond); // respond clList->netClient->SendPacket(); LogString("Sending lobby login respond"); break; } } break; case USER_MES_CHAT: // Skip sequences mes.ReadShort(); mes.ReadShort(); // Read chat text strcpy(chatter, mes.ReadString()); // Send the chat text to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList-> netClient->message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte (USER_MES_CHAT); // type clList->netClient->message.AddSequences (clList->netClient); // sequences clList->netClient->message.WriteString (chatter); // text } networkServer->SendPackets();

Tutorial 4 / Creating the Game Lobby

413

break; case USER_MES_CREATEGAME: // Skip sequences mes.ReadShort(); mes.ReadShort(); // Read game name strcpy(name, mes.ReadString()); AddGame(name); // Find the correct client by comparing addresses clList = clientList; for(; clList != NULL; clList = clList->next) { if(memcmp(clList->netClient-> GetSocketAddress(), &address, sizeof(address)) == 0) { ind = clList->netClient->GetIndex(); break; } } // Send to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList-> netClient->message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte (USER_MES_CREATEGAME); // type clList->netClient->message.AddSequences (clList->netClient); // sequences clList->netClient->message.WriteString(name); // game name clList->netClient->message.WriteShort(ind); // host's index clList->netClient->message.WriteByte(0); // in progress? } networkServer->SendPackets(); break; case USER_MES_REMOVEGAME: // Skip sequences mes.ReadShort();

414

Tutorial 4 / Creating the Game Lobby

mes.ReadShort(); // Read game name strcpy(name, mes.ReadString()); LogString("REMOVING %s", name); RemoveGame(name); // Send to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList-> netClient->message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_ MES_REMOVEGAME); // type clList->netClient->message.AddSequences (clList->netClient); // sequences clList->netClient->message.WriteString(name); // game name } networkServer->SendPackets(); break; } } } catch(...) { LogString("Unknown Exception caught in Lobby ReadPackets loop"); #ifdef Win32 MessageBox(NULL, "Unknown Exception caught in Lobby ReadPackets loop", "Error", MB_OK | MB_TASKMODAL); #endif } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobbyServer::SendExitNotification(void) { clientLoginData *toClient = clientList; for(; toClient != NULL; toClient = toClient->next) {

Tutorial 4 / Creating the Game Lobby

415

toClient->netClient->message.Init(toClient->netClient-> message.outgoingData, sizeof(toClient->netClient-> message.outgoingData)); toClient->netClient->message.WriteByte(USER_MES_SERVEREXIT); // type toClient->netClient->message.AddSequences(toClient->netClient); // sequences } networkServer->SendPackets(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobbyServer::AddClient(void) { // First get a pointer to the beginning of the client list clientLoginData *list = clientList; clientLoginData *prev; dreamClient *netList = networkServer->GetClientList(); // No clients yet, adding the first one if(clientList == NULL) { LogString("App: Server: Adding first client"); clientList = (clientLoginData *) calloc(1, sizeof(clientLoginData)); clientList->netClient = netList; clientList->next = NULL; } else { LogString("App: Server: Adding another client"); prev = list; list = clientList->next; netList = netList->next; while(list != NULL) { prev = list; list = list->next; netList = netList->next; } list = (clientLoginData *) calloc(1, sizeof(clientLoginData));

416

Tutorial 4 / Creating the Game Lobby

list->netClient = netList; list->next = NULL; prev->next = list; } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobbyServer::RemoveClient(struct sockaddr *address) { clientLoginData *list = clientList; clientLoginData *prev = NULL; clientLoginData *next = NULL; for(; list != NULL; list = list->next) { if(memcmp(list->netClient->GetSocketAddress(), address, sizeof(address)) == 0) { if(prev != NULL) { prev->next = list->next; } break; } prev = list; } if(list == clientList) { if(list) { next = list->next; free(list); } list = NULL; clientList = next; } else { if(list) { next = list->next; free(list); } list = next;

Tutorial 4 / Creating the Game Lobby

417

} } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobbyServer::RemoveClients(void) { clientLoginData *list = clientList; clientLoginData *next; while(list != NULL) { if(list) { next = list->next; free(list); } list = next; } clientList = NULL; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobbyServer::AddGame(char *name) { } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobbyServer::RemoveGame(char *name) { } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CLobbyServer::RemoveGames(void) { } //--------------------------------------------------------------------------// Name: empty() // Desc: //---------------------------------------------------------------------------

418

Tutorial 4 / Creating the Game Lobby

void CLobbyServer::Frame(int msec) { ReadPackets(); }

ReadPackets Function
This function reads the packets from the lobby clients.
void CLobbyServer::ReadPackets(void) { char data[1400]; int type; int ret; // Some incoming data char name[30]; char password[50]; char chatter[50]; int respond; int ind; struct sockaddr address; clientLoginData *clList; dreamMessage mes; mes.Init(data, sizeof(data)); // Get the packet from the socket try { while(ret = networkServer->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading(); type = mes.ReadByte(); // Check the type of the message switch(type) { case DREAMSOCK_MES_CONNECT: AddClient(); break; case DREAMSOCK_MES_DISCONNECT: RemoveClient(&address); if(clientList == NULL) RemoveGames();

Tutorial 4 / Creating the Game Lobby

419

break; case USER_MES_LOGIN: // Skip sequences mes.ReadShort(); mes.ReadShort(); strcpy(name, mes.ReadString()); strcpy(password, mes.ReadString()); LogString("Lobby: Player %s logged in", name); try { // -> Create a connection to the database Connection con("onlinedata", "127.0.0.1"); // -> Create a query object that is bound // to our connection Query query = con.query(); // -> Assign the query to that object query << "SELECT id,firstname,password FROM playerdata WHERE nickname = \"" << name << "\""; // -> Store the results from the query Result res = query.store(); Result::iterator i; Row row; i = res.begin(); if(i!=res.end()) { row = *i; if(!strcmp(password, row["password"])) { // -> Update the 'lastlogin' // field to current date and time query << "UPDATE playerdata SET lastlogin = NULL WHERE id = " << row["id"]; query.execute(); // -> Set the player to 'online' query << "UPDATE playerdata SET online = 1 WHERE id = " << row["id"]; query.execute(); // -> Player login successful! respond = LOBBYLOGIN_RESULT_ ACCEPTED;

420

Tutorial 4 / Creating the Game Lobby

} else { // -> Password did not match respond = LOBBYLOGIN_RESULT_ PASSWORDBAD; } } else { // -> Nickname could not be found respond = LOBBYLOGIN_RESULT_USERNAMEBAD; } } catch (BadQuery er) // handle any connection errors { // -> MySQL server not running? respond = LOBBYLOGIN_RESULT_MYSQLERROR; } // Find the correct client by comparing addresses clList = clientList; for(; clList != NULL; clList = clList->next) { if(memcmp(clList->netClient-> GetSocketAddress(), &address, sizeof(address)) == 0) { clList->netClient->message.Init(clList-> netClient->message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte (USER_MES_LOGIN); // type clList->netClient->message.AddSequences (clList->netClient); // sequences clList->netClient->message.WriteShort (respond); // respond clList->netClient->SendPacket(); LogString("Sending lobby login respond"); break; } } break; case USER_MES_CHAT: // Skip sequences mes.ReadShort();

Tutorial 4 / Creating the Game Lobby

421

mes.ReadShort(); // Read chat text strcpy(chatter, mes.ReadString()); // Send the chat text to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList-> netClient->message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte (USER_MES_CHAT); // type clList->netClient->message.AddSequences (clList->netClient); // sequences clList->netClient->message.WriteString (chatter); // text } networkServer->SendPackets(); break; case USER_MES_CREATEGAME: // Skip sequences mes.ReadShort(); mes.ReadShort(); // Read game name strcpy(name, mes.ReadString()); AddGame(name); // Find the correct client by comparing addresses clList = clientList; for(; clList != NULL; clList = clList->next) { if(memcmp(clList->netClient-> GetSocketAddress(), &address, sizeof(address)) == 0) { ind = clList->netClient->GetIndex(); break; } } // Send to everybody for(clList = clientList; clList != NULL; clList = clList->next) {

422

Tutorial 4 / Creating the Game Lobby

clList->netClient->message.Init(clList-> netClient->message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_ MES_CREATEGAME); // type clList->netClient->message.AddSequences (clList->netClient); // sequences clList->netClient->message.WriteString(name); // game name clList->netClient->message.WriteShort(ind); // host's index clList->netClient->message.WriteByte(0); // in progress? } networkServer->SendPackets(); break; case USER_MES_REMOVEGAME: // Skip sequences mes.ReadShort(); mes.ReadShort(); // Read game name strcpy(name, mes.ReadString()); LogString("REMOVING %s", name); RemoveGame(name); // Send to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList-> netClient->message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_ MES_REMOVEGAME); // type clList->netClient->message.AddSequences (clList->netClient); // sequences clList->netClient->message.WriteString(name); // game name } networkServer->SendPackets(); break; }

Tutorial 4 / Creating the Game Lobby

423

} } catch(...) { LogString("Unknown Exception caught in Lobby ReadPackets loop"); #ifdef Win32 MessageBox(NULL, "Unknown Exception caught in Lobby ReadPackets loop", "Error", MB_OK | MB_TASKMODAL); #endif } }

In case of a chat message, we forward the message to every client and do nothing else.
case USER_MES_CHAT: // Skip sequences mes.ReadShort(); mes.ReadShort(); // Read chat text strcpy(chatter, mes.ReadString()); // Send the chat text to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList->netClient-> message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_MES_CHAT); // type clList->netClient->message.AddSequences(clList->netClient); // sequences clList->netClient->message.WriteString(chatter); // text } networkServer->SendPackets(); break;

If a client sends us the create game message, we first create a game on the server side and then forward the message to every client, so they can add a game too. We find out the index number of the game creator, so that player will open a different kind of dialog than the rest (where he or she can start the game).
case USER_MES_CREATEGAME: // Skip sequences mes.ReadShort(); mes.ReadShort();

424

Tutorial 4 / Creating the Game Lobby

// Read game name strcpy(name, mes.ReadString()); AddGame(name); // Find the correct client by comparing addresses clList = clientList; for(; clList != NULL; clList = clList->next) { if(memcmp(clList->netClient->GetSocketAddress(), &address, sizeof(address)) == 0) { ind = clList->netClient->GetIndex(); break; } } // Send to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList->netClient-> message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_MES_CREATEGAME); // type clList->netClient->message.AddSequences(clList->netClient); // sequences clList->netClient->message.WriteString(name); // game name clList->netClient->message.WriteShort(ind); // host's index clList->netClient->message.WriteByte(0); // in progress? } networkServer->SendPackets(); break;

If we get a remove game message, the server first removes the game locally, and then tells the clients about it so they will remove it too.
case USER_MES_REMOVEGAME: // Skip sequences mes.ReadShort(); mes.ReadShort(); // Read game name strcpy(name, mes.ReadString()); LogString("REMOVING %s", name); RemoveGame(name);

Tutorial 4 / Creating the Game Lobby

425

// Send to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList->netClient-> message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_MES_REMOVEGAME); // type clList->netClient->message.AddSequences(clList->netClient); // sequences clList->netClient->message.WriteString(name); // game name } networkServer->SendPackets(); break;

Unimplemented Functions
The server also has some unimplemented functions (just like the lobby client) at this point. These cannot be implemented until we have the game data structure.
void CLobbyServer::AddGame(char *name) { } void CLobbyServer::RemoveGame(char *name) { } void CLobbyServer::RemoveGames(void) { }

Summary
In this tutorial we learned how to create a working game lobby. We learned how to handle players who are logging in, create games, and send chat messages. Now we have only one final step to take: making the actual game. This is covered in the final tutorial, Creating Your Online Game.

This page intentionally left blank.

Tutorial 5

Creating Your Online Game

Introduction
This is the last tutorial for our project of creating our very own online game. In this tutorial we learn how to make our game handle player input and show the result on the local and remote clients. Basically we learn how we should send the player movement and other command data to the other clients. The server will route all the commands to all the clients, and we can then see the movement of the players and the flag on our screens. We will also learn how to create a dead reckoning system (also known as client prediction). This is a method of making the players move smoothly even when the server is not responding very quickly. All the hosts participating in the network game must produce exactly the same frames. If that is not the case, the servers frame is used. But the rule of thumb is that the server and a client always produce the same frames. Obviously when we send this over the network to the remote clients, they try their best to match it but cannot do it perfectly. So when a client moves a remote clients player on the screen, it is possible that the result is wrong. In this case, the server tells us where to put it.

427

428

Tutorial 5 / Creating Your Online Game

Designing the Functionality

There are some methods to make our game work better over a network. These methods should be kept in mind when designing our games network functionality.

Frame Time
All computer games should keep track of the frame time. This is especially important in a multiplayer game because all the computers could be running the game at slightly different speeds (depending on how powerful the computer is). So when we move something in the game world, we should scale the velocity of the movement with the current frame time. This way the movement is smooth on every frame, and no client is moving the objects faster than any other.

Compressing Messages
While there are obviously ways to compress almost any kind of data, we are not talking about compressing the message byte by byte. Of course, one could implement that kind of method if required, but the packets usually are so small that it is not needed (or even useful). We are talking about delta compression. Delta means change over a time. For example, if we press a button down now, we need to tell that to the server only once, right? Why should we send that same information every frame when it already knows that the key is pressed down? Then again, if we release the key now, we need to send that information to the server. Between pressing down and releasing the key, we do not need to inform the server about that key in anyway. That is called delta compression. Another good example is sending the player coordinates from the server to a client. Why send them if they have not changed? Delta compression is achieved by using single bits of a byte (or any length of data) to tell what is included in a message. This is called flagging, and each bit in the byte is a flag. For example, if the key flag is up, it means that this message contains key data. Flags are defined in values of power of two. The first one is 1, the second one is 2, then 4, 8, 16, 32, 64, etc. A byte consists of 8 bits, so it can have eight of these values. Here are the command flags used in this tutorial game:
#define #define #define #define CMD_KEY CMD_HEADING CMD_ORIGIN CMD_BULLET 1 2 4 8

Tutorial 5 / Creating Your Online Game

429

#define CMD_FLAG #define CMD_KILL

16 32

The following code shows how to set the key flag up and down:
flags |= CMD_KEY; flags &= ~CMD_KEY; // Set the flag UP // Set the flag DOWN

The following code shows how to check if a flag is up:

if(flags & CMD_KEY)

So setting a flag up or down is actually setting a bit to 1 or 0. And because a byte is the smallest amount of data you can send over a network, this is a good way send lots of boolean values in one byte.

Dead Reckoning
Dead reckoning, or client prediction, is an advanced method of scaling down our required network bandwith. The server does not have to inform the clients as often as it would have to without dead reckoning. Dead reckoning is a method used in various places (i.e., not only in network games), such as aviation. It is used to predict our position at a given time, using our current position and velocity to calculate it. Normally when we press a key or do any action in the game, information about this is sent to the server, and the server sends it to each client. When the information reaches the clients, they make the action happen. Now, as you can imagine, there is a slight delay before the clients receive that info and make the action happen. There is nothing we can do about the delay concerning remote clients, but the local client is a totally different matter. Obviously if we press a key and send it to the server, we already know that we did it. So why wait for the server to tell us that? The idea of dead reckoning is this: When we press a key to make our player move forward, we send this information to the server, but we also immediately start to move our player forward on the local machine. When the server has processed the message and sent us information, we look at the position where the server thinks the player is, and compare it to what we thought it to be at that time (we have to save the old results for comparison). If they match, there is no problem. If they do not match, we obey the server and move the player to the position it told us. Sometimes this may be seen as player warping because the server is always right. Not the client. No matter what people tell you. When we use dead reckoning, the server does not have to send data to us as often. We save lots of bandwidth this way, and that is nothing but a good thing.

430

Tutorial 5 / Creating Your Online Game

Frame History
There is no need to store every frame in the history array. We store only the last 64 frames. To get the correct index in the array, we use the following method:
totalFrame & (COMMAND_HISTORY_SIZE1)

This returns a value from 0 to a maximum of 63 (COMMAND_ HISTORY_SIZE1). We use the message sequences as our frame counter, so usually you will see something like this:
outgoingSequence & (COMMAND_HISTORY_SIZE1)

Handling Messages
Because we use dead reckoning, we can save some network bandwidth by making the server send messages in 100 ms (or any other suitable time) intervals. Why can we do this then? Because the clients will try to predict the frames during this 100 ms by just moving the objects in the direction they were moving in the last known frame (using the last known velocity). But the server must get information about every client frame, so all the clients have to send frame data every frame. The server reads incoming packets every frame, and when it receives a message containg frame data, it runs the game logic for that client. So the server is not constantly running the game logic for all the clients; it only runs it when a frame data message is received. Keeping in mind that the clients send frame data on every frame, we end up running the game logic about every frame on the server too. This method keeps the server and clients in sync, even if the other is running on a significantly faster computer.

Game Server Code

In contrast to the previous tutorials, here we begin with the server-side code because it may be a bit easier to understand. And when we know how the server works, it is easier to understand the client code. So lets see how the game server code works. Like before, the basic network communication methods are the same, so we will not focus on them anymore. We need some new source code files: server.cpp, network.cpp, and server.h. Also, some of the old files have changed a little bit.

Tutorial 5 / Creating Your Online Game

431

Figure 1

server.h File
This header file contains the application-specific data structures. This means that the player data structures are here. Also the games main class is located in this file. This class has the network interface methods that we learned to create in the other tutorials. A CArmyWarServer object is created each time a new game is created, and this object processes that games functions.
#ifndef SERVER_H #define SERVER_H #include <string.h> #define #define #define #define #define #define #define #define NORTH NORTHEAST EAST SOUTHEAST SOUTH SOUTHWEST WEST NORTHWEST 0 45 90 135 180 225 270 315

#define BLUE_TEAM 0 #define RED_TEAM 1 typedef struct { float x; float y; } VECTOR2D; typedef struct bullet_t { VECTOR2D vel; VECTOR2D origin;

// Velocity // Position

432

Tutorial 5 / Creating Your Online Game

bool int } bullet_t; typedef struct { int key; int heading;

shot; lifetime;

// Is the bullet in the air? // Lifetime in ms

// Pressed keys // Heading // Velocity // Position // Bullet // How long to run command (in ms)

VECTOR2D vel; VECTOR2D origin; bullet_t bullet; int msec; } command_t;

typedef struct clientData { command_t frame[COMMAND_HISTORY_SIZE]; command_t serverFrame; command_t command; long processedFrame; dreamClient *netClient; VECTOR2D startPos; bool team; bool diedThisFrame; clientData *next; } clientData;

class CArmyWarServer { private: dreamServer *networkServer; clientData int int int float char bool bool int float *clientList; clients; realtime; servertime; frametime; gamename[32]; inProgress; mapdata[100][100]; index; flagX; // Client list // Number of clients // Real server up-time in ms // Server frame * 100 ms // Frame time in seconds

Tutorial 5 / Creating Your Online Game

433

float clientData bool bool int int long

flagY; *playerWithFlag; updateFlag; updateKill; redScore; blueScore; framenum;

public: CArmyWarServer(); ~CArmyWarServer(); // Network.cpp void ReadPackets(void); void SendCommand(void); void SendExitNotification(void); void ReadDeltaMoveCommand(dreamMessage *mes, clientData *client); void BuildMoveCommand(dreamMessage *mes, clientData *client); void BuildDeltaMoveCommand(dreamMessage *mes, clientData *client); // Server.cpp int InitNetwork(int gameAmount); void ShutdownNetwork(void); void GenerateRandomMap(void); void CalculateVelocity(command_t *command, float frametime); void CalculateHeading(command_t *command); void CalculateBulletVelocity(command_t *command); void MovePlayers(void); void MovePlayer(clientData *client); void CheckFlagCollisions(void); void void void void AddClient(void); RemoveClient(struct sockaddr *address); RemoveClients(void); Frame(int msec);

clientData *GetClientList(void) {return clientList;} void SetName(char *n) char *GetName(void) void SetInProgress(bool p) bool GetInProgress(void) void SetIndex(int ind) int GetIndex(void) CArmyWarServer *next; }; #endif {strcpy(gamename, n);} {return gamename;} {inProgress = p;} {return inProgress;} {index = ind;} {return index;}

434

Tutorial 5 / Creating Your Online Game

Each frames commands are kept in a structure, which is seen below. All the variables are easy to understand except for one perhaps, and that is the integer msec. It is basically that frames frame time, and hence it is the length of the command (in milliseconds).
int key; int heading; VECTOR2D vel; VECTOR2D origin; bullet_t bullet; int msec; // Pressed keys // Heading // Velocity // Position // Bullet // How long to run command (in ms)

The players game-specific data (such as their origin) is stored in the clientData structure. As you can see, there is more than one command_t object. One is for the current frame, one is for the last known server frame, and one is a history of the last 64 frames.
command_t frame[COMMAND_HISTORY_SIZE]; command_t serverFrame; command_t command; long processedFrame; dreamClient *netClient; VECTOR2D startPos; bool team; bool diedThisFrame; clientData *next;

The command_t frame is an array (history) of the last 64 frames. The command_t serverFrame is the last known server frame. The command_t command is the current frame. The long processedFrame tells us the last frame we have processed. This is a frame history indexed value (0 to 63). The dreamClient netClient is a pointer to the dreamClient network client that matches with this player. The VECTOR2D startPos is the start position for the player. The boolean team is the players team (red = 1 or blue = 0). The boolean diedThisFrame is a flag that indicates if the player died this frame (used to inform all the clients which players died during the current frame). The clientData next is a pointer to the next client. CArmyWarServer also has some member variables that we should look at before we move on:

Tutorial 5 / Creating Your Online Game

435

private: dreamServer *networkServer; clientData int int int float char bool bool int float float clientData bool bool int int long *clientList; clients; realtime; servertime; frametime; gamename[32]; inProgress; mapdata[100][100]; index; flagX; flagY; *playerWithFlag; updateFlag; updateKill; redScore; blueScore; framenum; // Client list // Number of clients // Real server up-time in ms // Server frame * 100 ms // Frame time in seconds

The dreamServer networkServer is the dreamServer network server object. The clientData clientList is the game-specific client list holding the players data such as their current origin. The integer clients tells us how many clients have joined this game. The integer realtime tells us in real time how long this server has been running (in milliseconds). The integer servertime is used to make each network frame last for 100 milliseconds. The formula for this variable is: server frame number * 100 ms. The char gamename stores the games name. The boolean inProgress is a flag that tells us if the game is in progress or not. The boolean mapdata is an array that holds the map information. If a node is true, a tree exists in that position. If the value is false, there is nothing but grass. The integer index tells us the index number of the game. The floats flagX and flagY store the flag origin. The clientData playerWithFlag is a pointer to the player with the flag (if any). The boolean updateFlag is a flag that is set when the flag information should be sent to each client. The boolean updateKill is a flag that is set when someone dies. This information should be sent to each client. The integers redScore and blueScore keep track of the team scores. The long frameNum counts the frames.

436

Tutorial 5 / Creating Your Online Game

network.h File
Here is the final version of the server-side network.h. There are some new definitions here, such as command history size and the keyboard commands.
#ifndef NETWORK_H #define NETWORK_H #include "dreamSock.h" #define COMMAND_HISTORY_SIZE #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define KEY_UP KEY_DOWN KEY_LEFT KEY_RIGHT KEY_WEAPON CMD_KEY CMD_HEADING CMD_ORIGIN CMD_BULLET CMD_FLAG CMD_KILL USER_MES_FRAME USER_MES_NONDELTAFRAME USER_MES_SERVEREXIT USER_MES_LOGIN USER_MES_SIGNIN USER_MES_CHAT USER_MES_CREATEGAME USER_MES_REMOVEGAME USER_MES_GAMEDATA USER_MES_STARTGAME USER_MES_MAPDATA 64 1 2 4 8 16 1 2 4 8 16 32 1 2 3 4 5 6 7 8 9 10 11

typedef struct clientLoginData { dreamClient netClient; clientLoginData next; } clientLoginData; #endif

main.cpp File
The servers main source code file has not changed much. Only the applications main functions (WinMain and main) have something new. We now run each games frame right after the lobby and sign-in frames. We go through the whole game list and run every games frames.

Tutorial 5 / Creating Your Online Game

437

Lobby.Frame(time); Signin.Frame(time); CArmyWarServer *list = Lobby.GetGameList(); for(; list != NULL; list = list->next) { list->Frame(time); }

network.cpp File CArmyWarServer Class Part 1

Network.cpp holds the network part of the CArmyWarServer class. Here we receive, process, build, and send packets.
/******************************************/ /* Programming Multiplayer Games */ /* Tutorial game server */ /* Programming: */ /* Teijo Hakala */ /******************************************/ #include "common.h" #include <math.h> #ifdef Win32 #include <windows.h> #include <mmsystem.h> #include <assert.h> #endif //--------------------------------------------------------------------------// Name: InitNetwork() // Desc: Initialize network //--------------------------------------------------------------------------int CArmyWarServer::InitNetwork(int gameAmount) { if(dreamSock_Initialize() != 0) { LogString("Error initializing Communication Library!"); return 1; } LogString("Initializing game %d", gameAmount); // Create the game servers on new ports, starting from 30004 int ret = networkServer->Initialize("", 30004 + gameAmount); if(ret == DREAMSOCK_SERVER_ERROR) { #ifdef Win32 char text[64];

438

Tutorial 5 / Creating Your Online Game

sprintf(text, "Could not open server on port %d", networkServer->GetPort()); MessageBox(NULL, text, "Error", MB_OK); #else LogString("Could not open server on port %d", networkServer->GetPort()); #endif return 1; } return 0; } //--------------------------------------------------------------------------// Name: ShutdownNetwork() // Desc: Shut down network //--------------------------------------------------------------------------void CArmyWarServer::ShutdownNetwork(void) { LogString("Shutting down game server..."); RemoveClients(); networkServer->Uninitialize(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::ReadPackets(void) { char data[1400]; int type; int ret; struct sockaddr address; clientData *clList; dreamMessage mes; mes.Init(data, sizeof(data)); // Get the packet from the socket try { while(ret = networkServer->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading();

Tutorial 5 / Creating Your Online Game

439

type = mes.ReadByte(); // Check the type of the message switch(type) { case DREAMSOCK_MES_CONNECT: AddClient(); break; case DREAMSOCK_MES_DISCONNECT: RemoveClient(&address); break; case USER_MES_FRAME: // Skip sequences mes.ReadShort(); mes.ReadShort(); // Find the correct client by comparing addresses clList = clientList; for(; clList != NULL; clList = clList->next) { if(memcmp(clList->netClient-> GetSocketAddress(), &address, sizeof(address)) == 0) { ReadDeltaMoveCommand(&mes, clList); MovePlayer(clList); break; } } break; case USER_MES_NONDELTAFRAME: clList = clientList; clientData *dataClient; // Find the correct client by comparing addresses for(; clList != NULL; clList = clList->next) { if(memcmp(clList->netClient-> GetSocketAddress(), &address, sizeof(address)) == 0) { break; } }

440

Tutorial 5 / Creating Your Online Game

clList->netClient->message.Init(clList->netClient-> message.outgoingData, sizeof(clList-> netClient->message.outgoingData)); clList->netClient->message.WriteByte(USER_MES_ NONDELTAFRAME); clList->netClient->message.WriteShort(clList-> netClient->GetOutgoingSequence()); clList->netClient->message.WriteShort(clList-> netClient->GetIncomingSequence()); for(dataClient = clientList; dataClient != NULL; dataClient = dataClient->next) { BuildMoveCommand(&clList->netClient->message, dataClient); } clList->netClient->SendPacket(); break; case USER_MES_STARTGAME: // Skip sequences mes.ReadShort(); mes.ReadShort(); // Send to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList-> netClient->message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_ MES_MAPDATA); // type clList->netClient->message.AddSequences (clList->netClient); // sequences for(int i = 0; i < 100; i++) { for(int j = 0; j < 100; j++) { if(mapdata[i][j] == true) { clList->netClient-> message.WriteByte(i); clList->netClient-> message.WriteByte(j); } } }

Tutorial 5 / Creating Your Online Game

441

} networkServer->SendPackets(); // Send to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList-> netClient->message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_ MES_STARTGAME); // type clList->netClient->message.AddSequences (clList->netClient); // sequences } networkServer->SendPackets(); break; } } } catch(...) { LogString("Unknown Exception caught in Lobby ReadPackets loop"); #ifdef Win32 MessageBox(NULL, "Unknown Exception caught in Lobby ReadPackets loop", "Error", MB_OK | MB_TASKMODAL); #endif } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::AddClient(void) { // First get a pointer to the beginning of the client list clientData *list = clientList; clientData *prev; dreamClient *netList = networkServer->GetClientList(); // No clients yet, adding the first one if(clientList == NULL) { LogString("App: Server: Adding first client"); clientList = (clientData *) calloc(1, sizeof(clientData));

442

Tutorial 5 / Creating Your Online Game

clientList->netClient = netList; if(clients % 2 == 0) { clientList->team = RED_TEAM; clientList->startPos.x = 46.0f * 32.0f + ((clients/2) * 32.0f); clientList->startPos.y = 96.0f * 32.0f; } else { clientList->team = BLUE_TEAM; clientList->startPos.x = 46.0f * 32.0f + ((clients/2) * 32.0f); clientList->startPos.y = 4.0f * 32.0f; } clientList->command.origin.x = clientList->startPos.x; clientList->command.origin.y = clientList->startPos.y; clientList->next = NULL; } else { LogString("App: Server: Adding another client"); prev = list; list = clientList->next; netList = netList->next; while(list != NULL) { prev = list; list = list->next; netList = netList->next; } list = (clientData *) calloc(1, sizeof(clientData)); list->netClient = netList; if(clients % 2 == 0) { list->team = RED_TEAM; list->startPos.x = 46.0f * 32.0f + ((clients/2) * 32.0f); list->startPos.y = 96.0f * 32.0f; } else { list->team = BLUE_TEAM;

Tutorial 5 / Creating Your Online Game

443

list->startPos.x = 46.0f * 32.0f + ((clients/2) * 32.0f); list->startPos.y = 4.0f * 32.0f; } list->command.origin.x = list->startPos.x; list->command.origin.y = list->startPos.y; list->next = NULL; prev->next = list; } clients++; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::RemoveClient(struct sockaddr *address) { clientData *list = clientList; clientData *prev = NULL; clientData *next = NULL; for(; list != NULL; list = list->next) { if(memcmp(list->netClient->GetSocketAddress(), address, sizeof(address)) == 0) { if(prev != NULL) { prev->next = list->next; } break; } prev = list; } // Drop the flag if player with flag exits the game if(list == playerWithFlag) { playerWithFlag = NULL; updateFlag = true; } if(list == clientList) { if(list) { next = list->next; free(list);

444

Tutorial 5 / Creating Your Online Game

} list = NULL; clientList = next; } else { if(list) { next = list->next; free(list); } list = next; } clients--; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::RemoveClients(void) { clientData *list = clientList; clientData *next; while(list != NULL) { if(list) { next = list->next; free(list); } list = next; } clientList = NULL; clients = 0; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::SendCommand(void) { clientData *toClient; clientData *dataClient; // Fill messages for(toClient = clientList; toClient != NULL; toClient = toClient->next)

Tutorial 5 / Creating Your Online Game

445

{ toClient->netClient->message.Init(toClient->netClient-> message.outgoingData, sizeof(toClient->netClient-> message.outgoingData)); toClient->netClient->message.WriteByte(USER_MES_FRAME); // type toClient->netClient->message.AddSequences(toClient->netClient); // sequences for(dataClient = clientList; dataClient != NULL; dataClient = dataClient->next) { BuildDeltaMoveCommand(&toClient->netClient->message, dataClient); } } // Send messages to all clients networkServer->SendPackets(); // Store the sent command in history for(toClient = clientList; toClient != NULL; toClient = toClient->next) { int i = (toClient->netClient->GetOutgoingSequence() 1) & (COMMAND_HISTORY_SIZE1); memcpy(&toClient->frame[i], &toClient->command, sizeof(command_t)); } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::SendExitNotification(void) { clientData *toClient = clientList; for(; toClient != NULL; toClient = toClient->next) { toClient->netClient->message.Init(toClient->netClient-> message.outgoingData, sizeof(toClient->netClient-> message.outgoingData)); toClient->netClient->message.WriteByte(USER_MES_SERVEREXIT); // type toClient->netClient->message.AddSequences(toClient->netClient); // sequences } networkServer->SendPackets(); }

446

Tutorial 5 / Creating Your Online Game

//--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::ReadDeltaMoveCommand(dreamMessage *mes, clientData *client) { int flags = 0; // Flags flags = mes->ReadByte(); // Key if(flags & CMD_KEY) { client->command.key = mes->ReadByte(); LogString("Client %d: read CMD_KEY (%d)", client->netClient-> GetIndex(), client->command.key); } // Read time to run command client->command.msec = mes->ReadByte(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::BuildMoveCommand(dreamMessage *mes, clientData *client) { // Add to the message // Key mes->WriteByte(client->command.key); // Heading mes->WriteShort(client->command.heading); // Origin mes->WriteFloat(client->command.origin.x); mes->WriteFloat(client->command.origin.y); mes->WriteFloat(client->command.vel.x); mes->WriteFloat(client->command.vel.y); mes->WriteFloat(client->command.bullet.origin.x); mes->WriteFloat(client->command.bullet.origin.y); mes->WriteFloat(client->command.bullet.vel.x); mes->WriteFloat(client->command.bullet.vel.y); mes->WriteShort(client->command.bullet.lifetime); mes->WriteByte(client->command.bullet.shot); // Flag & points if(playerWithFlag) mes->WriteShort(playerWithFlag->netClient->GetIndex());

Tutorial 5 / Creating Your Online Game

447

else mes->WriteShort(1); mes->WriteFloat(flagX); mes->WriteFloat(flagY); mes->WriteByte(redScore); mes->WriteByte(blueScore); mes->WriteByte(client->command.msec); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::BuildDeltaMoveCommand(dreamMessage *mes, clientData *client) { int flags = 0; int last = (client->netClient->GetOutgoingSequence() 1) & (COMMAND_HISTORY_SIZE1); // Check what needs to be updated if(client->frame[last].key != client->command.key) { flags |= CMD_KEY; } if(client->frame[last].origin.x != client->command.origin.x || client->frame[last].origin.y != client->command.origin.y) { flags |= CMD_ORIGIN; } if(client->command.bullet.shot) { if(client->frame[last].bullet.origin.x != client-> command.bullet.origin.x || client->frame[last].bullet.origin.y != client-> command.bullet.origin.y) { flags |= CMD_BULLET; } } if(client->frame[last].bullet.shot != client->command.bullet.shot) { flags |= CMD_BULLET; } if(updateFlag == true) {

448

Tutorial 5 / Creating Your Online Game

flags |= CMD_FLAG; } if(updateKill == true) { flags |= CMD_KILL; }

// Add to the message // Flags mes->WriteByte(flags); // Key if(flags & CMD_KEY) { mes->WriteByte(client->command.key); } if(flags & CMD_ORIGIN || flags & CMD_BULLET) { mes->WriteByte(client->processedFrame & (COMMAND_HISTORY_SIZE1)); } // Origin if(flags & CMD_ORIGIN) { mes->WriteFloat(client->command.origin.x); mes->WriteFloat(client->command.origin.y); mes->WriteFloat(client->command.vel.x); mes->WriteFloat(client->command.vel.y); } // Origin if(flags & CMD_BULLET) { mes->WriteFloat(client->command.bullet.origin.x); mes->WriteFloat(client->command.bullet.origin.y); mes->WriteFloat(client->command.bullet.vel.x); mes->WriteFloat(client->command.bullet.vel.y); mes->WriteByte(client->command.bullet.shot); } // Flag & points if(flags & CMD_FLAG) { if(playerWithFlag) mes->WriteShort(playerWithFlag->netClient->GetIndex()); else

Tutorial 5 / Creating Your Online Game

449

mes->WriteShort(0); mes->WriteFloat(flagX); mes->WriteFloat(flagY); mes->WriteByte(redScore); mes->WriteByte(blueScore); } // Someone died if(flags & CMD_KILL) { mes->WriteByte(client->diedThisFrame); } mes->WriteByte(client->command.msec); }

InitNetwork Function
This function has only one thing different from the previous implementations the port number and the way we choose it. Every created game opens its own port, and we start from port 30004. The next game would open port number 30005 and the next one 30006 and so on.
// Create the game servers on new ports, starting from 30004 int ret = networkServer->Initialize("", 30004 + gameAmount);

ReadPackets Function
If a frame message is received from a client, the commands are read using the ReadDeltaMoveCommand function. The commands are delta-compressed, meaning only the values that have changed are included in the packet. Once the commands are read, they are put into action by running the MovePlayer function.
case USER_MES_FRAME: // Skip sequences mes.ReadShort(); mes.ReadShort(); // Find the correct client by comparing addresses clList = clientList; for(; clList != NULL; clList = clList->next) { if(memcmp(clList->netClient->GetSocketAddress(), &address, sizeof(address)) == 0) { ReadDeltaMoveCommand(&mes, clList); MovePlayer(clList); break;

450

Tutorial 5 / Creating Your Online Game

} } break;

If we received a non-delta frame message, it means that a client wants us to send it the current absolute values of the game. The BuildMoveCommand function is used to send the absolute values.
case USER_MES_NONDELTAFRAME: clList = clientList; clientData *dataClient; // Find the correct client by comparing addresses for(; clList != NULL; clList = clList->next) { if(memcmp(clList->netClient->GetSocketAddress(), &address, sizeof(address)) == 0) { break; } } clList->netClient->message.Init(clList->netClient->message.outgoingData, sizeof(clList->netClient->message.outgoingData)); clList->netClient->message.WriteByte(USER_MES_NONDELTAFRAME); clList->netClient->message.WriteShort(clList->netClient-> GetOutgoingSequence()); clList->netClient->message.WriteShort(clList->netClient-> GetIncomingSequence()); for(dataClient = clientList; dataClient != NULL; dataClient = dataClient->next) { BuildMoveCommand(&clList->netClient->message, dataClient); } clList->netClient->SendPacket(); break;

Receiving a start game message means that the game host wants to start the game. We send the map data to each client and inform them to start the game.
case USER_MES_STARTGAME: // Skip sequences mes.ReadShort(); mes.ReadShort(); // Send to everybody for(clList = clientList; clList != NULL; clList = clList->next)

Tutorial 5 / Creating Your Online Game

451

{ clList->netClient->message.Init(clList->netClient-> message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_MES_MAPDATA); // type clList->netClient->message.AddSequences(clList->netClient); // sequences for(int i = 0; i < 100; i++) { for(int j = 0; j < 100; j++) { if(mapdata[i][j] == true) { clList->netClient->message.WriteByte(i); clList->netClient->message.WriteByte(j); } } } } networkServer->SendPackets(); // Send to everybody for(clList = clientList; clList != NULL; clList = clList->next) { clList->netClient->message.Init(clList->netClient-> message.outgoingData, sizeof(clList->netClient-> message.outgoingData)); clList->netClient->message.WriteByte(USER_MES_STARTGAME); // type clList->netClient->message.AddSequences(clList->netClient); // sequences } networkServer->SendPackets(); break;

SendCommand Function
This function sends the current commands to each client. The commands are then stored in frame history.
void CArmyWarServer::SendCommand(void) { clientData *toClient; clientData *dataClient; // Fill messages for(toClient = clientList; toClient != NULL; toClient = toClient->next) {

452

Tutorial 5 / Creating Your Online Game

ReadDeltaMoveCommand Function
This function reads a delta-compressed command from a client. A client only sends information about keypresses.
void CArmyWarServer::ReadDeltaMoveCommand(dreamMessage *mes, clientData *client) { int flags = 0; // Flags flags = mes->ReadByte(); // Key if(flags & CMD_KEY) { client->command.key = mes->ReadByte(); LogString("Client %d: read CMD_KEY (%d)", client->netClient->GetIndex(), client->command.key); }

Tutorial 5 / Creating Your Online Game

453

// Read time to run command client->command.msec = mes->ReadByte(); }

The function takes two parameters (dreamMessage *mes and clientData *client). The first one is a pointer to the message to read, and the second one is a pointer to the client that owns this message. We first read the flags byte, which is always there. Depending on the value of this byte, we read the key command. Finally, we read the frame time of this command.

BuildMoveCommand Function
This function builds a non-delta-compressed command message for a client.
void CArmyWarServer::BuildMoveCommand(dreamMessage *mes, clientData *client) { // Add to the message // Key mes->WriteByte(client->command.key); // Heading mes->WriteShort(client->command.heading); // Origin mes->WriteFloat(client->command.origin.x); mes->WriteFloat(client->command.origin.y); mes->WriteFloat(client->command.vel.x); mes->WriteFloat(client->command.vel.y); mes->WriteFloat(client->command.bullet.origin.x); mes->WriteFloat(client->command.bullet.origin.y); mes->WriteFloat(client->command.bullet.vel.x); mes->WriteFloat(client->command.bullet.vel.y); mes->WriteShort(client->command.bullet.lifetime); mes->WriteByte(client->command.bullet.shot); // Flag & points if(playerWithFlag) mes->WriteShort(playerWithFlag->netClient->GetIndex()); else mes->WriteShort(1); mes->WriteFloat(flagX); mes->WriteFloat(flagY); mes->WriteByte(redScore); mes->WriteByte(blueScore); mes->WriteByte(client->command.msec); }

454

Tutorial 5 / Creating Your Online Game

BuildDeltaMoveCommand Function
This function builds a delta-compressed command message for a client.
void CArmyWarServer::BuildDeltaMoveCommand(dreamMessage *mes, clientData *client) { int flags = 0; int last = (client->netClient->GetOutgoingSequence() 1) & (COMMAND_HISTORY_SIZE1); // Check what needs to be updated if(client->frame[last].key != client->command.key) { flags |= CMD_KEY; } if(client->frame[last].origin.x != client->command.origin.x || client->frame[last].origin.y != client->command.origin.y) { flags |= CMD_ORIGIN; } if(client->command.bullet.shot) { if(client->frame[last].bullet.origin.x != client->command.bullet.origin.x || client->frame[last].bullet.origin.y != client->command.bullet.origin.y) { flags |= CMD_BULLET; } } if(client->frame[last].bullet.shot != client->command.bullet.shot) { flags |= CMD_BULLET; } if(updateFlag == true) { flags |= CMD_FLAG; } if(updateKill == true) {

Tutorial 5 / Creating Your Online Game

455

flags |= CMD_KILL; }

456

Tutorial 5 / Creating Your Online Game

mes->WriteByte(blueScore); } // Someone died if(flags & CMD_KILL) { mes->WriteByte(client->diedThisFrame); } mes->WriteByte(client->command.msec); }

The function takes two parameters (dreamMessage *mes and clientData *client). The first one is a pointer to the message to write to, and the second is a pointer to the client that owns the message. First we check what has changed since the last frame and set the correct flags:
int flags = 0; int last = (client->netClient->GetOutgoingSequence() 1) & (COMMAND_HISTORY_SIZE1); // Check what needs to be updated if(client->frame[last].key != client->command.key) { flags |= CMD_KEY; } if(client->frame[last].origin.x != client->command.origin.x || client->frame[last].origin.y != client->command.origin.y) { flags |= CMD_ORIGIN; } if(client->command.bullet.shot) { if(client->frame[last].bullet.origin.x != client->command.bullet.origin.x || client->frame[last].bullet.origin.y != client->command.bullet.origin.y) { flags |= CMD_BULLET; } } if(client->frame[last].bullet.shot != client->command.bullet.shot) { flags |= CMD_BULLET; } if(updateFlag == true) { flags |= CMD_FLAG;

Tutorial 5 / Creating Your Online Game

457

} if(updateKill == true) { flags |= CMD_KILL; }

We write the flags to the message and then we add the changed values to the message by checking which flags are up:
// Add to the message // Flags mes->WriteByte(flags); // Key if(flags & CMD_KEY) { mes->WriteByte(client->command.key); } if(flags & CMD_ORIGIN || flags & CMD_BULLET) { mes->WriteByte(client->processedFrame & (COMMAND_HISTORY_SIZE1)); } // Origin if(flags & CMD_ORIGIN) { mes->WriteFloat(client->command.origin.x); mes->WriteFloat(client->command.origin.y); mes->WriteFloat(client->command.vel.x); mes->WriteFloat(client->command.vel.y); } // Origin if(flags & CMD_BULLET) { mes->WriteFloat(client->command.bullet.origin.x); mes->WriteFloat(client->command.bullet.origin.y); mes->WriteFloat(client->command.bullet.vel.x); mes->WriteFloat(client->command.bullet.vel.y); mes->WriteByte(client->command.bullet.shot); } // Flag & points if(flags & CMD_FLAG) { if(playerWithFlag) mes->WriteShort(playerWithFlag->netClient->GetIndex()); else mes->WriteShort(0);

458

Tutorial 5 / Creating Your Online Game

mes->WriteFloat(flagX); mes->WriteFloat(flagY); mes->WriteByte(redScore); mes->WriteByte(blueScore); } // Someone died if(flags & CMD_KILL) { mes->WriteByte(client->diedThisFrame); } mes->WriteByte(client->command.msec);

server.cpp File CArmyWarServer Class Part 2

This file contains the last part of the CArmyWarServer class. This part is the logic part where we process the commands.
/******************************************/ /* Programming Multiplayer Games */ /* Tutorial game server */ /* Programming: */ /* Teijo Hakala */ /******************************************/ #include #include #include #include #include "common.h" <fstream> <math.h> <malloc.h> <stdlib.h>

//--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------float VectorLength(VECTOR2D *vec) { return (float) sqrt(vec->x*vec->x + vec->y*vec->y); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------VECTOR2D VectorSubtract(VECTOR2D *vec1, VECTOR2D *vec2) { VECTOR2D vec; vec.x = vec1->x vec2->x; vec.y = vec1->y vec2->y;

Tutorial 5 / Creating Your Online Game

459

return vec; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CArmyWarServer::CArmyWarServer() { networkServer = new dreamServer; clientList clients realtime servertime inProgress index next flagX flagY playerWithFlag updateFlag redScore blueScore framenum } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CArmyWarServer::~CArmyWarServer() { delete networkServer; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::GenerateRandomMap(void) { // Make all land passable for(int i = 0; i < 100; i++) { for(int j = 0; j < 100; j++) { mapdata[i][j] = false; = NULL; = 0; = 0; = 0; = false; = 0; = NULL; = = = = 0.0f; 0.0f; NULL; false;

= 0; = 0; = 0;

460

Tutorial 5 / Creating Your Online Game

} } // Use the game's index number for random seed srand(index); // Place some random trees (avoiding the players start locations) for(int m = 0; m < 300; m++) { mapdata[rand()%100][(rand()%80)+10] = true; } // Set the flag position flagX = 49*32; flagY = 49*32; playerWithFlag = NULL; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::CalculateVelocity(command_t *command, float frametime) { int checkX; int checkY; float multiplier = 100.0f; command->vel.x = 0.0f; command->vel.y = 0.0f; if(command->key & KEY_UP) { checkX = (int) (command->origin.x/32.0f); checkY = (int) ((command->origin.y multiplier * frametime) / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.y += multiplier * frametime; } if(command->key & KEY_DOWN) { checkX = (int) (command->origin.x / 32.0f); checkY = (int) ((command->origin.y + multiplier * frametime) / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.y += multiplier * frametime; } if(command->key & KEY_LEFT)

Tutorial 5 / Creating Your Online Game

461

{ checkX = (int) ((command->origin.x multiplier * frametime) / 32.0f); checkY = (int) (command->origin.y / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.x += multiplier * frametime; } if(command->key & KEY_RIGHT) { checkX = (int) ((command->origin.x + multiplier * frametime) / 32.0f); checkY = (int) (command->origin.y / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.x += multiplier * frametime; } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::CalculateHeading(command_t *command) { // Right if((command->vel.x > 0.0f) && (command->vel.y == 0.0f)) { command->heading = EAST; } // Left if((command->vel.x < 0.0f) && (command->vel.y == 0.0f)) { command->heading = WEST; } // Down if((command->vel.y > 0.0f) && (command->vel.x == 0.0f)) { command->heading = SOUTH; } // Up if((command->vel.y < 0.0f) && (command->vel.x == 0.0f)) { command->heading = NORTH; }

462

Tutorial 5 / Creating Your Online Game

// Down-Right if((command->vel.x > 0.0f) && (command->vel.y > 0.0f)) { command->heading = SOUTHEAST; } // Up-Right if((command->vel.x > 0.0f) && (command->vel.y < 0.0f)) { command->heading = NORTHEAST; } // Down-Left if((command->vel.x < 0.0f) && (command->vel.y > 0.0f)) { command->heading = SOUTHWEST; } // Up-Left if((command->vel.x < 0.0f) && (command->vel.y < 0.0f)) { command->heading = NORTHWEST; } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::CalculateBulletVelocity(command_t *command) { command->bullet.shot = true; if(command->heading == NORTH) { command->bullet.vel.x = command->bullet.vel.y = } if(command->heading == SOUTH) { command->bullet.vel.x = command->bullet.vel.y = } if(command->heading == EAST) { command->bullet.vel.x = command->bullet.vel.y = } if(command->heading == WEST) {

0.0f; 200.0f;

200.0f; 0.0f;

Tutorial 5 / Creating Your Online Game

463

command->bullet.vel.x = 200.0f; command->bullet.vel.y = 0.0f; } if(command->heading == NORTHEAST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } if(command->heading == NORTHWEST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } if(command->heading == SOUTHEAST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } if(command->heading == SOUTHWEST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::MovePlayer(clientData *client) { float clientFrametime; float multiplier = 100.0f; clientFrametime = client->command.msec / 1000.0f;; CalculateVelocity(&client->command, clientFrametime); CalculateHeading(&client->command); // Move the client based on the commands client->command.origin.x += client->command.vel.x; client->command.origin.y += client->command.vel.y; // Bullet if(client->command.bullet.shot == false) { client->command.bullet.origin.x = client->command.origin.x; client->command.bullet.origin.y = client->command.origin.y; } else {

464

Tutorial 5 / Creating Your Online Game

client->command.bullet.lifetime += (int) (clientFrametime * 1000.0f); if(client->command.bullet.lifetime > 2000) { client->command.bullet.shot = false; client->command.bullet.lifetime = 0; client->command.bullet.origin.x = client->command.origin.x; client->command.bullet.origin.y = client->command.origin.y; } } if(client->command.key & KEY_WEAPON && client->command.bullet.shot == false) { CalculateBulletVelocity(&client->command); } if(client->command.bullet.shot) { client->command.bullet.origin.x += client->command.bullet.vel.x * clientFrametime; client->command.bullet.origin.y += client->command.bullet.vel.y * clientFrametime; } // Check for bullet hits if(client->command.bullet.shot) { for(clientData *client2 = clientList; client2 != NULL; client2 = client2->next) { if(client == client2) continue; client2->diedThisFrame = false; VECTOR2D pos = client2->command.origin; pos.x += 16.0f; pos.y += 16.0f; VECTOR2D vec = VectorSubtract(&client-> command.bullet.origin, &pos); float distance = VectorLength(&vec); if(distance < 16.0f) { // Player dies client2->command.origin.x = client2->startPos.x; client2->command.origin.y = client2->startPos.y; client2->diedThisFrame = true; if(client2 == playerWithFlag)

Tutorial 5 / Creating Your Online Game

465

{ playerWithFlag = NULL; updateFlag = true; } updateKill = true; client->command.bullet.shot = false; break; } } } int f = client->netClient->GetIncomingSequence() & (COMMAND_HISTORY_SIZE1); client->processedFrame = f; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::CheckFlagCollisions(void) { if(playerWithFlag != NULL) { // Move the flag with the player flagX = playerWithFlag->command.origin.x; flagY = playerWithFlag->command.origin.y; // Check if the player is at home base if((playerWithFlag) && (playerWithFlag->team == BLUE_TEAM)) { if(playerWithFlag->command.origin.x+16 > (49*32) && playerWithFlag->command.origin.x+16 < (50*32) && playerWithFlag->command.origin.y+16 > (3*32) && playerWithFlag->command.origin.y+16 < (4*32)) { flagX = 49*32; flagY = 49*32; playerWithFlag = NULL; blueScore++; updateFlag = true; } } if((playerWithFlag) && (playerWithFlag->team == RED_TEAM)) { if(playerWithFlag->command.origin.x+16 > (49*32) && playerWithFlag->command.origin.x+16 < (50*32) && playerWithFlag->command.origin.y+16 > (97*32) && playerWithFlag->command.origin.y+16 < (98*32)) {

466

Tutorial 5 / Creating Your Online Game

flagX = 49*32; flagY = 49*32; playerWithFlag = NULL; redScore++; updateFlag = true; } } } else { // Check if anyone is in contact with the flag clientData *list = clientList; for(; list != NULL; list = list->next) { if(list->command.origin.x+16 > flagX && list-> command.origin.x+16 < flagX+32 && list->command.origin.y+16 > flagY && list-> command.origin.y+16 < flagY+32) { char team[10]; if(list->team == RED_TEAM) strcpy(team, "RED team"); else strcpy(team, "BLUE team"); LogString("FLAG hit : player %d: %s, %s", list-> netClient->GetIndex(), list->netClient-> GetName(), team); playerWithFlag = list; updateFlag = true; return; } } } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWarServer::Frame(int msec) { realtime += msec; frametime = msec / 1000.0f; // Read packets from clients ReadPackets(); if(inProgress == false)

Tutorial 5 / Creating Your Online Game

467

return; // Check if someone hit the flag CheckFlagCollisions(); // Wait full 100 ms before allowing to send if(realtime < servertime) { // never let the time get too far off if(servertime realtime > 100) { realtime = servertime 100; } return; } // Bump frame number, and calculate new servertime framenum++; servertime = framenum * 100; if(servertime < realtime) realtime = servertime; SendCommand(); // Reset update flags updateFlag = false; updateKill = false; }

GenerateRandomMap Function
Here we generate the map by randomizing the tree positions. This is slightly controlled by setting the game index number as the random seed. The original idea was to randomize exactly the same map on each client by using the same random seed, but it seems that the Unix and Windows random seeds are not always compatible. So the map data must be sent over the network to the clients (shown in the ReadPackets function).
void CArmyWarServer::GenerateRandomMap(void) { // Make all land passable for(int i = 0; i < 100; i++) { for(int j = 0; j < 100; j++) { mapdata[i][j] = false; } } // Use the game's index number for random seed

468

Tutorial 5 / Creating Your Online Game

srand(index); // Place some random trees (avoiding the players start locations) for(int m = 0; m < 300; m++) { mapdata[rand()%100][(rand()%80)+10] = true; } // Set the flag position flagX = 49*32; flagY = 49*32; playerWithFlag = NULL; }

CalculateVelocity Function
This function calculates the velocity on a given command and frame time.
void CArmyWarServer::CalculateVelocity(command_t *command, float frametime) { int checkX; int checkY; float multiplier = 100.0f; command->vel.x = 0.0f; command->vel.y = 0.0f; if(command->key & KEY_UP) { checkX = (int) (command->origin.x / 32.0f); checkY = (int) ((command->origin.y multiplier * frametime) / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.y += multiplier * frametime; } if(command->key & KEY_DOWN) { checkX = (int) (command->origin.x / 32.0f); checkY = (int) ((command->origin.y + multiplier * frametime) / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.y += multiplier * frametime; } if(command->key & KEY_LEFT) { checkX = (int) ((command->origin.x multiplier * frametime) / 32.0f);

Tutorial 5 / Creating Your Online Game

469

checkY = (int) (command->origin.y / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.x += multiplier * frametime; } if(command->key & KEY_RIGHT) { checkX = (int) ((command->origin.x + multiplier * frametime) / 32.0f); checkY = (int) (command->origin.y / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.x += multiplier * frametime; } }

The function takes two parameters (command_t *command and float frametime). The first one is a pointer to the command to use, and the second one is the frame time value. The velocity is calculated based on the command and frame time, but we also check if we are colliding with a tree. We first calculate the new position to which the object would move after the command commences. If a tree exists in that position, we do nothing and the object does not move (it collides with the tree). If there is no tree, the object moves normally.
checkX = (int) (command->origin.x / 32.0f); checkY = (int) ((command->origin.y multiplier * frametime) / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.y += multiplier * frametime;

CalculateHeading Function
This function is usually run after CalculateVelocity, as we need the velocity values here to calculate the heading for the given command.
void CArmyWarServer::CalculateHeading(command_t *command) { // Right if((command->vel.x > 0.0f) && (command->vel.y == 0.0f)) { command->heading = EAST; } // Left if((command->vel.x < 0.0f) && (command->vel.y == 0.0f)) { command->heading = WEST;

470

Tutorial 5 / Creating Your Online Game

} // Down if((command->vel.y > 0.0f) && (command->vel.x == 0.0f)) { command->heading = SOUTH; } // Up if((command->vel.y < 0.0f) && (command->vel.x == 0.0f)) { command->heading = NORTH; } // Down-Right if((command->vel.x > 0.0f) && (command->vel.y > 0.0f)) { command->heading = SOUTHEAST; } // Up-Right if((command->vel.x > 0.0f) && (command->vel.y < 0.0f)) { command->heading = NORTHEAST; } // Down-Left if((command->vel.x < 0.0f) && (command->vel.y > 0.0f)) { command->heading = SOUTHWEST; } // Up-Left if((command->vel.x < 0.0f) && (command->vel.y < 0.0f)) { command->heading = NORTHWEST; } }

The function takes one parameter (command_t *command). It is a pointer to the command to use. The function simply looks at the velocity values and determines the heading of the object.

CalculateBulletVelocity Function
This function calculates the velocity for the bullet on a given command. This function is run only when firing the bullet, so the velocity here

Tutorial 5 / Creating Your Online Game

471

sets the initial heading of the bullet. The velocity is not scaled with frame time because this function is not called each frame. Frame time scaling is done when moving the bullet.
void CArmyWarServer::CalculateBulletVelocity(command_t *command) { command->bullet.shot = true; if(command->heading == NORTH) { command->bullet.vel.x = 0.0f; command->bullet.vel.y = 200.0f; } if(command->heading == SOUTH) { command->bullet.vel.x = 0.0f; command->bullet.vel.y = 200.0f; } if(command->heading == EAST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 0.0f; } if(command->heading == WEST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 0.0f; } if(command->heading == NORTHEAST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } if(command->heading == NORTHWEST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } if(command->heading == SOUTHEAST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } if(command->heading == SOUTHWEST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } }

The function takes one parameter (command_t *command) that is the pointer to the command to use. This function needs to run after the

472

Tutorial 5 / Creating Your Online Game

players heading has been calculated, because the bullets heading will be the same.

MovePlayer Function
This function does the actual moving of the players and bullets. Note that the formulas we use to calculate the velocities and positions must match the ones used on the client side. This function is run every time we receive a frame message from a client (a client sends one on every frame, so we end up running the same amount of frames as the client).
void CArmyWarServer::MovePlayer(clientData *client) { float clientFrametime; float multiplier = 100.0f; clientFrametime = client->command.msec / 1000.0f; CalculateVelocity(&client->command, clientFrametime); CalculateHeading(&client->command); // Move the client based on the commands client->command.origin.x += client->command.vel.x; client->command.origin.y += client->command.vel.y; // Bullet if(client->command.bullet.shot == false) { client->command.bullet.origin.x = client->command.origin.x; client->command.bullet.origin.y = client->command.origin.y; } else { client->command.bullet.lifetime += (int) (clientFrametime * 1000.0f); if(client->command.bullet.lifetime > 2000) { client->command.bullet.shot = false; client->command.bullet.lifetime = 0; client->command.bullet.origin.x = client->command.origin.x; client->command.bullet.origin.y = client->command.origin.y; } } if(client->command.key & KEY_WEAPON && client->command.bullet.shot == false) {

Tutorial 5 / Creating Your Online Game

473

CalculateBulletVelocity(&client->command); } if(client->command.bullet.shot) { client->command.bullet.origin.x += client->command.bullet.vel.x * clientFrametime; client->command.bullet.origin.y += client->command.bullet.vel.y * clientFrametime; } // Check for bullet hits if(client->command.bullet.shot) { for(clientData *client2 = clientList; client2 != NULL; client2 = client2->next) { if(client == client2) continue; client2->diedThisFrame = false; VECTOR2D pos = client2->command.origin; pos.x += 16.0f; pos.y += 16.0f; VECTOR2D vec = VectorSubtract(&client-> command.bullet.origin, &pos); float distance = VectorLength(&vec); if(distance < 16.0f) { // Player dies client2->command.origin.x = client2->startPos.x; client2->command.origin.y = client2->startPos.y; client2->diedThisFrame = true; if(client2 == playerWithFlag) { playerWithFlag = NULL; updateFlag = true; } updateKill = true; client->command.bullet.shot = false; break; } } } int f = client->netClient->GetIncomingSequence() & (COMMAND_HISTORY_SIZE1);

474

Tutorial 5 / Creating Your Online Game

client->processedFrame = f; }

The function takes one parameter (clientData *client). This is a pointer to the client to move. We use the clients own frame time, because we must run exactly the same frames as the client. The frame time is given to us in the packet the client sent us.
clientFrametime = client->command.msec / 1000.0f;

The bullet lives for 2 seconds before it is removed. The following piece of code shows how we calculate the lifetime. We simply add the frame time to the bullets lifetime and then check if 2 seconds have passed.
client->command.bullet.lifetime += (int) (clientFrametime * 1000.0f); if(client->command.bullet.lifetime > 2000) { client->command.bullet.shot = false; client->command.bullet.lifetime = 0; client->command.bullet.origin.x = client->command.origin.x; client->command.bullet.origin.y = client->command.origin.y; }

The following code shows how to check for bullet hits. First of all, the bullet must be shot in order to do that. Then we loop through all the players except the one who owns the bullet and check if the bullet position is within 16 pixels of a players position. If it is, the bullet hit and killed that player. If that player was carrying the flag, it is dropped. The flags are set up to indicate that a player died and we should update the clients, and more precisely that this player died.
// Check for bullet hits if(client->command.bullet.shot) { for(clientData *client2 = clientList; client2 != NULL; client2 = client2->next) { if(client == client2) continue; client2->diedThisFrame = false; VECTOR2D pos = client2->command.origin; pos.x += 16.0f; pos.y += 16.0f; VECTOR2D vec = VectorSubtract(&client->command.bullet.origin, &pos); float distance = VectorLength(&vec);

Tutorial 5 / Creating Your Online Game

475

if(distance < 16.0f) { // Player dies client2->command.origin.x = client2->startPos.x; client2->command.origin.y = client2->startPos.y; client2->diedThisFrame = true; if(client2 == playerWithFlag) { playerWithFlag = NULL; updateFlag = true; } updateKill = true; client->command.bullet.shot = false; break; } } }

The last thing we do in this function is set the processedFrame variable for the client. This is a history array indexed value, and it is sent to the client for comparison reasons. (The client compares this frame to the one it processed itself, and this number is the index number that identifies the frame.)
int f = client->netClient->GetIncomingSequence() & (COMMAND_HISTORY_SIZE1); client->processedFrame = f;

CheckFlagCollisions Function
This function checks for possible flag collisions, both if a player picked up the flag and if a player carried the flag to the target area.
void CArmyWarServer::CheckFlagCollisions(void) { if(playerWithFlag != NULL) { // Move the flag with the player flagX = playerWithFlag->command.origin.x; flagY = playerWithFlag->command.origin.y; // Check if the player is at home base if((playerWithFlag) && (playerWithFlag->team == BLUE_TEAM)) { if(playerWithFlag->command.origin.x+16 > (49*32) && playerWithFlag->command.origin.x+16 < (50*32) && playerWithFlag->command.origin.y+16 > (3*32) && playerWithFlag->command.origin.y+16 < (4*32)) { flagX = 49*32; flagY = 49*32;

476

Tutorial 5 / Creating Your Online Game

playerWithFlag = NULL; blueScore++; updateFlag = true; } } if((playerWithFlag) && (playerWithFlag->team == RED_TEAM)) { if(playerWithFlag->command.origin.x+16 > (49*32) && playerWithFlag->command.origin.x+16 < (50*32) && playerWithFlag->command.origin.y+16 > (97*32) && playerWithFlag->command.origin.y+16 < (98*32)) { flagX = 49*32; flagY = 49*32; playerWithFlag = NULL; redScore++; updateFlag = true; } } } else { // Check if anyone is in contact with the flag clientData *list = clientList; for(; list != NULL; list = list->next) { if(list->command.origin.x+16 > flagX && list-> command.origin.x+16 < flagX+32 && list->command.origin.y+16 > flagY && list-> command.origin.y+16 < flagY+32) { char team[10]; if(list->team == RED_TEAM) strcpy(team, "RED team"); else strcpy(team, "BLUE team"); LogString("FLAG hit : player %d: %s, %s", list->netClient->GetIndex(), list-> netClient->GetName(), team); playerWithFlag = list; updateFlag = true; return; } } } }

Tutorial 5 / Creating Your Online Game

477

Frame Function
This function runs the frame on the server. It reads the packets from the clients, moves the players if they send us something, and sends the commands to all the clients. Commands are sent in 100-millisecond intervals to save required bandwidth. Dead reckoning will take care of the clients moving even when they do not receive a packet from the server.
void CArmyWarServer::Frame(int msec) { realtime += msec; frametime = msec / 1000.0f; // Read packets from clients ReadPackets(); if(inProgress == false) return; // Check if someone hit the flag CheckFlagCollisions(); // Wait full 100 ms before allowing to send if(realtime < servertime) { // never let the time get too far off if(servertime realtime > 100) { realtime = servertime 100; } return; } // Bump frame number, and calculate new servertime framenum++; servertime = framenum * 100; if(servertime < realtime) realtime = servertime; SendCommand(); // Reset update flags updateFlag = false; updateKill = false; }

This function takes one parameter (int msec), the frame time of the server. Packets are read every frame, but they are not sent every frame.

478

Tutorial 5 / Creating Your Online Game

Here we check if enough time has passed to send packets. If not, the function returns and does not reach the sending function.
// Wait full 100 ms before allowing to send if(realtime < servertime) { // Never let the time get too far off if(servertime realtime > 100) { realtime = servertime 100; } return; }

A new server time is calculated when the old one is reached:

// Bump frame number, and calculate new servertime framenum++; servertime = framenum * 100;

Commands are sent to each client:

SendCommand();

lobby.cpp File
Now that we have the game data structures, we can implement the functions we introduced in Tutorial 4.

AddGame Function
This function adds a game to the servers game list. If the list does not exist yet, it is created. This means that no games exist yet, so the game we are adding is the first one. Otherwise, we just add a new game to the list. The games information is filled in (name, index number, etc.). Then the function tries to open a game server for the new game, using the InitNetwork function. Each game uses its own port number, which is based on the games index number. Finally, the games counter is increased.
void CLobbyServer::AddGame(char *name) { // First get a pointer to the beginning of client list CArmyWarServer *list = gameList; CArmyWarServer *prev; // No clients yet, adding the first one if(gameList == NULL) {

Tutorial 5 / Creating Your Online Game

479

gameList = new CArmyWarServer; gameList->SetName(name); gameList->next = NULL; gameList->SetIndex(gameAmount); gameList->GenerateRandomMap(); if(gameList->InitNetwork(GetGameAmount()) != 0) { LogString("Could not create game server"); } } else { prev = list; list = gameList->next; while(list != NULL) { prev = list; list = list->next; } list = new CArmyWarServer; list->SetName(name); list->next = NULL; list->SetIndex(gameAmount); list->GenerateRandomMap(); if(list->InitNetwork(GetGameAmount()) != 0) { LogString("Could not create game server"); } prev->next = list; } gameAmount++; }

RemoveGame Function
This function removes the selected game from the servers game list. First, the correct game is looked up by going through the game list and comparing the names. Once the game is found, the games network is shut down and the game list is updated. If the game is the last in the list, the list is marked empty by pointing it to NULL (or as you can see in the code, by pointing it to the next game in the list, which is NULL). The game counter is then decreased.

480

Tutorial 5 / Creating Your Online Game

void CLobbyServer::RemoveGame(char *name) { CArmyWarServer *list = gameList; CArmyWarServer *prev = NULL; CArmyWarServer *next = NULL; for( ; list != NULL; list = list->next) { if(strcmp(name, list->GetName()) == 0) { if(prev != NULL) { prev->next = list->next; } break; } prev = list; } if(list == gameList) { if(list) { list->ShutdownNetwork(); next = list->next; delete list; } list = NULL; gameList = next; } else { if(list) { list->ShutdownNetwork(); next = list->next; delete list; } list = next; } gameAmount--; }

Tutorial 5 / Creating Your Online Game

481

RemoveGames Function
This function simply removes all the games in the game list.
void CLobbyServer::RemoveGames(void) { CArmyWarServer *list = gameList; CArmyWarServer *next; while(list != NULL) { if(list) { list->ShutdownNetwork(); next = list->next; delete list; } list = next; } gameList = NULL; gameAmount = 0; }

Summary of Server Code

That is it for the server-side code. We now have a working server for our game that receives data and moves the players, and sends only in 100-millisecond intervals to save bandwidth. Now all we need is a client connecting to the server.

Game Client Code

Now to the final part of this tutorial, the client-side code. We have some new source code files: client.cpp, network.cpp, and client.h. Also some of the old client-side files have changed a little bit.

482

Tutorial 5 / Creating Your Online Game

Figure 2

client.h File
Like on the server side (server.h), this header file contains the application-specific data structures. This is where the player data structures are, along with the games main class and some definitions. The main class has the network interface methods that we learned to create in the previous tutorials (like ReadPackets() and so on).
#ifndef CLIENT_H #define CLIENT_H #include #include #include #include <gl/gl.h> <gl/glu.h> <gl/glaux.h> <2dlib.h>

#include "network.h" #define #define #define #define #define #define #define #define NORTH NORTHEAST EAST SOUTHEAST SOUTH SOUTHWEST WEST NORTHWEST 0 45 90 135 180 225 270 315

#define BLUE_TEAM 0 #define RED_TEAM 1 typedef struct {

Tutorial 5 / Creating Your Online Game

483

float x; float y; } VECTOR2D; typedef struct bullet_t { VECTOR2D vel; VECTOR2D origin; VECTOR2D predictedOrigin; bool int } bullet_t; typedef struct { int int VECTOR2D VECTOR2D VECTOR2D bullet_t int } command_t; shot; lifetime;

key; heading; vel; origin; predictedOrigin; bullet; msec;

typedef struct clientData { command_t frame[COMMAND_HISTORY_SIZE]; command_t serverFrame; command_t int int VECTOR2D bool char char clientData } clientData; command; index; processedFrame; startPos; team; nickname[30]; password[30]; *next;

// // // //

frame history the latest frame from server current frame's commands

// The main application class interface class CArmyWar { private: // Methods // Client.cpp

484

Tutorial 5 / Creating Your Online Game

void InitializeEngine(void); void DrawMap(void); void CheckVictory(void); void KillPlayer(int index); clientData *GetClientPointer(int index); void void void void void void void CheckPredictionError(int a); CheckBulletPredictionError(int a); CalculateVelocity(command_t *command, float frametime); CalculateHeading(command_t *command); CalculateBulletVelocity(command_t *command); PredictMovement(int prevFrame, int curFrame); MoveObjects(void);

void AddClient(int local, int index, char *name); void RemoveClient(int index); void RemoveClients(void); // Network.cpp void ReadPackets(void); void SendCommand(void); void SendRequestNonDeltaFrame(void); void ReadMoveCommand(dreamMessage *mes, clientData *client); void ReadDeltaMoveCommand(dreamMessage *mes, clientData *client); void BuildDeltaMoveCommand(dreamMessage *mes, clientData *theClient);

// Variables // Network variables dreamClient *networkClient; clientData *clientList; clientData *localClient; int clients; clientData inputClient; // Graphic declarations GFX_IMAGE2D grass; GFX_IMAGE2D redman; GFX_IMAGE2D blueman; GFX_IMAGE2D tree; GFX_IMAGE2D redtarget; GFX_IMAGE2D bluetarget; GFX_IMAGE2D flag; GFX_IMAGE2D rednumbers[10]; GFX_IMAGE2D bluenumbers[10]; float frametime; char gamename[32]; bool inProgress; // Handles all keyboard input // Client list // Pointer to the local client in the // client list

// // // // // // // // //

variable variable variable variable variable variable variable variable variable

to to to to to to to to to

hold hold hold hold hold hold hold hold hold

graphic graphic graphic graphic graphic graphic graphic graphic graphic

Tutorial 5 / Creating Your Online Game

485

bool init; // Tile scroll positions int scrollX; int scrollY; int tileScrollX; int tileScrollY; bool mapdata[100][100]; int gameIndex; float targetRotation; int redScore; int blueScore; float flagX; float flagY; clientData *playerWithFlag; public: CArmyWar(); ~CArmyWar(); // Client.cpp void Shutdown(void); void CheckKeys(void); void Frame(void); void RunNetwork(int msec); // Network.cpp void StartConnection(int ind); void Connect(void); void Disconnect(void); void SendStartGame(void); void SetName(char *n) char *GetName(void) {strcpy(gamename, n);} {return gamename;} // variable to rotate the target images

void SetGameIndex(int index) {gameIndex = index;} int GetGameIndex(void) {return gameIndex;} clientData *GetClientList(void) {return clientList;} void SetInProgress(bool p) bool GetInProgress(void) CArmyWar *next; }; #endif {inProgress = p;} {return inProgress;}

486

Tutorial 5 / Creating Your Online Game

The command data structures are the same as on the server side, so we will skip that part now. The client data structure is slightly different, but is very easy to understand. The directions in which a player can move are given in the following definitions:
#define #define #define #define #define #define #define #define NORTH NORTHEAST EAST SOUTHEAST SOUTH SOUTHWEST WEST NORTHWEST 0 45 90 135 180 225 270 315

The CArmyWar class has some variables that need explaining, so here goes.
// Network variables dreamClient *networkClient; clientData *clientList; clientData *localClient; int clients; clientData inputClient; // Graphic declarations GFX_IMAGE2D grass; GFX_IMAGE2D redman; GFX_IMAGE2D blueman; GFX_IMAGE2D tree; GFX_IMAGE2D redtarget; GFX_IMAGE2D bluetarget; GFX_IMAGE2D flag; GFX_IMAGE2D rednumbers[10]; GFX_IMAGE2D bluenumbers[10]; float frametime; char gamename[32]; bool inProgress; bool init; // Tile scroll positions int scrollX; int scrollY; int tileScrollX; int tileScrollY; bool mapdata[100][100]; int gameIndex; // Handles all keyboard input // Client list // Pointer to the local client in the // client list

// // // // // // // // //

variable variable variable variable variable variable variable variable variable

to to to to to to to to to

hold hold hold hold hold hold hold hold hold

graphic graphic graphic graphic graphic graphic graphic graphic graphic

Tutorial 5 / Creating Your Online Game

487

float targetRotation; int redScore; int blueScore; float flagX; float flagY; clientData *playerWithFlag;

// variable to rotate the target images

The dreamClient networkClient is the dreamClient network client object. The clientData *clientList is the list of all the clients connected on the same server as we are. The clientData *localClient is the local client on that list. The clientData inputClient is used to handle the keyboard input. The ints scrollX and scrollY are used to store the amount the screen has scrolled (in pixels). The ints tileScrollX and tileScrollY are used to store the amount the screen has scrolled (in tiles). A tile is 32 pixels in height and width. The bool mapdata[100][100] stores the map data. A 0 means plain grass and a 1 means a tree. The floats flagX and flagY store the flags coordinates. The clientData *playerWithFlag is a pointer to the player who is carrying the flag. If no one is carrying the flag, this pointer is NULL.

network.h File
The network.h file on the client side now looks like this:
#ifndef NETWORK_H #define NETWORK_H #define COMMAND_HISTORY_SIZE #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define KEY_UP KEY_DOWN KEY_LEFT KEY_RIGHT KEY_WEAPON CMD_KEY CMD_HEADING CMD_ORIGIN CMD_BULLET CMD_FLAG CMD_KILL USER_MES_FRAME USER_MES_NONDELTAFRAME USER_MES_SERVEREXIT USER_MES_LOGIN USER_MES_SIGNIN 64 1 2 4 8 16 1 2 4 8 16 32 1 2 3 4 5

488

Tutorial 5 / Creating Your Online Game

#define #define #define #define #define #define #define

USER_MES_CHAT USER_MES_CREATEGAME USER_MES_REMOVEGAME USER_MES_GAMEDATA USER_MES_STARTGAME USER_MES_MAPDATA USER_MES_KEEPALIVE

6 7 8 9 10 11 12

typedef struct clientLoginData { int index; char nickname[30]; clientLoginData *next; } clientLoginData; #endif

common.h File
This header file is just a header file wrapper. So you only need to include this header, and all the required header files are provided for you in the correct order.
#ifndef __COMMON_H__ #define __COMMON_H__ #include "dreamSock.h" #include #include #include #include #include #endif "client.h" "network.h" "lobby.h" "signin.h" "main.h"

main.cpp File
Our games base initialization happens in this file, which includes some new functions. All the windows are created and handled here. This file does not include game logic.

VectorLength and VectorSubtract Functions

These functions are helper functions to do some vector calculations. The first one returns the length of a vector, and the second one subtracts one vector from another.
float VectorLength(VECTOR2D *vec) { return (float) sqrt(vec->x*vec->x + vec->y*vec->y);

Tutorial 5 / Creating Your Online Game

489

} VECTOR2D VectorSubtract(VECTOR2D *vec1, VECTOR2D *vec2) { VECTOR2D vec; vec.x = vec1->x vec2->x; vec.y = vec1->y vec2->y; return vec; }

ApplicationProc Function
This function now handles keyboard input and maintains the Join Game buttons state.
case WM_KEYDOWN: { keys[wParam] = TRUE; break; } case WM_KEYUP: { keys[wParam] = FALSE; break; } default: if(!Lobby.GetGameAmount()) { EnableWindow(GetDlgItem(hWnd_LobbyDialog, IDC_JOINGAME), FALSE); } break;

Dialog Procedures
The dialog procedure functions have also changed, as we now actually send the data that goes with the dialogs. They are very simple additions, so they are not listed here.

Main Loop
The games main loop is almost the same as in the earlier tutorials, but this time it also handles network and keyboard input/output for the game itself. The new part can be seen here:
// If we have a local game, run the frames for it if(Lobby.GetLocalGame() != NULL) { Lobby.GetLocalGame()->RunNetwork(time); Lobby.GetLocalGame()->CheckKeys();

490

Tutorial 5 / Creating Your Online Game

Lobby.GetLocalGame()->Frame(); }

network.cpp File
This file contains the network-only code for the client. The connection can be started with these methods and data can be sent and received. Also, the client list is updated with these functions. There are many functions that are similar to the ones on the server side, so those functions will not be explained here.
#include "common.h" //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::StartConnection(int ind) { LogString("StartConnection %d", ind); gameIndex = ind; int ret = networkClient->Initialize("", serverIP, 30004 + gameIndex); if(ret == DREAMSOCK_CLIENT_ERROR) { char text[64]; sprintf(text, "Could not open client socket"); MessageBox(NULL, text, "Error", MB_OK); } Connect(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::ReadPackets(void) { char data[1400]; struct sockaddr address; clientData *clList; int int int int type; ind; local; ret;

Tutorial 5 / Creating Your Online Game

491

char name[50]; dreamMessage mes; mes.Init(data, sizeof(data)); while(ret = networkClient->GetPacket(mes.data, &address)) { mes.SetSize(ret); mes.BeginReading(); type = mes.ReadByte(); switch(type) { case DREAMSOCK_MES_ADDCLIENT: local = mes.ReadByte(); ind = mes.ReadByte(); strcpy(name, mes.ReadString()); AddClient(local, ind, name); break; case DREAMSOCK_MES_REMOVECLIENT: ind = mes.ReadByte(); LogString("Got removeclient %d message", ind); RemoveClient(ind); if(clientList == NULL) { LogString("clientList == NULL, sending remove game %s", gamename); Lobby.SendRemoveGame(gamename); } break; case USER_MES_FRAME: // Skip sequences mes.ReadShort(); mes.ReadShort(); for(clList = clientList; clList != NULL; clList = clList->next) { LogString("Reading DELTAFRAME for client %d", clList->index); ReadDeltaMoveCommand(&mes, clList); } break; case USER_MES_NONDELTAFRAME: // Skip sequences

492

Tutorial 5 / Creating Your Online Game

mes.ReadShort(); mes.ReadShort(); clList = clientList; for(clList = clientList; clList != NULL; clList = clList->next) { LogString("Reading NONDELTAFRAME for client %d", clList->index); ReadMoveCommand(&mes, clList); } break; case USER_MES_SERVEREXIT: MessageBox(NULL, "Server disconnected", "Info", MB_OK); Disconnect(); break; case USER_MES_STARTGAME: // Skip sequences mes.ReadShort(); mes.ReadShort(); DestroyWindow(hWnd_JoinGameDialog); InitializeEngine(); break; case USER_MES_MAPDATA: // Skip sequences mes.ReadShort(); mes.ReadShort(); for(int m = 0; m < 300; m++) { int i = mes.ReadByte(); int j = mes.ReadByte(); mapdata[i][j] = true; } break; } } } //--------------------------------------------------------------------------// Name: empty() // Desc: //---------------------------------------------------------------------------

Tutorial 5 / Creating Your Online Game

493

void CArmyWar::AddClient(int local, int ind, char *name) { // First get a pointer to the beginning of the client list clientData *list = clientList; clientData *prev; LogString("App: Client: Adding client with index %d", ind); // No clients yet, adding the first one if(clientList == NULL) { LogString("App: Client: Adding first client"); clientList = (clientData *) calloc(1, sizeof(clientData)); if(local) { LogString("App: Client: This one is local"); localClient = clientList; } clientList->index = ind; strcpy(clientList->nickname, name); if(clients % 2 == 0) clientList->team = RED_TEAM; else clientList->team = BLUE_TEAM; clientList->next = NULL; } else { LogString("App: Client: Adding another client"); prev = list; list = clientList->next; while(list != NULL) { prev = list; list = list->next; } list = (clientData *) calloc(1, sizeof(clientData)); if(local) { LogString("App: Client: This one is local"); localClient = list; } list->index = ind;

494

Tutorial 5 / Creating Your Online Game

strcpy(list->nickname, name); if(clients % 2 == 0) list->team = RED_TEAM; else list->team = BLUE_TEAM; list->next = NULL; prev->next = list; } clients++; // If we just joined the game, request a non-delta compressed frame if(local) SendRequestNonDeltaFrame(); Lobby.RefreshJoinedPlayersList(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::RemoveClient(int ind) { clientData *list = clientList; clientData *prev = NULL; clientData *next = NULL; // Look for correct client and update list for(; list != NULL; list = list->next) { if(list->index == ind) { if(prev != NULL) { prev->next = list->next; } break; } prev = list; } // First entry if(list == clientList) { if(list) { next = list->next; free(list);

Tutorial 5 / Creating Your Online Game

495

} list = NULL; clientList = next; } // Other else { if(list) { next = list->next; free(list); } list = next; } clients--; Lobby.RefreshJoinedPlayersList(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::RemoveClients(void) { clientData *list = clientList; clientData *next; while(list != NULL) { if(list) { next = list->next; free(list); } list = next; } clientList = NULL; clients = 0; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::SendCommand(void) { if(networkClient->GetConnectionState() != DREAMSOCK_CONNECTED)

496

Tutorial 5 / Creating Your Online Game

return; dreamMessage message; char data[1400]; int i = networkClient->GetOutgoingSequence() & (COMMAND_HISTORY_SIZE1); message.Init(data, sizeof(data)); message.WriteByte(USER_MES_FRAME); message.AddSequences(networkClient); // Build delta-compressed move command BuildDeltaMoveCommand(&message, &inputClient); // Send the packet networkClient->SendPacket(&message); // Store the command to the input client's history memcpy(&inputClient.frame[i], &inputClient.command, sizeof(command_t)); clientData *clList = clientList; // Store the commands to the clients' history for(; clList != NULL; clList = clList->next) { memcpy(&clList->frame[i], &clList->command, sizeof(command_t)); } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::SendStartGame(void) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_STARTGAME); message.AddSequences(networkClient); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::SendRequestNonDeltaFrame(void) { char data[1400]; dreamMessage message;

// type // sequences

Tutorial 5 / Creating Your Online Game

497

message.Init(data, sizeof(data)); message.WriteByte(USER_MES_NONDELTAFRAME); message.AddSequences(networkClient); networkClient->SendPacket(&message); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::Connect(void) { if(init) { LogString("ArmyWar already initialized"); return; } LogString("CArmyWar::Connect"); init = true; networkClient->SendConnect(Lobby.GetLocalClient()->nickname); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::Disconnect(void) { if(!init) return; LogString("CArmyWar::Disconnect"); init = false; localClient = NULL; memset(&inputClient, 0, sizeof(clientData)); networkClient->SendDisconnect(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::ReadMoveCommand(dreamMessage *mes, clientData *client) { // Key client->serverFrame.key = mes->ReadByte();

498

Tutorial 5 / Creating Your Online Game

// Heading client->serverFrame.heading // Origin client->serverFrame.origin.x client->serverFrame.origin.y client->serverFrame.vel.x client->serverFrame.vel.y client->serverFrame.bullet.origin.x client->serverFrame.bullet.origin.y client->serverFrame.bullet.vel.x client->serverFrame.bullet.vel.y client->serverFrame.bullet.lifetime client->serverFrame.bullet.shot int playerWithFlagIndex

= mes->ReadShort();

= = = = = = = = = =

mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadShort(); mes->ReadByte();

= mes->ReadShort();

if(playerWithFlagIndex != 1) { playerWithFlag = GetClientPointer(playerWithFlagIndex); } flagX = mes->ReadFloat(); flagY = mes->ReadFloat(); redScore = mes->ReadByte(); blueScore = mes->ReadByte(); // Read time to run command client->serverFrame.msec = mes->ReadByte(); memcpy(&client->command, &client->serverFrame, sizeof(command_t)); // Fill the history array with the position we got for(int f = 0; f < COMMAND_HISTORY_SIZE; f++) { client->frame[f].predictedOrigin.x = client->command.origin.x; client->frame[f].predictedOrigin.y = client->command.origin.y; client->frame[f].bullet.predictedOrigin.x = client-> command.bullet.origin.x; client->frame[f].bullet.predictedOrigin.y = client-> command.bullet.origin.y; } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::ReadDeltaMoveCommand(dreamMessage *mes, clientData *client) { int processedFrame;

Tutorial 5 / Creating Your Online Game

499

int flags = 0; // Flags flags = mes->ReadByte(); // Key if(flags & CMD_KEY) { client->serverFrame.key = mes->ReadByte(); client->command.key = client->serverFrame.key; LogString("Client %d: Read key %d", client->index, client-> command.key); } if(flags & CMD_ORIGIN || flags & CMD_BULLET) { processedFrame = mes->ReadByte(); client->processedFrame = processedFrame; } // Origin if(flags & CMD_ORIGIN) { client->serverFrame.origin.x = mes->ReadFloat(); client->serverFrame.origin.y = mes->ReadFloat(); client->serverFrame.vel.x = mes->ReadFloat(); client->serverFrame.vel.y = mes->ReadFloat(); if(client == localClient) { CheckPredictionError(processedFrame); } else { client->command.origin.x = client->serverFrame.origin.x; client->command.origin.y = client->serverFrame.origin.y; } } if(flags & CMD_BULLET) { client->serverFrame.bullet.origin.x = mes->ReadFloat(); client->serverFrame.bullet.origin.y = mes->ReadFloat(); client->serverFrame.bullet.vel.x = mes->ReadFloat(); client->serverFrame.bullet.vel.y = mes->ReadFloat(); client->serverFrame.bullet.shot = mes->ReadByte(); client->command.bullet.shot = client->serverFrame.bullet.shot; if(client == localClient) { CheckBulletPredictionError(processedFrame);

500

Tutorial 5 / Creating Your Online Game

} else { client->command.bullet.origin.x = client-> serverFrame.bullet.origin.x; client->command.bullet.origin.y = client-> serverFrame.bullet.origin.y; } } // Flag & points if(flags & CMD_FLAG) { int playerWithFlagIndex = mes->ReadShort(); if(playerWithFlagIndex != 0) { LogString("FLAG playerWithFlagIndex %d", playerWithFlagIndex); playerWithFlag = GetClientPointer(playerWithFlagIndex); } else { playerWithFlag = NULL; } flagX = mes->ReadFloat(); flagY = mes->ReadFloat(); redScore = mes->ReadByte(); blueScore = mes->ReadByte(); CheckVictory(); } // Someone died if(flags & CMD_KILL) { int died = mes->ReadByte(); if(died) KillPlayer(client->index); } // Read time to run command client->command.msec = mes->ReadByte(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::BuildDeltaMoveCommand(dreamMessage *mes, clientData *theClient) {

Tutorial 5 / Creating Your Online Game

501

int flags = 0; int last = (networkClient->GetOutgoingSequence() 1) & (COMMAND_HISTORY_SIZE1); // Check what needs to be updated if(theClient->frame[last].key != theClient->command.key) flags |= CMD_KEY; // Add to the message // Flags mes->WriteByte(flags); // Key if(flags & CMD_KEY) { mes->WriteByte(theClient->command.key); } mes->WriteByte(theClient->command.msec); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::RunNetwork(int msec) { static int time = 0; time += msec; // Framerate is too high if(time < (1000 / 60)) return; frametime = time / 1000.0f; time = 0; // Read packets from server, and send new commands ReadPackets(); SendCommand(); int ack = networkClient->GetIncomingAcknowledged(); int current = networkClient->GetOutgoingSequence(); // Check that we haven't gone too far if(current ack > COMMAND_HISTORY_SIZE) return; // Predict the frames that we are waiting on from the server for(int a = ack + 1; a < current; a++) { int prevframe = (a1) & (COMMAND_HISTORY_SIZE1); int frame = a & (COMMAND_HISTORY_SIZE1);

502

Tutorial 5 / Creating Your Online Game

PredictMovement(prevframe, frame); } MoveObjects(); }

StartConnection Function
This function starts a specified connection. The parameter ind is used to specify the index number of the game to which we want to connect. Each game uses its own port, which is 30004 plus the games index number. So the first game is run on port 30004, the second one on port 30005, and so on.
void CArmyWar::StartConnection(int ind) { LogString("StartConnection %d", ind); gameIndex = ind; int ret = networkClient->Initialize("", serverIP, 30004 + gameIndex); if(ret == DREAMSOCK_CLIENT_ERROR) { char text[64]; sprintf(text, "Could not open client socket"); MessageBox(NULL, text, "Error", MB_OK); } Connect(); }

SendCommand Function
The SendCommand function is used to send the commands to the server. This function is run every frame, so every frames commands are sent to the server.
void CArmyWar::SendCommand(void) { if(networkClient->GetConnectionState() != DREAMSOCK_CONNECTED) return; dreamMessage message; char data[1400]; int i = networkClient->GetOutgoingSequence() & (COMMAND_HISTORY_SIZE1); message.Init(data, sizeof(data)); message.WriteByte(USER_MES_FRAME); message.AddSequences(networkClient);

// type // sequences

Tutorial 5 / Creating Your Online Game

503

// Build delta-compressed move command BuildDeltaMoveCommand(&message, &inputClient); // Send the packet networkClient->SendPacket(&message); // Store the command to the input client's history memcpy(&inputClient.frame[i], &inputClient.command, sizeof(command_t)); clientData *clList = clientList; // Store the commands to the clients' history for(; clList != NULL; clList = clList->next) { memcpy(&clList->frame[i], &clList->command, sizeof(command_t)); } }

A message is always filled on every frame and contains that frames commands. The message is built using the BuildDeltaMoveCommand() function. That function will be explained later in this tutorial. When the message is built, it is ready to be sent to the server. Remember that this happens on every frame, even if nothing has changed from the previous frame. The BuildDeltaMoveCommand() function takes care of keeping track of any changes.
message.Init(data, sizeof(data)); message.WriteByte(USER_MES_FRAME); message.AddSequences(networkClient); // Build delta-compressed move command BuildDeltaMoveCommand(&message, &inputClient); // Send the packet networkClient->SendPacket(&message); // type // sequences

Every frame, the message is stored into message history as seen below. An index number for the history table is calculated by looking at the outgoing sequence number of the client. Every frame, this number is one bigger than the last time. Remember that the index number starts at 0 and ends at COMMAND_HISTORY_SIZE 1 (COMMAND_ HISTORY_SIZE is defined as 64 in network.h). After 64 frames, the history starts to get overwritten as that data is too old anyway.
int i = networkClient->GetOutgoingSequence() & (COMMAND_HISTORY_SIZE1); ... // Store the command to the input client's history memcpy(&inputClient.frame[i], &inputClient.command, sizeof(command_t)); clientData *clList = clientList;

504

Tutorial 5 / Creating Your Online Game

// Store the commands to the clients' history for(; clList != NULL; clList = clList->next) { memcpy(&clList->frame[i], &clList->command, sizeof(command_t)); }

SendStartGame Function
This function builds a start game message and sends it to the server.
void CArmyWar::SendStartGame(void) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_STARTGAME); message.AddSequences(networkClient); networkClient->SendPacket(&message); }

SendRequestNonDeltaFrame Function
This function builds a request non-delta frame message and sends it to the server.
void CArmyWar::SendRequestNonDeltaFrame(void) { char data[1400]; dreamMessage message; message.Init(data, sizeof(data)); message.WriteByte(USER_MES_NONDELTAFRAME); message.AddSequences(networkClient); networkClient->SendPacket(&message); }

Connect Function
This function sends a connect message to the server, telling the server we want start a connection with it. UDP protocol does not really connect to a server, so we sort of fake it this way.
void CArmyWar::Connect(void) { if(init) { LogString("ArmyWar already initialized"); return; }

Tutorial 5 / Creating Your Online Game

505

LogString("CArmyWar::Connect"); init = true; networkClient->SendConnect(Lobby.GetLocalClient()->nickname); }

Disconnect Function
This function disconnects from the server by sending a disconnect message. Some local variables are reset so they can be reused.
void CArmyWar::Disconnect(void) { if(!init) return; LogString("CArmyWar::Disconnect"); init = false; localClient = NULL; memset(&inputClient, 0, sizeof(clientData)); networkClient->SendDisconnect(); }

ReadMoveCommand Function
This function reads the non-delta (absolute values) commands of a message. The parameters define the message to read and the client that owns the commands.
void CArmyWar::ReadMoveCommand(dreamMessage *mes, clientData *client) { // Key client->serverFrame.key = mes->ReadByte(); // Heading client->serverFrame.heading // Origin client->serverFrame.origin.x client->serverFrame.origin.y client->serverFrame.vel.x client->serverFrame.vel.y client->serverFrame.bullet.origin.x client->serverFrame.bullet.origin.y client->serverFrame.bullet.vel.x client->serverFrame.bullet.vel.y client->serverFrame.bullet.lifetime client->serverFrame.bullet.shot

= mes->ReadShort();

= = = = = = = = = =

mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadFloat(); mes->ReadShort(); mes->ReadByte();

506

Tutorial 5 / Creating Your Online Game

int playerWithFlagIndex

= mes->ReadShort();

ReadDeltaMoveCommand Function
This function reads the delta (change from last known value) commands. First the flags are read to see which commands are included in this message. Then each included command is read and stored to the serverFrame structure.
void CArmyWar::ReadDeltaMoveCommand(dreamMessage *mes, clientData *client) { int processedFrame; int flags = 0; // Flags flags = mes->ReadByte(); // Key if(flags & CMD_KEY) { client->serverFrame.key = mes->ReadByte(); client->command.key = client->serverFrame.key;

Tutorial 5 / Creating Your Online Game

507

LogString("Client %d: Read key %d", client->index, client-> command.key); } if(flags & CMD_ORIGIN || flags & CMD_BULLET) { processedFrame = mes->ReadByte(); client->processedFrame = processedFrame; } // Origin if(flags & CMD_ORIGIN) { client->serverFrame.origin.x = mes->ReadFloat(); client->serverFrame.origin.y = mes->ReadFloat(); client->serverFrame.vel.x = mes->ReadFloat(); client->serverFrame.vel.y = mes->ReadFloat(); if(client == localClient) { CheckPredictionError(processedFrame); } else { client->command.origin.x = client->serverFrame.origin.x; client->command.origin.y = client->serverFrame.origin.y; } } if(flags & CMD_BULLET) { client->serverFrame.bullet.origin.x = mes->ReadFloat(); client->serverFrame.bullet.origin.y = mes->ReadFloat(); client->serverFrame.bullet.vel.x = mes->ReadFloat(); client->serverFrame.bullet.vel.y = mes->ReadFloat(); client->serverFrame.bullet.shot = mes->ReadByte(); client->command.bullet.shot = client->serverFrame.bullet.shot; if(client == localClient) { CheckBulletPredictionError(processedFrame); } else { client->command.bullet.origin.x = client-> serverFrame.bullet.origin.x; client->command.bullet.origin.y = client-> serverFrame.bullet.origin.y; } } // Flag & points

508

Tutorial 5 / Creating Your Online Game

if(flags & CMD_FLAG) { int playerWithFlagIndex = mes->ReadShort(); if(playerWithFlagIndex != 0) { LogString("FLAG playerWithFlagIndex %d", playerWithFlagIndex); playerWithFlag = GetClientPointer(playerWithFlagIndex); } else { playerWithFlag = NULL; } flagX = mes->ReadFloat(); flagY = mes->ReadFloat(); redScore = mes->ReadByte(); blueScore = mes->ReadByte(); CheckVictory(); } // Someone died if(flags & CMD_KILL) { int died = mes->ReadByte(); if(died) KillPlayer(client->index); } // Read time to run command client->command.msec = mes->ReadByte(); }

BuildDeltaMoveCommand Function
This function is used to build the command message based on local inputs. It is pretty simple; if a key has been pressed this frame, that key command is written into the message. Then the frame time is written. The parameters define the message to write to and the input data to use (normally inputClient).
void CArmyWar::BuildDeltaMoveCommand(dreamMessage *mes, clientData *theClient) { int flags = 0; int last = (networkClient->GetOutgoingSequence() 1) & (COMMAND_HISTORY_SIZE1); // Check what needs to be updated

Tutorial 5 / Creating Your Online Game

509

if(theClient->frame[last].key != theClient->command.key) flags |= CMD_KEY; // Add to the message // Flags mes->WriteByte(flags); // Key if(flags & CMD_KEY) { mes->WriteByte(theClient->command.key); } mes->WriteByte(theClient->command.msec); }

RunNetwork Function
This function runs the network so the data will flow. Because this is client-side code, we can and should keep the frame rate under control or we might end up with clients that are too fast. The function reads and sends packets, and then finally moves the local objects so the player sees what is happening in the game.
void CArmyWar::RunNetwork(int msec) { static int time = 0; time += msec; // Framerate is too high if(time < (1000 / 60)) return; frametime = time / 1000.0f; time = 0; // Read packets from server and send new commands ReadPackets(); SendCommand(); int ack = networkClient->GetIncomingAcknowledged(); int current = networkClient->GetOutgoingSequence(); // Check that we haven't gone too far if(current ack > COMMAND_HISTORY_SIZE) return; // Predict the frames that we are waiting on from the server for(int a = ack + 1; a < current; a++) { int prevframe = (a1) & (COMMAND_HISTORY_SIZE1); int frame = a & (COMMAND_HISTORY_SIZE1);

510

Tutorial 5 / Creating Your Online Game

PredictMovement(prevframe, frame); } MoveObjects(); }

You can see how the network is run below. First we read packets from the server, then we send new commands back to the server. After that, the interesting part begins: client prediction or dead reckoning. The local client will produce its own versions of the frames that it is waiting on from the server. The client first produces the frame that we last got from the server, so the client and the server agree on all the player positions and so on. Now that we are sending packets to the server each frame, but the server only sends packets to us in 100 ms intervals, we must guess what the frames in between look like. So the client produces these frames by using the known player velocities and so on. Once we get a new frame from the server, it all starts from the beginning.
// Read packets from server and send new commands ReadPackets(); SendCommand(); int ack = networkClient->GetIncomingAcknowledged(); int current = networkClient->GetOutgoingSequence(); // Check that we haven't gone too far if(current ack > COMMAND_HISTORY_SIZE) return;

The current frame is produced with the following code. We must produce some past frames first, because we do not know if there was a prediction error (a difference between the servers frame and ours). If there was an error in one of the past frames, it becomes fixed once the server tells us the real frame data. It might take some time for the server to do that though, so we must keep producing the old frames until we know the real values of the frame data (player positions and movement).
// Predict the frames that we are waiting on from the server for(int a = ack + 1; a < current; a++) { int prevframe = (a1) & (COMMAND_HISTORY_SIZE1); int frame = a & (COMMAND_HISTORY_SIZE1); PredictMovement(prevframe, frame); } MoveObjects();

Tutorial 5 / Creating Your Online Game

511

client.cpp File
This file contains the game logic, meaning that here we have the functions that move the players and check for bullet or flag hits. The file in its entirety is listed here:
#include "common.h" //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CArmyWar::CArmyWar() { networkClient = new dreamClient; clientList = NULL; localClient = NULL; clients = 0; memset(&inputClient, 0, sizeof(clientData)); memset(&mapdata, 0, sizeof(mapdata)); frametime inProgress init scrollX scrollY tileScrollX tileScrollY gameIndex targetRotation redScore blueScore playerWithFlag next } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------CArmyWar::~CArmyWar() { delete networkClient; = 0.0f; = false; = false; = 0; = 0; = 0; = 0; = 0; = 0.0f; = 0; = 0; = NULL; = NULL;

512

Tutorial 5 / Creating Your Online Game

} //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::InitializeEngine(void) { // Init the graphics engine GFX_Init("Army War Engine v2.0", 640, 480, 16, 0, ApplicationProc); // Load required graphics GFX_LoadBitmap(&grass, GFX_LoadBitmap(&redman, GFX_LoadBitmap(&blueman, GFX_LoadBitmap(&tree, GFX_LoadBitmap(&redtarget, GFX_LoadBitmap(&bluetarget, GFX_LoadBitmap(&flag,

"gfx\\grass.bmp"); "gfx\\redman.bmp"); "gfx\\blueman.bmp"); "gfx\\tree.bmp"); "gfx\\redtarget.bmp"); "gfx\\bluetarget.bmp"); "gfx\\flag.bmp"); "gfx\\red0.bmp"); "gfx\\red1.bmp"); "gfx\\red2.bmp"); "gfx\\red3.bmp"); "gfx\\red4.bmp"); "gfx\\red5.bmp"); "gfx\\red6.bmp"); "gfx\\red7.bmp"); "gfx\\red8.bmp"); "gfx\\red9.bmp"); "gfx\\blue0.bmp"); "gfx\\blue1.bmp"); "gfx\\blue2.bmp"); "gfx\\blue3.bmp"); "gfx\\blue4.bmp"); "gfx\\blue5.bmp"); "gfx\\blue6.bmp"); "gfx\\blue7.bmp"); "gfx\\blue8.bmp"); "gfx\\blue9.bmp");

GFX_LoadBitmap(&rednumbers[0], GFX_LoadBitmap(&rednumbers[1], GFX_LoadBitmap(&rednumbers[2], GFX_LoadBitmap(&rednumbers[3], GFX_LoadBitmap(&rednumbers[4], GFX_LoadBitmap(&rednumbers[5], GFX_LoadBitmap(&rednumbers[6], GFX_LoadBitmap(&rednumbers[7], GFX_LoadBitmap(&rednumbers[8], GFX_LoadBitmap(&rednumbers[9],

GFX_LoadBitmap(&bluenumbers[0], GFX_LoadBitmap(&bluenumbers[1], GFX_LoadBitmap(&bluenumbers[2], GFX_LoadBitmap(&bluenumbers[3], GFX_LoadBitmap(&bluenumbers[4], GFX_LoadBitmap(&bluenumbers[5], GFX_LoadBitmap(&bluenumbers[6], GFX_LoadBitmap(&bluenumbers[7], GFX_LoadBitmap(&bluenumbers[8], GFX_LoadBitmap(&bluenumbers[9], // Set the scroll positions scrollX = 40*32;

if(localClient->team == RED_TEAM) scrollY = 90*32; else scrollY = 0; // Set the flag position flagX = 49*32; flagY = 49*32;

Tutorial 5 / Creating Your Online Game

513

playerWithFlag = NULL; // Reset score counters redScore = 0; blueScore = 0; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::Shutdown(void) { Disconnect(); GFX_Begin(); GFX_Shutdown(); GFX_End(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::DrawMap(void) { int heading = 0; // Work out how many tiles have been scrolled tileScrollX = scrollX/32; tileScrollY = scrollY/32; for(int i = (tileScrollX)2; i < (tileScrollX)+21; i++) { for(int j = (tileScrollY)+15; j > (tileScrollY)2; j) { GFX_Blit(&grass, (32*i)(scrollX),(32*j)(scrollY), 32, 32, 0); // Draw a tree if required if(mapdata[i][j] == true) { GFX_Blit(&tree, (32*i+16)(scrollX),(32*j+16) (scrollY), 32, 32, 0); } // Draw the static targets if(i==49 && j==3) { // draw the blue target GFX_Blit(&bluetarget, (32*i)(scrollX),(32*j) (scrollY), 32, 32, targetRotation); }

514

Tutorial 5 / Creating Your Online Game

if(i==49 && j==97) { // draw the red target GFX_Blit(&redtarget, (32*i)(scrollX),(32*j) (scrollY), 32, 32, targetRotation); } } } // Render the flag GFX_Blit(&flag, ((int) flagX)(scrollX), ((int) flagY)(scrollY), 32, 32, 0); // Render players clientData *list = clientList; for(; list != NULL; list = list->next) { if(list->team == RED_TEAM) { GFX_Blit(&redman, ((int) list->command.origin.x)(scrollX), ((int) list->command.origin.y)(scrollY), 32, 32, (float) list->command.heading); } if(list->team == BLUE_TEAM) { GFX_Blit(&blueman, ((int) list->command.origin.x) (scrollX), ((int) list->command.origin.y) (scrollY), 32, 32, (float) list->command.heading); } // Render bullets if(list->command.bullet.shot) { if(list->team == RED_TEAM) { GFX_RectFill(((int) list->command.bullet.origin.x2) (scrollX), ((int) list-> command.bullet.origin.y2)(scrollY), ((int) list->command.bullet.origin.x+2) (scrollX), ((int) list-> command.bullet.origin.y+2)(scrollY), 200, 0, 0); } if(list->team == BLUE_TEAM) { GFX_RectFill(((int) list->command.bullet.origin.x2) (scrollX), ((int) list-> command.bullet.origin.y2)(scrollY), ((int) list->command.bullet.origin.x+2) (scrollX), ((int) list->

Tutorial 5 / Creating Your Online Game

515

command.bullet.origin.y+2)(scrollY), 0, 0, 200); } } } // Finally, render the team scores GFX_Blit(&rednumbers[redScore], 5, 410, 64, 64, 0); GFX_Blit(&bluenumbers[blueScore], 570, 410, 64, 64, 0); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::Frame(void) { if(!localClient) return; // Scroll the map to follow the local player if((localClient->command.origin.x scrollX) > 340) { if(scrollX <= 3200(19*32)2) scrollX += 2; } if((localClient->command.origin.x scrollX) < 300) { if(scrollX >= 2) scrollX = 2; } if((localClient->command.origin.y scrollY) > 260) { if(scrollY <= 3200(15*32)2) scrollY += 2; } if((localClient->command.origin.y scrollY) < 220) { if(scrollY >= 2) scrollY = 2; } // Move the flag with the player if(playerWithFlag) { flagX = playerWithFlag->command.origin.x; flagY = playerWithFlag->command.origin.y; } // Rotate the target images if(targetRotation < 360)

516

Tutorial 5 / Creating Your Online Game

targetRotation += 1; else targetRotation = targetRotation; // Draw map GFX_Begin(); { DrawMap(); } GFX_End(); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::CheckVictory(void) { if(localClient == NULL) return; // Check team scores if(redScore > 1) { if(localClient->team == RED_TEAM) { MessageBox(NULL, "Your team (RED) won!", "Victory", MB_OK); } else { MessageBox(NULL, "The other team (RED) won", "Failure", MB_OK); } Shutdown(); } if(blueScore > 1) { if(localClient->team == BLUE_TEAM) { MessageBox(NULL, "Your team (BLUE) won!", "Victory", MB_OK); } else { MessageBox(NULL, "The other team (BLUE) won", "Failure", MB_OK); } Shutdown(); } }

Tutorial 5 / Creating Your Online Game

517

//--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::KillPlayer(int index) { LogString("Player %d died", index); clientData *client = GetClientPointer(index); if(client == NULL) return; client->command.origin.x = client->startPos.x; client->command.origin.y = client->startPos.y; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------clientData *CArmyWar::GetClientPointer(int index) { for(clientData *clList = clientList; clList != NULL; clList = clList->next) { if(clList->index == index) return clList; } return NULL; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::CheckKeys(void) { inputClient.command.key = 0; if(keys[VK_ESCAPE]) { Shutdown(); keys[VK_ESCAPE] = false; } if(keys[VK_DOWN]) { inputClient.command.key |= KEY_DOWN; } if(keys[VK_UP])

518

Tutorial 5 / Creating Your Online Game

{ inputClient.command.key |= KEY_UP; } if(keys[VK_LEFT]) { inputClient.command.key |= KEY_LEFT; } if(keys[VK_RIGHT]) { inputClient.command.key |= KEY_RIGHT; } if(keys[VK_SPACE]) { inputClient.command.key |= KEY_WEAPON; } inputClient.command.msec = (int) (frametime * 1000); } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::CheckPredictionError(int a) { if(a < 0 && a > COMMAND_HISTORY_SIZE) return; float errorX = localClient->serverFrame.origin.x localClient->frame[a].predictedOrigin.x; float errorY = localClient->serverFrame.origin.y localClient->frame[a].predictedOrigin.y; // Fix the prediction error if((errorX != 0.0f) || (errorY != 0.0f)) { localClient->frame[a].predictedOrigin.x = localClient-> serverFrame.origin.x; localClient->frame[a].predictedOrigin.y = localClient-> serverFrame.origin.y; localClient->frame[a].vel.x = localClient->serverFrame.vel.x; localClient->frame[a].vel.y = localClient->serverFrame.vel.y; LogString("Prediction error for frame %d: %f, %f\n", a, errorX, errorY); } }

Tutorial 5 / Creating Your Online Game

519

//--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::CheckBulletPredictionError(int a) { if(a < 0 && a > COMMAND_HISTORY_SIZE) return; float errorX = localClient->serverFrame.bullet.origin.x localClient-> frame[a].bullet.predictedOrigin.x; float errorY = localClient->serverFrame.bullet.origin.y localClient-> frame[a].bullet.predictedOrigin.y; // Fix the prediction error if((errorX != 0.0f) || (errorY != 0.0f)) { localClient->frame[a].bullet.predictedOrigin.x = localClient-> serverFrame.bullet.origin.x; localClient->frame[a].bullet.predictedOrigin.y = localClient-> serverFrame.bullet.origin.y; localClient->frame[a].bullet.vel.x = localClient-> serverFrame.bullet.vel.x; localClient->frame[a].bullet.vel.y = localClient-> serverFrame.bullet.vel.y; LogString("Bullet prediction error for frame %d: %f, %f\n", a, errorX, errorY); } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::CalculateVelocity(command_t *command, float frametime) { int checkX; int checkY; float multiplier = 100.0f; command->vel.x = 0.0f; command->vel.y = 0.0f; if(command->key & KEY_UP) { checkX = (int) (command->origin.x / 32.0f); checkY = (int) ((command->origin.y multiplier * frametime) / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.y += multiplier * frametime;

520

Tutorial 5 / Creating Your Online Game

} if(command->key & KEY_DOWN) { checkX = (int) (command->origin.x / 32.0f); checkY = (int) ((command->origin.y + multiplier * frametime) / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.y += multiplier * frametime; } if(command->key & KEY_LEFT) { checkX = (int) ((command->origin.x multiplier * frametime) / 32.0f); checkY = (int) (command->origin.y / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.x += multiplier * frametime; } if(command->key & KEY_RIGHT) { checkX = (int) ((command->origin.x + multiplier * frametime) / 32.0f); checkY = (int) (command->origin.y / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.x += multiplier * frametime; } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::CalculateHeading(command_t *command) { // Right if((command->vel.x > 0.0f) && (command->vel.y == 0.0f)) { command->heading = EAST; } // Left if((command->vel.x < 0.0f) && (command->vel.y == 0.0f)) { command->heading = WEST; }

Tutorial 5 / Creating Your Online Game

521

// Down if((command->vel.y > 0.0f) && (command->vel.x == 0.0f)) { command->heading = SOUTH; } // Up if((command->vel.y < 0.0f) && (command->vel.x == 0.0f)) { command->heading = NORTH; } // Down-Right if((command->vel.x > 0.0f) && (command->vel.y > 0.0f)) { command->heading = SOUTHEAST; } // Up-Right if((command->vel.x > 0.0f) && (command->vel.y < 0.0f)) { command->heading = NORTHEAST; } // Down-Left if((command->vel.x < 0.0f) && (command->vel.y > 0.0f)) { command->heading = SOUTHWEST; } // Up-Left if((command->vel.x < 0.0f) && (command->vel.y < 0.0f)) { command->heading = NORTHWEST; } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::CalculateBulletVelocity(command_t *command) { command->bullet.shot = true; if(command->heading == NORTH) { command->bullet.vel.x = 0.0f;

522

Tutorial 5 / Creating Your Online Game

command->bullet.vel.y = } if(command->heading == SOUTH) { command->bullet.vel.x = command->bullet.vel.y = } if(command->heading == EAST) { command->bullet.vel.x = command->bullet.vel.y = } if(command->heading == WEST) { command->bullet.vel.x = command->bullet.vel.y = }

200.0f;

0.0f; 200.0f;

200.0f; 0.0f;

if(command->heading == NORTHEAST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } if(command->heading == NORTHWEST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } if(command->heading == SOUTHEAST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } if(command->heading == SOUTHWEST) { command->bullet.vel.x = 200.0f; command->bullet.vel.y = 200.0f; } } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::PredictMovement(int prevFrame, int curFrame) { if(!localClient) return; float frametime = inputClient.frame[curFrame].msec / 1000.0f; localClient->frame[curFrame].key = inputClient.frame[curFrame].key; //

Tutorial 5 / Creating Your Online Game

523

// Player -> // // Process commands CalculateVelocity(&localClient->frame[curFrame], frametime); CalculateHeading(&localClient->frame[curFrame]); // Calculate new predicted origin localClient->frame[curFrame].predictedOrigin.x = localClient->frame[prevFrame].predictedOrigin.x + localClient->frame[curFrame].vel.x; localClient->frame[curFrame].predictedOrigin.y = localClient->frame[prevFrame].predictedOrigin.y + localClient->frame[curFrame].vel.y; // Copy values to "current" values localClient->command.predictedOrigin.x = localClient-> frame[curFrame].predictedOrigin.x; localClient->command.predictedOrigin.y = localClient-> frame[curFrame].predictedOrigin.y; localClient->command.vel.x = localClient->frame[curFrame].vel.x; localClient->command.vel.y = localClient->frame[curFrame].vel.y; localClient->command.heading = localClient->frame[curFrame].heading;

// // Bullet -> // // First set the previous values localClient->frame[curFrame].bullet.shot = localClient-> frame[prevFrame].bullet.shot; localClient->frame[curFrame].bullet.vel.x = localClient-> frame[prevFrame].bullet.vel.x; localClient->frame[curFrame].bullet.vel.y = localClient-> frame[prevFrame].bullet.vel.y; localClient->frame[curFrame].bullet.lifetime = localClient-> frame[prevFrame].bullet.lifetime; // The bullet is carried by the player if(localClient->frame[curFrame].bullet.shot == false) { localClient->frame[curFrame].bullet.predictedOrigin.x = localClient->frame[curFrame].predictedOrigin.x; localClient->frame[curFrame].bullet.predictedOrigin.y = localClient->frame[curFrame].predictedOrigin.y; } else { localClient->frame[curFrame].bullet.lifetime += (int) (frametime * 1000.0f); if(localClient->frame[curFrame].bullet.lifetime > 2000)

524

Tutorial 5 / Creating Your Online Game

{ localClient->frame[curFrame].bullet.shot = false; localClient->frame[curFrame].bullet.lifetime = 0; localClient->frame[curFrame].bullet.predictedOrigin.x = localClient->frame[curFrame].predictedOrigin.x; localClient->frame[curFrame].bullet.predictedOrigin.y = localClient->frame[curFrame].predictedOrigin.y; } } // Calculate the heading for the bullet only when firing if(localClient->frame[curFrame].key & KEY_WEAPON && localClient->frame[curFrame].bullet.shot == false) { CalculateBulletVelocity(&localClient->frame[curFrame]); } // If the bullet is in the air (shot), update its origin if(localClient->frame[curFrame].bullet.shot) { localClient->frame[curFrame].bullet.predictedOrigin.x = localClient->frame[prevFrame].bullet.predictedOrigin.x + localClient->frame[curFrame].bullet.vel.x * frametime; localClient->frame[curFrame].bullet.predictedOrigin.y = localClient->frame[prevFrame].bullet.predictedOrigin.y + localClient->frame[curFrame].bullet.vel.y * frametime; } // Copy values to "current" values localClient->command.bullet.predictedOrigin.x = localClient-> frame[curFrame].bullet.predictedOrigin.x; localClient->command.bullet.predictedOrigin.y = localClient-> frame[curFrame].bullet.predictedOrigin.y; localClient->command.bullet.vel.x = localClient-> frame[curFrame].bullet.vel.x; localClient->command.bullet.vel.y = localClient-> frame[curFrame].bullet.vel.y; } //--------------------------------------------------------------------------// Name: empty() // Desc: //--------------------------------------------------------------------------void CArmyWar::MoveObjects(void) { if(!localClient) return; clientData *client = clientList; for(; client != NULL; client = client->next) {

Tutorial 5 / Creating Your Online Game

525

// Remote players if(client != localClient) { CalculateVelocity(&client->command, frametime); CalculateHeading(&client->command); client->command.origin.x += client->command.vel.x; client->command.origin.y += client->command.vel.y; client->command.bullet.origin.x += serverFrame.bullet.vel.x * client->command.bullet.origin.y += serverFrame.bullet.vel.y * } // Local player else { client->command.origin.x = client-> command.predictedOrigin.x; client->command.origin.y = client-> command.predictedOrigin.y; client->command.bullet.origin.x = client-> command.bullet.predictedOrigin.x; client->command.bullet.origin.y = client-> command.bullet.predictedOrigin.y; } } } client-> frametime; client-> frametime;

CArmyWar Constructor and Destructor Functions

These two functions initialize and uninitialize the CArmyWar object that contains game-specific data. The constructor allocates memory and sets all the variables to initial states. The destructor frees the allocated memory, which is very important (otherwise we would have a memory leak).
CArmyWar::CArmyWar() { networkClient clientList localClient clients

= = = =

new dreamClient; NULL; NULL; 0;

memset(&inputClient, 0, sizeof(clientData)); memset(&mapdata, 0, sizeof(mapdata)); frametime inProgress = 0.0f; = false;

526

Tutorial 5 / Creating Your Online Game

init scrollX scrollY tileScrollX tileScrollY gameIndex targetRotation redScore blueScore playerWithFlag next }

= false; = 0; = 0; = 0; = 0; = 0; = 0.0f; = 0; = 0; = NULL; = NULL;

CArmyWar::~CArmyWar() { delete networkClient; }

InitializeEngine Function
We are still initializing things with this function. This ones purpose is to initialize 2DLIB our graphics engine and load the graphics into memory. It also sets the flag to its correct initial position and moves the screen (scrolls) to the correct position, based on which team the local player is on. Notice how the team score bitmaps are loaded into an array, so we can use the team scores as an index to that array to get the correct bitmap number.
void CArmyWar::InitializeEngine(void) { // Init the graphics engine GFX_Init("Army War Engine v2.0", 640, 480, 16, 0, ApplicationProc); // Load required graphics GFX_LoadBitmap(&grass, GFX_LoadBitmap(&redman, GFX_LoadBitmap(&blueman, GFX_LoadBitmap(&tree, GFX_LoadBitmap(&redtarget, GFX_LoadBitmap(&bluetarget, GFX_LoadBitmap(&flag,

GFX_LoadBitmap(&rednumbers[0], GFX_LoadBitmap(&rednumbers[1], GFX_LoadBitmap(&rednumbers[2], GFX_LoadBitmap(&rednumbers[3],

Tutorial 5 / Creating Your Online Game

527

GFX_LoadBitmap(&rednumbers[4], GFX_LoadBitmap(&rednumbers[5], GFX_LoadBitmap(&rednumbers[6], GFX_LoadBitmap(&rednumbers[7], GFX_LoadBitmap(&rednumbers[8], GFX_LoadBitmap(&rednumbers[9],

"gfx\\red4.bmp"); "gfx\\red5.bmp"); "gfx\\red6.bmp"); "gfx\\red7.bmp"); "gfx\\red8.bmp"); "gfx\\red9.bmp"); "gfx\\blue0.bmp"); "gfx\\blue1.bmp"); "gfx\\blue2.bmp"); "gfx\\blue3.bmp"); "gfx\\blue4.bmp"); "gfx\\blue5.bmp"); "gfx\\blue6.bmp"); "gfx\\blue7.bmp"); "gfx\\blue8.bmp"); "gfx\\blue9.bmp");

if(localClient->team == RED_TEAM) scrollY = 90*32; else scrollY = 0; // Set the flag position flagX = 49*32; flagY = 49*32; playerWithFlag = NULL; // Reset score counters redScore = 0; blueScore = 0; }

Shutdown Function
The Shutdown function disconnects from the server (if connected) and shuts down the graphics engine.
void CArmyWar::Shutdown(void) { Disconnect(); GFX_Begin(); GFX_Shutdown(); GFX_End(); }

528

Tutorial 5 / Creating Your Online Game

DrawMap Function
Now we finally get to do something besides initializing things. This function draws the game map, trees, grass, players, flag, and bullets. First we need to figure out their positions on the screen by looking at the scroll coordinates. One tile is 32 pixels in height and width, so it is easy to calculate which tiles to draw. Just divide the actual pixel scroll values by 32, and you get the first tiles to draw (top-left corner). Then we loop through the visible tiles and draw what lies in that tile. To get the correct position on screen, you need to multiply the tile scroll value by 32 and then subtract the pixel scroll value from that.
void CArmyWar::DrawMap(void) { int heading = 0; // Work out how many tiles have been scrolled tileScrollX = scrollX/32; tileScrollY = scrollY/32; for(int i = (tileScrollX)2; i < (tileScrollX)+21; i++) { for(int j = (tileScrollY)+15; j > (tileScrollY)2; j) { GFX_Blit(&grass, (32*i)(scrollX),(32*j)(scrollY), 32, 32, 0); // Draw a tree if required if(mapdata[i][j] == true) { GFX_Blit(&tree, (32*i+16)(scrollX),(32*j+16) (scrollY), 32, 32, 0); } // Draw the static targets if(i==49 && j==3) { // draw the blue target GFX_Blit(&bluetarget, (32*i)(scrollX),(32*j) (scrollY), 32, 32, targetRotation); } if(i==49 && j==97) { // draw the red target GFX_Blit(&redtarget, (32*i)(scrollX),(32*j) (scrollY), 32, 32, targetRotation); } } }

Tutorial 5 / Creating Your Online Game

529

// Render the flag GFX_Blit(&flag, ((int) flagX)(scrollX), ((int) flagY)(scrollY), 32, 32, 0); // Render players clientData *list = clientList; for(; list != NULL; list = list->next) { if(list->team == RED_TEAM) { GFX_Blit(&redman, ((int) list->command.origin.x)(scrollX), ((int) list->command.origin.y)(scrollY), 32, 32, (float) list->command.heading); } if(list->team == BLUE_TEAM) { GFX_Blit(&blueman, ((int) list->command.origin.x) (scrollX), ((int) list->command.origin.y) (scrollY), 32, 32, (float) list->command.heading); } // Render bullets if(list->command.bullet.shot) { if(list->team == RED_TEAM) { GFX_RectFill(((int) list->command.bullet.origin.x2) (scrollX), ((int) list-> command.bullet.origin.y2)(scrollY), ((int) list->command.bullet.origin.x+2) (scrollX), ((int) list-> command.bullet.origin.y+2)(scrollY), 200, 0, 0); } if(list->team == BLUE_TEAM) { GFX_RectFill(((int) list->command.bullet.origin.x2) (scrollX), ((int) list-> command.bullet.origin.y2)(scrollY), ((int) list-> command.bullet.origin.x+2)(scrollX), ((int) list->command.bullet.origin.y+2) (scrollY), 0, 0, 200); } } } // Finally, render the team scores GFX_Blit(&rednumbers[redScore], 5, 410, 64, 64, 0); GFX_Blit(&bluenumbers[blueScore], 570, 410, 64, 64, 0); }

530

Tutorial 5 / Creating Your Online Game

Here we draw a tree by looking at the map data. If the value on that tile is 1 (or true in other words), we draw a tree. Otherwise, we draw only plain grass.
// Draw a tree if required if(mapdata[i][j] == true) { GFX_Blit(&tree, (32*i+16)(scrollX),(32*j+16)(scrollY), 32, 32, 0); }

Frame Function
This functions purpose is to handle the game logic. It scrolls the screen by trying to follow the players position. Also, if someone is carrying the flag, the flag is moved with the player. The teams base indicators are rotated to add some graphical effects. Finally, the map is drawn with the current positions.
void CArmyWar::Frame(void) { if(!localClient) return; // Scroll the map to follow the local player if((localClient->command.origin.x scrollX) > 340) { if(scrollX <= 3200(19*32)2) scrollX += 2; } if((localClient->command.origin.x scrollX) < 300) { if(scrollX >= 2) scrollX = 2; } if((localClient->command.origin.y scrollY) > 260) { if(scrollY <= 3200(15*32)2) scrollY += 2; } if((localClient->command.origin.y scrollY) < 220) { if(scrollY >= 2) scrollY = 2; } // Move the flag with the player if(playerWithFlag) { flagX = playerWithFlag->command.origin.x; flagY = playerWithFlag->command.origin.y;

Tutorial 5 / Creating Your Online Game

531

} // Rotate the target images if(targetRotation < 360) targetRotation += 1; else targetRotation -= targetRotation; // Draw map GFX_Begin(); { DrawMap(); } GFX_End(); }

CheckVictory Function
This function checks the victory conditions. If they are met, it notifies the player whose team won the game and then shuts down the game engine after the player presses the OK button.
void CArmyWar::CheckVictory(void) { if(localClient == NULL) return; // Check team scores if(redScore > 1) { if(localClient->team == RED_TEAM) { MessageBox(NULL, "Your team (RED) won!", "Victory", MB_OK); } else { MessageBox(NULL, "The other team (RED) won", "Failure", MB_OK); } Shutdown(); } if(blueScore > 1) { if(localClient->team == BLUE_TEAM) { MessageBox(NULL, "Your team (BLUE) won!", "Victory", MB_OK); } else { MessageBox(NULL, "The other team (BLUE) won", "Failure", MB_OK);

532

Tutorial 5 / Creating Your Online Game

} Shutdown(); } }

KillPlayer Function
The KillPlayer function makes sure that when a player is shot, the player is moved back to the start position. This function fixes the prediction error that would occur if the server only sent the new position (without this function, the client would keep the shot player at the position where he or she was shot).
void CArmyWar::KillPlayer(int index) { LogString("Player %d died", index); clientData *client = GetClientPointer(index); if(client == NULL) return; client->command.origin.x = client->startPos.x; client->command.origin.y = client->startPos.y; }

GetClientPointer Function
This function returns a pointer to a player in the client list, and the player is chosen with the player index number given as a parameter. If no such index number is found, the function returns NULL, so we must check the result before using it.
clientData *CArmyWar::GetClientPointer(int index) { for(clientData *clList = clientList; clList != NULL; clList = clList->next) { if(clList->index == index) return clList; } return NULL; }

CheckKeys Function
Here we check the input keys and set the commands to correspond to keypresses. Notice that we use inputClient here. At the beginning of the function we reset the commands because we check the commands every frame.

Tutorial 5 / Creating Your Online Game

533

void CArmyWar::CheckKeys(void) { inputClient.command.key = 0; if(keys[VK_ESCAPE]) { Shutdown(); keys[VK_ESCAPE] = false; } if(keys[VK_DOWN]) { inputClient.command.key |= KEY_DOWN; } if(keys[VK_UP]) { inputClient.command.key |= KEY_UP; } if(keys[VK_LEFT]) { inputClient.command.key |= KEY_LEFT; } if(keys[VK_RIGHT]) { inputClient.command.key |= KEY_RIGHT; } if(keys[VK_SPACE]) { inputClient.command.key |= KEY_WEAPON; } inputClient.command.msec = (int) (frametime * 1000); }

CheckPredictionError Function
This function might seem a little bit odd at first, but it really is not odd at all. It compares the frame received from the server and the frame produced by the local client. If there is any difference between them, the servers frame is used. In other words, we always use the servers frame, but because the server sends us frame data in 100 ms intervals we need to produce the frames in between ourselves. As there is no way to determine when exactly the server gives us the next frame, we check every frame the server gives us. When the server gives us a new frame, it is actually old already (because of network lag and other processing times). So we have to compare an old frame that is stored in our frame history (or command history). If there is an error, the client

534

Tutorial 5 / Creating Your Online Game

fixes the error in the past frame and reproduces all the frames all the way to the current one. This is seen in the RunNetwork() function (network.cpp).
void CArmyWar::CheckPredictionError(int a) { if(a < 0 && a > COMMAND_HISTORY_SIZE) return; float errorX = localClient->serverFrame.origin.x localClient-> frame[a].predictedOrigin.x; float errorY = localClient->serverFrame.origin.y localClient-> frame[a].predictedOrigin.y; // Fix the prediction error if((errorX != 0.0f) || (errorY != 0.0f)) { localClient->frame[a].predictedOrigin.x = localClient-> serverFrame.origin.x; localClient->frame[a].predictedOrigin.y = localClient-> serverFrame.origin.y; localClient->frame[a].vel.x = localClient->serverFrame.vel.x; localClient->frame[a].vel.y = localClient->serverFrame.vel.y; LogString("Prediction error for frame %d: %f, %f\n", a, errorX, errorY); } }

CheckBulletPredictionError Function
This function is similar to the CheckPredictionError() function, except that this one is for bullets.
void CArmyWar::CheckBulletPredictionError(int a) { if(a < 0 && a > COMMAND_HISTORY_SIZE) return; float errorX = localClient->serverFrame.bullet.origin.x localClient-> frame[a].bullet.predictedOrigin.x; float errorY = localClient->serverFrame.bullet.origin.y localClient-> frame[a].bullet.predictedOrigin.y; // Fix the prediction error if((errorX != 0.0f) || (errorY != 0.0f)) { localClient->frame[a].bullet.predictedOrigin.x = localClient-> serverFrame.bullet.origin.x; localClient->frame[a].bullet.predictedOrigin.y = localClient->

Tutorial 5 / Creating Your Online Game

535

serverFrame.bullet.origin.y; localClient->frame[a].bullet.vel.x = localClient-> serverFrame.bullet.vel.x; localClient->frame[a].bullet.vel.y = localClient-> serverFrame.bullet.vel.y; LogString("Bullet prediction error for frame %d: errorX, errorY); } } %f, %f\n", a,

CalculateVelocity Function
This function calculates the local players movement velocity (or direction really, as there is no acceleration). The function does some simple collision detection by looking at the map data at the position to which the player is moving. If there is a tree, the player moves nowhere. Notice that we use frametime to make the player move the correct amount in any given time. This is very, very important, especially in a network game. It makes the players move the same amount on all the computers in any given time (for example, if you keep the up key pressed for one second).
void CArmyWar::CalculateVelocity(command_t *command, float frametime) { int checkX; int checkY; float multiplier = 100.0f; command->vel.x = 0.0f; command->vel.y = 0.0f; if(command->key & KEY_UP) { checkX = (int) (command->origin.x / 32.0f); checkY = (int) ((command->origin.y multiplier * frametime) / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.y += multiplier * frametime; } if(command->key & KEY_DOWN) { checkX = (int) (command->origin.x / 32.0f); checkY = (int) ((command->origin.y + multiplier * frametime) / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.y += multiplier * frametime;

536

Tutorial 5 / Creating Your Online Game

} if(command->key & KEY_LEFT) { checkX = (int) ((command->origin.x multiplier * frametime) / 32.0f); checkY = (int) (command->origin.y / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.x += multiplier * frametime; } if(command->key & KEY_RIGHT) { checkX = (int) ((command->origin.x + multiplier * frametime) / 32.0f); checkY = (int) (command->origin.y / 32.0f); if(mapdata[checkX][checkY] == false) command->vel.x += multiplier * frametime; } }

CalculateHeading Function
This function is used to convert the players velocity to simple heading values (in degrees).
void CArmyWar::CalculateHeading(command_t *command) { // Right if((command->vel.x > 0.0f) && (command->vel.y == 0.0f)) { command->heading = EAST; } // Left if((command->vel.x < 0.0f) && (command->vel.y == 0.0f)) { command->heading = WEST; } // Down if((command->vel.y > 0.0f) && (command->vel.x == 0.0f)) { command->heading = SOUTH; } // Up if((command->vel.y < 0.0f) && (command->vel.x == 0.0f))

Tutorial 5 / Creating Your Online Game

537

{ command->heading = NORTH; } // Down-Right if((command->vel.x > 0.0f) && (command->vel.y > 0.0f)) { command->heading = SOUTHEAST; } // Up-Right if((command->vel.x > 0.0f) && (command->vel.y < 0.0f)) { command->heading = NORTHEAST; } // Down-Left if((command->vel.x < 0.0f) && (command->vel.y > 0.0f)) { command->heading = SOUTHWEST; } // Up-Left if((command->vel.x < 0.0f) && (command->vel.y < 0.0f)) { command->heading = NORTHWEST; } }

PredictMovement Function
This function produces the local frames, or in other words, predicts player movement. It produces a frame and hopes it is the same as what the server comes up with (and it will be the same if all the data reaches the server and the client). What the function does is take the previous frame and then add current velocities to the previous positions. And that is mostly it. It also updates some other variables, like the bullet lifetime, but it is updated the same way as the positions: You take the previous lifetime value and add current frametime.
void CArmyWar::PredictMovement(int prevFrame, int curFrame) { if(!localClient) return; float frametime = inputClient.frame[curFrame].msec / 1000.0f;

538

Tutorial 5 / Creating Your Online Game

localClient->frame[curFrame].key = inputClient.frame[curFrame].key; // // Player -> // // Process commands CalculateVelocity(&localClient->frame[curFrame], frametime); CalculateHeading(&localClient->frame[curFrame]); // Calculate new predicted origin localClient->frame[curFrame].predictedOrigin.x = localClient-> frame[prevFrame].predictedOrigin.x + localClient-> frame[curFrame].vel.x; localClient->frame[curFrame].predictedOrigin.y = localClient-> frame[prevFrame].predictedOrigin.y + localClient-> frame[curFrame].vel.y; // Copy values to "current" values localClient->command.predictedOrigin.x = localClient-> frame[curFrame].predictedOrigin.x; localClient->command.predictedOrigin.y = localClient-> frame[curFrame].predictedOrigin.y; localClient->command.vel.x = localClient->frame[curFrame].vel.x; localClient->command.vel.y = localClient->frame[curFrame].vel.y; localClient->command.heading = localClient->frame[curFrame].heading;

Tutorial 5 / Creating Your Online Game

539

localClient->frame[curFrame].bullet.lifetime += (int) (frametime * 1000.0f); if(localClient->frame[curFrame].bullet.lifetime > 2000) { localClient->frame[curFrame].bullet.shot = false; localClient->frame[curFrame].bullet.lifetime = 0; localClient->frame[curFrame].bullet.predictedOrigin.x = localClient->frame[curFrame].predictedOrigin.x; localClient->frame[curFrame].bullet.predictedOrigin.y = localClient->frame[curFrame].predictedOrigin.y; } } // Calculate the heading for the bullet only when firing if(localClient->frame[curFrame].key & KEY_WEAPON && localClient-> frame[curFrame].bullet.shot == false) { CalculateBulletVelocity(&localClient->frame[curFrame]); } // If the bullet is in the air (shot), update its origin if(localClient->frame[curFrame].bullet.shot) { localClient->frame[curFrame].bullet.predictedOrigin.x = localClient->frame[prevFrame].bullet.predictedOrigin.x + localClient->frame[curFrame].bullet.vel.x * frametime; localClient->frame[curFrame].bullet.predictedOrigin.y = localClient->frame[prevFrame].bullet.predictedOrigin.y + localClient->frame[curFrame].bullet.vel.y * frametime; } // Copy values to "current" values localClient->command.bullet.predictedOrigin.x = localClient-> frame[curFrame].bullet.predictedOrigin.x; localClient->command.bullet.predictedOrigin.y = localClient-> frame[curFrame].bullet.predictedOrigin.y; localClient->command.bullet.vel.x = localClient-> frame[curFrame].bullet.vel.x; localClient->command.bullet.vel.y = localClient-> frame[curFrame].bullet.vel.y; }

MoveObjects Function
This function moves all the remote players and the local player to their new positions. Then the map can be drawn to show the player positions. The remote players are moved by adding their last known velocity to their current origin. The server tells us the real origin in time, and there may be some warping if the clients origin differs from the servers origin.

540

Tutorial 5 / Creating Your Online Game

This is one way to do it, but to get smoother movement, you could add dead reckoning or client prediction on the remote players too. But we will leave that to you.
void CArmyWar::MoveObjects(void) { if(!localClient) return; clientData *client = clientList; for(; client != NULL; client = client->next) { // Remote players if(client != localClient) { CalculateVelocity(&client->command, frametime); CalculateHeading(&client->command); client->command.origin.x += client->command.vel.x; client->command.origin.y += client->command.vel.y; client->command.bullet.origin.x += serverFrame.bullet.vel.x * client->command.bullet.origin.y += serverFrame.bullet.vel.y * } // Local player else { client->command.origin.x = client-> command.predictedOrigin.x; client->command.origin.y = client-> command.predictedOrigin.y; client->command.bullet.origin.x = client-> command.bullet.predictedOrigin.x; client->command.bullet.origin.y = client-> command.bullet.predictedOrigin.y; } } } client-> frametime; client-> frametime;

Tutorial 5 / Creating Your Online Game

541

lobby.cpp File
Now that we have the game data structures, we can implement the functions we introduced in Tutorial 4.

RefreshGameList Function
This function refreshes the game list in the lobby dialog. First the contents of the list are completely removed, and then everything is added again to match the current game list. If a game is in progress, text is added after the games name to inform us of that.
void CLobby::RefreshGameList(void) { char temp[128]; SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_GAMELIST), LB_RESETCONTENT, 0, 0); CArmyWar *list = gameList; for( ; list != NULL; list = list->next) { strcpy(temp, list->GetName()); if(list->GetInProgress()) { strcat(temp, " (in progress)"); } SendMessage(GetDlgItem(hWnd_LobbyDialog, IDC_GAMELIST), LB_ADDSTRING, 0, (LPARAM) temp); } }

RefreshJoinedPlayersList Function
Here we refresh the list of joined players. The list is first reset so that it can be filled again with the current information. Note that this function updates the list for both the game host and the players who joined the game.
void CLobby::RefreshJoinedPlayersList(void) { if(localGame == NULL) return; SendMessage(GetDlgItem(hWnd_CreateViewPlayersDialog, IDC_PLAYERSINGAME), LB_RESETCONTENT, 0, 0);

542

Tutorial 5 / Creating Your Online Game

SendMessage(GetDlgItem(hWnd_JoinGameDialog, IDC_JOINPLAYERSINGAME), LB_RESETCONTENT, 0, 0); clientData *list = localGame->GetClientList(); for( ; list != NULL; list = list->next) { SendMessage(GetDlgItem(hWnd_CreateViewPlayersDialog, IDC_PLAYERSINGAME), LB_ADDSTRING, 0, (LPARAM) list->nickname); SendMessage(GetDlgItem(hWnd_JoinGameDialog, IDC_JOINPLAYERSINGAME), LB_ADDSTRING, 0, (LPARAM) list->nickname); } }

Other Unimplemented Functions

The other unimplemented functions (AddGame, RemoveGame, and RemoveGames), work the same way as they do on the server side, so there is no need to explain them here. Only the AddGame function is slightly different, as it has some new parameters. These parameters only contain game information, such as the index number and inprogress flag. These parameters do not affect the functionality of the function; they are only stored in the games structure.

Summary
All good things must come to an end, and so ends this tutorial. Here we covered the final version of the Army War v2.0 server and client. We learned how to move the players and other objects in the game world, not just by moving them on our local screen but also by moving them on the server. We learned that the server is always right, not the client. We also now know that by using frame time to calculate the amount of movement in one frame, we can keep all the clients and the server in sync. Now that the 2D Army War game has been fully covered in this tutorial and once you understand the concept of multiplayer game programming, you can put your skills to the test by programming the 3D Army War engine (located on the companion CD) to include multiplayer code. Try to understand which parts are essential to make the code work on any game, and you will be able to make a 3D multiplayer game. All you need is patience and time. Good luck!

Index
* indicates the reference includes code listing. 2D library, adding to workspace, 183 creating windowed application with, 184-189* graphical functions, 190-193 2D library example, loading graphics, 200-202* moving primitives, 195-199* rotating graphics, 200-202* 2D positions, displaying, 190

B
BeginReading function, 257* big-endian order, 136 bind function, 125-126 blocking I/O, 169-170 blocking socket, 144 Broadcast function, 176-177* broadcasting, 175 BuildDeltaMoveCommand function, 454-458*, 508-509* BuildMoveCommand function, 453-454* bullets, calculating velocity of, 470-472* comparing, 534-535* moving, 472-475* byte ordering, 136

A
accept function, 128 account, creating, 316-319* AddClient function, 293-295*, 333-334*, 366-367* AddGame function, 478-479* address classes, 107 address conversion, 133 functions, 133-134 address structures, generic, 123 IPv4, 121-122 IPv6, 122 AddSequences function, 255* adduser.tpl.html, 97* alias naming restrictions, 20 ALTER TABLE command, 26-28 Apache web server, installing PHP4 on, 60-63 setting up, 57-60 application, creating, 184-189* creating with dreamSock, 305-369* application layer, 105 ApplicationProc function, 489* arrays, using in PHP, 70

C
CalculateBulletVelocity function, 470-472* CalculateHeading function, 469-470*, 536-537* CalculateVelocity function, 468-469*, 535-536* CArmyWar class, constructor, 525-526* destructor, 525-526* initializing, 525-526* uninitializing, 525-526* CArmyWarServer class methods, 437-449*, 458-467* CheckBulletPredictionError function, 534-535* CheckFlagCollisions function, 475-476* CheckForTimeout function, 299-301* CheckKeys function, 532-533* CheckPredictionError function, 533-534* CheckVictory function, 531-532*

543

Index

Clear function, 253* client application, creating, 306-338* client list, 208 returning pointer to player in, 532* client prediction, 429 client.cpp, 511-525* client.h, 482-487* ClientDlgProc callback procedure, 7-8 ClientProcess function TCP, 155-157 UDP, 159-160 clients, 135-136 adding to client list, 293-295*, 333-334*, 366-367* checking for timing out, 299-301* connecting with server, 126 creating, 150, 481-542* creating for game lobby, 371-405* creating TCP, 151-153* creating UDP, 157-159* initializing, 268-269* keeping alive, 336* pinging, 293* receiving packets from, 301-303* removing from client list, 295-297*, 334-336*, 366-368 resetting, 270* sending messages to, 290-293* uninitializing, 269* CLobby class methods, 389-398* CLobbyServer class methods, 408-418* close function, 129 closesocket function, 129 code, structuring, 73-78 colors, specifying, 190 column naming restrictions, 20 commands, building, 453-458* reading, 452-453*, 505-508* sending, 451-452*, 502-504* common.h, 310*, 340*, 488 concurrent server, 112, 135 conditional statements, using in PHP, 68-70 using in SQL, 37-39 connect function, 126 Connect function, 337*, 403*, 504-505* connected socket, 128

connection, starting, 502* connectionless protocol, 109 connection-oriented protocol, 108 core.php, 74*, 76* CREATE DATABASE command, 20-21 CREATE TABLE command, 25 CreateAccountDialogProc function, 316-319* CreateGameDialogProc function, 380-381* CreateThread function, 168-169 CreateViewPlayersDialogProc function, 380* CSignin class, 308-309* constructor, 329* destructor, 329* global variables, 316 methods, 323-329* CSigninServer class, 339-340* constructor, 361* destructor, 361* global variables, 347 methods, 355-361*

D
daemon, 18 daemonInit function, 351-352* data, adding to table, 30-32, 41 deleting from table, 34-35 importing, 44-46 linking, 42-43 modifying in table, 32-34 reading, 257-260* receiving, 243-244* receiving in socket, 131-133 relating, 42-43 removing, 84-90* requesting, 401* retrieving, 50-53* sending, 208-209, 244-245* sending to socket, 129-132 storing, 79-84* updating, 53-55*, 84-90* writing, 255*, 257* data buffer, dumping, 270* retrieving point from, 254* writing to, 255*

544

Index

data definition language, 19 using, 19-29 data input, 44-46 data link layer, 105 data manipulation language, 19, 29 using, 29-39 data types, and platforms, 121 defining, 212-213 in MySQL, 22-23 database, adding table to, 25 backing up to file, 47-49 connecting to, 50-53 connecting to with PHP script, 78-79* creating, 20-21 creating relational, 40 dropping, 21-22 dropping table from, 28-29 modifying table in, 26-28 naming restrictions, 20 relational, 40 removing data from, 84-90* restoring, 49 storing data in, 79-84* updating data in, 84-90* viewing, 19-20 database, relational, 40 see also database datagram socket, 111-112 dead reckoning, 429 debug version, 12 DELETE command, 34-35 delta compression, 428-429 DESCRIBE command, 26 DialogBox function, 7 dialogs, see also windows creating, 6-7* creating for game lobby, 372-377 Disconnect function, 337*, 403-404*, 505* doAddUser function, 84 Domain Name Service, 106 DrawMap function, 528-530* dreamClient class, 226-228* constructor, 268* destructor, 268* functionality of, 262-268* variables, 260-262

dreamConsole constructor, 215 destructor, 215 dreamMessage class, 225-226* functionality of, 249-253* variables, 249 dreamServer class, 228-229* constructor, 289* destructor, 289-290* functionality of, 280-288* variables, 280 dreamsock, 206 dreamSock, creating network application with, 305-369* dreamSock application, planning functionality of, 306 dreamSock network library, creating, 248-303* dreamSock.h, 220-225* dreamSock_Broadcast function, 245-246* dreamSock_CloseSocket function, 243* dreamSock_GetCurrentSystemTime function, 246* dreamSock_GetPacket function, 243-244* dreamSock_Initialize function, 229-230* dreamSock_InitializeWinSock function, 230-231* dreamSock_Linux_ GetCurrentSystemTime function, 246-247* dreamSock_OpenUDPSocket function, 241-243* dreamSock_SendPacket function, 244-245* dreamSock_SetBroadcasting function, 239-240* dreamSock_SetNonBlocking function, 239* dreamSock_Shutdown function, 231-232* dreamSock_Socket function, 238-239* dreamSock_StringToSockaddr function, 240* dreamSock_Win_ GetCurrentSystemTime function, 247-248* DROP DATABASE command, 21-22 DROP TABLE command, 28-29 DumpBuffer function, 270*

E
echo application, running, 160-161 echo TCP client, creating, 151-153* echo TCP server, creating, 139-142*

545

Index

echo UDP client, creating, 157-159* echo UDP server, creating, 146-150* engine, initializing, 526-527* shutting down, 527* ephemeral ports, 110 errno variable, using to retrieve error number, 248 error handling, 120 error values, retrieving, 248 event object, 166-167 events, detecting network, 163 processing, 186-187 Excel, using to enter data, 45-46 exceptions, catching, 306

F
FastTemplate, 90 converting to, 95-101 using, 91-95* flag collisions, checking for, 475-476* footer.php, 75-76* Frame function, 477-478*, 530-531* frame history, 430 frame time, monitoring, 428 frames, comparing, 533-534* running, 477-478* storing, 430 ft.php, 91-92* ft2.php, 94* ftcore.php, 98* ftusers.php, 99-101* functions, global, 229-248* graphical, 190-193 using in PHP, 70-71

G
game, adding to game list, 478-479* creating log system for, 214-219* removing from game list, 479-481* game client, creating, 481-542* game list, adding game to, 478-479* refreshing, 541* removing game from, 479-481*

game lobby, creating, 371-425* creating dialogs for, 372-377, 380-386* game logic, handling, 530-531* game map, drawing, 528-530* generating, 467-468* game platform, choosing, 113 game server, creating, 430-481* reasons for creating, 15 game tutorial code client.cpp, 511-525* client.h, 482-487* common.h, 310*, 340*, 488 dreamSock.h, 220-225* lobby.cpp, 389-398*, 408-418*, 478-481*, 541-542* lobby.h, 377-378*, 406-407* main.cpp, 310-316*, 341-347*, 380-389*, 407-408*, 436-437*, 488-490* main.h, 309*, 379* network.cpp, 437-449*, 490-502* network.h, 309*, 340*, 379*, 407*, 436*, 487-488* server.cpp, 458-467* server.h, 431-435* signin.cpp, 323-329*, 355-361* signin.h, 307-308*, 338-339* GenerateRandomMap function, 467-468* GET method, 73 GetClientPointer function, 532* GetNewPoint function, 254* GetPacket function, 275-276*, 301-303* getsockopt function, 173-174 GFX_Blit function, 192-193 GFX_Line function, 191 GFX_LoadBitmap function, 192 GFX_Pixel function, 191 GFX_Rect function, 191 GFX_RectFill function, 191 GFX_Tri function, 191 GFX_TriFill function, 191 global functions, 229-248* graphical functions, 190-193

546

Index

graphics, loading, 200-202* rotating, 200-202* graphics engine, initializing, 526-527* shutting down, 527*

H
header files, setting up for network library, 219-220* header.php, 75* heading, calculating, 469-470* hosts, identifying, 207-208 sending data to, 208-209 HWND, 4

ioctl function, 172-173 ioctlsocket function, 172-173 IP, 106-108 address classes, 107 addresses, 107 IPv4, 106-107 address structure, 121-122 IPv6, 107-108 address structure, 122 ISO, 104 iterative server, 112, 134

J
JoinGameDialogProc function, 382*

K
key, checking for status of, 352* keyboard input, 193-195 checking for, 532-533* handling, 489* keyPress function, 352* KillPlayer function, 532*

I
I/O, controlling, 172 I/O strategies, 169 images, displaying, 192-193 loading, 192, 200-202* rotating, 200-202* Include folder, 10 index.php, 64* index2.php, 65* index3.php, 67* index4.php, 68-69* index5.php, 70* inet_aton function, 134 Init function, 253* initialization, Windows, 115-120 Initialize function, 268-269*, 290* InitializeEngine function, 526-527* InitializeWinSock function, 117-119* InitNetwork function, 362*, 449* InitSockets function TCP, 142-145, 154-155 UDP, 148-149, 159 input, checking for, 532-533* handling, 489* input.php, 71-72* input/output functions, 172-175 INSERT command, 30-32 International Organization for Standardization, see ISO Internet Protocol, see IP

L
LAN, 136 LAN server, searching for, 175-176 latency, calculating, 209 Lib folder, 10 LIKE command, 39 line, drawing, 191 link table, 40 creating, 40-41 Linux, building network library in, 211-212 listen function, 127 little-endian order, 136 LOAD command, 44 lobby.cpp, 389-398*, 408-418*, 478-481*, 541-542* lobby.h, 377-378*, 406-407* LobbyDialogProc function, 384-386* local area network, see LAN log system, 213-214 creating, 214-219* LoginDialogProc function, 382-383* LogString function, 217-218* loops, using in PHP, 67-68

547

Index

M
macros, socket descriptor, 164-165 main function, 353-355* TCP, 153-154 main.cpp, 310-316*, 341-347*, 380-389*, 407-408*, 436-437*, 488-490* main.h, 309*, 379* main.tpl.html, 96* mainbody.tpl.html, 93* map, drawing, 528-530* generating, 467-468* message loop, creating, 184-185* messages, building, 508-509* compressing, 428-429 functions for working with, 271-279* handling, 430 notification, 366* parsing, 297-299*, 329-333* reading, 362-366*, 399-401*, 418-425*, 449-451* sending, 402-403*, 504-505* sending to clients, 303* sending to server, 336* system vs. user, 270-271 update, 8-9 messaging system, in Windows, 3-4 MoveObjects function, 539-540* MovePlayer function, 472-475* multicast IP addresses, 107 multiplay, 134 multiple templates, using, 93-95* multiplexing, 171 multiplexing I/O, 171 multitasking, 4 multithreading, 4, 167-168 MySQL, 15 connecting to database with, 78-79* data types, 22-23 installing, 16-19 storing data with, 79-84* using, 78-90* MySQL C++ interface, see MySQL++ MySQL example, connecting to database and retrieving data, 50-53*

updating data from an application, 53-55* MySQL server, running, 19 MySQL++, 50 mysql1.php, 79* mysqlcore.php, 80*, 85* mysqldump utility, 47

N
NET_WinSockInitialize function, TCP, 142 network, initializing, 449* running, 509-510* network application, creating with dreamSock, 305-369* network byte order, 136 network events, detecting, 163 network functionality, designing, 428-430 network latency, calculating, 209 network layer, 105 network library, building, 209-212 creating dreamSock, 248-303* initializing, 229-230* planning, 206-207 reasons for creating, 205-206 setting up files for, 219-220* shutting down, 231-232* network protocol, see protocol network system, running, 337-338*, 369*, 404-405* network.cpp, 437-449*, 490-502* network.h, 309*, 340*, 379*, 407*, 436*, 487-488* non-blocking I/O, 170 non-blocking socket, 144

O
Open Systems Interconnection model, see OSI model OpenGL libraries, adding to workspace, 183 operators, using in PHP, 67-68 OSI model, 104 layers, 104-105 usage of, 105 output.php, 72*

548

Index

P
page1.php, 77 page2.php, 77 ParsePacket function, 273-275*, 297-299* PHP, arrays in, 70 conditional statements in, 68-70 connecting to database with, 78-79* functions in, 70-71 operators and loops in, 67-68 testing installation of, 62 updating data with, 84-90* user input in, 71-73 using, 63-73 variables in, 65-66 PHP4, installing on Apache web server, 60-63 physical layer, 105 pinging, 209 pixel, plotting, 191 platforms, and data types, 121, 212-213 and sockets, 113 and sockets API, 112-113 choosing, 113 Unix, 113 Windows, 113 player, calculating heading of, 536-537* calculating velocity of, 535-536* moving, 472-475*, 539-540* moving back to start position, 532* predicting movement of, 537-539* returning pointer to, 532* player list, refreshing, 541-542* updating, 398-399* port numbers, 109 ports, 109-110 POST method, 73 PredictMovement function, 537-539* presentation layer, 105 primary key, 40 primitives, moving, 195-199* println function, 215-216* program, running as daemon, 351-352* project, adding static libraries to, 183

creating, 182-189 protocol, 103-104 connectionless, 109 connection-oriented, 108 pthread_create function, 169

R
Read function, 257-258* ReadByte function, 258* ReadDeltaMoveCommand function, 452-453*, 506-508* ReadFloat function, 259* ReadLong function, 259* ReadMoveCommand function, 505-506* ReadPackets function, 329-333*, 362-366*, 399-401*, 418-425*, 449-451* ReadShort function, 258-259* ReadString function, 259-260* rectangle, drawing, 191 recv function, 131-132 recvfrom function, 133 RefreshGameList function, 541* RefreshJoinedPlayersList function, 541-542* RefreshPlayerList function, 398-399* relational database, 40-41 see also database release version, 12 RemoveClient function, 295-297*, 334-335*, 367-368* RemoveClients function, 335-336*, 368* RemoveGame function, 479-480* RemoveGames function, 481* RequestGameData function, 401* Reset function, 270* RunNetwork function, 337-338*, 369*, 404-405*, 509-510*

S
select function, 163-164 SELECT statements, conditional, 37-39 using, 35-39 send function, 129-131 SendAddClient function, 290-292* SendChat function, 402* SendCommand function, 451-452*, 502-504* SendConnect function, 271-272* SendCreateGame function, 402*

549

Index

SendDisconnect function, 272* SendDlgItemMessage function, 9 SendExitNotification function, 366* SendKeepAlive function, 336* SendPacket function, 276-279* SendPackets function, 303* SendPing function, 272-273*, 293* SendRemoveClient function, 292-293* SendRemoveGame function, 402* SendRequestNonDeltaFrame function, 504* SendSignIn function, 336* SendStartGame function, 403*, 504* sendto function, 132 server application, creating, 338-369* server.cpp, 458-467* server.h, 431-435* ServerProcess function TCP, 145-146* UDP, 149-150 servers, 134-135 connecting to, 337*, 403* connecting with client, 126 creating, 136-137, 430-481* creating for game lobby, 406-425* creating TCP, 139-142* creating UDP, 146-150* disconnecting from, 337*, 403-404*, 527* initializing, 362* reasons for storing game data on, 15 searching for, 175-176 types of, 112, 134-135 uninitializing, 362* session layer, 105 setsockopt function, 173-174 SHOW DATABASES command, 19-22 SHOW TABLES command, 25-26 shutdown function, 174-175 Shutdown function, 527* ShutdownNetwork function, 362* signal-driven I/O, 170-171 sign-in system, creating, 306-369* signin.cpp, 323-329*, 355-361* signin.h, 307-308*, 338-339* simple.tpl.html, 91* skeleton project, creating, 182-189 socket descriptor macros, 164-165

socket function, 124 socket functions, global, 232-248* sockets, 110-111 addresses of, 112 binding address to, 125-126 blocking, 144 broadcasting, 245-246* closing, 129, 243*, 290* connected, 128 controlling I/O mode of, 172-173 converting address string of, 240* creating, 124, 238-239*, 290* data types of, 121 disabling, 174-175 getting options for, 173-174 initializing, 142-145, 148-149, 154-155, 159 non-blocking, 144 opening UDP, 241-243* receiving data in, 131-133 sending data to, 129-132 setting blocking mode of, 239* setting for incoming connections, 127 setting options for, 173-174 setting to broadcast, 239-240* types of, 111-112 sockets API, 110 and platforms, 112-113 sockets functions, basic, 124-129 input/output, 129-133 source files, adding to project, 183 setting up for network library, 219-220* SQL, 19 StartConnection function, 502* StartLog function, 216-217* StartLogConsole function, 214 static libraries, adding to project, 183 static link library, 9-10 creating, 10-13 structure of, 10 using, 13-14 StopLog function, 218-219* stream socket, 111 structured query language, see SQL

550

Index

system messages vs. user messages, 270-271 system time, checking, 246-248

Unix, as game platform, 113 building network library in, 211-212 input/output functions, 172-175 multithreading in, 169 sockets functions, 124-133 UPDATE command, 32-34 update message, 8-9 USE command, 25 User Datagram Protocol, see UDP user input, in PHP, 71-73 user messages vs. system messages, 270-271 userlist.tpl.html, 96-97* userlist_row.tpl.html, 97* users.php, 81-82*, 85-88*

T
tablerow.tpl.html, 94* tables, adding data to, 30-32, 41 adding to relational database, 40-41 creating, 24-26 deleting data from, 34-35 dropping, 28-29 modifying, 26-28 modifying data in, 32-34 naming restrictions, 20 viewing columns in, 26 TCP, 106, 108 TCP client, 150-151 creating, 151-153* TCP connection, accepting incoming, 128 listening for, 127 TCP server, 137-138 creating, 139-142* TCP/IP, 106 templates, 90 creating, 95-96* using, 91-95* using multiple, 93-95* test.php, 62* text file, using to enter data, 44-45 thread, 134-135 creating, 168-169 timing out, 209 Transmission Control Protocol, see TCP Transmission Control Protocol/Internet Protocol, see TCP/IP transport layer, 105, 108 triangle, drawing, 191 typecasting, 123

V
variables, initializing, 253* resetting, 235* using in PHP, 65-66 VectorLength function, 488-489* vectors, working with, 488-489* VectorSubtract function, 488-489* velocity, calculating, 468-472* victory conditions, checking, 531-532* Visual Studio, configuring, 181-182 creating static link library with, 10-13 creating window with, 4-6 finding static link library with, 13-14

W
web server, see Apache web server welcome.php, 74*, 76-77* well-known ports, 109 WHERE command, 37-39 Win32 functions, input/output, 172-175 sockets, 124-133 windowed application, creating, 184-189* WindowProc function, 347* windows, see also dialogs creating, 4-6 handle, 4 updating, 8-9

U
UDP, 106, 109 UDP client, 151 creating, 157-159* UDP server, 138-139 creating, 146-150* UDP socket, opening, 241-243* Uninitialize function, 269*, 290*

551

Index

Windows, as game platform, 113 building network library in, 210-211 callback procedure, see WndProc messaging system, 3-4 multithreading in, 167-168 Windows Sockets, see WinSock WinMain function, 5, 7-8, 184-185*, 319-323*, 347-351*, 386-389* WinSock, 112-113 initializing, 115-120 WinSock API, initializing, 230-231* WM_CLOSE event, 186-187 WM_KEYDOWN event, 187 WM_KEYUP event, 187 WM_SIZE event, 187 WndProc, 5, 185-186*

workspace, adding static libraries to, 183 creating, 182-183 Write function, 255* WriteByte function, 255-256* WriteFloat function, 256* WriteLong function, 256* WriteShort function, 256* WriteString function, 257* WSAAsyncSelect function, 165 WSACleanup function, 116 WSAEnumProtocols function, 117, 120 WSAEventSelect function, 165-166 WSAGetLastError function, 120 using to retrieve error number, 248 WSAStartup function, 115-116 WSAWaitForMultipleEvents function, 166

552

This page intentionally left blank.

Looking
Check out Wordwares marketfeaturing the following new

Introduction to 3D Game Programming with DirectX 9.0

1-55622-913-5 $49.95 6 x 9 424 pp.

Advanced 3D Game Programming with DirectX 9.0

1-55622-968-2 $59.95 6 x 9 552 pp.

Learn Vertex and Pixel Shader Programming with DirectX 9

1-55622-287-4 $34.95 6 x 9 304 pp.

ShaderX2: Introductions & Tutorials with DirectX 9

1-55622-902-X $44.95 6 x 9 384 pp.

ShaderX2: Shader Programming Tips & Tricks with DirectX 9

1-55622-988-7 $59.95 6 x 9 728 pp.

DirectX 9 User Interfaces: Design and Implementation

1-55622-249-1 $44.95 6 x 9 376 pp.

DirectX 9 Audio Exposed: Interactive Audio Development

1-55622-288-2 $59.95 6 x 9 568 pp.

Strategy Game Programming with DirectX 9.0

1-55622-922-4 $59.95 6 x 9 560 pp.

Visit us online at

www.wordware.com for more information.

for more?
leading Game Developer s Library releases and backlist titles.

Java 1.4 Game Programming

1-55622-963-1 $59.95 6 x 9 672 pp.

Vector Game Math Processors

1-55622-921-6 $59.95 6 x 9 528 pp.

Game Design: Theory and Practice

1-55622-735-3 $49.95 7 x 9 608 pp.

3D Math Primer for Graphics and Game Development

1-55622-911-9 $49.95 7 x 9 448 pp.

Game Design Foundations

1-55622-973-9 $39.95 6 x 9 400 pp.

CGI Filmmaking: The Creation of Ghost Warrior Game Development and Production
1-55622-951-8 $49.95 6 x 9 432 pp.

Games That Sell!

1-55622-950-X $34.95 6 x 9 336 pp.

1-55622-227-0 $49.95 9 x 7 344 pp.

Use the following coupon code for online specials:

pmg0766

Looking for more?

Check out Wordwares market-leading Game Developer s Library featuring the following new releases, backlist, and upcoming titles.

LightWave 3D 7 Character Animation

1-55622-901-1 $49.95 7 x 9 360 pp.

Essential LightWave 3D 7.5

1-55622-226-2 $44.95 6 x 9 424 pp.

LightWave 3D 7.5 Lighting

1-55622-354-4 $69.95 6 x 9 496 pp.

LightWave 3D 8: 1001 Tips & Tricks

1-55622-090-1 $39.95 6 x 9 500 pp.

Modeling a Character in 3DS Max (2nd Edition)

1-55622-088-X $44.95 6 x 9 550 pp.

LightWave 3D 8 Cartoon Character Creation

1-55622-083-9 $49.95 6 x 9 500 pp.

Advanced Lighting and Materials with Shaders

1-55622-292-0 $59.95 9 x 7 500 pp.

Fundamentals of Character Animation

1-55622-248-3 $49.95 9 x 7 400 pp.

Design First for 3D Animators

1-55622-085-5 $49.95 9 x 7 350 pp.

Visit us online at

www.wordware.com for more information.

pmg0766

Use the following coupon code for online specials:

About the CD
The companion CD contains all the source code from the book as well as many useful tools and applications discussed in the book. The CD will autorun when you insert it in your CD drive. Click the Continue button on the page that appears to view the contents. (If the CD does not autorun, simply use Windows Explorer to browse the CD.) The directories and their contents are: n Software MySQL for Linux and Windows, DBI modules for Perl, WinZip 8.0 SHAREWARE EVALUATION version (not the registered version), the Apache package, the FastTemplate class, and Adobe Acrobat Reader Source Source code from the book Libraries dreamSock, OpenGL, 2DLIB, 3DLIB, and MySQL++

n n

CD/Source Code Usage License Agreement

Please read the following CD/Source Code usage license agreement before opening the CD and using the contents therein: 1. By opening the accompanying software package, you are indicating that you have read and agree to be bound by all terms and conditions of this CD/Source Code usage license agreement. The compilation of code and utilities contained on the CD and in the book are copyrighted and protected by both U.S. copyright law and international copyright treaties, and is owned by Wordware Publishing, Inc. Individual source code, example programs, help files, freeware, shareware, utilities, and evaluation packages, including their copyrights, are owned by the respective authors. No part of the enclosed CD or this book, including all source code, help files, shareware, freeware, utilities, example programs, or evaluation programs, may be made available on a public forum (such as a World Wide Web page, FTP site, bulletin board, or Internet news group) without the express written permission of Wordware Publishing, Inc. or the author of the respective source code, help files, shareware, freeware, utilities, example programs, or evaluation programs. You may not decompile, reverse engineer, disassemble, create a derivative work, or otherwise use the enclosed programs, help files, freeware, shareware, utilities, or evaluation programs except as stated in this agreement. The software, contained on the CD and/or as source code in this book, is sold without warranty of any kind. Wordware Publishing, Inc. and the authors specifically disclaim all other warranties, express or implied, including but not limited to implied warranties of merchantability and fitness for a particular purpose with respect to defects in the disk, the program, source code, sample files, help files, freeware, shareware, utilities, and evaluation programs contained therein, and/or the techniques described in the book and implemented in the example programs. In no event shall Wordware Publishing, Inc., its dealers, its distributors, or the authors be liable or held responsible for any loss of profit or any other alleged or actual private or commercial damage, including but not limited to special, incidental, consequential, or other damages. One (1) copy of the CD or any source code therein may be created for backup purposes. The CD and all accompanying source code, sample files, help files, freeware, shareware, utilities, and evaluation programs may be copied to your hard drive. With the exception of freeware and shareware programs, at no time can any part of the contents of this CD reside on more than one computer at one time. The contents of the CD can be copied to another computer, as long as the contents of the CD contained on the original computer are deleted. You may not include any part of the CD contents, including all source code, example programs, shareware, freeware, help files, utilities, or evaluation programs in any compilation of source code, utilities, help files, example programs, freeware, shareware, or evaluation programs on any media, including but not limited to CD, disk, or Internet distribution, without the express written permission of Wordware Publishing, Inc. or the owner of the individual source code, utilities, help files, example programs, freeware, shareware, or evaluation programs. You may use the source code, techniques, and example programs in your own commercial or private applications unless otherwise noted by additional usage agreements as found on the CD.

WARNING By opening the CD package, you accept the terms and conditions of the CD/Source Code Usage License Agreement. Additionally, opening the CD package makes this book nonreturnable.