Fundamentals of

C++
Programming
A F T
DR
Richard L. Halterman
School of Computing
Southern Adventist University

January 18, 2015

Copyright 20082015 Richard L. Halterman. All rights reserved.
 i

Contents

1 The Context of Software Development 1

1.1 Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Development Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Learning Programming with C++ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 Writing a C++ Program 7

2.1 General Structure of a Simple C++ Program . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Editing, Compiling, and Running the Program . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Variations of our simple program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.4 Template for simple C++ programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3 Values and Variables 15

3.1 Integer Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Variables and Assignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3 Identifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.4 Additional Integer Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.5 Floating-point Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.6 Constants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.7 Other Numeric Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.8 Characters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.9 Enumerated Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.10 Type Inference with auto . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

2015 Richard L. Halterman Draft date: January 18, 2015

CONTENTS ii

3.11 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.12 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4 Expressions and Arithmetic 37

4.1 Expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2 Mixed Type Expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Operator Precedence and Associativity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.4 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.5 Formatting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.6 Errors and Warnings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.6.1 Compile-time Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.6.2 Run-time Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.6.3 Logic Errors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.6.4 Compiler Warnings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.7 Arithmetic Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.8 Integers vs. Floating-point Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.8.1 Integer Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.8.2 Floating-point Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.9 More Arithmetic Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.10 Bitwise Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.11 Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.12 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.13 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

5 Conditional Execution 89
5.1 Type bool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.2 Boolean Expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
5.3 The Simple if Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.4 Compound Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.5 The if/else Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.6 Compound Boolean Expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.7 Nested Conditionals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.8 Multi-way if/else Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.9 Errors in Conditional Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.10 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

2015 Richard L. Halterman Draft date: January 18, 2015

CONTENTS iii

5.11 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

6 Iteration 127
6.1 The while Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.2 Nested Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6.3 Abnormal Loop Termination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
6.3.1 The break statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
6.3.2 The goto Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
6.3.3 The continue Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
6.4 Infinite Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
6.5 Iteration Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
6.5.1 Drawing a Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
6.5.2 Printing Prime Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
6.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160

7 Other Conditional and Iterative Statements 165

7.1 The switch Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
7.2 The Conditional Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
7.3 The do/while Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
7.4 The for Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
7.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
7.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181

8 Using Functions 185

8.1 Introduction to Using Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
8.2 Standard Math Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
8.3 Maximum and Minimum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
8.4 clock Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
8.5 Character Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
8.6 Random Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
8.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
8.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204

9 Writing Functions 207

9.1 Function Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208

2015 Richard L. Halterman Draft date: January 18, 2015

CONTENTS iv

9.2 Using Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216

9.3 Pass by Value . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
9.4 Function Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
9.4.1 Better Organized Prime Generator . . . . . . . . . . . . . . . . . . . . . . . . . . 222
9.4.2 Command Interpreter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
9.4.3 Restricted Input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
9.4.4 Better Die Rolling Simulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
9.4.5 Tree Drawing Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
9.4.6 Floating-point Equality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
9.4.7 Multiplication Table with Functions . . . . . . . . . . . . . . . . . . . . . . . . . 232
9.5 Commenting Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
9.6 Custom Functions vs. Standard Functions . . . . . . . . . . . . . . . . . . . . . . . . . . 237
9.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
9.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240

10 Managing Functions and Data 245

10.1 Global Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
10.2 Static Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
10.3 Overloaded Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
10.4 Recursion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
10.5 Making Functions Reusable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
10.6 Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
10.7 Reference Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
10.8 Pass by Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
10.8.1 Pass by Reference via Pointers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
10.8.2 Pass by Reference via References . . . . . . . . . . . . . . . . . . . . . . . . . . 277
10.9 Higher-order Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
10.10Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
10.11Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282

11 Aggregate Data 289

11.1 Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
11.1.1 Declaring and Using Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
11.1.2 Traversing a Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
11.1.3 Vector Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299

2015 Richard L. Halterman Draft date: January 18, 2015

CONTENTS v

11.1.4 Vectors and Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302

11.1.5 Multidimensional Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
11.2 Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
11.3 Prime Generation with a Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
11.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
11.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322

12 Sorting and Searching 325

12.1 Sorting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
12.2 Flexible Sorting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
12.3 Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
12.3.1 Linear Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 330
12.3.2 Binary Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
12.4 Vector Permutations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
12.5 Randomly Permuting a Vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
12.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
12.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 356

13 Standard C++ Classes 359

13.1 String Objects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
13.2 Input/Output Streams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
13.3 File Streams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
13.4 Complex Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
13.5 Better Pseudorandom Number Generation . . . . . . . . . . . . . . . . . . . . . . . . . . 372
13.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380
13.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381

14 Custom Objects 383

14.1 Object Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
14.2 Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
14.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390
14.4 Constructors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397
14.5 Destructors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 400
14.6 Defining a New Numeric Type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401
14.7 Encapsulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403

2015 Richard L. Halterman Draft date: January 18, 2015

CONTENTS vi

14.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 406

14.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407

15 Fine Tuning Objects 413

15.1 Pointers to Objects and Object Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
15.2 The this Pointer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
15.3 const Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
15.4 Separating Method Declarations and Definitions . . . . . . . . . . . . . . . . . . . . . . . 420
15.5 Preventing Multiple Inclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426
15.6 Overloaded Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430
15.6.1 Operator Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 430
15.6.2 Operator Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433
15.7 static Members . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434
15.8 Classes vs. structs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
15.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
15.10Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439

16 Building some Useful Classes 441

16.1 A Better Rational Number Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 441
16.2 Stopwatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
16.3 Sorting with Logging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
16.4 Linked Lists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453
16.5 Automating Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463
16.6 Convenient High-quality Pseudorandom Numbers . . . . . . . . . . . . . . . . . . . . . . 468
16.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470
16.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 470

17 Inheritance and Polymorphism 473

17.1 I/O Stream Inheritance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
17.2 Inheritance Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475
17.3 Uses of Inheritance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478
17.4 Polymorphism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485
17.5 Protected Members . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 491
17.6 Fine Tuning Inheritance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499
17.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509

2015 Richard L. Halterman Draft date: January 18, 2015

CONTENTS vii

17.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 510

18 Generic Programming 513

18.1 Function Templates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
18.2 Class Templates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524
18.3 Iterators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 534
18.4 Iterator Ranges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 538
18.5 The Standard Template Library . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546
18.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546
18.7 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 546

19 Handling Exceptions 547

19.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 547
19.2 Exception Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 548
19.3 Custom Exceptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 556
19.4 Catching Multiple Exceptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 558
19.5 Exception Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561
19.6 Using Exceptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
19.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
19.8 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566

20 Concurrent Programming 567

20.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 567
20.2 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 567
20.3 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 567

Appendices 571

A Using Visual Studio 2013 to Develop C++ Programs 571

B Visual C++ Command Line Development 577

C Developing C++ Programs with the GNU Tools 581

D Arrays 585
D.1 Declaring and Using Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 585
D.2 Arrays and Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 592
D.3 Prime Generation with an Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595

2015 Richard L. Halterman Draft date: January 18, 2015

CONTENTS viii

D.4 Multidimensional Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596

D.5 Pointers and Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 599
D.6 Array Ranges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 603
D.7 C Strings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604
D.8 Dynamic Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 608
D.9 The sizeof Operator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 612
D.10 Copying an Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 614
D.11 Memory Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 616
D.12 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 620
D.13 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 621

Bibliography 624

Index 625

2015 Richard L. Halterman Draft date: January 18, 2015

 ix

Preface

Legal Notices and Information

Permission is hereby granted to make hardcopies and freely distribute the material herein under the
following conditions:

The copyright and this legal notice must appear in any copies of this document made in whole or in
part.
None of material herein can be sold or otherwise distributed for commercial purposes without written
permission of the copyright holder.

Instructors at any educational institution may freely use this document in their classes as a primary
or optional textbook under the conditions specified above.

A local electronic copy of this document may be made under the terms specified for hard copies:

The copyright and these terms of use must appear in any electronic representation of this document
made in whole or in part.
None of material herein can be sold or otherwise distributed in an electronic form for commercial
purposes without written permission of the copyright holder.

Instructors at any educational institution may freely store this document in electronic form on a local
server as a primary or optional textbook under the conditions specified above.

Additionally, a hardcopy or a local electronic copy must contain the uniform resource locator (URL)
providing a link to the original content so the reader can check for updated and corrected content. The
current standard URL is http://python.cs.southern.edu/cppbook/progcpp.pdf.

2015 Richard L. Halterman Draft date: January 18, 2015

 1

Chapter 1

The Context of Software Development

A computer program, from one perspective, is a sequence of instructions that dictate the flow of electri-
cal impulses within a computer system. These impulses affect the computers memory and interact with
the display screen, keyboard, mouse, and perhaps even other computers across a network in such a way
as to produce the magic that permits humans to perform useful tasks, solve high-level problems, and
play games. One program allows a computer to assume the role of a financial calculator, while another
transforms the machine into a worthy chess opponent. Note the two extremes here:

at the lower, more concrete level electrical impulses alter the internal state of the computer, while

at the higher, more abstract level computer users accomplish real-world work or derive actual plea-
sure.

So well is the higher-level illusion achieved that most computer users are oblivious to the lower-level
activity (the machinery under the hood, so to speak). Surprisingly, perhaps, most programmers today write
software at this higher, more abstract level also. An accomplished computer programmer can develop
sophisticated software with little or no interest or knowledge of the actual computer system upon which it
runs. Powerful software construction tools hide the lower-level details from programmers, allowing them
to solve problems in higher-level terms.
The concepts of computer programming are logical and mathematical in nature. In theory, computer
programs can be developed without the use of a computer. Programmers can discuss the viability of a
program and reason about its correctness and efficiency by examining abstract symbols that correspond
to the features of real-world programming languages but appear in no real-world programming language.
While such exercises can be very valuable, in practice computer programmers are not isolated from their
machines. Software is written to be used on real computer systems. Computing professionals known
as software engineers develop software to drive particular systems. These systems are defined by their
underlying hardware and operating system. Developers use concrete tools like compilers, debuggers, and
profilers. This chapter examines the context of software development, including computer systems and
tools.

2015 Richard L. Halterman Draft date: January 18, 2015

1.1. SOFTWARE 2

1.1 Software

A computer program is an example of computer software. Software makes a computer a truly universal
machine transforming it into the proper tool for the task at hand. One can refer to a program as a piece of
software as if it were a tangible object, but software is actually quite intangible. It is stored on a medium. A
hard drive, a CD, a DVD, and a USB pen drive are all examples of media upon which software can reside.
The CD is not the software; the software is a pattern on the CD. In order to be used, software must be stored
in the computers memory. Typically computer programs are loaded into memory from a medium like the
computers hard disk. An electromagnetic pattern representing the program is stored on the computers hard
drive. This pattern of electronic symbols must be transferred to the computers memory before the program
can be executed. The program may have been installed on the hard disk from a CD or from the Internet. In
any case, the essence that was transferred from medium to medium was a pattern of electronic symbols that
direct the work of the computer system.
These patterns of electronic symbols are best represented as a sequence of zeroes and ones, digits from
the binary (base 2) number system. An example of a binary program sequence is

10001011011000010001000001001110

To the underlying computer hardware, specifically the processor, a zero here and three ones there might
mean that certain electrical signals should be sent to the graphics device so that it makes a certain part of
the display screen red. Unfortunately, only a minuscule number of people in the world would be able to
produce, by hand, the complete sequence of zeroes and ones that represent the program Microsoft Word
for an Intel-based computer running the Windows 8 operating system. Further, almost none of those who
could produce the binary sequence would claim to enjoy the task.
The Word program for older Mac OS X computers using a PowerPC processor works similarly to the
Windows version and indeed is produced by the same company, but the program is expressed in a com-
pletely different sequence of zeroes and ones! The Intel Core i7 processor in the Windows machine accepts
a completely different binary language than the PowerPC processor in the Mac. We say the processors have
their own machine language.

1.2 Development Tools

If very few humans can (or want) to speak the machine language of the computers processors and software
is expressed in this language, how has so much software been developed over the years?
Software can be represented by printed words and symbols that are easier for humans to manage than
binary sequences. Tools exist that automatically convert a higher-level description of what is to be done
into the required lower-level code. Higher-level programming languages like C++ allow programmers to
express solutions to programming problems in terms that are much closer to a natural language like English.
Some examples of the more popular of the hundreds of higher-level programming languages that have been
devised over the past 60 years include FORTRAN, COBOL, Lisp, Haskell, C, Perl, Python, Java, and C#.
Most programmers today, especially those concerned with high-level applications, usually do not worry
about the details of underlying hardware platform and its machine language.
One might think that ideally such a conversion tool would accept a description in a natural language,
such as English, and produce the desired executable code. This is not possible today because natural
languages are quite complex compared to computer programming languages. Programs called compilers
that translate one computer language into another have been around for over 60 years, but natural language

2015 Richard L. Halterman Draft date: January 18, 2015

1.2. DEVELOPMENT TOOLS 3

processing is still an active area of artificial intelligence research. Natural languages, as they are used
by most humans, are inherently ambiguous. To understand properly all but a very limited subset of a
natural language, a human (or artificially intelligent computer system) requires a vast amount of background
knowledge that is beyond the capabilities of todays software. Fortunately, programming languages provide
a relatively simple structure with very strict rules for forming statements that can express a solution to any
problem that can be solved by a computer.
Consider the following program fragment written in the C++ programming language:
subtotal = 25;
tax = 3;
total = subtotal + tax;

These three lines do not make up a complete C++ program; they are merely a piece of a program. The
statements in this program fragment look similar to expressions in algebra. We see no sequence of bi-
nary digits. Three words, subtotal, tax, and total, called variables, are used to hold information.
Mathematicians have used variables for hundreds of years before the first digital computer was built. In
programming, a variable represents a value stored in the computers memory. Familiar operators (= and +)
are used instead of some cryptic binary digit sequence that instructs the processor to perform the operation.
Since this program is expressed in the C++ language, not machine language, it cannot be executed directly
on any processor. A C++ compiler is used to translate the C++ code into machine code.
The higher-level language code is called source code. The compiled machine language code is called
the target code. The compiler translates the source code into the target machine language.
The beauty of higher-level languages is this: the same C++ source code can be compiled to different
target platforms. The target platform must have a C++ compiler available. Minor changes in the source code
may be required because of architectural differences in the platforms, but the work to move the program
from one platform to another is far less than would be necessary if the program for the new platform had
to be rewritten by hand in the new machine language. Just as importantly, when writing the program the
human programmer is free to think about writing the solution to the problem in C++, not in a specific
machine language.
Programmers have a variety of tools available to enhance the software development process. Some
common tools include:

Editors. An editor allows the user to enter the program source code and save it to files. Most pro-
gramming editors increase programmer productivity by using colors to highlight language features.
The syntax of a language refers to the way pieces of the language are arranged to make well-formed
sentences. To illustrate, the sentence

The tall boy runs quickly to the door.

uses proper English syntax. By comparison, the sentence

Boy the tall runs door to quickly the.

is not correct syntactically. It uses the same words as the original sentence, but their arrangement
does not follow the rules of English.
Similarly, programmers must follow strict syntax rules to create well-formed computer programs.
Only well-formed programs are acceptable and can be compiled and executed. Some syntax-aware
editors can use colors or other special annotations to alert programmers of syntax errors before the
program is compiled.

2015 Richard L. Halterman Draft date: January 18, 2015

1.2. DEVELOPMENT TOOLS 4

Concept of
(Design
problem
program logic)
solution

Editor (Edit) Programmers

responsibility

#include <io

Source code
using namespace
std;
int main()
{
srand(23);
int n;
n = rand();

Library proc(n);

declarations
(source code)

Preprocessor
istream cin;
ostream cout;
int rand();
void sand();
(Preprocess)
typedef unsigned U
#define NULL (0)

#include <io

Enhanced using namespace

std;
int main()
{

source code srand(23);

int n;
n = rand();
proc(n);

Automated
Compiler (Compile)
by tools

Object code 101100010101

000010001100
1100001111010100
0011101101110011
1000000010000110
0111000000111111
1100111011001001

Pre-compiled 0000100001111000
0001110111101101
1101111011111010

libraries
(object code)
11000011110
00111011011
1000000010000110
0111000000111111
1100111011001001
Linker (Link)
0000100001111000
0001110111101101
1101111011111010

101100010101
000010001100

Executable 1100001111010100
0011101101110011
1000000010000110
0111000000111111

program 1100111011001001
0000100001111000
0001110111101101
1101111011111010

Figure 1.1: Source code to target code sequence

Compilers. A compiler translates the source code to target code. The target code may be the machine
language for a particular platform or embedded device. The target code could be another source
language; for example, the earliest C++ compiler translated C++ into C, another higher-level language.
The resulting C code was then processed by a C compiler to produce an executable program. C++
compilers today translate C++ directly into machine language.
The complete set of build tools for C++ includes a preprocessor, compiler, and linker:

Preprocessoradds to or modifies the contents of the source file before the compiler begins
processing the code. We use the services of the preprocessor mainly to #include information
about library routines our programs use.
Compilertranslates C++ source code to machine code.
Linkercombines the compiler-generated machine code with precompiled library code or
compiled code from other sources to make a complete executable program. Most compiled
C++ code is incapable of running by itself and needs some additional machine code to make a
complete executable program. The missing machine code has been precompiled and stored in
a repository of code called a library. A program called a linker combines the programmers
compiled code and the library code to make a complete program.

2015 Richard L. Halterman Draft date: January 18, 2015

1.3. LEARNING PROGRAMMING WITH C++ 5

We generally do not think about the preprocessor, compiler, and linker working as three separate
programs (although they do); the tools we use make it appear as only one process is taking place:
translating our source code to an executable program.

Debuggers. A debugger allows a programmer to more easily trace a programs execution in order
to locate and correct errors in the programs implementation. With a debugger, a developer can
simultaneously run a program and see which line in the source code is responsible for the programs
current actions. The programmer can watch the values of variables and other program elements to see
if their values change as expected. Debuggers are valuable for locating errors (also called bugs) and
repairing programs that contain errors. (See Section 4.6 for more information about programming
errors.)

Profilers. A profiler collects statistics about a programs execution allowing developers to tune ap-
propriate parts of the program to improve its overall performance. A profiler indicates how many
times a portion of a program is executed during a particular run, and how long that portion takes to
execute. Profilers also can be used for testing purposes to ensure all the code in a program is actually
being used somewhere during testing. This is known as coverage. It is common for software to fail
after its release because users exercise some part of the program that was not executed anytime during
testing. The main purpose of profiling is to find the parts of a program that can be improved to make
the program run faster.

The programming components of the development process are illustrated in Figure 1.1.
Many developers use integrated development environments (IDEs). An IDE includes editors, debug-
gers, and other programming aids in one comprehensive program. Examples of IDEs for C++ include
Microsofts Visual Studio 2013, the Eclipse Foundations Eclipse CDT, and Apples XCode.
Despite the plethora of tools (and tool vendors claims), the programming process for all but trivial
programs is not automatic. Good tools are valuable and certainly increase the productivity of developers,
but they cannot write software. There are no substitutes for sound logical thinking, creativity, common
sense, and, of course, programming experience.

1.3 Learning Programming with C++

Bjarne Stroustrup of AT&T Bell Labs created C++ in the mid 1980s. C++ is an extension of the programming
language C, a product of AT&T Bell Labs from the early 1970s. C was developed to write the Unix
operating system, and C is widely used for systems-level software and embedded systems development.
C++ initially provided object-oriented programming features (see Chapter 13 and Chapter 14) and later
added generic programming capabilities. C++s close relationship to C allows C++ programs to utilize a
large collection of code developed in C.
C++ is widely used in industry for commercial software development. It is an industrial strength pro-
gramming language used for developing complex systems in business, science, and engineering. Examples
of software written in C++ include Microsoft Windows 8, Microsoft Office, Mac OS X, and Adobe Creative
Suite.
In order to meet the needs of commercial software development and accomplish all that it does, C++
itself is complex. While experienced programmers can accomplish great things with C++, beginners some-
times have a difficult time with it. Professional software developers enjoy the flexible design options that
C++ permits, but beginners need more structure and fewer options so they can master simpler concepts
before moving on to more complex ones.

2015 Richard L. Halterman Draft date: January 18, 2015

1.4. SUMMARY 6

This book does not attempt to cover all the facets of the C++ programming language. Experienced
programmers should look elsewhere for books that cover C++ in much more detail. The focus here is on
introducing programming techniques and developing good habits. To that end, our approach avoids some of
the more esoteric features of C++ and concentrates on the programming basics that transfer directly to other
imperative programming languages such as Java, C#, and Python. We stick with the basics and explore
more advanced features of C++ only when necessary to handle the problem at hand.

1.4 Summary

Computers require both hardware and software to operate. Software consists of instructions that
control the hardware.
At the lowest level, the instructions for a computer program can be represented as a sequence of zeros
and ones. The pattern of zeros and ones determine the instructions performed by the processor.
Two different kinds of processors can have different machine languages.
Application software can be written largely without regard to the underlying hardware. A tool called
a compiler translates the higher-level, abstract language into the machine language required by the
hardware.
Programmers develop software using tools such as editors, compilers, debuggers, and profilers.
C++ is a higher-level programming language.
An IDE is an integrated development environmentone program that provides all the tools that
developers need to write software.

1.5 Exercises

1. What is a compiler?
2. How is compiled code different from source code?
3. What tool does a programmer use to produce C++ source code?
4. What tool(s) does a programmer use to convert C++ source code into executable machine code?
5. What does the linker do?
6. Does the linker deal with files containing source code or or machine language code?
7. What does the preprocessor do to source code?
8. List several advantages developing software in a higher-level language has over developing software
in machine language.
9. How can an IDE improve a programmers productivity?
10. Name a popular C++ IDE is used by programmers developing for Microsoft Windows.
11. Name a popular C++ IDE is used by programmers developing for Apple Mac OS X.

2015 Richard L. Halterman Draft date: January 18, 2015

 7

Chapter 2

Writing a C++ Program

Properly written C++ programs have a particular structure. The syntax must be correct, or the compiler
will generate error messages and not produce executable machine language. This chapter introduces C++
by providing some simple example programs and associated fundamental concepts. Most of the concepts
presented in this chapter are valid in many other programming languages as well. While other languages
may implement the concepts using slightly different syntax, the ideas are directly transferable to other
languages like C, Java, C#, and Ada.

2.1 General Structure of a Simple C++ Program

Listing 2.1 (simple.cpp) is one of the simplest C++ programs that does something:

Listing 2.1: simple.cpp

#include <iostream>

using namespace std;

int main() {
cout << "This is a simple C++ program!" << endl;
}

You can type the text as shown in Listing 2.1 (simple.cpp) into an editor and save it to a file named
simple.cpp. The actual name of the file is irrelevant, but the name simple accurately describes the nature
of this program. The extension .cpp is a common extension used for C++ source code.
After creating this file with a text editor and compiling it, you can run the program. The program prints
the message

This is a simple C++ program!

Listing 2.1 (simple.cpp) contains five non-blank lines of code:

2015 Richard L. Halterman Draft date: January 18, 2015

2.1. GENERAL STRUCTURE OF A SIMPLE C++ PROGRAM 8

#include <iostream>
This line is a preprocessing directive. All preprocessing directives within C++ source code begin with
a # symbol. This one directs the preprocessor to add some predefined source code to our existing
source code before the compiler begins to process it. This process is done automatically and is
invisible to us.
Here we want to use some parts of the iostream library, a collection precompiled C++ code that C++
programs (like ours) can use. The iostream library contains routines that handle input and output
(I/O) that include functions such as printing to the display, getting user input from the keyboard, and
dealing with files.
Two items used in Listing 2.1 (simple.cpp), cout and endl, are not part of the C++ language itself.
These items, along with many other things related to input and output, were developed in C++, com-
piled, and stored in the iostream library. The compiler needs to be aware of these iostream
items so it can compile our program. The #include directive specifies a file, called a header, that
contains the specifications for the library code. The compiler checks how we use cout and endl
within our code against the specifications in the <iostream> header to ensure that we are using
the library code correctly.
Most of the programs we write use this #include <iostream> directive, and some programs
we will write in the future will #include other headers as well.

using namespace std;

The two items our program needs to display a message on the screen, cout and endl, have longer
names: std::cout and std::endl. This using namespace std directive allows us to
omit the std:: prefix and use their shorter names. This directive is optional, but if we omit it, we
must use the longer names. Listing 2.2 (simple2.cpp) shows how the longer names are used. The
name std stands for standard, and the using namespace std line indicates that some of the
names we use in our program are part of the so-called standard namespace.

int main() {
This specifies the real beginning of our program. Here we are declaring a function named main. All
C++ programs must contain this function to be executable. Details about the meaning of int and
the parentheses will appear in later chapters. More general information about functions appear in
Chapter 8 and Chapter 9.
The opening curly brace at the end of the line marks the beginning of the body of a function. The
body of a function contains the statements the function is to execute.

cout << "This is a simple C++ program!"<< endl;

The body of our main function contains only one statement. This statement directs the executing
program to print the message This is a simple C++ program! on the screen. A statement is the
fundamental unit of execution in a C++ program. Functions contain statements that the compiler
translates into executable machine language instructions. C++ has a variety of different kinds of
statements, and the chapters that follow explore these various kinds of statements. All statements in
C++ end with a semicolon (;). A more detailed explanation of this statement appears below.

}
The closing curly brace marks the end of the body of a function. Both the open curly brace and close
curly brace are required for every function definition.

2015 Richard L. Halterman Draft date: January 18, 2015

2.2. EDITING, COMPILING, AND RUNNING THE PROGRAM 9

Note which lines in the program end with a semicolon (;) and which do not. Do
not put a semicolon after the #include preprocessor directive. Do not put a
semicolon on the line containing main, and do not put semicolons after the curly
braces.

2.2 Editing, Compiling, and Running the Program

C++ programmers have two options for C++ development environments. One option involves a command-
line environment with a collection of independent tools. The other option is to use an IDE (see Section 1.2)
which combines all the tools into a convenient package. Visual Studio is the dominant IDE on the Microsoft
Windows platform, and Apple Mac developers often use the XCode IDE. Appendix A provides an overview
of how to use the Visual Studio 2013 IDE to develop a simple C++ program.
The myriad of features and configuration options in these powerful IDEs can be bewildering to those
learning how to program. In a command-line environment the programmer needs only type a few simple
commands into a console window to edit, compile, and execute programs. Some developers prefer the
simplicity and flexibility of command-line build environments, especially for less complex projects.
One prominent command-line build system is the GNU Compiler Collection (http://gcc.gnu.org),
or GCC for short. The GCC C++ compiler, called g++, is one of most C++ standards conforming compilers
available. The GCC C++ compiler toolset is available for the Microsoft Windows, Apple Mac, and Linux
platforms, and it is a free, open-source software project with a world-wide development team. Appendix C
provides an overview of how to use the GCC C++ compiler.
Visual Studio and XCode offer command line development options as well. Appendix B provides an
overview of the Visual Studio command line development process.

2.3 Variations of our simple program

The two items Listing 2.1 (simple.cpp) needed to display a message on the screen, cout and endl, have
longer names: std::cout and std::endl. The using namespace std directive allows us to
omit the std:: prefixes and use their shorter names. This directive is optional, but if it is omitted, the
longer names are required. For example, Listing 2.2 (simple2.cpp) shows an alternative way of writing
Listing 2.1 (simple.cpp).

Listing 2.2: simple2.cpp

#include <iostream>

int main() {
std::cout << "This is a simple C++ program!" << std::endl;
}

Listing 2.3 (simple3.cpp) shows another way to use the shorter names for cout and endl within a C++
program.

Listing 2.3: simple3.cpp

2015 Richard L. Halterman Draft date: January 18, 2015

2.3. VARIATIONS OF OUR SIMPLE PROGRAM 10

#include <iostream>

using std::cout;
using std::endl;

int main() {
cout << "This is a simple C++ program!" << endl;
}

We generally will not use the third approach, although you will encounter it in published C++ code. The
compiler will generate the same machine language code for all three versions. We generally will write
programs that use the using namespace directive and, therefore, use the shorter names.
The statement in the main function in any of the three versions of our program uses the services of an
object called std::cout. The std::cout object prints text on the computers screen. The text of the
message as it appears in the C++ source code is called a string, for string of characters. Strings are enclosed
within quotation marks ("). The symbols << make up the insertion operator. You can think of the message
to be printed as being inserted into the cout object. The cout object represents the output stream; that
is, text that the program prints to the console window. The endl word means the end of line of printed
text, and it causes the cursor to move down to the next line so any subsequent output will appear on the
next line. If you read the statement from left to right, the cout object, which is responsible for displaying
text on the screen, first receives the text to print and then receives the end-of-line directive to move to the
next line.
For simplicity, well refer to this type of statement as a print statement, even though the word print does
not appear anywhere in the statement.
With minor exceptions, any statement in C++ must appear within a function definition. Our single print
statement appears within the function named main.
Any function, including main, may contain multiple statements. In Listing 2.4 (arrow.cpp), six print
statements draw an arrow on the screen:

Listing 2.4: arrow.cpp

#include <iostream>

using namespace std;

int main() {
cout << " * " << endl;
cout << " *** " << endl;
cout << " ***** " << endl;
cout << " * " << endl;
cout << " * " << endl;
cout << " * " << endl;
}

The output of Listing 2.4 (arrow.cpp) is

2015 Richard L. Halterman Draft date: January 18, 2015

2.3. VARIATIONS OF OUR SIMPLE PROGRAM 11

*
***
*****
*
*
*

Each print statement draws a horizontal slice of the arrow. The six statements
cout << " * " << endl;
cout << " *** " << endl;
cout << " ***** " << endl;
cout << " * " << endl;
cout << " * " << endl;
cout << " * " << endl;

constitute the body of the main function. The body consists of all the statements between the open curly
brace ({) and close curly brace (}). We say that the curly braces delimit the body of the function. The word
delimit means to determine the boundaries or limits of something. The { symbol determines the beginning
of the functions body, and the } symbol specifies the end of the functions body.
We can rewrite Listing 2.4 (arrow.cpp) to achieve the same effect with only one long print statement as
Listing 2.5 (arrow2.cpp) shows.

Listing 2.5: arrow2.cpp

#include <iostream>

using namespace std;

int main() {
cout << " * " << endl
<< " *** " << endl
<< " ***** " << endl
<< " * " << endl
<< " * " << endl
<< " * " << endl;
}

At first, Listing 2.4 (arrow.cpp) and Listing 2.5 (arrow2.cpp) may appear to be identical, but upon closer
inspection of this new program we see that cout appears only once within main, and only one semicolon
(;) appears within main. Since semicolons in C++ terminate statements, there really is only one statement.
Notice that a single statement can be spread out over several lines. The statement within main appearing
as
cout << " * " << endl
<< " *** " << endl
<< " ***** " << endl
<< " * " << endl
<< " * " << endl
<< " * " << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

2.4. TEMPLATE FOR SIMPLE C++ PROGRAMS 12

could have just as easily been written as

cout << " * " << endl << " *** "
<< endl << " ***** " << endl
<< " * " << endl
<< " * " << endl << " * " << endl;

but the first way of expressing it better portrays how the output will appear. Read this second version
carefully to convince yourself that the printed pieces will indeed flow to the cout printing object in the
proper sequence to produce the same picture of the arrow.

Consider the mistake of putting semicolons at the end of each of the lines in the
one statement version:
cout << " * " << endl;
<< " *** " << endl;
<< " ***** " << endl;
<< " * " << endl;
<< " * " << endl;
<< " * " << endl;
If we put this code fragment in main, the program will not compile. The reason
is simplethe semicolon at the end of the first line terminates the statement on
that line. The compiler expects a new statement on the next line, but
<< " *** " << endl;
is not a complete legal C++ statement since the << operator is missing the cout
object. The string " *** " and the end-of-line marker has nothing to flow
into.

Listing 2.6 (empty.cpp) is even simpler than Listing 2.1 (simple.cpp).

Listing 2.6: empty.cpp

int main() {
}

Since Listing 2.6 (empty.cpp) uses neither the cout object nor endl, it does not need the #include and
using directives. While it is legal and sometimes even useful in C++ to write functions with empty bodies,
such functions will do nothing when they execute. Listing 2.6 (empty.cpp) with its empty main function
is, therefore, truly the simplest executable C++ program we can write, but it does nothing when we runs it!
In general, a C++ program may contain multiple functions, but we defer such generality until Chapter 9.
For now, we will restrict our attention to programs with only a main function.

2.4 Template for simple C++ programs

For our immediate purposes all the programs we write will have the form shown in Figure 2.1.

2015 Richard L. Halterman Draft date: January 18, 2015

2.5. SUMMARY 13

#include <iostream>

using namespace std;

int main() {

program statements

Figure 2.1: The general structure of a very simple C++ program.

Our programs generally will print something, so we need the #include directive. Since we will use
the short names for the printing objects, we also include the using namespace directive. The main
function definition is required for an executable program, and we will fill its body with statements that
make our program do as we wish. Later, our programs will become more sophisticated, and we will need
to augment this simple template.

2.5 Summary

Any programs that print to the screen must use the #include directive to so that the program can
make use of the precompiled iostream library code to do I/O (input and output).
All preprocessor directives begin with the # symbol.
Messages can be printed on the screen by sending them to the std::cout object using the <<
operator.
The << operator used in conjunction with the cout stream object is known as the insertion operator.
The end of a printed line is signified by sending std::endl to the std::cout object.
The shorter names cout and endl can be used if the using namespace std appears at the top
of the programs source code.
An executable C++ program must contain a function named main.
Functions contain statements that are executed by the computer when the program runs. (The program
must be compiled for the computer to be able to execute the statements.)
A functions body consists of all the statements inside the delimiting curly braces ({}).
A functions body may be empty; however, such a function performs no useful activity when it
executes.
Statements are terminated with semicolons (;).
Within the Visual Studio IDE a program is developed within a project.

2015 Richard L. Halterman Draft date: January 18, 2015

2.6. EXERCISES 14

Visual Studios command prompt environment can be used to develop simple C++ programs without
the overhead of creating a project.
In a printing statement the characters within the quotation marks (") are printed literally on the screen.

2.6 Exercises

1. What preprocessor directive is necessary to use statements with the std::cout printing stream
object?
2. What statement allows the short names cout and endl to be used instead of std::cout and
std::endl?

3. What does the name std stand for?

4. All C++ programs must have a function named what?
5. The body of main is enclosed within what symbols?

6. What operator directs information to the std::cout output stream?

7. Write a C++ program that prints your name in the console window.
8. Write a C++ program that prints your first and last name in the console window. Your first name
should appear on one line, and your last name appear on the next line.

9. What other files must you distribute with your executable file so that your program will run on a
Windows PC without Visual Studio installed?
10. Can a single statement in C++ span multiple lines in the source code?

2015 Richard L. Halterman Draft date: January 18, 2015

 15

Chapter 3

Values and Variables

In this chapter we explore some building blocks that are used to develop C++ programs. We experiment
with the following concepts:

numeric values
variables
declarations
assignment
identifiers
reserved words

In the next chapter we will revisit some of these concepts in the context of other data types.

3.1 Integer Values

The number four (4) is an example of a numeric value. In mathematics, 4 is an integer value. Integers
are whole numbers, which means they have no fractional parts, and an integer can be positive, negative, or
zero. Examples of integers include 4, 19, 0, and 1005. In contrast, 4.5 is not an integer, since it is not a
whole number.
C++ supports a number of numeric and non-numeric values. In particular, C++ programs can use integer
values. It is easy to write a C++ program that prints the number four, as Listing 3.1 (number4.cpp) shows.

Listing 3.1: number4.cpp

#include <iostream>

using namespace std;

int main() {
cout << 4 << endl;
}

2015 Richard L. Halterman Draft date: January 18, 2015

3.1. INTEGER VALUES 16

Notice that unlike the programs we saw earlier, Listing 3.1 (number4.cpp) does not use quotation marks
("). Compare Listing 3.1 (number4.cpp) to Listing 3.2 (number4-alt.cpp).

Listing 3.2: number4-alt.cpp

#include <iostream>

using namespace std;

int main() {
cout << "4" << endl;
}

Both programs behave identically, but Listing 3.1 (number4.cpp) prints the value of the number four, while
Listing 3.2 (number4-alt.cpp) prints a message containing the digit four. The distinction here seems unim-
portant, but we will see in Section 3.2 that the presence or absence of the quotes can make a big difference
in the output.
In C++ source code, integers may not contain commas. This means we must write the number two
thousand, four hundred sixty-eight as 2468, not 2,468. In mathematics, integers are unbounded; said
another way, the set of mathematical integers is infinite. In C++ the range of integers is limited because all
computers have a finite amount of memory. The exact range of integers supported depends on the computer
system and particular C++ compiler. C++ on most 32-bit computer systems can represent integers in the
range 2,147,483,648 to +2,147,483,647.
What happens if you exceed the range of C++ integers? Try Listing 3.3 (exceed.cpp) on your system.

Listing 3.3: exceed.cpp

#include <iostream>

using namespace std;

int main() {
cout << -3000000000 << endl;
}

Negative three billion is too large for 32-bit integers, however, and the programs output is obviously wrong:

1294967296

The number printed was not even negative! Most C++ compilers will issue a warning about this statement.
secrefsec:expressionsarithmetic.errors explores errors vs. warnings in more detail. If the compiler finds an
error in the source, it will not generate the executable code. A warning indicates a potential problem and
does not stop the compiler from producing an executable program. Here we see that the programmer should
heed this warning because the programs execution produces meaningless output.
This limited range of values is common among programming languages since each number is stored in
a fixed amount of memory. Larger numbers require more storage in memory. In order to model the infinite
set of mathematical integers an infinite amount of memory would be needed! As we will see later, C++
supports an integer type with a greater range. Section 4.8.1 provides some details about the implementation
of C++ integers.

2015 Richard L. Halterman Draft date: January 18, 2015

3.2. VARIABLES AND ASSIGNMENT 17

3.2 Variables and Assignment

In algebra, variables are used to represent numbers. The same is true in C++, except C++ variables also can
represent values other than numbers. Listing 3.4 (variable.cpp) uses a variable to store an integer value and
then prints the value of the variable.

Listing 3.4: variable.cpp

#include <iostream>

using namespace std;

int main() {
int x;
x = 10;
cout << x << endl;
}

The main function in Listing 3.4 (variable.cpp) contains three statements:

int x;
This is a declaration statement. All variables in a C++ program must be declared. A declaration
specifies the type of a variable. The word int indicates that the variable is an integer. The name
of the integer variable is x. We say that variable x has type int. C++ supports types other than
integers, and some types require more or less space in the computers memory. The compiler uses
the declaration to reserve the proper amount of memory to store the variables value. The declaration
enables the compiler to verify the programmer is using the variable properly within the program; for
example, we will see that integers can be added together just like in mathematics. For some other
data types, however, addition is not possible and so is not allowed. The compiler can ensure that a
variable involved in an addition operation is compatible with addition. It can report an error if it is
not.
The compiler will issue an error if a programmer attempts to use an undeclared variable. The com-
piler cannot deduce the storage requirements and cannot verify the variables proper usage if it not
declared. Once declared, a particular variable cannot be redeclared in the same context. A variable
may not change its type during its lifetime.

x = 10;
This is an assignment statement. An assignment statement associates a value with a variable. The
key to an assignment statement is the symbol = which is known as the assignment operator. Here the
value 10 is being assigned to the variable x. This means the value 10 will be stored in the memory
location the compiler has reserved for the variable named x. We need not be concerned about where
the variable is stored in memory; the compiler takes care of that detail.
After we declare a variable we may assign and reassign it as often as necessary.

cout << x << endl;

This statement prints the variable xs current value.

2015 Richard L. Halterman Draft date: January 18, 2015

3.2. VARIABLES AND ASSIGNMENT 18

Note that the lack of quotation marks here is very important. If x has the value
10, the statement
cout << x << endl;
prints 10, the value of the variable x, but the statement
cout << "x" << endl;
prints x, the message containing the single letter x.

The meaning of the assignment operator (=) is different from equality in mathematics. In mathematics,
= asserts that the expression on its left is equal to the expression on its right. In C++, = makes the variable
on its left take on the value of the expression on its right. It is best to read x = 5 as x is assigned the
value 5, or x gets the value 5. This distinction is important since in mathematics equality is symmetric:
if x = 5, we know 5 = x. In C++, this symmetry does not exist; the statement
5 = x;

attempts to reassign the value of the literal integer value 5, but this cannot be done, because 5 is always 5
and cannot be changed. Such a statement will produce a compiler error:

error C2106: = : left operand must be l-value

Variables can be reassigned different values as needed, as Listing 3.5 (multipleassignment.cpp) shows.

Listing 3.5: multipleassignment.cpp

#include <iostream>

using namespace std;

int main() {
int x;
x = 10;
cout << x << endl;
x = 20;
cout << x << endl;
x = 30;
cout << x << endl;
}

Observe the each print statement in Listing 3.5 (multipleassignment.cpp) is identical, but when the program
runs the print statements produce different results.
A variable may be given a value at the time of its declaration; for example, Listing 3.6 (variable-init.cpp)
is a variation of Listing 3.4 (variable.cpp).

Listing 3.6: variable-init.cpp

#include <iostream>

using namespace std;

2015 Richard L. Halterman Draft date: January 18, 2015

3.2. VARIABLES AND ASSIGNMENT 19

int main() {
int x = 10;
cout << x << endl;
}

Notice that in Listing 3.6 (variable-init.cpp) the declaration and assignment of the variable x is performed in
one statement instead of two. This combined declaration and immediate assignment is called initialization.
C++ supports another syntax for initializing variables as shown in Listing 3.7 (alt-variable-init.cpp).

Listing 3.7: alt-variable-init.cpp

#include <iostream>

using namespace std;

int main() {
int x{10};
cout << x << endl;
}

This alternate form is not commonly used for simple variables, but it necessary for initializing more com-
plicated kinds of variables called objects. We introduce objects in Chapter 13 and Chapter 14.
Multiple variables of the same type can be declared and, if desired, initialized in a single statement. The
following statements declare three variables in one declaration statement:
int x, y, z;

The following statement declares three integer variables and initializes two of them:
int x = 0, y, z = 5;

Here ys value is undefined. The declarations may be split up into multiple declaration statements:
int x = 0;
int y;
int z = 5;

In the case of multiple declaration statements the type name (here int) must appear in each statement.
The compiler maps a variable to a location in the computers memory. We can visualize a variable and
its corresponding memory location as a box as shown in Figure 3.1.

a
5
Figure 3.1: Representing a variable and its memory location as a box

2015 Richard L. Halterman Draft date: January 18, 2015

3.3. IDENTIFIERS 20

We name the box with the variables name. Figure 3.2 shows how the following sequence of C++ code
affects memory.
int a, b;
a = 2;
b = 5;
a = b;
b = 4;

a b
int a, b; ? ?
a b
a = 2; 2 ?
a b
b = 5; 2 5

a b
a = b; 5 5
a b
b = 4; 5 4

Figure 3.2: How memory changes during variable assignment

Importantly, the statement

a = b;

does not mean a and b refer to the same box (memory location). After this statement a and b still refer
to separate boxes (memory locations). It simply means the value stored in bs box (memory location) has
been copied to as box (memory location). a and b remain distinct boxes (memory locations). The original
value found in as box is overwritten when the contents of bs box are copied into a. After the assignment
of b to a, the reassignment of b to 4 does not affect a.

3.3 Identifiers

While mathematicians are content with giving their variables one-letter names like x, programmers should
use longer, more descriptive variable names. Names such as altitude, sum, and user_name are much

2015 Richard L. Halterman Draft date: January 18, 2015

3.3. IDENTIFIERS 21

better than the equally permissible a, s, and u. A variables name should be related to its purpose within the
program. Good variable names make programs more readable by humans. Since programs often contain
many variables, well-chosen variable names can render an otherwise obscure collection of symbols more
understandable.
C++ has strict rules for variable names. A variable name is one example of an identifier. An identifier
is a word used to name things. One of the things an identifier can name is a variable. We will see in later
chapters that identifiers name other things such as functions and classes. Identifiers have the following
form:

Identifiers must contain at least one character.

The first character must be an alphabetic letter (upper or lower case) or the underscore
ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz_
The remaining characters (if any) may be alphabetic characters (upper or lower case), the underscore,
or a digit
ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz_0123456789
No other characters (including spaces) are permitted in identifiers.
A reserved word cannot be used as an identifier (see Table 3.1).

Here are some examples of valid and invalid identifiers:

All of the following words are valid identifiers and so qualify as variable names: x, x2, total,
port_22, and FLAG.
None of the following words are valid identifiers: sub-total (dash is not a legal symbol in an
identifier), first entry (space is not a legal symbol in an identifier), 4all (begins with a digit),
#2 (pound sign is not a legal symbol in an identifier), and class (class is a reserved word).

C++ reserves a number of words for special use that could otherwise be used as identifiers. Called
reserved words or keywords, these words are special and are used to define the structure of C++ programs
and statements. Table 3.1 lists all the C++ reserved words.
The purposes of many of these reserved words are revealed throughout this book.
You may not use any of the reserved words in Table 3.1 as identifiers. Fortunately, if you accidentally
attempt to use one of the reserved words in a program as a variable name, the compiler will issue an error
(see Section 4.6 for more on compiler errors).
In Listing 2.1 (simple.cpp) we used several reserved words: using, namespace, and int. Notice
that include, cout, endl, and main are not reserved words.
Some programming languages do not require programmers to declare variables before they are used;
the type of a variable is determined by how the variable is used. Some languages allow the same variable
to assume different types as its use differs in different parts of a program. Such languages are known as
dynamically-typed languages. C++ is a statically-typed language. In a statically-typed language, the type of
a variable must be explicitly specified before it is used by statements in a program. While the requirement
to declare all variables may initially seem like a minor annoyance, it offers several advantages:

When variables must be declared, the compiler can catch typographical errors that dynamically-typed
languages cannot detect. For example, consider the following section of code:

2015 Richard L. Halterman Draft date: January 18, 2015

3.3. IDENTIFIERS 22

alignas decltype namespace struct

alignof default new switch
and delete noexcept template
and_eq double not this
asm do not_eq thread_local
auto dynamic_cast nullptr throw
bitand else operator true
bitor enum or try
bool explicit or_eq typedef
break export private typeid
case extern protected typename
catch false public union
char float register unsigned
char16_t for reinterpret_cast using
char32_t friend return virtual
class goto short void
compl if signed volatile
const inline sizeof wchar_t
constexpr int static while
const_cast long static_assert xor
continue mutable static_cast xor_eq

Table 3.1: C++ reserved words. C++ reserves these words for specific purposes in program construction. None of the
words in this list may be used as an identifier; thus, you may not use any of these words to name a variable.

int ZERO;
ZER0 = 1;

The identifier in the first line ends with a capital Oh. In the second line, the identifier ends with the
digit zero. The distinction may be difficult or impossible to see in a particular editor or printout of
the code. A C++ compiler would immediately detect the typo in the second statement, since ZER0
(last letter a zero) has not been declared. A dynamically-typed language would create two variables:
ZERO and ZER0.
When variables must be declared, the compiler can catch invalid operations. For example, a variable
may be declared to be of type int, but the programmer may accidentally assign a non-numeric
value to the variable. In a dynamically-typed language, the variable would silently change its type
introducing an error into the program. In C++, the compiler would report the improper assignment as
error, since once declared a C++ variable cannot change its type.
Ideally, requiring the programmer to declare variables forces the programmer to plan ahead and think
more carefully about the variables a program might require. The purpose of a variable is tied to its
type, so the programmer must have a clear notion of the variables purpose before declaring it. When
variable declarations are not required, a programmer can make up variables as needed as the code is
written. The programmer need not do the simple double check of the variables purpose that writing
the variables declaration requires. While declaring the type of a variable specifies its purpose in only
a very limited way, any opportunity to catch such errors is beneficial.
Statically-typed languages are generally more efficient than dynamically-typed languages. The com-
piler knows how much storage a variable requires based on its type. The space for that variables
value will not change over the life of the variable, since its type cannot change. In a dynamically

2015 Richard L. Halterman Draft date: January 18, 2015

3.4. ADDITIONAL INTEGER TYPES 23

typed language that allows a variable to change its type, if a variables type changes during program
execution, the storage it requires may change also, so memory for that variable must be allocated
elsewhere to hold the different type. This memory reallocation at run time slows down the programs
execution.

C++ is a case-sensitive language. This means that capitalization matters. if is a reserved word, but
none of If, IF, or iF are reserved words. Identifiers are case sensitive also; the variable called Name is
different from the variable called name.
Since it can be confusing to human readers, you should not distinguish variables merely by names that
differ in capitalization. For the same reason, it is considered poor practice to give a variable the same name
as a reserved word with one or more of its letters capitalized.

3.4 Additional Integer Types

C++ supports several other integer types. The type short int, which may be written as just short,
represents integers that may occupy fewer bytes of memory than the int type. If the short type occupies
less memory, it necessarily must represent a smaller range of integer values than the int type. The C++
standard does not require the short type to be smaller than the int type; in fact, they may represent the
same set of integer values. The long int type, which may be written as just long, may occupy more
storage than the int type and thus be able to represent a larger range of values. Again, the standard does
not require the long type to be bigger then the int type. Finally, the long long int type, or just
long long, may be larger than a long. The C++ standard guarantees the following relative ranges of
values hold:

short int int long int long long int

On a small embedded device, for example, all of these types may occupy the exact same amount of memory
and, thus, there would be no advantage of using one type over another. On most systems, however, there
will some differences in the ranges.
C++ provides integer-like types that exclude negative numbers. These types include the word unsigned
in their names, meaning they do not allow a negative sign. The unsigned types come in various potential
sizes in the same manner as the signed types. The C++ standard guarantees the following relative ranges of
unsigned values:

unsigned short unsigned unsigned long unsigned long long

Table 3.2 lists the differences among the signed and unsigned integer types in Visual C++. Notice that
the corresponding signed and unsigned integer times occupy the same amount of memory. As a result,
the unsigned types provide twice the range of positive values available to their signed counterparts. For
applications that do not require negative numbers, the unsigned type may be a more appropriate option.
Within the source code, any unadorned numerical literal without a decimal point is interpreted as an
int literal; for example, in the statement
int x = 4456;

the literal value 4456 is an int. In order to represent 4456 as an long, append an L, as in
long x = 4456L;

2015 Richard L. Halterman Draft date: January 18, 2015

3.5. FLOATING-POINT TYPES 24

Type Name Short Name Storage Smallest Magnitude Largest Magnitude

short int short 2 bytes 32, 768 32, 767
int int 4 bytes 2, 147, 483, 648 2, 147, 483, 647
long int long 4 bytes 2, 147, 483, 648 2, 147, 483, 647
long long int long long 8 bytes 9, 223, 372, 036, 854, 775, 808 9, 223, 372, 036, 854, 775, 807
unsigned short unsigned short 2 bytes 0 65, 535
unsigned int unsigned 4 bytes 0 4, 294, 967, 295
unsigned long int unsigned long 4 bytes 0 4, 294, 967, 295
unsigned long long int unsigned long long 8 bytes 0 18, 446, 744, 073, 709, 551, 615

Table 3.2: Characteristics of Visual C++ Integer Types

Type Storage Smallest Magnitude Largest Magnitude Minimum Precision

float 4 bytes 1.17549 1038 3.40282 10+38 6 digits
double 8 bytes 2.22507 10308 1.79769 10+308 15 digits
long double 8 bytes 2.22507 10308 1.79769 10+308 15 digits

Table 3.3: Characteristics of Floating-point Numbers on 32-bit Computer Systems

C++ also permits the lower-case l (elle), as in

long x = 4456l;

but you should avoid it since on many display and printer fonts it looks too much like the digit 1 (one).
Use the LL suffix for long long literals. The suffixes for the unsigned integers are u (unsigned), us
(unsigned short), uL (unsigned long), and uLL (unsigned long long). The capitalization
is unimportant, although capital Ls are preferred.

Within C++ source code all integer literals are int values unless an L or l is
appended to the end of the number; for example, 2 is an int literal, while 2L is
a long literal.

3.5 Floating-point Types

Many computational tasks require numbers that have fractional parts. For example, the formula from math-
ematics to compute the area of a circle given the circles radius, involves the value , which is approximately
3.14159. C++ supports such non-integer numbers, and they are called floating-point numbers. The name
comes from the fact that during mathematical calculations the decimal point can move or float to various
positions within the number to maintain the proper number of significant digits. The types float and
double represent different types of floating-point numbers. The type double is used more often, since it
stands for double-precision floating-point, and it can represent a wider range of values with more digits of
precision. The float type represents single-precision floating-point values that are less precise. Table 3.3
provides some information about floating-point values as commonly implemented on 32-bit computer sys-
tems. Floating point numbers can be both positive and negative.
As you can see from Table 3.3, doubles provide more precision at the cost of using more memory.
Listing 3.8 (pi-print.cpp) prints an approximation of the mathematical value .

2015 Richard L. Halterman Draft date: January 18, 2015

3.5. FLOATING-POINT TYPES 25

Listing 3.8: pi-print.cpp

#include <iostream>

using namespace std;

int main() {
double pi = 3.14159;
cout << "Pi = " << pi << endl;
cout << "or " << 3.14 << " for short" << endl;
}

The first line in Listing 3.8 (pi-print.cpp) declares a variable named pi and assigns it a value. The second
line in Listing 3.8 (pi-print.cpp) prints the value of the variable pi, and the third line prints a literal value.
Any literal numeric value with a decimal point in a C++ program automatically has the type double, so
3.14

has type double. To make a literal floating-point value a float, you must append an f or F to the
number, as in
3.14f

(The f or F suffix is used with literal values only; you cannot change a double variable into a float
variable by appending an f. Attempting to do so would change the name of the variable!)

All floating-point literals are double values unless an f or F is appended to the

end of the number; for example, 2.0 is a double literal, while 2.0f is a float
literal.

Floating-point numbers are an approximation of mathematical real numbers. As in the case of the
int data type, the range of floating-point numbers is limited, since each value requires a fixed amount of
memory. In some ways, though, ints are very different from doubles. Any integer within the range of
the int data type can be represented exactly. This is not true for the floating-point types. Consider the
real number . Since contains an infinite number of digits, a floating-point number with finite precision
can only approximate its value. Since the number of digits available is limited, even numbers with a finite
number of digits have no exact representation; for example, the number 23.3123400654033989 contains too
many digits for the double type and must be approximated as 23.3023498654034. Section 4.8.2 contains
more information about the consequences of the inexact nature of floating-point numbers.
We can express floating-point numbers in scientific notation. Since most programming editors do not
provide superscripting and special symbols like , C++ slightly alters the normal scientific notation. The
number 6.022 1023 is written 6.022e23. The number to the left of the e (we can use capital E as well)
is the mantissa, and the number to the right of the e is the exponent of 10. As another example, 5.1 104
is expressed in C++ as -5.1e-4. Listing 3.9 (scientificnotation.cpp) prints some scientific constants using
scientific notation.

Listing 3.9: scientificnotation.cpp

#include <iostream>

2015 Richard L. Halterman Draft date: January 18, 2015

3.6. CONSTANTS 26

using namespace std;

int main() {
double avogadros_number = 6.022e23, c = 2.998e8;
cout << "Avogadro's number = " << avogadros_number << endl;
cout << "Speed of light = " << c << endl;
}

Section 4.8.2 provides some insight into the implementation of C++ floating-point values and explains how
internally all floating-point numbers are stored in exponential notation with a mantissa and exponent.

3.6 Constants

In Listing 3.9 (scientificnotation.cpp), Avogadros number and the speed of light are scientific constants;
that is, to the degree of precision to which they have been measured and/or calculated, they do not vary. C++
supports named constants. Constants are declared like variables with the addition of the const keyword:

const double PI = 3.14159;

Once declared and initialized, a constant can be used like a variable in all but one waya constant may not
be reassigned. It is illegal for a constant to appear on the left side of the assignment operator (=) outside its
declaration statement. A subsequent statement like

PI = 2.5;

would cause the compiler to issue an error message:

error C3892: PI : you cannot assign to a variable that is const

and fail to compile the program. Since the scientific constants do not change, Listing 3.10 (const.cpp) is a
better version of Listing 3.9 (scientificnotation.cpp).

Listing 3.10: const.cpp

#include <iostream>

using namespace std;

int main() {
const double avogadros_number = 6.022e23, c = 2.998e8;
cout << "Avogadro's number = " << avogadros_number << endl;
cout << "Speed of light = " << c << endl;
}

Since it is illegal to assign a constant outside of its declaration statement, all constants must initialized
where they are declared.
By convention, C++ programmers generally express constant names in all capital letters; in this way,
within the source code a human reader can distinguish a constant quickly from a variable.

2015 Richard L. Halterman Draft date: January 18, 2015

3.7. OTHER NUMERIC TYPES 27

3.7 Other Numeric Types

C++ supports several other numeric data types:

long inttypically provides integers with a greater range than the int type; its abbreviated
name is long. It is guaranteed to provide a range of integer values at least as large as the int type.
An integer literal with a L suffix, as in 19L, has type long. A lower case elle (l) is allowed as a
suffix as well, but you should not use it because it is difficult for human readers to distinguish between
l (lower case elle) and 1 (digit one). (The L suffix is used with literal values only; you cannot change
an int variable into a long by appending an L. Attempting to do so would change the name of the
variable!)

short inttypically provides integers with a smaller range than the int type; its abbreviated
name is short. It is guaranteed that the range of ints is at least as big as the range of shorts.

unsigned intis restricted to non-negative integers; its abbreviated name is unsigned. While
the unsigned type is limited in non-negative values, it can represent twice as many positive values
as the int type. (The name int is actually the short name for signed int and int can be
written as signed.)

long doublecan extend the range and precision of the double type.

While the C++ language standard specifies minimum ranges and precision for all the numeric data types,
a particular C++ compiler may exceed the specified minimums.
C++ provides such a variety of numeric types for specialized purposes usually related to building highly
efficient programs. We will have little need to use many of these types. Our examples will use mainly
the numeric types int for integers, double for an approximation of real numbers, and, less frequently,
unsigned when non-negative integral values are needed.

3.8 Characters

The char data type is used to represent single characters: letters of the alphabet (both upper and lower
case), digits, punctuation, and control characters (like newline and tab characters). Most systems support the
American Standard Code for Information Interchange (ASCII) character set. Standard ASCII can represent
128 different characters. Table 3.4 lists the ASCII codes for various characters.
In C++ source code, characters are enclosed by single quotes ('), as in
char ch = 'A';

Standard (double) quotes (") are reserved for strings, which are composed of characters, but strings and
chars are very different. C++ strings are covered in Section D.7. The following statement would produce
a compiler error message:
ch = "A";

since a string cannot be assigned to a character variable.

Internally, chars are stored as integer values, and C++ permits assigning numeric values to char
variables and assigning characters to numeric variables. The statement

2015 Richard L. Halterman Draft date: January 18, 2015

3.8. CHARACTERS 28

0 '\0' 16 - 32 '' 48 '0' 64 @@ 80 'P' 96 '`' 112 'p'

1 - 17 - 33 '!' 49 '1' 65 'A' 81 'Q' 97 'a' 113 'q'
2 - 18 - 34 '"' 50 '2' 66 'B' 82 'R' 98 'b' 114 'r'
3 - 19 - 35 '#' 51 '3' 67 'C' 83 'S' 99 'c' 115 's'
4 - 20 - 36 '$' 52 '4' 68 'D' 84 'T' 100 'd' 116 't'
5 - 21 - 37 '%' 53 '5' 69 'E' 85 'U' 101 'e' 117 'u'
6 - 22 - 38 & 54 '6' 70 'F' 86 'V' 102 'f' 118 'v'
7 '\a' 23 - 39 '\'' 55 '7' 71 'G' 87 'W' 103 'g' 119 'w'
8 '\b' 24 - 40 '(' 56 '8' 72 'H' 88 'X' 104 'h' 120 'x'
9 '\t' 25 - 41 ')' 57 '9' 73 'I' 89 'Y' 105 'i' 121 'y'
10 '\n' 26 - 42 '*' 58 ':' 74 'J' 90 'Z' 106 'j' 122 'z'
11 - 27 - 43 '+' 59 ';' 75 'K' 91 '[' 107 'k' 123 '{'
12 '\f' 28 - 44 ',' 60 '<' 76 'L' 92 '\\' 108 'l' 124 '|'
13 '\r' 29 - 45 '-' 61 '=' 77 'M' 93 ']' 109 'm' 125 '}'
14 - 30 - 46 '.' 62 '>' 78 'N' 94 '' 110 'n' 126 ''
15 - 31 - 47 '/' 63 '?' 79 'O' 95 '_' 111 'o' 127 -

Table 3.4: ASCII codes for characters

ch = 65;

assigns a number to a char variable to show that this perfectly legal. The value 65 is the ASCII code for
the character A. If ch is printed, as in
ch = 65;
cout << ch;

the corresponding character, A, would be printed because chs declared type is char, not int or some
other numeric type.
Listing 3.11 (charexample.cpp) shows how characters can be used within a program.

Listing 3.11: charexample.cpp

#include <iostream>

using namespace std;

int main() {
char ch1, ch2;
ch1 = 65;
ch2 = 'A';
cout << ch1 << ", " << ch2 << ", " << 'A' << endl;
}

The program displays

A, A, A

The first A is printed because the statement

ch1 = 65;

assigns the ASCII code for A to ch1. The second A is printed because the statement
ch2 = 'A';

2015 Richard L. Halterman Draft date: January 18, 2015

3.8. CHARACTERS 29

assigns the literal character A to ch2. The third A is printed because the literal character 'A' is sent directly
to the output stream.
Integers and characters can be freely assigned to each other, but the range of chars is much smaller
than the range of ints, so care must be taken when assigning an int value to a char variable.
Some characters are non-printable characters. The ASCII chart lists several common non-printable
characters:

'\n'the newline character

'\r'the carriage return character

'\b'the backspace character

'\a'the alert character (causes a beep sound on many systems)

'\t'the tab character

'\f'the formfeed character

'\0'the null character (used in C strings, see Section D.7)

These special non-printable characters begin with a backslash (\) symbol. The backslash is called
an escape symbol, and it signifies that the symbol that follows has a special meaning and should not be
interpreted literally. This means the literal backslash character must be represented as two backslashes:
'\\'.
These special non-printable character codes can be embedded within strings. To embed a backslash
within a string, you must escape it; for example, the statement
cout << "C:\\Dev\\cppcode" << endl;

would print

C:\Dev\cppcode

See what this statement prints:

cout << "AB\bCD\aEF" << endl;

The following two statements behave identically on most computer systems:

cout << "End of line" << endl;
cout << "End of line\n";

On the Microsoft Windows platform, the character sequence "\r\n" (carriage return, line feed) appears
at the end of lines in text files. Under Unix and Linux, lines in text files end with '\n' (line feed). On
Apple Macintosh systems, text file lines end with the '\r' (carriage return) character. The compilers that
adhere to the C++ standard will ensure that the '\n' character in a C++ program when sent to the output
stream will produce the same results as the endl manipulator.

2015 Richard L. Halterman Draft date: January 18, 2015

3.9. ENUMERATED TYPES 30

3.9 Enumerated Types

C++ allows a programmer to create a new, very simple type and list all the possible values of that type. Such
a type is called an enumerated type, or an enumeration type. The enum keyword introduces an enumerated
type:
enum Color { Red, Orange, Yellow, Green, Blue, Violet };

Here, the new type is named Color, and a variable of type Color may assume one of the values that
appears in the list of values within the curly braces. The semicolon following the close curly brace is
required. Sometimes the enumerated type definition is formatted as
enum Color {
Red,
Orange,
Yellow,
Green,
Blue,
Violet
};

but the formatting makes no different to the compiler.

The values listed with the curly braces constitute all the values that a variable of the enumerated type
can attain. The name of each value of an enumerated type must be a valid C++ identifier (see Section 3.3).
Given the Color type defined as above, we can declare and use variables of the enum type as shown
by the following code fragment:
Color myColor;
myColor = Orange;

Here the variable myColor has our custom type Color, and its value is Orange.
It is illegal to reuse an enumerated value name within another enumerated type within the same program.
In the following code, the enumerated value Light is used in both Shade and Weight:
enum Shade { Dark, Dim, Light, Bright };
enum Weight { Light, Medium, Heavy };

These two enumerated types are incompatible because they share the value Light, and so the compiler
will issue an error.
The value names within an enum type must be unique. The convention in C++ is to captialize the first
letter of an enum type and its associated values, although the language does not enforce this convention.
An enum type is handy for representing a small number of discrete, non-numeric options. For example,
consider a program that controls the movements made by a small robot. The allowed orientations are
forward, backward, left, right, up, and down. The program could encode these movements as integers,
where 0 means left, 1 means backward, etc. While that implementation will work, it is not ideal. Integers
may assume many more values than just the six values expected. The compiler cannot ensure that an integer
variable representing a robot move will stay in the range 0...5. What if the programmer makes a mistake and
under certain rare circumstances assigns a value outside of the range 0...5? The program then will contain
an error that may result in erratic behavior by the robot. With enum types, if the programmer uses only the
named values of the enum type, the compiler will ensure that such a mistake cannot happen.

2015 Richard L. Halterman Draft date: January 18, 2015

3.10. TYPE INFERENCE WITH AUTO 31

A particular enumerated type necessarily has far fewer values than a type such as int. Imagine making
an integer enum type and having to list all of its values! (The standard 32-bit int type represents over four
billion values.) Enumerated types, therefore, are practical only for types that have a relatively small range
of values.

3.10 Type Inference with auto

C++ requires that a variable be declared before it is used. Ordinarily this means specifying the variables
type, as in
int count;
char ch;
double limit;

A variable may be initialized when it is declared:

int count = 0;
char ch = 'Z';
double limit = 100.0;

Each of the values has a type: 0 is an int, 'Z' is a char, and 0.0 is a double. The auto keyword
allows the compiler to automatically deduce the type of a variable if it is initialized when it is declared:
auto count = 0;
auto ch = 'Z';
auto limit = 100.0;

The auto keyword may not be used without an accompanying initialization; for example, the following
declaration is illegal:
auto x;

because the compiler cannot deduce xs type.

Automatic type inference is supported only by compilers that comply with the lat-
est C++11 standard. Programmers using older compilers must specify a variables
exact type during the variables declaration.

Automatic type deduction with auto is not useful to beginning C++ programmers. It is just as easy to
specify the variables type. The value of auto will become clearer when we consider some of the more
advanced features of C++ (see Section 18.3).

3.11 Summary

C++ supports several kinds of numeric values and variables.

In C++ commas cannot be used when expressing numeric literals.

2015 Richard L. Halterman Draft date: January 18, 2015

3.11. SUMMARY 32

Numbers represented on a computer have limitations based on the finite nature of computer systems.

All numeric types in C++ have limited ranges. This is because they must be stored in the computers
memory and be manipulated by the processor. The memory and processor deal with fixed-size quan-
tities, and in order to handle an unlimited range of values memory would have to be infinite.

Variables are used to store values.

All variables in C++ must be declared before they are used within the program.

How can you declare a variable named total to be of type integer?

The = operator means assignment, not mathematical equality.

A variable can be reassigned at any time.

A variable can be given an initial value at the time of its declaration.

Multiple variables can be declared in one declaration statement.

A variable represents a location in memory capable of storing a value.

The statement a = b; copies the value stored in variable b into variable a. a and b still refer to
there own memory locations.

A variable name is an example of an identifier.

The name of a variable must follow the identifier naming rules.

All identifiers must consist of at least one character. The first symbol must be an alphabetic letter or
the underscore. Remaining symbols (if any) must be alphabetic letters, the underscore, or digits.

Reserved words have special meaning within a C++ program and cannot be used as identifiers.

Descriptive variable names are preferred over one-letter names.

C++ is case sensitive; the name X is not the same as the name x.

Floating numbers approximate mathematical real numbers.

The float type represents single-precision floating-point numbers, and the double type represents
double-precision floating-point numbers.

There are many values that floating-point numbers cannot represent exactly.

The symbol f or F can be appended to a floating-point literal to signify that it is a float value
instead of a double value.

In C++ we express scientific notation literals of the form 1.0 101 as 1.0e1.0.

The qualifier const renders a variable unchangeable; it signifies a constant.

On some systems longs may have extended range compared to normal ints.

On some systems shorts may have reduced range compared to normal ints.

The unsigned type represents non-negative integers.

On some systems long doubles may have extended precision compared to doubles.

2015 Richard L. Halterman Draft date: January 18, 2015

3.12. EXERCISES 33

The char type represents characters.

Characters are stored internally as 8-bit integers. Their values correspond to the ASCII code.

A char value can be assigned to an int variable, and an int value can be assigned to a char
variable. chara and ints are readily interchangeable in C++.

Character literals appear within single quote marks ('), unlike string literals that appear within double
quote marks (").

Special non-printable control codes like newline and tab are prefixed with the backslash escape char-
acter (\).

The '\n' character is an alternative to using endl to end a printed line.

The C++ enumerated type feature allows programmers to design their own custom type and specify
the values that make up that type.

A value of an enumerated type must be an identifier.

An enum type is handy for representing a small number of discrete, non-numeric options.

The auto keyword used during a variables declaration allows the compiler to deduce the variables
type based on its initial value.

3.12 Exercises

1. Will the following lines of code print the same thing? Explain why or why not.
cout << 6 << endl;
cout << "6" << endl;

2. Will the following lines of code print the same thing? Explain why or why not.
cout << x << endl;
cout << "x" << endl;

3. What is the largest int available on your system?

4. What is the smallest int available on your system?

5. What is the largest double available on your system?

6. What is the smallest double available on your system?

7. What C++ data type represents non-negative integers?

8. What happens if you attempt to use a variable within a program, and that variable is not declared?

9. What is wrong with the following statement that attempts to assign the value ten to variable x?
10 = x;

10. Once a variable has been properly declared and initialized can its value be changed?

2015 Richard L. Halterman Draft date: January 18, 2015

3.12. EXERCISES 34

11. What is another way to write the following declaration and initialization?
int x = 10;

12. In C++ can you declare more than variable in the same declaration statement? If so, how?

13. In the declaration

int a;
int b;

do a and b represent the same memory location?

14. Classify each of the following as either a legal or illegal C++ identifier:

(a) fred
(b) if
(c) 2x
(d) -4
(e) sum_total
(f) sumTotal
(g) sum-total
(h) sum total
(i) sumtotal
(j) While
(k) x2
(l) Private
(m) public
(n) $16
(o) xTwo
(p) _static
(q) _4
(r) ___
(s) 10%
(t) a27834
(u) wilma's

15. What can you do if a variable name you would like to use is the same as a reserved word?

16. Why does C++ require programmers to declare a variable before using it? What are the advantages of
declaring variables?

17. What is the difference between float and double?

18. How can a programmer force a floating-point literal to be a float instead of a double?

19. How is the value 2.45 105 expressed as a C++ literal?

2015 Richard L. Halterman Draft date: January 18, 2015

3.12. EXERCISES 35

20. How can you ensure that a variables value can never be changed after its initialization?
21. How can you extend the range of int on some systems?
22. How can you extend the range and precision of double on some systems?

23. Write a program that prints the ASCII chart for all the values from 0 to 127.
24. Is "i" a string literal or character literal?
25. Is 'i' a string literal or character literal?
26. Is it legal to assign a char value to an int variable?

27. Is it legal to assign an int value to a char variable?

28. What is printed by the following code fragment?
int x;
x = 'A';
cout << x << endl;

29. What is the difference between the character 'n' and the character '\n'?
30. Write a C++ program that simply emits a beep sound when run.

31. Rewrite the following code fragment so that the code behaves exactly the same but does not use
endl.
cout << endl << "Aye!" << endl << "Bye!" << endl;

32. Create an enumerated type that represents the days of the week.

33. Create an enumerated type that represents the months of the year.
34. Determine the exact type of each of the following variables:
(a) auto a = 5;
(b) auto b = false;
(c) auto c = 9.3;
(d) auto d = 5.1f;
(e) auto e = 5L;

2015 Richard L. Halterman Draft date: January 18, 2015

3.12. EXERCISES 36

2015 Richard L. Halterman Draft date: January 18, 2015

 37

Chapter 4

Expressions and Arithmetic

This chapter uses the C++ numeric types introduced in Chapter 3 to build expressions and perform arith-
metic. Some other important concepts are covereduser input, source formatting, comments, and dealing
with errors.

4.1 Expressions

A literal value like 34 and a properly declared variable like x are examples of simple expressions. We can
use operators to combine values and variables and form more complex expressions. Listing 4.1 (adder.cpp)
shows how the addition operator (+) is used to add two integers.

Listing 4.1: adder.cpp

#include <iostream>

using namespace std;

int main() {
int value1, value2, sum;
cout << "Please enter two integer values: ";
cin >> value1 >> value2;
sum = value1 + value2;
cout << value1 << " + " << value2 << " = " << sum << endl;
}

In Listing 4.1 (adder.cpp):

int value1, value2, sum;

This statement declares three integer variables, but it does not initialize them. As we examine the rest
of the program we will see that it would be superfluous to assign values to the variables here.

cout << "Please enter two integer values: ";

This statement prompts the user to enter some information. This statement is our usual print state-
ment, but it is not terminated with the end-of-line marker endl. This is because we want the cursor

2015 Richard L. Halterman Draft date: January 18, 2015

4.1. EXPRESSIONS 38

to remain at the end of the printed line so when the user types in values they appear on the same line
as the message prompting for the values. When the user presses the enter key to complete the input,
the cursor will automatically move down to the next line.
cin >> value1 >> value2;
This statement causes the programs execution to stop until the user types two numbers on the key-
board and then presses enter. The first number entered will be assigned to value1, and the second
number entered will be assigned to value2. Once the user presses the enter key, the value entered is
assigned to the variable. The user may choose to type one number, press enter, type the second num-
ber, and press enter again. Instead, the user may enter both numbers separated by one of more spaces
and then press enter only once. The program will not proceed until the user enters two numbers.

The cin input stream object can assign values to multiple variables in one state-
ment, as shown here:
int num1, num2, num3;
cin >> num1 >> num2 >> num3;
A common beginners mistake is use commas to separate the variables, as in
int num1, num2, num3;
cin >> num1, num2, num3;
The compiler will not generate an error message, because it is legal C++ code. The
statement, however, will not assign the three variables from user input as desired.
The comma operator in C++ has different meanings in different contexts, and here
it is treated like a statement separator; thus, the variables num2 and num3 are
not involved with the cin input stream object. We will have no need to use the
comma operator in this way, but you should be aware of this potential pitfall.

cin is a object that can be used to read input from the user. The >> operatoras used here in the
context of the cin objectis known as the extraction operator. Notice that it is backwards from
the << operator used with the cout object. The cin object represents the input streaminformation
flowing into the program from user input from the keyboard. The >> operator extracts the data from
the input stream cin and assigns the pieces of the data, in order, to the various variables on its right.
sum = value1 + value2;
This is an assignment statement because it contains the assignment operator (=). The variable sum
appears to the left of the assignment operator, so sum will receive a value when this statement exe-
cutes. To the right of the assignment operator is an arithmetic expression involving two variables and
the addition operator. The expression is evaluated by adding together the values of the two variables.
Once the expressions value has been determined, that value can be assigned to the sum variable.

All expressions have a value. The process of determining the expressions value is called evaluation.
Evaluating simple expressions is easy. The literal value 54 evaluates to 54. The value of a variable named
x is the value stored in the memory location reserved for x. The value of a more complex expression is
found by evaluating the smaller expressions that make it up and combining them with operators to form
potentially new values.
Table 4.1 lists the main C++ arithmetic operators. Table 4.1. The common arithmetic operations, ad-
dition, subtraction, and multiplication, behave in the expected way. All these operators are classified as
binary operators because they operate on two operands. In the statement

2015 Richard L. Halterman Draft date: January 18, 2015

4.1. EXPRESSIONS 39

Operator Meaning
+ addition
- subtraction
* multiplication
/ division
% modulus

Table 4.1: The simple C++ arithmetic operators

x = y + z;

the right side is an addition expression y + z. The two operands of the + operator are y and z.
Two of the operators above, + and -, serve also as unary operators. A unary operator has only one
operand. The - unary operator expects a single numeric expression (literal number, variable, or complex
numeric expression within parentheses) immediately to its right; it computes the additive inverse of its
operand. If the operand is positive (greater than zero), the result is a negative value of the same magnitude;
if the operand is negative (less than zero), the result is a positive value of the same magnitude. Zero is
unaffected. For example, the following code sequence
int x = 3;
int y = -4;
int z = 0;
cout << -x << " " << -y << " " -z << endl;

within a program would print

-3 4 0

The following statement

cout << -(4 - 5) << endl;

within a program would print

The unary + operator is present only for completeness; when applied to a numeric value, variable, or
expression, the resulting value is no different from the original value of its operand. Omitting the unary +
operator from the following statement
x = +y;

does not change the statements behavior.

All the arithmetic operators are subject to the limitations of the data types on which they operate; for
example, on a system in which the largest int is 2,147,483,647, the expression
2147483647 + 1

2015 Richard L. Halterman Draft date: January 18, 2015

4.1. EXPRESSIONS 40

will not evaluate to the correct answer since the correct answer falls outside the range of ints.
If you add, subtract, multiply, is divide two ints, the result is an integer. As long as the operation does
not exceed the range of ints, the arithmetic works as expected. Division, however, is another matter. The
statement
cout << 10/3 << " " << 3/10 << endl;

prints

3 0

because in the first case 10 divided by 3 is 3 with a remainder of 1, and in the second case 3 divided by
10 is 0 with a remainder of 3. Since integers are whole numbers, any fractional part of the answer must
be discarded. The process of discarding the fractional part leaving only the whole number part is called
truncation. 10 divided by 3 should be 3.3333..., but that value is truncated to 3. Truncation is not rounding;
for example, 11 divided by 3 is 3.6666..., but it also truncates to 3.

Truncation simply removes any fractional part of the value. It does not round.
Both 10.01 and 10.999 truncate to 10.

The modulus operator (%) computes the remainder of integer division; thus,
cout << 10%3 << " " << 3%10 << endl;

prints

1 3

since 10 divided by 3 is 3 with a remainder of 1, and 3 divided by 10 is 0 with a remainder of 3. Figure 4.1
uses long division for a more hands on illustration of how the integer division and modulus operators work.

The modulus operator is more useful than it may first appear. Listing 4.10 (timeconv.cpp) shows how
we can use it to convert a given number of seconds to hours, minutes, and seconds.
In contrast to integer arithmetic, floating-point arithmetic with doubles behaves as expected:
cout << 10.0/3.0 << " " << 3.0/10.0 << endl;

prints

3.33333 0.3

2015 Richard L. Halterman Draft date: January 18, 2015

4.2. MIXED TYPE EXPRESSIONS 41

8 25/3
3)25
-24
1 25%3

Figure 4.1: Integer division vs. integer modulus. Integer division produces the quotient, and modulus produces the
remainder. In this example, 25/3 is 8, and 25%3 is 1.

Since a char is stored internally as a number (see Section 3.8), we can perform arithmetic on charac-
ters. We will have little need to apply mathematics to characters, but sometimes it is useful. As an example,
the lower-case letters of the alphabet az occupy ASCII values 97123, with a = 97, b = 98, etc. The upper-
case letters AZ are coded as 6591, with A = 65, B = 66, etc. To capitalize any lower-case letter, you need
only subtract 32, as in
char lower = 'd', upper = lower - 32;
cout << upper << endl;

This section of code would print D. If you do not remember the offset of 32 between upper- and lower-case
letter, you can compute it with the letters themselves:
upper = lower - ('a' - 'A');

In this case, if lower has been assigned any value in the range 'a' to 'z', the statement will assign to
upper the capitalized version of lower. On the other hand, if lowers value is outside of that range,
upper will not receive a meaningful value.

4.2 Mixed Type Expressions

Expressions may contain mixed elements; for example, the following program fragment
int x = 4;
double y = 10.2, sum;
sum = x + y;

adds an int to a double, and the result is being assigned to a double. How is the arithmetic performed?
As shown in Figure 4.2, the range of ints falls completely within the range of doubles; thus, any
int value can represented by a double. The int 4 also can be expressed as the double 4.0. In fact,
since the largest int on most systems is 2,147,483,647, the minimum 15 digits of double precision are
more than adequate to represent all integers exactly. This means that any int value can be represented by
a double. The converse is not true, however. 2,200,000,000 can be represented by a double but it is too
big for the int type. We say that the double type is wider than the int type and that the int type is
narrower than the double type.

2015 Richard L. Halterman Draft date: January 18, 2015

4.2. MIXED TYPE EXPRESSIONS 42

Range of double
Range of int

+1.8 10308 -2.1 109 0 +2.1 109 +1.8 10308

Figure 4.2: Range of ints vs. range of doubles

It would be reasonable, then, to be able to assign int values to double variables. The process is
called widening, and it is always safe to widen an int to a double. The following code fragment
double d1;
int i1 = 500;
d1 = i1;
cout << "d1 = " << d1 << endl;

is legal C++ code, and when part of a complete program it would display

d1 = 500

Assigning a double to an int variable is not always possible, however, since the double value may
not be in the range of ints. Furthermore, if the double variable falls within the range of ints but is
not a whole number, the int variable is unable to the manage fractional part. Consider the following code
fragment:
double d = 1.6;
int i = d;

The second line assigns 1 to i. Truncation loses the 0.6 fractional part (see Section 4.1). Note that proper
rounding is not done. The Visual C++ compiler will warn us of a potential problem:

warning C4244: = : conversion from double to int, possible loss of data

This warning reminds us that some information may be lost in the assignment. While the compiler and
linker will generate an executable program when warnings are present, you should carefully scrutinize all
warnings. This warning is particularly useful, since it is easy for errors due to the truncation of floating-
point numbers to creep into calculations.
Converting from a wider type to a narrower type (like double to int) is called narrowing. It often is
necessary to assign a floating-point value to an integer variable. If we know the value to assign is within the
range of ints, and the value has no fractional parts or its truncation would do no harm, the assignment is
safe. To perform the assignment without a warning from the compiler, we use a procedure called a cast, also
called a type cast. The cast forces the compiler to accept the assignment without a issuing a warning. The
following statement convinces the compiler to accept the double-to-int assignment without a warning:

2015 Richard L. Halterman Draft date: January 18, 2015

4.2. MIXED TYPE EXPRESSIONS 43

i = static_cast<int>(d);

The reserved word static_cast performs the narrowing conversion and silences the compiler warning.
The item to convert (in this case the variable d) is placed in the parentheses, and the desired type (in this
case the type int) appears in the angle brackets. The statement
i = static_cast<int>(d);

does not change the type of the variable d; d is declared to be a double and so must remain a double
variable. The statement makes a copy of ds value in a temporary memory location, converting it to its
integer representation during the process.
We also can cast literal values and expressions:
i = static_cast<int>(1.6);
i = static_cast<int>(x + 2.1);

Narrowing a floating-point value to an integer discards any fractional part. Nar-

rowing truncates; it does not round. For example, the double value 1.7 narrows
to the int value 1.

The widening conversion is always safe, so a type cast is not required. Narrowing is a potentially
dangerous operation, and using an explicit cast does not remove the dangerit simply silences the compiler.
For example, consider Listing 4.2 (badnarrow.cpp).

Listing 4.2: badnarrow.cpp

#include <iostream>

using namespace std;

int main() {
double d = 2200000000.0;
int i = d;
cout << "d = " << d << ", i = " << i << endl;
}

The Visual C++ compiler issues a warning about the possible loss of precision when assigning d to i.
Silencing the warning with a type cast in this case is a bad idea; the programs output indicates that the
warning should be heeded:

d = 2.2e+009, i = -2147483648

The printed values of i and d are not even close, nor can they be because it is impossible to represent the
value 2,200,000,000 as an int on a system that uses 32-bit integers. When assigning a value of a wider
type to a variable of a narrower type, the programmer must assume the responsibility to ensure that the

2015 Richard L. Halterman Draft date: January 18, 2015

4.2. MIXED TYPE EXPRESSIONS 44

actual value to be narrowed is indeed within the range of the narrower type. The compiler cannot ensure
the safety of the assignment.
Casts should be used sparingly and with great care because a cast creates a spot in the program that is
immune to the compilers type checking. A careless assignment can produce a garbage result introducing
an error into the program.
When we must perform mixed arithmeticsuch as adding an int to a doublethe compiler auto-
matically produces machine language code that copies the int value to a temporary memory location and
transforms it into its double equivalent. It then performs double-precision floating-point arithmetic to
compute the result.
Integer arithmetic occurs only when both operands are ints. 1/3 thus evaluates to 0, but 1.0/3.0,
1/3.0, and 1.0/3 all evaluate to 0.33333.
Since double is wider than int, we say that double dominates int. In a mixed type arithmetic
expression, the less dominant type is coerced into the more dominant type in order to perform the arithmetic
operation.
Section 3.9 introduced enumerated types. Behind the scenes, the compiler translates enumerated values
into integers. The first value in the enumeration is 0, the second value is 1, etc. The following code
demonstrates the relationship between enumerated types and integers:
enum Color { Red, Orange, Yellow, Green, Blue, Violet };
cout << Orange << " " << Green << endl;

The output of this code fragment is

1 3

C++ allows a programmer to assign an enum type to an integer variable; for example, the following state-
ment is legal:
int num = Orange;

Here, the variable num assumes the value 1. A programmer, however, may not directly assign an integer
value to a variable declared to be of an enum type; for example, the following statement:
Color col = 1;

is illegal. Sometimes it is convenient to convert an integer to an enumerated type, and a type cast enables
the assignment:
enum Color { Red, Orange, Yellow, Green, Blue, Violet };
Color myColor = static_cast<Color>(2);

Since the compiler encodes enumerated values with the integers 0, 1, 2, . . . in the order the values are spec-
ified in the enumeration list, the assignment statement above assigns the value Yellow to the variable
myColor. Casting, of course, can be dangerous; consider:
Color badColor = static_cast<Color>(45);

This statement assigns a meaningless value to the variable badColor, and the cast renders the compiler
powerless to help.

2015 Richard L. Halterman Draft date: January 18, 2015

4.3. OPERATOR PRECEDENCE AND ASSOCIATIVITY 45

Even though enumerated types are encoded as integers internally, programmers may not perform arith-
metic on enumerated types without involving casts. Such opportunities should be very rare; if you need to
perform arithmetic on a variable, it most likely should be a numerical type, not an enumerated type.

4.3 Operator Precedence and Associativity

When different operators are used in the same expression, the normal rules of arithmetic apply. All C++
operators have a precedence and associativity:

Precedencewhen an expression contains two different kinds of operators, which should be applied
first?
Associativitywhen an expression contains two operators with the same precedence, which should
be applied first?

To see how precedence works, consider the expression

2 + 3 * 4

Should it be interpreted as
(2 + 3) * 4

(that is, 20), or rather is

2 + (3 * 4)

(that is, 14) the correct interpretation? As in normal arithmetic, in C++ multiplication and division have
equal importance and are performed before addition and subtraction. We say multiplication and division
have precedence over addition and subtraction. In the expression
2 + 3 * 4

the multiplication is performed before addition, since multiplication has precedence over addition. The
result is 14. The multiplicative operators (*, /, and %) have equal precedence with each other, and the
additive operators (binary + and -) have equal precedence with each other. The multiplicative operators
have precedence over the additive operators.
As in standard arithmetic, in C++ if the addition is to be performed first, parentheses can override the
precedence rules. The expression
(2 + 3) * 4

evaluates to 20. Multiple sets of parentheses can be arranged and nested in any ways that are acceptable in
standard arithmetic.
To see how associativity works, consider the expression
2 - 3 - 4

The two operators are the same, so they have equal precedence. Should the first subtraction operator be
applied before the second, as in
(2 - 3) - 4

2015 Richard L. Halterman Draft date: January 18, 2015

4.3. OPERATOR PRECEDENCE AND ASSOCIATIVITY 46

(that is, 5), or rather is

2 - (3 - 4)

(that is, 3) the correct interpretation? The former (5) is the correct interpretation. We say that the subtrac-
tion operator is left associative, and the evaluation is left to right. This interpretation agrees with standard
arithmetic rules. All binary operators except assignment are left associative. Assignment is an exception; it
is right associative. To see why associativity is an issue with assignment, consider the statement
w = x = y = z;

This is legal C++ and is called chained assignment. Assignment can be used as both a statement and an
expression. The statement
x = 2;

assigns the value 2 to the variable x. The expression

x = 2

assigns the value 2 to the variable x and evaluates to the value that was assigned; that is, 2. Since assignment
is right associative, the chained assignment example should be interpreted as
w = (x = (y = z));

which behaves as follows:

The expression y = z is evaluated first. zs value is assigned to y, and the value of the expression
y = z is zs value.
The expression x = (y = z) is evaluated. The value of y = z, that is z, is assigned to x. The
overall value of the expression x = y = z is thus the value of z. Now the values of x, y, and z are
all equal (to z).
The expression w = (x = y = z) is evaluated. The value of the expression x = y = z is
equal to zs value, so zs value is assigned to w. The overall value of the expression w = x = y = z
is equal to z, and the variables w, x, y, and z are all equal (to z).

As in the case of precedence, we can use parentheses to override the natural associativity within an expres-
sion.
The unary operators have a higher precedence than the binary operators, and the unary operators are
right associative. This means the statements
cout << -3 + 2 << endl;
cout << -(3 + 2) << endl;

which display

-1
-5

behave as expected.
Table 4.2 shows the precedence and associativity rules for some C++ operators. The * operator also has
a unary form that has nothing to do with mathematics; it is covered in Section 10.6.

2015 Richard L. Halterman Draft date: January 18, 2015

4.4. COMMENTS 47

Arity Operators Associativity

Unary +, -
Binary *, /, % Left
Binary +, - Left
Binary = Right

Table 4.2: Operator precedence and associativity. The operators in each row have a higher precedence than the operators
below it. Operators within a row have the same precedence.

4.4 Comments

Good programmers annotate their code by inserting remarks that explain the purpose of a section of code or
why they chose to write a section of code the way they did. These notes are meant for human readers, not
the compiler. It is common in industry for programs to be reviewed for correctness by other programmers
or technical managers. Well-chosen identifiers (see Section 3.3) and comments can aid this assessment
process. Also, in practice, teams of programmers develop software. A different programmer may be
required to finish or fix a part of the program written by someone else. Well-written comments can help
others understand new code quicker and increase their productivity modifying old or unfinished code. While
it may seem difficult to believe, even the same programmer working on her own code months later can have
a difficult time remembering what various parts do. Comments can help greatly.
Any text contained within comments is ignored by the compiler. C++ supports two types of comments:
single line comments and block comments:

Single line commentthe first type of comment is useful for writing a single line remark:
// Compute the average of the values
avg = sum / number;

The first line here is a comment that comment explains what the statement that follows it is supposed
to do. The comment begins with the double forward slash symbols (//) and continues until the end
of that line. The compiler will ignore the // symbols and the contents of the rest of the line. This
type of comment is also useful for appending a short comment to the end of a statement:
avg = sum / number; // Compute the average of the values

Here, an executable statement and the comment appear on the same line. The compiler will read the
assignment statement here, but it will ignore the comment. The compiler generates the same machine
code for this example as it does for the preceding example, but this example uses one line of source
code instead of two.
Block commentthe second type of comment begins with the symbols /* and is in effect until
the */ symbols are encountered. The /* . . . */ symbols delimit the comment like parentheses
delimit a parenthetical expression. Unlike parentheses, however, these block comments cannot be
nested within other block comments.
The block comment is handy for multi-line comments:
/* After the computation is completed
the result is displayed. */
cout << result << endl;

What should be commented? Avoid making a remark about the obvious; for example:

2015 Richard L. Halterman Draft date: January 18, 2015

4.5. FORMATTING 48

result = 0; // Assign the value zero to the variable named result

The effect of this statement is clear to anyone with even minimal C++ programming experience. Thus, the
audience of the comments should be taken into account; generally, routine activities require no remarks.
Even though the effect of the above statement is clear, its purpose may need a comment. For example:
result = 0; // Ensures 'result' has a well-defined minimum value

This remark may be crucial for readers to completely understand how a particular part of a program works.
In general, programmers are not prone to providing too many comments. When in doubt, add a remark.
The extra time it takes to write good comments is well worth the effort.

4.5 Formatting

Program comments are helpful to human readers but ignored by the compiler. Another aspect of source
code that is largely irrelevant to the compiler but that people find valuable is its formatting. Imagine the
difficulty of reading a book in which its text has no indentation or spacing to separate one paragraph from
another. In comparison to the source code for a computer program, a books organization is quite simple.
Over decades of software construction programmers have established a small collection of source code
formatting styles that the industry finds acceptable.
The compiler allows a lot of leeway for source code formatting. Consider Listing 4.3 (reformattedvariable.cpp)
which is a reformatted version of Listing 3.4 (variable.cpp).

Listing 4.3: reformattedvariable.cpp

#include <iostream>
using
namespace
std
;
int
main
(
)
{
int
x
;
x
=
10
;
cout
<<
x
<<
endl
;
}

Listing 4.4 (reformattedvariable2.cpp) is another reformatted version of Listing 3.4 (variable.cpp).

2015 Richard L. Halterman Draft date: January 18, 2015

4.5. FORMATTING 49

Listing 4.4: reformattedvariable2.cpp

#include <iostream>
using namespace std;int main(){int x;x=10;cout<<x<<endl;}

Both reformatted programs are valid C++ and compile to the same machine language code as the orig-
inal version. Most would argue that the original version is easier to read and understand more quickly
than either of the reformatted versions. The elements in Listing 3.4 (variable.cpp) are organized bet-
ter. Experienced C++ programmers would find both Listing 4.3 (reformattedvariable.cpp) and Listing 4.4
(reformattedvariable2.cpp) visually painful.
What are some distinguishing characteristics of Listing 3.4 (variable.cpp)?

Each statement appears on its own line. A statement is not unnecessarily split between two lines of
text. Visually, one line of text implies one action (statement) to perform.

The close curly brace aligns vertically with the line above that contains the corresponding open curly
brace. This makes it easier to determine if the curly braces match and nest properly. It also better por-
trays the logical structure of the program. The ability to accurately communicate the logical structure
of a program becomes very important as write more complex programs. Programs with complex logic
frequently use multiple nested curly braces (for example, see Listing 5.11 (troubleshoot.cpp)). With-
out a consistent, organized arrangement of curly braces it can difficult to determine which opening
brace goes with a particular closing brace.

The statements that constitute the body of main are indented several spaces. This visually em-
phasizes the fact that the elements are indeed logically enclosed. As with curly brace alignment,
indentation to emphasize logical enclosure becomes more important as more complex programs are
considered.

Spaces are used to spread out statements and group pieces of the statement. Space around the op-
erators (=) makes it easier to visually separate the operands from the operators and comprehend the
details of the expression. Most people find the statement
total_sale = subtotal + tax;

much easier to read than

total_sale=subtotal+tax;

since the lack of space in the second version makes it more difficult to pick out the individual pieces
of the statement. In the first version with extra space, it is clearer where operators and variable names
begin and end.
In a natural language like English, a book is divided into distinct chapters, and chapters are composed
of paragraphs. One paragraph can be distinguished from another because the first line is indented or
an extra space appears between two paragraphs. Space is used to separate words in each sentence.
Consider how hard it would be to read a book if all the sentences were printed like this one:

Theboyranquicklytothetreetoseethestrandedcat.

Judiciously placed open space in a C++ program can greatly enhance its readability.

2015 Richard L. Halterman Draft date: January 18, 2015

4.5. FORMATTING 50

C++ gives the programmer a large amount of freedom in formatting source code.
The compiler reads the characters that make up the source code one symbol at a
time left to right within a line before moving to the next line. While extra space
helps readability, spaces are not allowed in some places:
Variable names and reserved words must appear as unbroken units.
Multi-symbol operators like << cannot be separated (< < is illegal).

One common coding convention that is universal in C++ programming is demonstrated in Listing 3.10
(const.cpp). While programmers usually use lower-case letters in variable names, they usually express
constant names with all capital letters; for example, PI is used for the mathematical constant instead of
pi. C++ does not require constants to be capitalized, but capitalizing them aids humans reading the source
code so they can quickly distinguish between variables and constants.
Figure 4.3 shows the three most common ways programmers use indentation and place curly braces
in C++ source code. The K&R and ANSI styles are the most popular in published C++ source code. The

int main() int main()

int main() { int main() {
{ {
Body Body
Body Body
} }
} }
K & R Style ANSI Style Whitesmith Style Banner Style

Figure 4.3: The most common C++ coding styles.

Whitesmith and Banner styles appear much less frequently. http://en.wikipedia.org/wiki/Indent_

style reviews the various ways to format C++ code. Observe that all the accepted formatting styles indent
the block of statements contained in the main function.
Most software development organizations adopt a set of style guidelines, sometimes called code conven-
tions. These guidelines dictate where to indent and by how many spaces, where to place curly braces, how
to assign names to identifiers, etc. Programmers working for the organization are required to follow these
style guidelines for the code they produce. This better enables any member of the development team to read
and understand more quickly code written by someone else. This is necessary when code is reviewed for
correctness or when code must be repaired or extended, and the original programmer is no longer with the
development team.
Even if you are not forced to use a particular style, it is important to use a consistent style throughout
the code you write. As our programs become more complex we will need to use additional curly braces
and various levels of indentation to organize the code we write. A consistent style (especially one of the
standard styles shown in Figure 4.3) makes it easier to read and verify that the code actually expresses our
intent. It also makes it easier to find and fix errors. Said another way, haphazard formatting increases the
time it takes to develop correct software because programmers mistakes hide better in poorly formatted
code.

2015 Richard L. Halterman Draft date: January 18, 2015

4.6. ERRORS AND WARNINGS 51

Good software development tools can boost programmer productivity, and many programming editors
have the ability to automatically format source code according to a standard style. Some of these editors
can correct the codes style as the programmer types in the text. A standalone program known as a pretty
printer can transform an arbitrarily formatted C++ source file into a properly formatted one.

4.6 Errors and Warnings

Beginning programmers make mistakes writing programs because of inexperience in programming in gen-
eral or because of unfamiliarity with a programming language. Seasoned programmers make mistakes due
to carelessness or because the proposed solution to a problem is faulty and the correct implementation of an
incorrect solution will not produce a correct program. Regardless of the reason, a programming error falls
under one of three categories:

compile-time error

run-time error

logic error

4.6.1 Compile-time Errors

A compile-time error results from the programmers misuse of the language. A syntax error is a common
compile-time error. For example, in English one can say

The boy walks quickly.

This sentence uses correct syntax. However, the sentence

The boy walk quickly.

is not correct syntactically: the number of the subject (singular form) disagrees with the number of the verb
(plural form). It contains a syntax error. It violates a grammatical rule of the English language. Similarly,
the C++ statement
x = y + 2;

is syntactically correct because it obeys the rules for the structure of an assignment statement described in
Section 3.2. However, consider replacing this assignment statement with a slightly modified version:
y + 2 = x;

If a statement like this one appears in a program and the variables x and y have been properly declared, the
compiler will issue an error message; for example, the Visual C++ compiler reports (among other things):

error C2106: = : left operand must be l-value

The syntax of C++ does not allow an expression like y + 2 to appear on the left side of the assignment
operator.

2015 Richard L. Halterman Draft date: January 18, 2015

4.6. ERRORS AND WARNINGS 52

(The term l-value in the error message refers to the left side of the assignment operator; the l is an
elle, not a one..)
The compiler may generate an error for a syntactically correct statement like
x = y + 2;

if either of the variables x or y has not been declared; for example, if y has not been declared, Visual C++
reports:

error C2065: y : undeclared identifier

Other common compile-time errors include missing semicolons at the end of statements, mismatched curly
braces and parentheses, and simple typographical errors.
Compile-time errors usually are the easiest to repair. The compiler pinpoints the exact location of the
problem, and the error does not depend on the circumstances under which the program executes. The exact
error can be reproduced by simply recompiling the same source code.
Compilers have the reputation for generating cryptic error messages. They seem to provide little help
as far as novice programmers are concerned. Sometimes a combination of errors can lead to messages that
indicate errors on lines that follow the line that contains the actual error. Once you encounter the same
error several times and the compiler messages become more familiar, you become better able to deduce the
actual problem from the reported message. Unfortunately C++ is such a complex language that sometimes a
simple compile-time error can result in a message that is incomprehensible to beginning C++ programmers.

4.6.2 Run-time Errors

The compiler ensures that the structural rules of the C++ language are not violated. It can detect, for exam-
ple, the malformed assignment statement and the use of a variable before its declaration. Some violations of
the language cannot be detected at compile time, however. A program may not run to completion but instead
terminate with an error. We commonly say the program crashed. Consider Listing 4.5 (dividedanger.cpp)
which under certain circumstances will crash.

Listing 4.5: dividedanger.cpp

// File dividedanger.cpp

#include <iostream>

using namespace std;

int main() {
int dividend, divisor;

// Get two integers from the user

cout << "Please enter two integers to divide:";
cin >> dividend >> divisor;
// Divide them and report the result
cout << dividend << "/" << divisor << " = "
<< dividend/divisor << endl;
}

The expression

2015 Richard L. Halterman Draft date: January 18, 2015

4.6. ERRORS AND WARNINGS 53

dividend/divisor

is potentially dangerous. If the user enters, for example, 32 and 4, the program works nicely

Please enter two integers to divide: 32 4

32/4 = 8

and displays the answer of 8. If the user instead types the numbers 32 and 0, the program reports an error
and terminates. Division by zero is undefined in mathematics, and integer division by zero in C++ is illegal.
When the program attempts the division at run time, the system detects the attempt and terminates the
program.
This particular program can fail in other ways as well; for example, outside of the C++ world, 32.0 looks
like a respectable integer. If the user types in 32.0 and 8, however, the program crashes because 32.0 is not
a valid way to represent an integer in C++. When the compiler compiles the source line
cin >> dividend >> divisor;

given that dividend has been declared to be an int, it generates slightly different machine language
code than it would if dividend has been declared to be a double instead. The compiled code expects
the text entered by the user to be digits with no extra decoration. Any deviation from this expectation results
in a run-time error. Similar results occur if the user enters text that does not represent an integer, like fred.
Observe that in either caseentry of a valid but inappropriate integer (zero) or entry of a non-integer
(32.0 or fred)it is impossible for the compiler to check for these problems at compile time. The compiler
cannot predict what the user will enter when the program is run. This means it is up to the programmer
to write code that can handle bad input that the user may provide. As we continue our exploration of
programming in C++, we will discover ways to make our programs more robust against user input (see
Listing 5.2 (betterdivision.cpp) in Chapter 5, for example). The solution involves changing the way the
program runs depending on the actual input provided by the user.

4.6.3 Logic Errors

Consider the effects of replacing the expression

dividend/divisor;

in Listing 4.5 (dividedanger.cpp) with the expression:

divisor/dividend;

The program compiles with no errors. It runs, and unless a value of zero is entered for the dividend, no
run-time errors arise. However, the answer it computes is not correct in general. The only time the correct
answer is printed is when dividend = divisor. The program contains an error, but neither the compiler
nor the run-time system is able detect the problem. An error of this type is known as a logic error.
Listing 4.19 (faultytempconv.cpp) is an example of a program that contains a logic error. Listing 4.19
(faultytempconv.cpp) compiles and does not generate any run-time errors, but it produces incorrect results.
Beginning programmers tend to struggle early on with compile-time errors due to their unfamiliarity
with the language. The compiler and its error messages are actually the programmers best friend. As

2015 Richard L. Halterman Draft date: January 18, 2015

4.6. ERRORS AND WARNINGS 54

the programmer gains experience with the language and the programs written become more complicated,
the number of compile-time errors decrease or are trivially fixed and the number of logic errors increase.
Unfortunately, both the compiler and run-time environment are powerless to provide any insight into the
nature and sometimes location of logic errors. Logic errors, therefore, tend to be the most difficult to find
and repair. Tools such as debuggers are frequently used to help locate and fix logic errors, but these tools
are far from automatic in their operation.
Errors that escape compiler detection (run-time errors and logic errors) are commonly called bugs.
Since the compiler is unable to detect these problems, such bugs are the major source of frustration for
developers. The frustration often arises because in complex programs the bugs sometimes only reveal
themselves in certain situations that are difficult to reproduce exactly during testing. You will discover this
frustration as your programs become more complicated. The good news is that programming experience
and the disciplined application of good programming techniques can help reduce the number logic errors.
The bad news is that since software development in an inherently human intellectual pursuit, logic errors
are inevitable. Accidentally introducing and later finding and eliminating logic errors is an integral part of
the programming process.

4.6.4 Compiler Warnings

A warning issued by the compiler does mark a violation of the rules in the C++ language, but it is an noti-
fication to the programmer that the program contains a construct that is a potential problem. In Listing 4.9
(tempconv.cpp) the programmer is attempting to print the value of a variable before it has been given a
known value.

Listing 4.6: uninitialized.cpp

// uninitialized.cpp

#include <iostream>

using namespace std;

int main() {
int n;
cout << n << endl;
}

An attempt to build Listing 4.6 (uninitialized.cpp) yields the following message from the Visual C++
compiler:

warning C4700: uninitialized local variable n used

The compiler issued a warning but still generated the executable file. When run, the program produces a
random result because it prints the value in memory associated with the variable, but the program does not
initialize that memory location.
Listing 4.7 (narrow.cpp) assigns a double value to an int variable, which we know from Section 4.1
truncates the result.

Listing 4.7: narrow.cpp

#include <iostream>

2015 Richard L. Halterman Draft date: January 18, 2015

4.6. ERRORS AND WARNINGS 55

using namespace std;

int main() {
int n;
double d = 1.6;
n = d;
cout << n << endl;
}

When compiled we see

warning C4244: = : conversion from double to int, possible loss of data

Since it is a warning and not an error, the compiler generates the executable, but the warning should prompt
us to stop and reflect about the correctness of the code. The enhanced warning level prevents the program-
mer from being oblivious to the situation.
The default Visual C++ warning level is 3 when compiling in the IDE and level 1 on the command line
(that is why we use the /W3 option on the command line); the highest warning level is 4. You can reduce
the level to 1 or 2 or disable warnings altogether, but that is not recommended. The only reason you might
want to reduce the warning level is to compile older existing C++ source code that does meet newer C++
standards. When developing new code, higher warning levels are preferred since they provide more help to
the programmer. Unless otherwise noted, all the complete program examples in this book compile cleanly
under Visual C++ set at warning level 3. Level 3 is helpful for detecting many common logic errors.
We can avoid most warnings by a simple addition to the code. Section 4.2 showed how we can use
static_cast to coerce a wider type to a narrower type. At Visual C++ warning Level 3, the compiler
issues a warning if the cast is not used. The little code that must be added should cause the programmer to
stop and reflect about the correctness of the construct. The enhanced warning level prevents the programmer
from being oblivious to the situation.

Use the strongest level of warnings available to your compiler. Treat all warnings
as problems that must be corrected. Do not accept as completed a program that
compiles with warnings.

We may assign a double literal to a float variable without any special type casting. The compiler
automatically narrows the double to a float as Listing 4.8 (assignfloat.cpp) shows:

Listing 4.8: assignfloat.cpp

#include <iostream>

using namespace std;

int main() {
float number;
number = 10.0; // OK, double literal assignable to a float
cout << "number = " << number << endl;
}

2015 Richard L. Halterman Draft date: January 18, 2015

4.7. ARITHMETIC EXAMPLES 56

The statement
number = 10.0;

assigns a double literal (10.0) to a float variable. You instead may explicitly use a float literal as:
number = 10.0f;

4.7 Arithmetic Examples

The kind of arithmetic to perform in a complex expression is determined on an operator by operator basis.
For example, consider Listing 4.9 (tempconv.cpp) that attempts to convert a temperature from degrees
Fahrenheit to degrees Celsius using the formula
5
C= ( F 32)
9

Listing 4.9: tempconv.cpp

// File tempconv.cpp

#include <iostream>

using namespace std;

int main() {
double degreesF, degreesC;
// Prompt user for temperature to convert
cout << "Enter the temperature in degrees F: ";
// Read in the user's input
cin >> degreesF;
// Perform the conversion
degreesC = 5/9*(degreesF - 32);
// Report the result
cout << degreesC << endl;
}

Listing 4.9 (tempconv.cpp) contains comments that document each step explaining the codes purpose.
An initial test is promising:

Enter the temperature in degrees F: 32

Degrees C = 0

Water freezes at 32 degrees Fahrenheit and 0 degrees Celsius, so the programs behavior is correct for this
test. Several other attempts are less favorableconsider

Enter the temperature in degrees F: 212

Degrees C = 0

2015 Richard L. Halterman Draft date: January 18, 2015

4.7. ARITHMETIC EXAMPLES 57

Water boils at 212 degrees Fahrenheit which is 100 degrees Celsius, so this answer is not correct.

Enter the temperature in degrees F: -40

Degrees C = 0

The value 40 is the point where the Fahrenheit and Celsius curves cross, so the result should be 40, not
zero. The first test was only coincidentally correct.
Unfortunately, the printed result is always zero regardless of the input. The problem is the division 5/9
in the statement
degreesC = 5/9*(degreesF - 32);

Division and multiplication have equal precedence, and both are left associative; therefore, the division is
performed first. Since both operands are integers, integer division is performed and the quotient is zero (5
divided by 9 is 0, remainder 5). Of course zero times any number is zero, thus the result. The fact that a
floating-point value is involved in the expression (degreesF) and the overall result is being assigned to a
floating-point variable, is irrelevant. The decision about the exact type of operation to perform is made on
an operator-by-operator basis, not globally over the entire expression. Since the division is performed first
and it involves two integer values, integer division is used before the other floating-point pieces become
involved.
One solution simply uses a floating-point literal for either the five or the nine, as in
degreesC = 5.0/9*(degreesF - 32);

This forces a double-precision floating-point division (recall that the literal 5.0 is a double). The correct
result, subject to rounding instead of truncation, is finally computed.
Listing 4.10 (timeconv.cpp) uses integer division and modulus to split up a given number of seconds to
hours, minutes, and seconds.

Listing 4.10: timeconv.cpp

// File timeconv.cpp

#include <iostream>

using namespace std;

int main() {
int hours, minutes, seconds;
cout << "Please enter the number of seconds:";
cin >> seconds;
// First, compute the number of hours in the given number
// of seconds
hours = seconds / 3600; // 3600 seconds = 1 hours
// Compute the remaining seconds after the hours are
// accounted for
seconds = seconds % 3600;
// Next, compute the number of minutes in the remaining
// number of seconds
minutes = seconds / 60; // 60 seconds = 1 minute

2015 Richard L. Halterman Draft date: January 18, 2015

4.7. ARITHMETIC EXAMPLES 58

// Compute the remaining seconds after the minutes are

// accounted for
seconds = seconds % 60;
// Report the results
cout << hours << " hr, " << minutes << " min, "
<< seconds << " sec" << endl;
}

If the user enters 10000, the program prints 2 hr, 46 min, 40 sec. Notice the assignments to
the seconds variable, such as
seconds = seconds % 3600

The right side of the assignment operator (=) is first evaluated. The remainder of seconds divided by
3,600 is assigned back to seconds. This statement can alter the value of seconds if the current value of
seconds is greater than 3,600. A similar statement that occurs frequently in programs is one like
x = x + 1;

This statement increments the variable x to make it one bigger. A statement like this one provides further
evidence that the C++ assignment operator does not mean mathematical equality. The following statement
from mathematics
x = x+1
is surely never true; a number cannot be equal to one more than itself. If that were the case, I would deposit
one dollar in the bank and then insist that I really had two dollars in the bank, since a number is equal to
one more than itself. That two dollars would become 3.00, then 4.00, etc., and soon I would be rich. In
C++, however, this statement simply means add one to x and assign the result back to x.
A variation on Listing 4.10 (timeconv.cpp), Listing 4.11 (enhancedtimeconv.cpp) performs the same
logic to compute the time pieces (hours, minutes, and seconds), but it uses more simple arithmetic to
produce a slightly different outputinstead of printing 11,045 seconds as 3 hr, 4 min, 5 sec, List-
ing 4.11 (enhancedtimeconv.cpp) displays it as 3:04:05. It is trivial to modify Listing 4.10 (timeconv.cpp)
so that it would print 3:4:5, but Listing 4.11 (enhancedtimeconv.cpp) includes some extra arithmetic to
put leading zeroes in front of single-digit values for minutes and seconds as is done on digital clock displays.

Listing 4.11: enhancedtimeconv.cpp

// File enhancedtimeconv.cpp

#include <iostream>

using namespace std;

int main() {
int hours, minutes, seconds;
cout << "Please enter the number of seconds:";
cin >> seconds;
// First, compute the number of hours in the given number
// of seconds
hours = seconds / 3600; // 3600 seconds = 1 hours
// Compute the remaining seconds after the hours are
// accounted for
seconds = seconds % 3600;
// Next, compute the number of minutes in the remaining

2015 Richard L. Halterman Draft date: January 18, 2015

4.8. INTEGERS VS. FLOATING-POINT NUMBERS 59

// number of seconds
minutes = seconds / 60; // 60 seconds = 1 minute
// Compute the remaining seconds after the minutes are
// accounted for
seconds = seconds % 60;
// Report the results
cout << hours << ":";
// Compute tens digit of minutes
int tens = minutes / 10;
cout << tens;
// Compute ones digit of minutes
int ones = minutes % 10;
cout << ones << ":";
// Compute tens digit of seconds
tens = seconds / 10;
cout << tens;
// Compute ones digit of seconds
ones = seconds % 10;
cout << ones << endl;
}

Listing 4.11 (enhancedtimeconv.cpp) uses the fact that if x is a one- or two-digit number, x / 10 is the
tens digit of x. If x / 10 is zero, x is necessarily a one-digit number.

4.8 Integers vs. Floating-point Numbers

CAUTION! SECTION UNDER CONSTRUCTION

Floating-point numbers offer some distinct advantages over integers. Floating-point numbers, especially
doubles have a much greater range of values than any integer type. Floating-point numbers can have
fractional parts and integers cannot. Integers, however, offer one big advantage that floating-point numbers
cannotexactness. To see why integers are exact and floating-point numbers are not, we will explore the
way computers store and manipulate the integer and floating-point types.
Computers store all data internally in binary form. The binary (base 2) number system is much simpler
than the familiar decimal (base 10) number system because it uses only two digits: 0 and 1. The decimal
system uses 10 digits: 0, 1, 2, 3, 4, 5, 6, 7, 8, and 9. Despite the lack of digits, every decimal integer has an
equivalent binary representation. Binary numbers use a place value system not unlike the decimal system.
Figure 4.4 shows how the familiar base 10 place value system works.
With 10 digits to work with, the decimal number system distinguishes place values with powers of 10.
Compare the base 10 system to the base 2 place value system shown in Figure 4.5.
With only two digits to work with, the binary number system distinguishes place values by powers of
two. Since both binary and decimal numbers share the digits 0 and 1, we will use the subscript 2 to indicate
a binary number; therefore, 100 represents the decimal value one hundred, while 1002 is the binary number
four. Sometimes to be very clear we will attach a subscript of 10 to a decimal number, as in 10010 .
In the decimal system, it is easy to add 3 + 5:
3
+ 5
8
2015 Richard L. Halterman Draft date: January 18, 2015
4.8. INTEGERS VS. FLOATING-POINT NUMBERS 60

4 7 3 4 0 6
105 104 103 102 101 100
100,000 10,000 1,000 100 10 1

473, 406 = 4 105 + 7 104 + 3 103 + 4 102 + 0 101 + 6 100

= 400, 000 + 70, 000 + 3, 000 + 400 + 0 + 6
= 473, 406

Figure 4.4: The base 10 place value system

1 0 0 1 1 1
25 24 23 22 21 20
32 16 8 4 2 1

1001112 = 1 25 + 0 24 + 0 23 + 1 22 + 1 21 + 1 20
= 32 + 0 + 0 + 4 + 2 + 1
= 39

Figure 4.5: The base 2 place value system

The sum 3 + 9 is a little more complicated, as early elementary students soon discover:
3
+ 9
The answer, of course, is 12, but there is no single digit that means 12it takes two digits, 1 and 2. The
sum is
1
03
+ 09
12
We can say 3 + 9 is 2, carry the 1. The rules for adding binary numbers are shorter and simpler than decimal
numbers:
02 + 02 = 02
02 + 12 = 12
12 + 02 = 12
12 + 12 = 102
We can say the sum 12 + 12 is 02 , carry the 12 . A typical larger sum would be
11
910 = 10012
+ 310 = 112
1210 = 11002

2015 Richard L. Halterman Draft date: January 18, 2015

4.8. INTEGERS VS. FLOATING-POINT NUMBERS 61

4.8.1 Integer Implementation

Mathematical integers are whole numbers (no fractional parts), both positive and negative. Standard C++
supports multiple integer types: int, short, long, and long long, unsigned, unsigned short,
unsigned long, and unsigned long long. These are distinguished by the number of bits required
to store the type, and, consequently, the range of values they can represent. Mathematical integers are in-
finite, but all of C++s integer types correspond to finite subsets of the mathematical integers. The most
commonly used integer type in C++ is int. All ints, regardless of their values, occupy the same amount
of memory and, therefore use the same number of bits. The exact number of bits in an int is processor
specific. A 32-bit processor, for example, is built to manipulate 32-bit integers very efficiently. A C++
compiler for such a system most likely would use 32-bit ints, while a compiler for a 64-bit machine might
represent ints with 64 bits. On a 32-bit computer, the numbers 4 and 1,320,002,912 both occupy 32 bits
of memory.
For simplicity, we will focus on unsigned integers, particularly the unsigned type. The unsigned
type in Visual C++ occupies 32 bits. With 32 bits we can represent 4, 294, 967, 296 different values, and so
Visual C++s unsigned type represents the integers 0 . . . 4, 294, 967, 295. The hardware in many computer
systems in the 1990s provided only 16-bit integer types, so it was common then for C++ compilers to
support 16-bit unsigned values with a range 0 . . . 32, 767. To simplify our exploration into the properties
of computer-based integers, we will consider an even smaller, mythical unsigned integer type that we will
call unsigned tiny. C++ has no such unsigned tiny type as it has a very small range of values
too small to be useful as an actual type in real programs. Our unsigned tiny type uses only five bits of
storage, and Table 4.3 shows all the values that a variable of type unsigned tiny can assume.

Binary Bit String Decimal Value

00000 0
00001 1
00010 2
00011 3
00100 4
00101 5
00110 6
00111 7
01000 8
01001 9
01010 10
01011 11
01100 12
01101 13
01110 14
01111 15
10000 16
10001 17
10010 18
10011 19
10100 20
10101 21
10110 22
10111 23
11000 24
11001 25
11010 26
11011 27
11100 28
11101 29
11110 30
11111 31

Table 4.3: The unsigned tiny values

Table 4.3 shows that the unsigned tiny type uses all the combinations of 0s and 1s in five bits. We

2015 Richard L. Halterman Draft date: January 18, 2015

4.8. INTEGERS VS. FLOATING-POINT NUMBERS 62

can derive the decimal number 6 directly from its bit pattern:
0 0110
00110 = = 0 16 + 0 8 + 1 4 + 1 2 + 0 1 = 6
16 8 4 2 1

To see that arithmetic works, try adding 4 + 3:

410 = 001002
+ 310 = 000112
710 = 001112

That was easy since involved no carries. Next we will try 3 + 1:

11
310 = 000112
+ 110 = 000012
410 = 001002

In the ones column (rightmost column), 12 + 12 = 102 , so write a 0 and carry the 1 to the top of the next
column to the left (that is, the twos column). In the twos column, 12 + 12 + 02 = 102 , so we must carry a 1
into the fours column as well.
The next example illustrates a limitation of our finite representation. Consider the sum 8 + 28:

11
810 = 010002
+ 2810 = 111002
410 = 1 001002

In the this sum we have a carry of 1 from the eights column to the 16s column, and we have a carry from the
16s column to nowhere. We need a sixth column (a 32s column), another place value, but our unsigned
tiny type is limited to five bits. That carry out from the 16s place is lost. The largest unsigned tiny
value is 31, but 28 + 8 = 36. It is not possible to store the value 36 in an unsigned tiny just as it is
impossible to store the value 5, 000, 000, 000 in a C++ unsigned variable.
Consider exceeding the capacity of the unsigned tiny type by just one:

11111
3110 = 111112
+ 110 = 000012
010 = 1 000002

Adding one to the largest possible unsigned tiny, 31, results in the smallest possible value, 0! This
mirrors the behavior of the actual C++ unsigned type, as Listing 4.12 (unsignedoverflow.cpp) demon-
strates.

Listing 4.12: unsignedoverflow.cpp

#include <iostream>

using namespace std;

int main() {
unsigned x = 4294967293; // Almost the largest possible unsigned value
cout << x << " + 1 = " << x + 1 << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

4.8. INTEGERS VS. FLOATING-POINT NUMBERS 63

cout << x << " + 2 = " << x + 2 << endl;

cout << x << " + 3 = " << x + 3 << endl;

Listing 4.12 (unsignedoverflow.cpp) prints

4294967293 + 1 = 4294967294
4294967293 + 2 = 4294967295
4294967293 + 3 = 0

In fact, Visual C++s 32-bit unsigneds follow the cyclic pattern shown in Figure 4.6. In the figure,

0
4,294,967,295 1
2
4,294,967,294
3
4,294,967,293
4
4
4,294,967,292

Subtract Add

2,147,483,650 2,147,483,645

2,147,483,649 2,147,483,646

2,147,483,648 2,147,483,647

Figure 4.6: The cyclic nature of 32-bit unsigned integers. Adding 1 to 4,294,967,295 produces 0, one position clockwise
from 4,294,967295. Subtracting 4 from 2 yields 4,294,967,294, four places counterclockwise from 2.

an addition moves a value clockwise around the circle, while a subtraction moves a value counterclockwise
around the circle. When the numeric limit is reached, the value rolls over like an automobile odometer.
Signed integers exibit a similar cyclic pattern as shown in Figure 4.7. In the case of signed integers,
as Figure 4.7 shows, adding one to the largest representable value produces the smallest negative value.
Listing 4.13 (integeroverflow.cpp) demonstrates.

Listing 4.13: integeroverflow.cpp

2015 Richard L. Halterman Draft date: January 18, 2015

4.8. INTEGERS VS. FLOATING-POINT NUMBERS 64

1 0
1
2 2

3 3

4
4 4

Subtract Add

2,147,483,645 2,147,483,645

2,147,483,646 2,147,483,646

2,147,483,647 2,147,483,647

2,147,483,648

Figure 4.7: The cyclic nature of 32-bit signed integers. Adding 1 to 2, 147, 483, 647 produces 2, 147, 483, 648, one
position clockwise from 2, 147, 483, 647. Subtracting 5 from 2, 147, 483, 645 yields 2, 147, 483, 646, five places coun-
terclockwise from 2, 147, 483, 645.

#include <iostream>

using namespace std;

int main() {
int x = 2147483645; // Almost the largest possible int value
cout << x << " + 1 = " << x + 1 << endl;
cout << x << " + 2 = " << x + 2 << endl;
cout << x << " + 3 = " << x + 3 << endl;

Listing 4.13 (integeroverflow.cpp) prints

2147483645 + 1 = 2147483646
2147483645 + 2 = 2147483647
2147483645 + 3 = -2147483648

Attempting to exceed the maximum limit of a numeric type results in overflow, and attempting to exceed
the minimum limit is called underflow. Integer arithmetic that overflow or underflow produces a valid, yet

2015 Richard L. Halterman Draft date: January 18, 2015

4.8. INTEGERS VS. FLOATING-POINT NUMBERS 65

incorrect integer result. The compiler does not check that a computation will result in exceeding the limit of
a type because it is impossible to do so in general (consider adding two integer variables whose values are
determined at run time). Also significantly, an overflow or underflow situation does not generate a run-time
error. It is, therefore, a logic error if a program performs an integral computation that, either as a final result
or an intermediate value, is outside the range of the integer type being used.

4.8.2 Floating-point Implementation

The standard C++ floating point types consist of float, double, and long double. Floating point
numbers can have fractional parts (decimal places), and the term floating point refers to the fact the dec-
imal point in a number can float left or right as necessary as the result of a calculation (for example,
2.5 3.3 = 8.25, two one-decimal place values produce a two-decimal place result). As with the integer
types, the different floating-point types may be distinguished by the number of bits of storage required and
corresponding range of values. The type float stands for single-precision floating-point, and double
stands for double-precision floating-point. Floating point numbers serve as rough approximations of math-
ematical real numbers, but as we shall see, they have some severe limitations compared to actual real
numbers.
On most modern computer systems floating-point numbers are stored internally in exponential form
according to the standard adopted by the Institute for Electrical and Electronic Engineers (IEEE 754). In
the decimal system, scientific notation is the most familiar form of exponential notation:

One mole contains 6.023 1023 molecules.

Here 6.023 is called the mantissa, and 23 is the exponent.

The IEEE 754 standard uses binary exponential notation; that is, the mantissa and exponent are binary
numbers. Single-precision floating-point numbers (type float) occupy 32 bits, distributed as follows:

Mantissa 24 bits
Exponent 7 bits
Sign 1 bit

Total 32 bits

Double-precision floating-point numbers (type double) require 64 bits:

Mantissa 52 bits
Exponent 11 bits
Sign 1 bit

Total 64 bits

The details of the IEEE 754 implementation are beyond the scope of this book, but a simplified example
serves to highlight the limitations of floating-point types in general. Recall the fractional place values in
the decimal system. The place values, from left to right, are

1 1 1 1
... 10, 000 1, 000 100 10 1 10 100 1000 10,000 ...
... 104 103 102 101 100 101 102 103 104 ...

2015 Richard L. Halterman Draft date: January 18, 2015

4.8. INTEGERS VS. FLOATING-POINT NUMBERS 66

Each place value is one-tenth the place value to its left. Move to the right, divide by ten; move to the left,
multiply by ten. In the binary system, the factor is two instead of ten:

1 1 1 1
... 16 8 4 2 1 2 4 8 16 ...
... 24 23 22 21 20 21 22 23 24 ...

As in our unsigned tiny example (see Section 4.8.1), consider a a binary exponential number that
consists of only five bitsfar fewer bits than either floats or doubles in C++. We will call our mythical
floating-point type tiny float. The first three bits of our 5-bit tiny float type will represent the
mantissa, and the remaining two bits store the exponent. The three bits of the mantissa all appear to the
right of the binary point. The base of the 2-bit exponent is, of course, two. Figure 4.8 illustrates such a
value. To simplify matters even more, neither the mantissa nor the exponent can be negative. Thus, with

. 21 20 21 2
1 0
1 0
22

1 2 3

Figure 4.8: A ting float simplified binary exponential value

three bits, the mantissa may assume one of eight possible values. Since two bits constitute the exponent
of tiny floats, the exponent may assume one of four possible values. Table 4.4 lists all the possible
values that tiny float mantissas and exponents may assume. The number shown in Figure 4.8 is thus

3-bit Mantissas
Bit String Binary Value Decimal Value
0 0 0
000 0.0002 + + = 0.000
2 4 8
0 0 1
001 0.0012 + + = 0.125
2 4 8 2-bit Exponents
0 1 0
010 0.0102 + + = 0.250 Bit String Binary Value Decimal Value
2 4 8
0 1 1 00 2002 20 = 1
011 0.0112 + + = 0.375 01 2012 21 = 2
2 4 8
1 0 0 10 2102 22 = 4
100 0.1002 + + = 0.500
2 4 8 11 2112 23 = 8
1 0 1
101 0.1012 + + = 0.625
2 4 8
1 1 0
110 0.1102 + + = 0.750
2 4 8
1 1 1
111 0.1112 + + = 0.875
2 4 8

Table 4.4: The eight possible mantissas and four possible exponents that make up all tiny float values

2015 Richard L. Halterman Draft date: January 18, 2015

4.8. INTEGERS VS. FLOATING-POINT NUMBERS 67

 
1 2 3 (121 +020 ) 1 1
(1 2 + 02 + 12 )2 = + 22
2 8
5
= 4
8
= 2.5

Table 4.5 combines the mantissas and exponents to reveal all possible tiny float values that we
can represent with the 32 different bit strings made up of five bits. The results are interesting.
Bit String Interpretation Decimal Equivalent Value
00000 .0002 2002 0.000 1 0.000
00001 .0002 2012 0.000 2 0.000
00010 .0002 2102 0.000 4 0.000
00011 .0002 2112 0.000 8 0.000
00100 .0012 2002 0.125 1 0.125
00101 .0012 2012 0.125 2 0.250
00110 .0012 2102 0.125 4 0.500
00111 .0012 2112 0.125 8 1.000
01000 .0102 2002 0.250 1 0.250
01001 .0102 2012 0.250 2 0.500
01010 .0102 2102 0.250 4 1.000
01011 .0102 2112 0.250 8 2.000
01100 .0112 2002 0.375 1 0.375
01101 .0112 2012 0.375 2 0.750
01110 .0112 2102 0.375 4 1.500
01111 .0112 2112 0.375 8 3.000
10000 .1002 2002 0.500 1 0.500
10001 .1002 2012 0.500 2 1.000
10010 .1002 2102 0.500 4 2.000
10011 .1002 2112 0.500 8 4.000
10100 .1012 2002 0.625 1 0.625
10101 .1012 2012 0.625 2 1.250
10110 .1012 2102 0.625 4 2.500
10111 .1012 2112 0.625 8 5.000
11000 .1102 2002 0.750 1 0.750
11001 .1102 2012 0.750 2 1.500
11010 .1102 2102 0.750 4 3.000
11011 .1102 2112 0.750 8 6.000
11100 .1112 2002 0.875 1 0.875
11101 .1112 2012 0.875 2 1.750
11110 .1112 2102 0.875 4 3.500
11111 .111 2 2112 0.875 8 7.000

Table 4.5: The tiny float values. The first three bits of the bit string constitute the mantissa, and the last three bits
represent the exponent. Notice that the 32 bit strings represent only 20 unique tiny float values.

The range of our tiny float numbers is 0 . . . 7. Just stating the range is misleading, however, since
1
it might give the impression that we may represent any value in between 0 and 7 down to the th place.
8
This, in fact, is not true. We can represent 2.5 with this scheme, but we have no way of expressing 2.25.
Figure 4.9 plots all the possible tiny float values on the real number line.
Table 4.5 and Figure 4.9 reveal several troubling issues about our tiny float type:

1. There are many gaps; for example, the value 2.4 is missing and thus cannot be represented exactly (2.5
is the closest approximation). As another example, 0.75 and 1.75 both appear, but 2.75 is missing.
2. The scheme duplicates some numbers; for example, three different bit patterns represent the decimal
value 0.5:
0.100 200 = 0.010 201 = 0.001 210 = 0.510

2015 Richard L. Halterman Draft date: January 18, 2015

4.8. INTEGERS VS. FLOATING-POINT NUMBERS 68

1 1 3 1 5 3 7 1 1 3 1 1
0 8 4 8 2 8 4 8 1 14 12 14 2 22 3 32 4 5 6 7

0 1 2 3 4 5 6 7

Figure 4.9: A plot of all the possible tiny float numbers on the real number line. Note that the numbers are more
dense near zero and become more sparse moving to the right. The precision in the range 0 . . . 1 is one-eighth. The
precision in the range 1 . . . 2 is only one-fourth, and over the range 2 . . . 4 it drops to one-half. Our tiny float type
can represent only whole numbers in the range 4 . . . 7.

This duplication limits the number of different values that can be represented by a given number of
bits. In our tiny float example 12 of the 32 bit strings (37.5%) are redundant.

3. The numbers are not uniformly dense. There are more values nearer to zero, and the numbers become
more sparse farther away from zero.

Our unsigned tiny type discussed in Section 4.8.1 exhibits none of these weaknesses. All integers
in a given range (0 . . . 31) are present, no two bit strings represent the same value, and the integers are
uniformly distributed across their specified range. While the standard integer types provided by C++ have
much greater ranges than our unsigned tiny type, they all share these same qualities: all values in
their ranges are present, and all bit strings represent unique integer values. The standard floating-point
types provided by C++ use many more bits than our tiny float type, yet they exhibit the same problems
shown to a much smaller degree: missing values, multiple bit patterns representing the same values, and
uneven distribution of values across their ranges. This is not solely a problem of C++s implementation
of floating-point numbers; all computer languages and hardware that adhere to the IEEE 754 standard
exhibit these problems. To overcome these problems and truly represent and compute with mathematical
real numbers we would need a computer with an infinite amount of memory along with an infinitely fast
processor.
Listing 4.14 (imprecisedifference.cpp) demonstrates the inexactness of floating-point arithmetic.

Listing 4.14: imprecisedifference.cpp

#include <iostream>
#include <iomanip>

using namespace std;

int main() {
double d1 = 2000.5;
double d2 = 2000.0;
cout << setprecision(16) << (d1 - d2) << endl;
double d3 = 2000.58;
double d4 = 2000.0;
cout << setprecision(16) << (d3 - d4) << endl;
}

The output of Listing 4.14 (imprecisedifference.cpp) is:

2015 Richard L. Halterman Draft date: January 18, 2015

4.8. INTEGERS VS. FLOATING-POINT NUMBERS 69

0.5
0.5799999999999272

The program uses an additional #include directive:

#include <iomanip>

This preprocessor directive allows us to use the setprecision output stream manipulator that directs
the cout output stream object to print more decimal places in floating-point values. During the programs
execution, the first subtraction yields the correct answer. We now know that some floating-point numbers
(like 0.5) have exact internal representations while others are only approximations. The exact answer for
the second subtraction should be 0.58, and if we round the reported result to 12 decimal places, the answer
matches. Floating-point arithmetic often produces results that are close approximations of the true answer.
Listing 4.15 (precise8th.cpp) computes zero in a roundabout way:

1 1 1 1 1 1 1 1
1 = 0
8 8 8 8 8 8 8 8

Listing 4.15: precise8th.cpp

#include <iostream>

using namespace std;

int main() {
double one = 1.0,
one_eighth = 1.0/8.0,
zero = one - one_eighth - one_eighth - one_eighth
- one_eighth - one_eighth - one_eighth
- one_eighth - one_eighth;

cout << "one = " << one << ", one_eighth = " << one_eighth
<< ", zero = " << zero << endl;

Listing 4.15 (precise8th.cpp) prints

one = 1, one_eighth = 0.125, zero = 0

1
The number has an exact decimal representation, 0.625. It also has an exact binary representation, 0.0012 .
8
1 1
Consider, however, . While = 0.2 has a finite representation in base 10, it has no finite representation in
5 5
base 2:
1
= 0.2 = 0.00110011001100112
5
2015 Richard L. Halterman Draft date: January 18, 2015
4.9. MORE ARITHMETIC OPERATORS 70

1
In the binary representation the 00112 bit sequence repeats without end. This means does not have an
5
1
exact floating-point representation. Listing 4.16 (imprecise5th.cpp) illustrates with arithmetic involving .
5
Listing 4.16: imprecise5th.cpp
#include <iostream>

using namespace std;

int main() {
double one = 1.0,
one_fifth = 1.0/5.0,
zero = one - one_fifth - one_fifth - one_fifth
- one_fifth - one_fifth;

cout << "one = " << one << ", one_fifth = " << one_fifth
<< ", zero = " << zero << endl;

one = 1, one_fifth = 0.2, zero = 5.55112e-017

Surely the reported answer (5.551122 1017 = 0.00000000000000005551122) is close to the correct
answer (zero). If you round it to the one-quadrillionth place (15 places behind the decimal point), it is
correct.
What are the ramifications for programmers of this inexactness of floating-point numbers? Section 9.4.6
shows how the misuse of floating-point values can lead to logic errors in programs.
Being careful to avoid overflow and underflow, integer arithmetic is exact and, on most computer sys-
tems, faster than floating-point arithmetic. If an application demands the absolute correct answer and inte-
gers are appropriate for the computation, you should choose integers. For example, in financial calculations
it is important to keep track of every cent. The exact nature of integer arithmetic makes integers an attractive
option. When dealing with numbers, an integer type should be the first choice of programmers.
The limitations of floating-point numbers are unavoidable since computers have finite resources. Com-
promise is inevitable even when when we do our best to approximate values with infinite characteristics in a
finite way. Despite their inexactness, double-precision floating-point numbers are used every day through-
out the world to solve sophisticated scientific and engineering problems; for example, the appropriate use
of floating-point numbers have enabled space probes to reach distant planets. In the example C++ programs
above that demonstrate the inexactness of floating-point numbers, the problems largely go away if we agree
that we must compute with the most digits possible and then round the result to fewer digits. Floating-point
numbers provide a good trade-off of precision for practicality.

4.9 More Arithmetic Operators

As Listing 4.11 (enhancedtimeconv.cpp) demonstrates, an executing program can alter a variables value
by performing some arithmetic on its current value. A variable may increase by one or decrease by five.

2015 Richard L. Halterman Draft date: January 18, 2015

4.9. MORE ARITHMETIC OPERATORS 71

The statement
x = x + 1;

increments x by one, making it one bigger than it was before this statement was executed. C++ has a shorter
statement that accomplishes the same effect:
x++;

This is the increment statement. A similar decrement statement is available:

x--; // Same as x = x - 1;

These statements are more precisely post-increment and post-decrement operators. There are also pre-
increment and pre-decrement forms, as in
--x; // Same as x = x - 1;
++y; // Same as y = y + 1;

When they appear alone in a statement, the pre- and post- versions of the increment and decrement
operators work identically. Their behavior is different when they are embedded within a more complex
statement. Listing 4.17 (prevspost.cpp) demonstrates how the pre- and post- increment operators work
slightly differently.

Listing 4.17: prevspost.cpp

#include <iostream>

using namespace std;

int main() {
int x1 = 1, y1 = 10, x2 = 100, y2 = 1000;
cout << "x1=" << x1 << ", y1=" << y1
<< ", x2=" << x2 << ", y2=" << y2 << endl;
y1 = x1++;
cout << "x1=" << x1 << ", y1=" << y1
<< ", x2=" << x2 << ", y2=" << y2 << endl;
y2 = ++x2;
cout << "x1=" << x1 << ", y1=" << y1
<< ", x2=" << x2 << ", y2=" << y2 << endl;
}

Listing 4.17 (prevspost.cpp) prints

x1=1, y1=10, x2=100, y2=1000

x1=2, y1=1, x2=100, y2=1000
x1=2, y1=1, x2=101, y2=101

If x1 has the value 1 just before the statement

y1 = x1++;

2015 Richard L. Halterman Draft date: January 18, 2015

4.9. MORE ARITHMETIC OPERATORS 72

then immediately after the statement executes x1 is 2 and y1 is 1.

If x1 has the value 1 just before the statement
y1 = ++x1;

then immediately after the statement executes x1 is 2 and y1 is also 2.

As you can see, the pre-increment operator uses the new value of the incremented variable when eval-
uating the overall expression. In contrast, the post-increment operator uses the original value of the in-
cremented variable when evaluating the overall expression. The pre- and post-decrement operator behaves
similarly.
For beginning programmers it is best to avoid using the increment and decrement operators within more
complex expressions. We will use them frequently as standalone statements since there is no danger of
misinterpreting their behavior when they are not part of a more complex expression.
C++ provides a more general way of simplifying a statement that modifies a variable through simple
arithmetic. For example, the statement
x = x + 5;

can be shorted to
x += 5;

This statement means increase x by five. Any statement of the form

x op= exp;

where

x is a variable.

op= is an arithmetic operator combined with the assignment operator; for our purposes, the ones most
useful to us are +=, -=, *=, /=, and %=.

exp is an expression compatible with the variable x.

Arithmetic reassignment statements of this form are equivalent to

x = x op exp;

This means the statement

x *= y + z;

is equivalent to
x = x * (y + z);

The version using the arithmetic assignment does not require parentheses. The arithmetic assignment is
especially handy if a variable with a long name is to be modified; consider
temporary_filename_length = temporary_filename_length / (y + z);

2015 Richard L. Halterman Draft date: January 18, 2015

4.10. BITWISE OPERATORS 73

versus
temporary_filename_length /= y + z;

Do not accidentally reverse the order of the symbols for the arithmetic assignment
operators, like in the statement
x =+ 5;
Notice that the + and = symbols have been reversed. The compiler interprets this
statement as if it had been written
x = +5;
that is, assignment and the unary operator. This assigns x to exactly five instead
of increasing it by five.

Similarly,
x =- 3;
would assign 3 to x instead of decreasing x by three.

Section 4.10 examines some additional operators available in C++.

4.10 Bitwise Operators

CAUTION! SECTION UNDER CONSTRUCTION

In addition to the common arithmetic operators introduced in Section 4.1, C++ provides a few other
special-purpose arithmetic operators. These special operators allow programmers to examine or manipulate
the individual bits that make up data values. They are known as the bitwise operators. These operators
consist of &, |, , , >>, and <<. Applications programmers generally do not need to use bitwise
operators very often, but bit manipulation is essential in many systems programming tasks.
Consider 32-bit unsigned integers. The bit positions usually are numbered right to left, starting with
zero. Figure 4.10 shows how the individual bit positions often are numbered.

31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0

Figure 4.10: The bit positions of a 32-bit C++ unsigned integer

The bitwise and operator, &, takes two integer subexpressions and computes an integer result. The
expression e1 & e2 is evaluated as follows:

If bit 0 in both e1 and e2 is 1, then bit 0 in the result is 1; otherwise, bit 0 in the result is 0.

2015 Richard L. Halterman Draft date: January 18, 2015

4.10. BITWISE OPERATORS 74

If bit 1 in both e1 and e2 is 1, then bit 1 in the result is 1; otherwise, bit 1 in the result is 0.

If bit 2 in both e1 and e2 is 1, then bit 2 in the result is 1; otherwise, bit 2 in the result is 0.

..
.

If bit 31 in both e1 and e2 is 1, then bit 31 in the result is 1; otherwise, bit 31 in the result is 0.

For example, the expression 13 & 14 evaluates to 12, since:

1310 = 000000000000000000000000000011012
& 1410 = 000000000000000000000000000011102
1210 = 000000000000000000000000000011002

Bits 2 and 3 are one for both 13 and 14; thus, bits 2 and 3 in the result must be one.
The bitwise or operator, |, takes two integer subexpressions and computes an integer result. The ex-
pression e1 | e2 is evaluated as follows:

If bit 0 in both e1 and e2 is 0, then bit 0 in the result is 0; otherwise, bit 0 in the result is 1.

If bit 1 in both e1 and e2 is 0, then bit 1 in the result is 0; otherwise, bit 1 in the result is 1.

If bit 2 in both e1 and e2 is 0, then bit 2 in the result is 0; otherwise, bit 2 in the result is 1.

..
.

If bit 31 in both e1 and e2 is 0, then bit 31 in the result is 0; otherwise, bit 31 in the result is 1.

For example, the expression 13 | 14 evaluates to 15, since:

1310 = 000000000000000000000000000011012
| 1410 = 000000000000000000000000000011102
1510 = 000000000000000000000000000011112

Bits 431 are zero in both 13 and 14. In bits 03 either 13 has a one or 14 has a one; therefore, the result
has ones in bits 03 and zeroes everywhere else.
The bitwise exclusive or (often refered to as xor) operator () takes two integer subexpressions and
computes an integer result. The expression e1 e2 is evaluated as follows:

If bit 0 in e1 is the same as bit 0 in e2 , then bit 0 in the result is 0; otherwise, bit 0 in the result is 1.

If bit 1 in e1 is the same as bit 1 in e2 , then bit 1 in the result is 0; otherwise, bit 1 in the result is 1.

If bit 2 in e1 is the same as bit 2 in e2 , then bit 2 in the result is 0; otherwise, bit 2 in the result is 1.

..
.

If bit 31 in e1 is the same as bit 31 in e2 , then bit 31 in the result is 0; otherwise, bit 31 in the result
is 1.

2015 Richard L. Halterman Draft date: January 18, 2015

4.10. BITWISE OPERATORS 75

For example, the expression 13 14 evaluates to 3, since:

1310 = 000000000000000000000000000011012
1410 = 000000000000000000000000000011102
310 = 000000000000000000000000000000112

Bits 0 and 1 differ in 13 and 14, so these bits are one in the result. The bits match in all the other positions,
so these positions must be set to zero in the result.
The bitwise negation operator ( ) is a unary operator that inverts all the bits of its expression. The
expression e is evaluated as follows:

If bit 0 in e is 0, then bit 0 in the result is 1; otherwise, bit 0 in the result is 0.

If bit 1 in e is 0, then bit 1 in the result is 1; otherwise, bit 1 in the result is 0.

If bit 2 in e is 0, then bit 2 in the result is 1; otherwise, bit 2 in the result is 0.

..
.

If bit 31 in e is 0, then bit 31 in the result is 1; otherwise, bit 31 in the result is 0.

For example, the unsigned expression 13u evaluates to 4, 294, 967, 282, since

1310 = 000000000000000000000000000011012
negate
1410 = 111111111111111111111111111100102

For signed integers the 31st bit stores the numbers sign. Signed integers use a representation called twos
complement binary, a slight variation of the standard binary layout. Suffice it to say that the int expression
13 evaluates to the same bit pattern as 13u, but as a signed integer it represents 14.
The shift operators move all the bits in an integer to the left or right:

Shift left (<<). The expression x << y, where x and y are integer types, shifts all the bits in x
to the left y places. Zeros fill vacated positions. The bits shifted off the left side are discarded.
The expression 5 << 2 evaluates to 20, since 510 = 1012 shifted two places to the left yields
101002 = 2010 . Observe that x << y is equal to x 2y .

Shift right (>>). The expression x >> y, where x and y are integer types, shifts all the bits in x to
the right y places. What fills the vacated bits on the left depends on whether the integer is signed or
unsigned (for example, int vs. unsigned):

For signed values the vacated bit positions are filled with the sign bit (the original leftmost bit).
For unsigned values the vacated bit positions are filled with zeros.

The bits shifted off the right side are discarded. The expression 5 >> 2 evaluates to 1, since
510 = 1012 shifted two places to the left yields 0012 = 2010 (the original bits in positions 1
and 0 are shifted off the end and lost). Observe that x >> y is equal to x 2y .

2015 Richard L. Halterman Draft date: January 18, 2015

4.10. BITWISE OPERATORS 76

Do not confuse the left shift operator (<<) with the output stream insertion opera-
tor (<<). The operators are identical, but the context differentiates them. If the left
operand is an integer type, << means left shift; if the left operand is a stream out-
put object like cout, << means send the right-hand operand to the output stream
object for display.
Similarly, the input stream object cin uses the >> for a different purpose from
the right shift operator used with integers.

Listing 4.18 (bitwiseoperators.cpp) experiments with bitwise operators.

Listing 4.18: bitwiseoperators.cpp

#include <iostream>

using namespace std;

int main() {
int x, y;
cout << "Please enter two integers: ";
cin >> x >> y;
cout << x << " & " << y << " = " << (x & y) << endl;
cout << x << " | " << y << " = " << (x | y) << endl;
cout << x << " " << y << " = " << (x y) << endl;
cout << "" << x << " = " << x << endl;
cout << x << " << " << 2 << " = " << (x << 2) << endl;
cout << x << " >> " << 2 << " = " << (x >> 2) << endl;
}

Developers use bitwise operations for a variety of systems-level programming tasks. For example, in a
graphical user interface (GUI), the user generates events by interacting with an application using the mouse
and a keyboard. One event might be clicking a mouse button over a particular graphical element (like a
button) within a window. Multiple pieces of information about this event can be stored in a single integer.
For example, bit 0 may indicate whether or not the Shift key was held down when the mouse button
was clicked. Bit 1 may be responsible for the Alt key, bit 3 for the Ctrl key, etc. Thus the integer value
5, which in binary is

00000000000000000000000000000101

means that when the mouse button was clicked both the Shift and Alt keys were being held down.
This might require a different action on the part of the program than if some other combination of keys
(or none) were being pressed. For example, suppose the int variable key_status holds information
about which keys the user was depressing during the most recent mouse click. Consider the expression
key_status & 1. The bit string

00000000000000000000000000000001

represents the expression 1, and the value of key_status is unknown, so key_status & 1 is
key status = ????????????????????????????????2
& 110 = 000000000000000000000000000000012
010 or 110 = 0000000000000000000000000000000?2
2015 Richard L. Halterman Draft date: January 18, 2015
4.10. BITWISE OPERATORS 77

If the answer is zero, this means bit 0 is 0 in key_status, and so the Shift key is not depressed. On
the other hand, if the answer is one, this means bit 0 is 1, and so the Shift key is depressed. In the
expression
key_status & 1

the 1 is called a mask, since it serves to mask off, or remove the first 31 bits of key_status.
Usually the GUI library will define constants that help programmers examine or modify particular bits;
for example, given the following constants:
const int SHIFT_DOWN = 1; // This is 0...0001
const int CTRL_DOWN = SHIFT_DOWN << 1; // This is 0...0010
const int ALT_DOWN = CTRL_DOWN << 1; // This is 0...0100

the expression
key_status & 1

is better written
key_status & SHIFT_DOWN

The expression
key_status & SHIFT_DOWN | ALT_DOWN

can test for both the Shift and Alt keys being down during the mouse event. Do you see how the
expression SHIFT_DOWN | ALT_DOWN means both keys are down simultaneously?
We can use masks to ensure that certain bits are on or off. To see how, consider the mask 5, which is

00000000000000000000000000000101

If x is a 32-bit integer variable, we can selectively turn on its bits 0 and 2 with the statement
x = x | 5;

Next, consider the unsigned value 4294967290u, which is 5u, or

11111111111111111111111111111010

If x is a 32-bit integer variable, we can selectively turn off its bits 0 and 2 with the statement
x = x & 4294967290u;

We cannot fully appreciate the utility of using bitwise operators for masking purposes until we consider
conditional execution in Chapter 5. Even then, since we concentrate on applications programming instead
of systems programming in this book, we will have little use for the bitwise operators except for a few
isolated situations. It is good to be aware of their presence, though, since their accidental use may lead to
difficult to diagnose compiler messages and logic errors.
Like the other C++ arithmetic operators that work on two operands, we may combine the bitwise binary
operators &, |, , <<, and >> with assignment to simplify the modification of a variable; for example, the
following statement

2015 Richard L. Halterman Draft date: January 18, 2015

4.11. ALGORITHMS 78

Assignment Short Cut

x = x & y; x &= y;
x = x | y; x |= y;
x = x y; x = y;
x = x << y; x <<= y;
x = x >> y; x >>= y;

Table 4.6: The bitwise assignment operators

x = x | y; // Turn on bits in x determined by

may be written as
x |= y; // Turn on bits in x determined by

Table 4.6 lists the possibilities.

4.11 Algorithms

An algorithm is a finite sequence of steps, each step taking a finite length of time, that solves a problem or
computes a result. A computer program is one example of an algorithm, as is a recipe to make lasagna. In
both of these examples, the order of the steps matter. In the case of lasagna, the noodles must be cooked
in boiling water before they are layered into the filling to be baked. It would be inappropriate to place the
raw noodles into the pan with all the other ingredients, bake it, and then later remove the already baked
noodles to cook them in boiling water separately. In the same way, the ordering of steps is very important
in a computer program. While this point may be obvious, consider the following sound argument:

1. The relationship between degrees Celsius and degrees Fahrenheit can be expressed as

5
C= ( F 32)
9

2. Given a temperature in degrees Fahrenheit, the corresponding temperature in degrees Celsius can be
computed.

Armed with this knowledge, Listing 4.19 (faultytempconv.cpp) follows directly.

Listing 4.19: faultytempconv.cpp

// File faultytempconv.cpp

#include <iostream>

using namespace std;

int main() {
double degreesF = 0, degreesC = 0;
// Define the relationship between F and C
degreesC = 5.0/9*(degreesF - 32);
// Prompt user for degrees F

2015 Richard L. Halterman Draft date: January 18, 2015

4.12. SUMMARY 79

cout << "Enter the temperature in degrees F: ";

// Read in the user's input
cin >> degreesF;
// Report the result
cout << degreesC << endl;
}

Unfortunately, the executing program always displays

-17.7778

regardless of the input provided. The English description provided above is correct. No integer division
problems lurk, as in Listing 4.9 (tempconv.cpp). The problem lies simply in statement ordering. The
statement
degreesC = 5.0/9*(degreesF - 32);

is an assignment statement, not a definition of a relationship that exists throughout the program. At the
point of the assignment, degreesF has the value of zero. The executing program computes and assigns
the degreesC variable before receiving degreesFs value from the user.
As another example, suppose x and y are two integer variables in some program. How would we
interchange the values of the two variables? We want x to have ys original value and y to have xs original
value. This code may seem reasonable:
x = y;
y = x;

The problem with this section of code is that after the first statement is executed, x and y both have the same
value (ys original value). The second assignment is superfluous and does nothing to change the values of
x or y. The solution requires a third variable to remember the original value of one the variables before it
is reassigned. The correct code to swap the values is
temp = x;
x = y;
y = temp;

This small example emphasizes the fact that algorithms must be specified precisely. Informal notions about
how to solve a problem can be valuable in the early stages of program design, but the coded program
requires a correct detailed description of the solution.
The algorithms we have seen so far have been simple. Statement 1, followed by Statement 2, etc. until
every statement in the program has been executed. Chapter 5 and Chapter 6 introduce some language
constructs that permit optional and repetitive execution of some statements. These constructs allow us to
build programs that do much more interesting things, but more complex algorithms are required to make it
happen. We must not lose sight of the fact that a complicated algorithm that is 99% correct is not correct.
An algorithms design and implementation can be derailed by inattention to the smallest of details.

4.12 Summary

The literal value 4 and integer sum are examples of simple C++ numeric expressions.

2015 Richard L. Halterman Draft date: January 18, 2015

4.12. SUMMARY 80

2*x + 4 is an example of a more complex C++ numeric expression.

Expressions can be printed via the cout output stream object and assigned to variables.

A binary operator performs an operation using two operands.

With regard to binary operators: + represents arithmetic addition; - represents arithmetic subtrac-
tion; * represents arithmetic multiplication; / represents arithmetic division; % represents arithmetic
modulus, or integer remainder after division.

The std::cin object can be used to assign user keyboard input to variables when the program is
executing.

The >> operator is called the extraction operator. It assigns to variables data extracted from the input
stream object cin.

A unary operator performs an operation using one operand.

The - unary operator represents the additive inverse of its operand.

The + unary operator has no effect on its operand.

Arithmetic applied to integer operands yields integer results.

With a binary operation, double-precision floating-point arithmetic is performed if at least one of its
operands is a floating-point number.

Floating-point arithmetic is inexact and subject to rounding errors because all floating-point values
have finite precision.

When a floating-point value is assigned to an integer variable, the value is truncated, not properly
rounded.

A mixed expression is an expression that contains values and/or variables of differing types.

Generally speaking, a type is narrower than another type if its range of values is smaller than the
other type; a type is wider than another type if its range of values is larger than the other type.

A wider type dominates a narrower type.

In mixed arithmetic with a binary operator, the type of arithmetic performed is determined by the
more dominant operand.

Floating point values can be assigned to integer variables, but any fractional part will be truncated.
Programmers must ensure that the value assigned falls within the range of integers.

In C++ a wider type may be assigned to a narrower type, but a floating-point to integer conversion
truncates, and a value outside the range of the narrower type results in a bogus value being assigned.
Programmers should use caution when assigning a wider type to a narrower type.

In C++, operators have both a precedence and an associativity.

With regard to the arithmetic operators, C++ uses the same precedence rules as standard arithmetic:
multiplication and division are applied before addition and subtraction unless parentheses dictate
otherwise.

The arithmetic operators associate left to right; assignment associates right to left.

2015 Richard L. Halterman Draft date: January 18, 2015

4.12. SUMMARY 81

Chained assignment can be used to assign the same value to multiple variables within one statement.
The unary operators + and - have precedence over the binary arithmetic operators *, /, and %,
which have precedence over the binary arithmetic operators + and -, which have precedence over the
assignment operator.
Comments are notes within the source code. The compiler ignores comments when compiling the
source code.
Comments inform human readers about the code.
Comments should not state the obvious, but it is better to provide too many comments rather than too
few.
A single line comment begins with the symbols // and continues until the end of the line.
A block comment begins with the symbols /* and continues in the source code until the symbols */
terminate it.
Source code should be formatted so that it is more easily read and understood by humans.
C++ has some general formatting guidelines that should be followed:
Each statement should appear on its own line.
Curly braces should be aligned in order to be matched visually more quickly.
The statements that comprise the body of a function should be indented; four spaces are gener-
ally accepted as ideal.
Spaces should be used within statements separating the logical pieces to make the statements
easier to read.
Rewrite Listing 4.1 (adder.cpp) in four different ways:
1. Use the K & R coding style
2. Use the ANSI coding style
3. Use the Whitesmith coding style
4. Use the Banner coding style
Compile-time errors are caused by the programmers misuse of the C++ language. Compile-time
errors are detected and reported by the compiler (and sometimes the linker).
Runtime errors are errors that are detected when the program is executing. When a run-time error
arises, the program terminates with an error message.
Logic errors elude detection by the compiler and run-time environment. A logic error is indicated
when the the program does not behave as expected.
The compiler should be set at the highest warning level to check for as many problems as possible.
Warnings should be taken seriously, and a program should not be considered finished while warnings
remain.
In complicated arithmetic expressions involving many operators and operands, the rules pertaining
to mixed arithmetic are applied on an operator-by-operator basis, following the precedence and asso-
ciativity laws, not globally over the entire expression.
The ++ and -- operators increment and decrement variables.

2015 Richard L. Halterman Draft date: January 18, 2015

4.13. EXERCISES 82

The family of op= operators (+=, -=, *=, /=, and %=) allow variables to be changed by a given
amount using a particular arithmetic operator.

C++ programs implement algorithms; as such, C++ statements do not declare statements of fact or
define relationships that hold throughout the programs execution; rather they indicate how the values
of variables change as the execution of the program progresses.

4.13 Exercises

1. Is the literal 4 a valid C++ expression?

2. Is the variable x a valid C++ expression?

3. Is x + 4 a valid C++ expression?

4. What affect does the unary + operator have when applied to a numeric expression?

5. Sort the following binary operators in order of high to low precedence: +, -, *, /, %, =.

6. Write a C++ program that receives two integer values from the user. The program then should print the
sum (addition), difference (subtraction), product (multiplication), quotient (division), and remainder
after division (modulus). Your program must use only integers.
A sample program run would look like (the user enters the 10 and the 2 after the colons, and the
program prints the rest):

Please enter the first number: 10

Please enter the second number: 2
10 + 2 = 12
10 - 2 = 8
10 * 2 = 20
10 / 2 = 5
10 % 2 = 0

Can you explain the results it produces for all of these operations?

7. Write a C++ program that receives two double-precision floating-point values from the user. The
program then should print the sum (addition), difference (subtraction), product (multiplication), and
quotient (division). Your program should use only integers.
A sample program run would look like (the user enters the 10 and the 2.5 after the colons, and the
program prints the rest):

Please enter the first number: 10

Please enter the second number: 2.5
10 + 2.5 = 12.5
10 - 2.5 = 7.5
10 * 2.5 = 25
10 / 2.5 = 4

2015 Richard L. Halterman Draft date: January 18, 2015

4.13. EXERCISES 83

Can you explain the results it produces for all these operations? What happens if you attempt to
compute the remainder after division (modulus) with double-precision floating-point values?

8. Given the following declaration:

int x = 2;

Indicate what each of the following C++ statements would print.

(a) cout << "x"<< endl;

(b) cout << 'x'<< endl;
(c) cout << x << endl;
(d) cout << "x + 1"<< endl;
(e) cout << 'x'+ 1 << endl;
(f) cout << x + 1 << endl;

9. Sort the following types in order from narrowest to widest: int, double, float, long, char.

10. Given the following declarations:

int i1 = 2, i2 = 5, i3 = -3;
double d1 = 2.0, d2 = 5.0, d3 = -0.5;

Evaluate each of the following C++ expressions.

(a) i1 + i2
(b) i1 / i2
(c) i2 / i1
(d) i1 * i3
(e) d1 + d2
(f) d1 / d2
(g) d2 / d1
(h) d3 * d1
(i) d1 + i2
(j) i1 / d2
(k) d2 / i1
(l) i2 / d1
(m) i1/i2*d1
(n) d1*i1/i2
(o) d1/d2*i1
(p) i1*d1/d2
(q) i2/i1*d1
(r) d1*i2/i1
(s) d2/d1*i1
(t) i1*d2/d1

2015 Richard L. Halterman Draft date: January 18, 2015

4.13. EXERCISES 84

11. What is printed by the following statement:

cout << /* 5 */ 3 << endl;

12. Given the following declarations:

int i1 = 2, i2 = 5, i3 = -3;
double d1 = 2.0, d2 = 5.0, d3 = -0.5;

Evaluate each of the following C++ expressions.

(a) i1 + (i2 * i3)
(b) i1 * (i2 + i3)
(c) i1 / (i2 + i3)
(d) i1 / i2 + i3
(e) 3 + 4 + 5 / 3
(f) (3 + 4 + 5) / 3
(g) d1 + (d2 * d3)
(h) d1 + d2 * d3
(i) d1 / d2 - d3
(j) d1 / (d2 - d3)
(k) d1 + d2 + d3 / 3
(l) (d1 + d2 + d3) / 3
(m) d1 + d2 + (d3 / 3)
(n) 3 * (d1 + d2) * (d1 - d3)
13. How are single-line comments different from block comments?
14. Can block comments be nested?
15. Which is better, too many comments or too few comments?
16. What is the purpose of comments?
17. The programs in Listing 3.4 (variable.cpp), Listing 4.3 (reformattedvariable.cpp), and Listing 4.4
(reformattedvariable2.cpp) compile to the same machine code and behave exactly the same. What
makes one of the programs clearly better than the others?
18. Why is human readability such an important consideration?
19. Consider the following program which contains some errors. You may assume that the comments
within the program accurately describe the programs intended behavior.
#include <iostream>

using namespace std;

int main() {
int n1, n2, d1; // 1
// Get two numbers from the user
cin << n1 << n2; // 2

2015 Richard L. Halterman Draft date: January 18, 2015

4.13. EXERCISES 85

// Compute sum of the two numbers

cout << n1 + n2 << endl; // 3
// Compute average of the two numbers
cout << n1+n2/2 << endl; // 4
// Assign some variables
d1 = d2 = 0; // 5
// Compute a quotient
cout << n1/d1 << endl; // 6
// Compute a product
n1*n2 = d1; // 7
// Print result
cout << d1 << endl; // 8
}

For each line listed in the comments, indicate whether or not a compile-time, run-time, or logic error
is present. Not all lines contain an error.

20. What distinguishes a compiler warning from a compiler error? Should you be concerned about
warnings? Why or why not?

21. What are the advantages to enhancing the warning reporting capabilities of the compiler?

22. Write the shortest way to express each of the following statements.

(a) x = x + 1;
(b) x = x / 2;
(c) x = x - 1;
(d) x = x + y;
(e) x = x - (y + 7);
(f) x = 2*x;
(g) number_of_closed_cases = number_of_closed_cases + 2*ncc;

23. What is printed by the following code fragment?

int x1 = 2, y1, x2 = 2, y2;
y1 = ++x1;
y2 = x2++;
cout << x1 << " " << x2 << endl;
cout << y1 << " " << y2 << endl;

Why does the output appear as it does?

24. Consider the following program that attempts to compute the circumference of a circle given the
radius entered by the user. Given a circles radius, r, the circles circumference, C is given by the
formula:

C = 2r

2015 Richard L. Halterman Draft date: January 18, 2015

4.13. EXERCISES 86

#include <iostream>

using namespace std;

int main() {
double C, r;
const double PI = 3.14159;
// Formula for the area of a circle given its radius
C = 2*PI*r;
// Get the radius from the user
cout >> "Please enter the circle's radius: ";
cin << r;
// Print the circumference
cout << "Circumference is " << C << endl;
}

(a) The compiler issues a warning. What is the warning?

(b) The program does not produce the intended result. Why?
(c) How can it be repaired so that it not only eliminates the warning but also removes the logic
error?

25. In mathematics, the midpoint between the two points (x1 , y1 ) and (x2 , y2 ) is computed by the formula
 
x1 + x2 y1 + y2
,
2 2

Write a C++ program that receives two mathematical points from the user and computes and prints
their midpoint.
A sample run of the program produces

Please enter the first point: (0,0)

Please enter the second point: (1,1)
The midpoint of (0,0) and (1,1) is (0.5,0.5)

The user literally enters (0,0) and (1,1) with the parentheses and commas as shown. To see how
to do this, suppose you want to allow a user to enter the point (2.3, 9), assigning the x component
of the point to a variable named x and the y component to a variable named y. You can add the
following code fragment to your program to achieve the desired effect:
double x, y;
char left_paren, comma, right_paren;
cin >> left_paren >> x >> comma >> y >> right_paren;

If the user literally types (2.3,9), the cin statement will assign the ( character to the variable
left_paren. It next will assign 2.3 to the variable x. It assigns the , character to the variable
named comma, the value 9 to the y variable, and the ) character to the right_paren variable.
The left_paren, comma, and right_paren variables are just placeholders for the users input
and are not used elsewhere within the program. In reality, the user can type in other characters in
place of the parentheses and comma as long as the numbers are in the proper location relative to the

2015 Richard L. Halterman Draft date: January 18, 2015

4.13. EXERCISES 87

Food Calories
Bean burrito 357
Salad w/dressing 185
Milkshake 388

Table 4.7: Calorie content of several fast food items

characters; for example, the user can type *2.3:9#, and the program will interpret the input as the
point (2.3, 9).
26. Table 4.7 lists the Calorie contents of several foods. Running or walking burns off about 100 Calories
per mile. Write a C++ program that requests three values from the user: the number of bean burritos,
salads, and shakes consumed (in that order). The program should then display the number of miles
that must be run or walked to burn off the Calories represented in that food. The program should run
as follows (the user types in the 3 2 1):

How many bean burritos, bowls of salad, and milkshakes did you consume? 3 2 1

You ingested 1829 Calories

You will have to run 18.29 miles to expend that much energy

Observe that the result is a floating-point value, so you should use floating-point arithmetic to com-
pute the answers for this problem.

2015 Richard L. Halterman Draft date: January 18, 2015

4.13. EXERCISES 88

2015 Richard L. Halterman Draft date: January 18, 2015

 89

Chapter 5

Conditional Execution

All the programs in the preceding chapters execute exactly the same statements regardless of the input, if
any, provided to them. They follow a linear sequence: Statement 1, Statement 2, etc. until the last statement
is executed and the program terminates. Linear programs like these are very limited in the problems they
can solve. This chapter introduces constructs that allow program statements to be optionally executed,
depending on the context (input) of the programs execution.

5.1 Type bool

Arithmetic expressions evaluate to numeric values; a Boolean expression, sometimes called a predicate,
evaluates to true or false. While Boolean expressions may appear very limited on the surface, they are
essential for building more interesting and useful programs.
C++ supports the non-numeric data type bool, which stands for Boolean. The term Boolean comes
from the name of the British mathematician George Boole. A branch of discrete mathematics called
Boolean algebra is dedicated to the study of the properties and the manipulation of logical expressions.
Compared to the numeric types, the bool type is very simple in that it can represent only two values:
true or false. Listing 5.1 (boolvars.cpp) is a simple program demonstrating the use of Boolean vari-
ables.

Listing 5.1: boolvars.cpp

#include <iostream>

using namespace std;

int main() {
// Declare some Boolean variables
bool a = true, b = false;
cout << "a = " << a << ", b = " << b << endl;
// Reassign a
a = false;
cout << "a = " << a << ", b = " << b << endl;
// Mix integers and Booleans
a = 1;
b = 1;

2015 Richard L. Halterman Draft date: January 18, 2015

5.2. BOOLEAN EXPRESSIONS 90

Operator Meaning
== Equal to
< Less than
> Greater than
<= Less than or equal to
>= Greater than or equal to
!= Not equal to

Table 5.1: C++ Relational operators

cout << "a = " << a << ", b = " << b << endl;
// Assign Boolean value to an integer
int x = a, y = true;
cout << "a = " << a << ", b = " << b
<< ", x = " << x << ", y = " << y << endl;
// More mixing
a = 1725; // Warning issued
b = -19; // Warning issued
cout << "a = " << a << ", b = " << b << endl;
}

As you can see from running Listing 5.1 (boolvars.cpp), the Boolean values false and true are
represented as integer 0 and integer 1. More precisely, zero represents the bool value false, and any
non-zero integer (positive or negative) means true. The direct assignment to a bool variable of an integer
other than 0 or 1 may result in a warning (Visual C++ reports truncation of int to bool), but the variable
is still interpreted as true. The data type bool is basically a convenience for programmers; any C++
program that uses bool variables can be rewritten using integers instead to achieve the same results. While
Boolean values and variables are freely compatible and interchangeable with integers, the bool type is
convenient and should be used when the context involves truth values instead of numbers.
It is important to note that the Visual C++ compiler issues warnings for the last two assignment state-
ments in Listing 5.1 (boolvars.cpp). Even though any non-zero value is considered true, 1 is the preferred
integer equivalent to true (as you can see when you attempt to print the literal value true). Since the
need to assign to a Boolean variable a value other than true or false or the equivalent 1 or 0 should be
extremely rare, the compilers message alerts the programmer to check to make sure the assignment is not
a mistake.

5.2 Boolean Expressions

The simplest Boolean expressions are false and true, the Boolean literals. A Boolean variable is also
a Boolean expression. An expression comparing numeric expressions for equality or inequality is also a
Boolean expression. The simplest kinds of Boolean expressions use relational operators to compare two
expressions. Table 5.1 lists the relational operators available in C++.
Table 5.2 shows some simple Boolean expressions with their associated values. An expression like
10 < 20 is legal but of little use, since the expression true is equivalent, simpler, and less likely to
confuse human readers. Boolean expressions are extremely useful when their truth values depend on the
values of one or more variables.
The relational operators are binary operators and are all left associative. They all have a lower prece-

2015 Richard L. Halterman Draft date: January 18, 2015

5.2. BOOLEAN EXPRESSIONS 91

Expression Value
10 < 20 always true
10 >= 20 always false
x == 10 true only if x has the value 10
X != y true unless x and y have the same values

Table 5.2: Relational operator examples

dence than any of the arithmetic operators; therefore, the expression

x + 2 < y / 10

is evaluated as if parentheses were placed as so:

(x + 2) < (y / 10)

C++ allows statements to be simple expressions; for example, the statement

x == 15;
may look like an attempt to assign the value 15 to the variable x, but it is not.
The = operator performs assignment, but the == operator checks for relational
equality. If you make a mistake and use == as shown here, Visual C++ will issue
a warning that includes the message
warning C4553: == : operator has no effect; did you intend =?

Recall from Section 4.6.4 that a compiler warning does not indicate a violation of
the rules of the language; rather it alerts the programmer to a possible trouble spot
in the code.

Another example of an expression used as a statement is

x + 15;
This statement is a legal (but useless) C++ statement, and the compiler notifies us
accordingly:
warning C4552: + : operator has no effect; expected operator with
side-effect

Why are expressions allowed as statements? Some simple expressions have side
effects that do alter the behavior of the program. One example of such an ex-
pression is x++. Listing 4.17 (prevspost.cpp) showed how x++ behaves both as a
standalone statement and as an expression within a larger statement. A more com-
mon example is the use of a function call (which is an expression) as standalone
a statement. (We introduce functions in Chapter 8.) In order to keep the structure
of the language as uniform as possible, C++ tolerates useless expressions as state-
ments to enable programmers to use the more useful expression-statements. For-
tunately, most compilers issue informative warnings about the useless expression-
statements to keep developers on track.

2015 Richard L. Halterman Draft date: January 18, 2015

5.3. THE SIMPLE IF STATEMENT 92

5.3 The Simple if Statement

The Boolean expressions described in Section 5.2 at first may seem arcane and of little use in practical
programs. In reality, Boolean expressions are essential for a program to be able to adapt its behavior at run
time. Most truly useful and practical programs would be impossible without the availability of Boolean
expressions.
The run-time exceptions mentioned in Section 4.6 arise from logic errors. One way that Listing 4.5
(dividedanger.cpp) can fail is when the user enters a zero for the divisor. Fortunately, programmers can take
steps to ensure that division by zero does not occur. Listing 5.2 (betterdivision.cpp) shows how it might be
done.

Listing 5.2: betterdivision.cpp

#include <iostream>

using namespace std;

int main() {
int dividend, divisor;

// Get two integers from the user

cout << "Please enter two integers to divide:";
cin >> dividend >> divisor;
// If possible, divide them and report the result
if (divisor != 0)
cout << dividend << "/" << divisor << " = "
<< dividend/divisor << endl;
}

The second cout statement may not always be executed. In the following run

Please enter two integers to divide: 32 8

32/8 = 4

it is executed, but if the user enters a zero as the second number:

Please enter two integers to divide: 32 0

the program prints nothing after the user enters the values.
The last statement in Listing 5.2 (betterdivision.cpp) begins with the reserved word if. The if state-
ment allows code to be optionally executed. In this case, the printing statement is executed only if the
variable divisors value is not zero.
The Boolean expression
divisor != 0

2015 Richard L. Halterman Draft date: January 18, 2015

5.3. THE SIMPLE IF STATEMENT 93

determines if the single statement that follows the right parenthesis is executed. If divisor is not zero,
the message is printed; otherwise, the program prints nothing.
Figure 5.1 shows how program execution flows through the if statement. of Listing 5.2 (betterdivision.cpp).

Is no
divisor 0?

yes

do the division
and print result

Figure 5.1: if flowchart

The general form of a simple if statement is

if ( condition )

statement

The reserved word if begins the if statement.

The Boolean expression condition determines whether or not the body will be executed. The Boolean
expression must be enclosed within parentheses as shown.
The statement is the statement to be executed if the Boolean expression is true. The statement makes
up the body of the if statement. Section 5.4 shows how the body can be composed of multiple
statements.

2015 Richard L. Halterman Draft date: January 18, 2015

5.3. THE SIMPLE IF STATEMENT 94

Good coding style dictates the body should be indented to emphasize the optional execution and improve
the programs readability. The indenting is not required by the compiler. Sometimes programmers will place
a short body on the same time as the if; for example, the following if statement optionally assigns y:

if (x < 10)
y = x;

and could be written as

if (x < 10) y = x;

but should not be written as

if (x < 10)
y = x;

because the lack of indentation hides the fact that the program optionally executes the assignment statement.
The compiler will accept it, but it is misleading to human readers accustomed to the indentation convention.
The compiler, of course, will accept the code written as

if(x<10)y=x;

but the lack of spaces makes it difficult for humans to read.

When the if statement is written the preferred way using two lines of source
code, it is important not to put a semicolon at the end of the first line:
if (x < 10); // No! Don't do this!
y = x;
Here, the semicolon terminates the if statement, but the indentation implies that
the second line is intended to be the body of the if statement. The compiler,
however, interprets the badly formatted if statement as if it were written as
if (x < 10)
; // This is what is really going on.
y = x;
This is legal in C++; it means the if statement has an empty body. In which
case the assignment is not part of the body. The assignment statement is after
the body and always will be executed regardless of the truth value of the Boolean
expression.

2015 Richard L. Halterman Draft date: January 18, 2015

5.4. COMPOUND STATEMENTS 95

When checking for equality, as in

if (x == 10)
cout << "ten";
be sure to use the relational equality operator (==), not the assignment opera-
tor (=). Since an assignment statement has a value (the value that is assigned,
see Section 4.3), C++ allows = within the conditional expression. It is, however,
almost always a mistake when beginning programmers use = in this context. Vi-
sual C++ at warning Level 4 checks for the use of assignment within a conditional
expression; the default Level 3 does not.

5.4 Compound Statements

Sometimes you need to optionally execute more than one statement based on a particular condition. List-
ing 5.3 (alternatedivision.cpp) shows how you must use curly braces to group multiple statements together
into one compound statement.

Listing 5.3: alternatedivision.cpp

#include <iostream>

using namespace std;

int main() {
int dividend, divisor, quotient;

// Get two integers from the user

cout << "Please enter two integers to divide:";
cin >> dividend >> divisor;
// If possible, divide them and report the result
if (divisor != 0) {
quotient = dividend / divisor;
cout << dividend << " divided by " << divisor << " is "
<< quotient << endl;
}
}

The assignment statement and printing statement are both a part of the body of the if statement. Given
the truth value of the Boolean expression divisor != 0 during a particular program run, either both
statements will be executed or neither statement will be executed.
A compound statement consists of zero or more statements grouped within curly braces. We say the
curly braces define a block of statements. As a matter of style many programmers always use curly braces
to delimit the body of an if statement even if the body contains only one statement:
if (x < 10) {
y = x;
}

2015 Richard L. Halterman Draft date: January 18, 2015

5.5. THE IF/ELSE STATEMENT 96

They do this because it is easy to introduce a logic error if additional statements are added to the body later
and the programmer forgets to add then required curly braces.

The format of the following code

if (x < 10)
y = x;
z = x + 5;
implies that both assignments are part of the body of the if statement. Since
multiple statements making up the body must be in a compound statement within
curly braces, the compiler interprets the code fragment as if it had been written
if (x < 10)
y = x;
z = x + 5;
Such code will optionally execute the first assignment statement and always exe-
cute the second assignment statement.

The programmer probably meant to write it as

if (x < 10) {
y = x;
z = x + 5;
}

The curly braces are optional if the body consists of a single statement. If the body consists of only one
statement and curly braces are not used, then the semicolon that terminates the statement in the body also
terminates the if statement. If curly braces are used to delimit the body, a semicolon is not required after
the bodys close curly brace.
An empty pair of curly braces represents an empty block. An empty block is a valid compound state-
ment.

5.5 The if/else Statement

One undesirable aspect of Listing 5.2 (betterdivision.cpp) is if the user enters a zero divisor, the program
prints nothing. It may be better to provide some feedback to the user to indicate that the divisor provided
cannot be used. The if statement has an optional else clause that is executed only if the Boolean expres-
sion is false. Listing 5.4 (betterfeedback.cpp) uses the if/else statement to provide the desired effect.

Listing 5.4: betterfeedback.cpp

#include <iostream>

using namespace std;

int main() {
int dividend, divisor;

2015 Richard L. Halterman Draft date: January 18, 2015

5.5. THE IF/ELSE STATEMENT 97

// Get two integers from the user

cout << "Please enter two integers to divide:";
cin >> dividend >> divisor;
// If possible, divide them and report the result
if (divisor != 0)
cout << dividend << "/" << divisor << " = "
<< dividend/divisor << endl;
else
cout << "Division by zero is not allowed" << endl;
}

A given program run will execute exactly one of either the if body or the else body. Unlike in
Listing 5.2 (betterdivision.cpp), a message is always displayed.

Please enter two integers to divide: 32 0

Division by zero is not allowed

The else clause contains an alternate body that is executed if the condition is false. The programs flow
of execution is shown in Figure 5.2.

yes Is no
divisor 0?

do the division
castigate user
and print result

Figure 5.2: if/else flowchart

2015 Richard L. Halterman Draft date: January 18, 2015

5.5. THE IF/ELSE STATEMENT 98

Listing 5.4 (betterfeedback.cpp) avoids the division by zero run-time error that causes the program to
terminate prematurely, but it still alerts the user that there is a problem. Another application may handle
the situation in a different way; for example, it may substitute some default value for divisor instead of
zero.
The general form of an if/else statement is

if ( condition )

statement 1

else

statement 2

The reserved word if begins the if/else statement.

The condition is a Boolean expression that determines whether the running program will execute
statement 1 or statement 2. As with the simple if statement, the condition must appear within
parentheses.
The program executes statement 1 if the condition is true. To make the if/else statement more
readable, indent statement 1 more spaces than the if line. This part of the if statement is sometimes
called the body of the if.
The reserved word else begins the second part of the if/else statement.
The program executes statement 2 if the condition is false. To make the if/else statement more
readable, indent statement 2 more spaces than the else line. This part of the if/else statement is
sometimes called the body of the else.

The body of the else clause of an if/else statement may be a compound statement:
if (x == y)
cout << x;
else {
x = 0;
cout << y;
}

or the if body alone may be a compound statement:

if (x == y) {
cout << x;
x = 0;
}
else
cout << y;

2015 Richard L. Halterman Draft date: January 18, 2015

5.5. THE IF/ELSE STATEMENT 99

or both parts may be compound:

if (x == y) {
cout << x;
x = 0;
}
else {
cout << y;
y = 0;
}

or, as in Listing 5.4 (betterfeedback.cpp), both the if body and the else body can be simple statements.

Remember, if you wish to associate more than one statement with the body of
the if or else, you must use a compound statement. Compound statements are
enclosed within curly braces ({}).

If you ever attempt to use an if/else statement and discover that you need to leave the else clause
empty, as in
if (x == 2)
cout << "x = " << x << endl;
else
; // Nothing to do otherwise

or, using a slightly different syntax, as

if (x == 2)
cout << "x = " << x << endl;
else {
} // Nothing to do otherwise

you instead should use a simple if statement:

if (x == 2)
cout << "x = " << x << endl;

The empty else clauses shown above do work, but they complicate the code and make it more difficult for
humans to read.
Due to the imprecise representation of floating-point numbers (see Listing 4.16 (imprecise5th.cpp) in
Section 4.1), programmers must use caution when using the equality operator (==) by itself to compare
floating-point expressions. Listing 5.5 (samedifferent.cpp) uses an if/else statement to demonstrate the
perils of using the equality operator with floating-point quantities.

Listing 5.5: samedifferent.cpp

#include <iostream>
#include <iomanip>

using namespace std;

2015 Richard L. Halterman Draft date: January 18, 2015

5.6. COMPOUND BOOLEAN EXPRESSIONS 100

int main() {
double d1 = 1.11 - 1.10,
d2 = 2.11 - 2.10;
cout << "d1 = " << d1 << endl;
cout << "d2 = " << d2 << endl;
if (d1 == d2)
cout << "Same" << endl;
else
cout << "Different" << endl;
cout << "d1 = " << setprecision(20) << d1 << endl;
cout << "d2 = " << setprecision(20) << d2 << endl;
}

In Listing 5.5 (samedifferent.cpp) the displayed values of d1 and d2 are rounded so they appear equivalent,
but internally the exact representations are slightly different. By including the header iomanip we can
use the setprecision stream manipulator to force cout to display more decimal places in the floating-
point number it prints. Observe from the output of Listing 5.5 (samedifferent.cpp) that the two quantities
that should be identically 0.01 are actually slightly different.

d1 = 0.01
d2 = 0.01
Different
d1 = 0.010000000000000009
d2 = 0.0099999999999997868

This result should not discourage you from using floating-point numbers where they truly are needed.
In Section 9.4.6 we will see how to handle floating-point comparisons properly.

5.6 Compound Boolean Expressions

Simple Boolean expressions, each involving one relational operator, can be combined into more complex
Boolean expressions using the logical operators && (and), || (or), and ! (not). A combination of two or
more Boolean expressions using logical operators is called a compound Boolean expression.
To introduce compound Boolean expressions, consider a computer science degree that requires, among
other computing courses, Operating Systems and Programming Languages. If we isolate those two courses,
we can say a student must successfully complete both Operating Systems and Programming Languages to
qualify for the degree. A student that passes Operating Systems but not Programming Languages will not
have met the requirements. Similarly, Programming Languages without Operating Systems is insufficient,
and a student completing neither Operating Systems nor Programming Languages surely does not qualify.
Logical AND works in exactly the same way. If e1 and e2 are two Boolean expressions, e1 && e2 is true
only if e1 and e2 are both true; if either one is false or both are false, the compound expression is false.
To illustrate logical OR, consider two mathematics courses, Differential Equations and Linear Algebra.
A computer science degree requires one of those two courses. A student who successfully completes
Differential Equations but does not take Linear Algebra meets the requirement. Similarly, a student may
take Linear Algebra but not Differential Equations. It is important to note the a student may elect to take

2015 Richard L. Halterman Draft date: January 18, 2015

5.6. COMPOUND BOOLEAN EXPRESSIONS 101

e1 e2 e1 && e2 e1 || e2 !e1
false false false false true
false true false true true
true false false true false
true true true true false

Table 5.3: Logical operatorse1 and e2 are Boolean expressions

both Differential Equations and Linear Algebra (perhaps on the way to a mathematics minor), but the
requirement is no less fulfilled.
Logical OR works in a similar fashion. Given our Boolean expressions e1 and e2 , the compound ex-
pression e1 || e2 is false only if e1 and e2 are both false; if either one is true or both are true, the compound
expression is true. Note that logical OR is an inclusive or, not an exclusive or. In informal conversion we
often imply exclusive or in a statement like Would you like cake or ice cream for dessert? The implica-
tion is one or the other, not both. In computer programming the or is inclusive; if both subexpressions in
an or expression are true, the or expression is true.
Logical NOT simply reverses the truth value of the expression to which it is applied. If e is a true
Boolean expression, !e is false; if e is false, !e is true.
Table 5.3 is called a truth table. It shows all the combinations of truth values for two simple expres-
sions and the values of compound Boolean expressions built from applying the &&, ||, and ! C++ logical
operators.
Both && and || are binary operators; that is, they require two operands, both of which must be Boolean
expressions. Logical not (!) is a unary operator (see Section 4.1); it requires a single Boolean operand
immediately to its right.
Operator ! has higher precedence than both && and ||. && has higher precedence than ||. && and ||
are left associative; ! is right associative. && and || have lower precedence than any other binary operator
except assignment. This means the expression

x <= y && x <= z

is evaluated

(x <= y) && (x <= z)

Some programmers prefer to use the parentheses as shown here even though they are not required. The
parentheses improve the readability of complex expressions, and the compiled code is no less efficient.

2015 Richard L. Halterman Draft date: January 18, 2015

5.6. COMPOUND BOOLEAN EXPRESSIONS 102

The relational operators such as < compare two operands. The result of the com-
parison is a Boolean value, which is freely convertible to an integer. The misappli-
cation of relational operators can lead to surprising results; consider, for example,
the expression
1 <= x <= 10
This expression is always true, regardless of the value of x! If the programmers
intent is to represent the mathematical notion of x falling within the range 1...10
inclusive, as in 1 x 10, the above C++ expression is not equivalent.

The expression
1 <= x <= 10
is evaluated as
(1 <= x) <= 10
If x is greater than or equal to one, the subexpression 1 <= x evaluates to
true, or integer 1. Integer 1, however, is always less than 10, so the overall expres-
sion is true. If instead x is less than one, the subexpression 1 <= x evaluates
to false, or integer 0. Integer 0 is always less than 10, so the overall expression is
true. The problem is due to the fact that C++ does not strictly distinguish between
Boolean and integer values.

A correct way to represent the mathematical notion of 1 x 10 is

1 <= x && x <= 10
In this case x must simultaneously be greater than or equal to 1 and less than
or equal to 10. The revised Boolean expression is a little more verbose than the
mathematical representation, but it is the correct formulation for C++.

The following section of code assigns the indicated values to a bool:

bool b;
int x = 10;
int y = 20;
b = (x == 10); // assigns true to b
b = (x != 10); // assigns false to b
b = (x == 10 && y == 20); // assigns true to b
b = (x != 10 && y == 20); // assigns false to b
b = (x == 10 && y != 20); // assigns false to b
b = (x != 10 && y != 20); // assigns false to b
b = (x == 10 || y == 20); // assigns true to b
b = (x != 10 || y == 20); // assigns true to b
b = (x == 10 || y != 20); // assigns true to b
b = (x != 10 || y != 20); // assigns false to b

Convince yourself that the following expressions are equivalent:

(x != y)

2015 Richard L. Halterman Draft date: January 18, 2015

5.6. COMPOUND BOOLEAN EXPRESSIONS 103

!(x == y)
(x < y || x > y)

In the expression e1 && e2 both subexpressions e1 and e2 must be true for the overall expression to be
true. Since the && operator evaluates left to right, this means that if e1 is false, there is no need to evaluate
e2 . If e1 is false, no value of e2 can make the expression e1 && e2 true. The logical and operator first tests the
expression to its left. If it finds the expression to be false, it does not bother to check the right expression.
This approach is called short-circuit evaluation. In a similar fashion, in the expression e1 || e2 , if e1 is
true, then it does not matter what value e2 hasa logical or expression is true unless both subexpressions
are false. The || operator uses short-circuit evaluation also.
Why is short-circuit evaluation important? Two situations show why it is important to consider:

The order of the subexpressions can affect performance. When a program is running, complex ex-
pressions require more time for the computer to evaluate than simpler expressions. We classify an
expression that takes a relatively long time to evaluate as an expensive expression. If a compound
Boolean expression is made up of an expensive Boolean subexpression and an less expensive Boolean
subexpression, and the order of evaluation of the two expressions does not affect the behavior of the
program, then place the more expensive Boolean expression second. If the first subexpression is false
and && is being used, then the expensive second subexpression is not evaluated; if the first subex-
pression is true and || is being used, then, again, the expensive second subexpression is avoided.

Subexpressions can be ordered to prevent run-time errors. This is especially true when one of the
subexpressions depends on the other in some way. Consider the following expression:

(x != 0) && (z/x > 1)

Here, if x is zero, the division by zero is avoided. If the subexpressions were switched, a run-time
error would result if x is zero.

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 104

Arity Operators Associativity

unary (post) ++, (post) --, static_cast
unary (pre) ++, (pre) --, !, +, -
binary *, /, % left
binary +, - left
binary <<, >> left
binary >, <, >=, <= left
binary ==, != left
binary && left
binary || left
binary =, +=, -=, *=, /=, %= right

Table 5.4: Precedence of C++ Operators (High to Low)

Suppose you wish to print the word OK if a variable x is 1, 2, or 3. An informal

translation from English might yield:
if (x == 1 || 2 || 3)
cout << "OK" << endl;
Unfortunately, xs value is irrelevant; the code always prints the word OK. Since
the == operator has lower precedence than ||, the expression
x == 1 || 2 || 3
is interpreted as
(x == 1) || 2 || 3
The expression x == 1 is either true or false, but integer 2 is always interpreted
as true, and integer 3 is interpreted as true is as well.

The correct statement would be

if (x == 1 || x == 2 || x == 3)
cout << "OK" << endl;
The revised Boolean expression is more verbose and less similar to the English
rendition, but it is the correct formulation for C++.

Our current list of C++ operators is shown in Table 5.4.

5.7 Nested Conditionals

The statements in the body of the if or the else may be any C++ statements, including other if/else
statements. We can use nested if statements to build arbitrarily complex control flow logic. Consider
Listing 5.6 (checkrange.cpp) that determines if a number is between 0 and 10, inclusive.

Listing 5.6: checkrange.cpp

#include <iostream>

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 105

using namespace std;

int main() {
int value;
cout << "Please enter an integer value in the range 0...10: ";
cin >> value;
if (value >= 0) // First check
if (value <= 10) // Second check
cout << "In range";
cout << "Done" << endl;
}

Listing 5.6 (checkrange.cpp) behaves as follows:

The program checks the value >= 0 condition first. If value is less than zero, the executing
program does not evaluate the second condition and does not print In range, but it immediately
executes the print statement following the outer if statement which prints Done.

If the executing program finds value to be greater than or equal to zero, it checks the second
condition. If the second condition is met, it displays the In range message; otherwise, it is not.
Regardless, the program prints Done before it terminates.

For the program to display the message In range both conditions of this nested if must be met. Said
another way, the first condition and the second condition must be met for the In range message to be printed.
From this perspective, we can rewrite the program to behave the same way with only one if statement, as
Listing 5.7 (newcheckrange.cpp) shows.

Listing 5.7: newcheckrange.cpp

#include <iostream>

using namespace std;

int main() {
int value;
cout << "Please enter an integer value in the range 0...10: ";
cin >> value;
if (value >= 0 && value <= 10)
cout << "In range";
cout << endl;
}

Listing 5.7 (newcheckrange.cpp) uses a logical && to check both conditions at the same time. Its logic
is simpler, using only one if statement, at the expense of a slightly more complex Boolean expression in
its condition. The second version is preferable here because simpler logic is usually a desirable goal.
Sometimes a programs logic cannot be simplified as in Listing 5.7 (newcheckrange.cpp). In Listing 5.8
(enhancedcheckrange.cpp) one if statement alone is insufficient to implement the necessary behavior.

Listing 5.8: enhancedcheckrange.cpp

#include <iostream>

using namespace std;

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 106

int main() {
int value;
cout << "Please enter an integer value in the range 0...10: ";
cin >> value;
if (value >= 0) // First check
if (value <= 10) // Second check
cout << value << " is acceptable";
else
cout << value << " is too large";
else
cout << value << " is too small";
cout << endl;
}

Listing 5.8 (enhancedcheckrange.cpp) provides a more specific message instead of a simple notification
of acceptance. The program prints exactly one of three messages based on the value of the variable. A single
if or if/else statement cannot choose from among more than two different execution paths.
Listing 5.9 (binaryconversion.cpp) uses a series of if statements to print a 10-bit binary string represent-
ing the binary equivalent of a decimal integer supplied by the user. (Section 4.8 provides some background
information about the binary number system.) We use if/else statements to print the individual digits
left to right, essentially assembling the sequence of bits that represents the binary number.

Listing 5.9: binaryconversion.cpp

#include <iostream>

using namespace std;

int main() {
int value;
// Get number from the user
cout << "Please enter an integer value in the range 0...1023: ";
cin >> value;
// Integer must be less than 1024
if (0 <= value && value < 1024) {
if (value >= 512) {
cout << 1;
value %= 512;
}
else
cout << 0;
if (value >= 256) {
cout << 1;
value %= 256;
}
else
cout << 0;
if (value >= 128) {
cout << 1;
value %= 128;
}
else
cout << 0;

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 107

if (value >= 64) {

cout << 1;
value %= 64;
}
else
cout << 0;
if (value >= 32) {
cout << 1;
value %= 32;
} else
cout << 0;
if (value >= 16) {
cout << 1;
value %= 16;
}
else
cout << 0;
if (value >= 8) {
cout << 1;
value %= 8;
}
else
cout << 0;
if (value >= 4) {
cout << 1;
value %= 4;
}
else
cout << 0;
if (value >= 2) {
cout << 1;
value %= 2;
}
else
cout << 0;
cout << value << endl;
}
}

In Listing 5.9 (binaryconversion.cpp):

The outer if checks to see if the value the use provides is in the proper range. The program works
only for non-negative integer values less than 1,024, so the range is 0-1023.
Each inner if compares the user-supplied entered integer against decreasing powers of two. If the
number is large enough, the program:
prints the digit 1 to the console, and
removes via the remainder operator that power of twos contribution to the value.
If the number is not at least as big as the given power of two, the program prints a 0 instead and
moves on without modifying the input value.
For the ones place at the end no check is necessarythe remaining value will be 0 or 1 and so the
program prints whatever remains.

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 108

The following shows a sample run of Listing 5.9 (binaryconversion.cpp):

Please enter an integer value in the range 0...1023: 805

1100100101

Figure 5.3 illustrates the execution of Listing 5.9 (binaryconversion.cpp) when the user enters 805.

Print prompt Remainder 293256 37

37 128? 5 4?
Please enter ...
Yes, 1
No, 0

Get value
37 64? Remainder 54 1
805 1 2?
No, 0
No, 0
0 805 1023? 37 32?

Yes Print remaining value

Yes, 1
1

805 512? Remainder 3732 5

5 16?
Yes, 1
No, 0
Remainder 805512 293
293 256? 5 8?

Yes, 1 No, 0

Figure 5.3: The process of the binary number conversion program when the user supplies 805 as the input value.

Listing 5.10 (simplerbinaryconversion.cpp) simplifies the logic of Listing 5.9 (binaryconversion.cpp) at

the expense of some additional arithmetic. It uses only one if statement.

Listing 5.10: simplerbinaryconversion.cpp

#include <iostream>

using namespace std;

int main() {
int value;
// Get number from the user
cout << "Please enter an integer value in the range 0...1023: ";
cin >> value;
// Integer must be less than 1024

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 109

if (0 <= value && value < 1024) {

cout << value/512;
value %= 512;
cout << value/256;
value %= 256;
cout << value/128;
value %= 128;
cout << value/64;
value %= 64;
cout << value/32;
value %= 32;
cout << value/16;
value %= 16;
cout << value/8;
value %= 8;
cout << value/4;
value %= 4;
cout << value/2;
value %= 2;
cout << value << endl;
}
}

The sole if statement in Listing 5.10 (simplerbinaryconversion.cpp) ensures that the user provides an inte-
ger in the proper range. The other if statements that originally appeared in Listing 5.9 (binaryconversion.cpp)
are gone. A clever sequence of integer arithmetic operations replace the original conditional logic. The two
programsbinaryconversion.cpp and simplerbinaryconversion.cppbehave identically but simplerbinarycon-
version.cpps logic is simpler.
Listing 5.11 (troubleshoot.cpp) implements a very simple troubleshooting program that (an equally
simple) computer technician might use to diagnose an ailing computer.

Listing 5.11: troubleshoot.cpp

#include <iostream>

using namespace std;

int main() {
cout << "Help! My computer doesn't work!" << endl;
char choice;
cout << "Does the computer make any sounds "
<< "(fans, etc.) or show any lights? (y/n):";
cin >> choice;
// The troubleshooting control logic
if (choice == 'n') { // The computer does not have power
cout << "Is it plugged in? (y/n):";
cin >> choice;
if (choice == 'n') { // It is not plugged in, plug it in
cout << "Plug it in. If the problem persists, "
<< "please run this program again." << endl;
}
else { // It is plugged in
cout << "Is the switch in the \"on\" position? (y/n):";
cin >> choice;

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 110

if (choice == 'n') { // The switch is off, turn it on!

cout << "Turn it on. If the problem persists, "
<< "please run this program again." << endl;
}
else { // The switch is on
cout << "Does the computer have a fuse? (y/n):";
cin >> choice;
if (choice == 'n') { // No fuse
cout << "Is the outlet OK? (y/n):";
if (choice == 'n') { // Fix outlet
cout << "Check the outlet's circuit "
<< "breaker or fuse. Move to a "
<< "new outlet, if necessary. "
<< "If the problem persists, "
<< "please run this program "
<< "again." << endl;
}
else { // Beats me!
cout << "Please consult a service "
<< "technician." << endl;
}
}
else { // Check fuse
cout << "Check the fuse. Replace if "
<< "necessary. If the problem "
<< "persists, then "
<< "please run this program again."
<< endl;
}
}
}
}
else { // The computer has power
cout << "Please consult a service technician." << endl;
}
}

This very simple troubleshooting program attempts to diagnose why a computer does not work. The
potential for enhancement is unlimited, but this version only deals with power issues that have simple fixes.
Notice that if the computer has power (fan or disk drive makes sounds or lights are visible), the program
directs the user to seek help elsewhere! The decision tree capturing the basic logic of the program is shown
in Figure 5.4. The steps performed are:

1. Is it plugged in? This simple fix is sometimes overlooked.

2. Is the switch in the on position? This is another simple fix.
3. If applicable, is the fuse blown? Some computer systems have a user-serviceable fuse that can blow
out during a power surge. (Most newer computers have power supplies that can handle power surges
and have no user-serviceable fuses.)
4. Is there power at the receptacle? Perhaps the outlets circuit breaker or fuse has a problem.

The program directs the user to make the easier checks first. It progressively introduces more difficult
checks as it continues. Based on your experience with troubleshooting computers that do not run properly,

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 111

Figure 5.4: Decision tree for troubleshooting a computer system

you may be able to think of many enhancements to this simple program.

Note that in Listing 5.11 (troubleshoot.cpp) curly braces are used in many places where they strictly
are not necessary. Their inclusion in Listing 5.11 (troubleshoot.cpp) improves the readability of the pro-
gram and makes the logic easier to understand. Even if you do not subscribe to the philosophy of using
curly braces for every if/else body, it is a good idea to use them in situations that improve the codes
readability.
Recall the time conversion program in Listing 4.10 (timeconv.cpp). If the user enters 10000, the
program runs as follows:

Please enter the number of seconds:10000

2 hr 46 min 40 sec

and if the user enters 9961, the program prints:

Please enter the number of seconds:9961

2 hr 46 min 1 sec

Suppose we wish to improve the English presentation by not using abbreviations. If we spell out hours,
minutes, and seconds, we must be careful to use the singular form hour, minute, or second when the

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 112

corresponding value is one. Listing 5.12 (timeconvcond1.cpp) uses if/else statements to express to
time units with the correct number.

Listing 5.12: timeconvcond1.cpp

// File timeconvcond1.cpp

#include <iostream>

using namespace std;

int main() {
// Some useful conversion constants
const int SECONDS_PER_MINUTE = 60,
SECONDS_PER_HOUR = 60*SECONDS_PER_MINUTE; // 3600
int hours, minutes, seconds;
cout << "Please enter the number of seconds:";
cin >> seconds;
// First, compute the number of hours in the given number
// of seconds
hours = seconds / SECONDS_PER_HOUR; // 3600 seconds = 1 hours
// Compute the remaining seconds after the hours are
// accounted for
seconds = seconds % SECONDS_PER_HOUR;
// Next, compute the number of minutes in the remaining
// number of seconds
minutes = seconds / SECONDS_PER_MINUTE; // 60 seconds = 1 minute
// Compute the remaining seconds after the minutes are
// accounted for
seconds = seconds % SECONDS_PER_MINUTE;
// Report the results
cout << hours;
// Decide between singular and plural form of hours
if (hours == 1)
cout << " hour ";
else
cout << " hours ";
cout << minutes;
// Decide between singular and plural form of minutes
if (minutes == 1)
cout << " minute ";
else
cout << " minutes ";
cout << seconds;
// Decide between singular and plural form of seconds
if (seconds == 1)
cout << " second";
else
cout << " seconds";
cout << endl;
}

The if/else statements within Listing 5.12 (timeconvcond1.cpp) are responsible for printing the correct
versionsingular or pluralfor each time unit. One run of Listing 5.12 (timeconvcond1.cpp) produces

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 113

Please enter the number of seconds:10000

2 hours 46 minutes 40 seconds

All the words are plural since all the value are greater than one. Another run produces

Please enter the number of seconds:9961

2 hours 46 minutes 1 second

Note the word second is singular as it should be.

Please enter the number of seconds:3601

1 hour 0 minutes 1 second

Here again the printed words agree with the number of the value they represent.
An improvement to Listing 5.12 (timeconvcond1.cpp) would not print a value and its associated time
unit if the value is zero. Listing 5.13 (timeconvcond2.cpp) adds this feature.

Listing 5.13: timeconvcond2.cpp

// File timeconvcond1.cpp

#include <iostream>

using namespace std;

int main() {
// Some useful conversion constants
const int SECONDS_PER_MINUTE = 60,
SECONDS_PER_HOUR = 60*SECONDS_PER_MINUTE; // 3600
int hours, minutes, seconds;
cout << "Please enter the number of seconds:";
cin >> seconds;
// First, compute the number of hours in the given number
// of seconds
hours = seconds / SECONDS_PER_HOUR; // 3600 seconds = 1 hours
// Compute the remaining seconds after the hours are
// accounted for
seconds = seconds % SECONDS_PER_HOUR;
// Next, compute the number of minutes in the remaining
// number of seconds
minutes = seconds / SECONDS_PER_MINUTE; // 60 seconds = 1 minute
// Compute the remaining seconds after the minutes are
// accounted for
seconds = seconds % SECONDS_PER_MINUTE;
// Report the results
if (hours > 0) { // Print hours at all?

2015 Richard L. Halterman Draft date: January 18, 2015

5.7. NESTED CONDITIONALS 114

cout << hours;

// Decide between singular and plural form of hours
if (hours == 1)
cout << " hour ";
else
cout << " hours ";
}
if (minutes > 0) { // Print minutes at all?
cout << minutes;
// Decide between singular and plural form of minutes
if (minutes == 1)
cout << " minute ";
else
cout << " minutes ";
}
// Print seconds at all?
if (seconds > 0 || (hours == 0 && minutes == 0 && seconds == 0)) {
cout << seconds;
// Decide between singular and plural form of seconds
if (seconds == 1)
cout << " second";
else
cout << " seconds";
}
cout << endl;
}

In Listing 5.13 (timeconvcond2.cpp) each code segment responsible for printing a time value and its English
word unit is protected by an if statement that only allows the code to execute if the time value is greater
than zero. The exception is in the processing of seconds: if all time values are zero, the program should
print 0 seconds. Note that each of the if/else statements responsible for determining the singular or
plural form is nested within the if statement that determines whether or not the value will be printed at all.
One run of Listing 5.13 (timeconvcond2.cpp) produces

Please enter the number of seconds:10000

2 hours 46 minutes 40 seconds

All the words are plural since all the value are greater than one. Another run produces

Please enter the number of seconds:9961

2 hours 46 minutes 1 second

Note the word second is singular as it should be.

Please enter the number of seconds:3601

1 hour 1 second

2015 Richard L. Halterman Draft date: January 18, 2015

5.8. MULTI-WAY IF/ELSE STATEMENTS 115

Here again the printed words agree with the number of the value they represent.

Please enter the number of seconds:7200

2 hours

Another run produces:

Please enter the number of seconds:60

1 minute

Finally, the following run shows that the program handles zero seconds properly:

Please enter the number of seconds:0

0 seconds

5.8 Multi-way if/else Statements

A simple if/else statement can select from between two execution paths. Suppose we wish to choose
one execution path from among several possible paths, as shown in Figure 5.5? Listing 5.8 (enhancedcheckrange.cpp)

Figure 5.5: Flowchart with multiple optional execution pathways

showed how to select from among three options. What if exactly one of many actions should be taken?

2015 Richard L. Halterman Draft date: January 18, 2015

5.8. MULTI-WAY IF/ELSE STATEMENTS 116

Nested if/else statements are required, and the form of these nested if/else statements is shown in
Listing 5.14 (digittoword.cpp).

Listing 5.14: digittoword.cpp

#include <iostream>

using namespace std;

int main() {
int value;
cout << "Please enter an integer in the range 0...5: ";
cin >> value;
if (value < 0)
cout << "Too small";
else
if (value == 0)
cout << "zero";
else
if (value == 1)
cout << "one";
else
if (value == 2)
cout << "two";
else
if (value == 3)
cout << "three";
else
if (value == 4)
cout << "four";
else
if (value == 5)
cout << "five";
else
cout << "Too large";
cout << endl;
}

Observe the following about Listing 5.14 (digittoword.cpp):

It prints exactly one of eight messages depending on the users input.

Notice that each if body contains a single printing statement and each else body, except the last
one, contains an if statement. The control logic forces the program execution to check each condi-
tion in turn. The first condition that matches wins, and its corresponding if body will be executed.
If none of the conditions are true, the last elses Too large message will be printed.

No curly braces are necessary to delimit the if or else bodies since each body contains only a
single statement (although a single deeply nested if/else statement is a mighty big statement).

Listing 5.14 (digittoword.cpp) is formatted according to the conventions used in earlier examples. As
a consequence, the mass of text drifts to the right as more conditions are checked. A commonly used
alternative style, shown in Listing 5.15 (restyleddigittoword.cpp), avoids this rightward drift.

2015 Richard L. Halterman Draft date: January 18, 2015

5.8. MULTI-WAY IF/ELSE STATEMENTS 117

Listing 5.15: restyleddigittoword.cpp

#include <iostream>

using namespace std;

int main() {
int value;
cout << "Please enter an integer in the range 0...5: ";
cin >> value;
if (value < 0)
cout << "Too small";
else if (value == 0)
cout << "zero";
else if (value == 1)
cout << "one";
else if (value == 2)
cout << "two";
else if (value == 3)
cout << "three";
else if (value == 4)
cout << "four";
else if (value == 5)
cout << "five";
else
cout << "Too large";
cout << endl;
}

Based on our experience so far, the formatting of Listing 5.15 (restyleddigittoword.cpp) somewhat hides
the true structure of the programs logic, but this style of formatting multi-way if/else statements is
so common that it is regarded as acceptable by most programmers. The sequence of else if lines all
indented to the same level identifies this construct as a multi-way if/else statement.
Listing 5.16 (datetransformer.cpp) uses a multi-way if/else to transform a numeric date in month/-
day format to an expanded US English form and an international Spanish form; for example, 2/14 would
be converted to February 14 and 14 febrero.

Listing 5.16: datetransformer.cpp

#include <iostream>

using namespace std;

int main() {
cout << "Please enter the month and day as numbers: ";
int month, day;
cin >> month >> day;
// Translate month into English
if (month == 1)
cout << "January";
else if (month == 2)
cout << "February";
else if (month == 3)
cout << "March";
else if (month == 4)

2015 Richard L. Halterman Draft date: January 18, 2015

5.8. MULTI-WAY IF/ELSE STATEMENTS 118

cout << "April";

else if (month == 5)
cout << "May";
else if (month == 6)
cout << "June";
else if (month == 7)
cout << "July";
else if (month == 8)
cout << "August";
else if (month == 9)
cout << "September";
else if (month == 10)
cout << "October";
else if (month == 11)
cout << "November";
else
cout << "December";
// Add the day
cout << " " << day << " or " << day << " de ";
// Translate month into Spanish
if (month == 1)
cout << "enero";
else if (month == 2)
cout << "febrero";
else if (month == 3)
cout << "marzo";
else if (month == 4)
cout << "abril";
else if (month == 5)
cout << "mayo";
else if (month == 6)
cout << "junio";
else if (month == 7)
cout << "julio";
else if (month == 8)
cout << "agosto";
else if (month == 9)
cout << "septiembre";
else if (month == 10)
cout << "octubre";
else if (month == 11)
cout << "noviembre";
else
cout << "diciembre";
cout << endl;
}

A sample run of Listing 5.16 (datetransformer.cpp) is shown here:

Please enter the month and day as numbers: 5 20

May 20 or 20 de mayo

2015 Richard L. Halterman Draft date: January 18, 2015

5.9. ERRORS IN CONDITIONAL STATEMENTS 119

Figure 5.6 compares the structure of the if/else statements in a program such as Listing 5.15
(restyleddigittoword.cpp) to those in a program like Listing 5.9 (binaryconversion.cpp). In a program like

Figure 5.6: The structure of the if statements in a program such as Listing 5.15 (restyleddigittoword.cpp) (left) vs.
those in a program like Listing 5.9 (binaryconversion.cpp) (right)

Listing 5.15 (restyleddigittoword.cpp), the if/else statements are nested, while in a program like List-
ing 5.9 (binaryconversion.cpp) the if/else statements are sequential.

5.9 Errors in Conditional Statements

Consider Listing 5.17 (badequality.cpp).

Listing 5.17: badequality.cpp

#include <iostream>

using namespace std;

2015 Richard L. Halterman Draft date: January 18, 2015

5.9. ERRORS IN CONDITIONAL STATEMENTS 120

int main() {
int input;
cout << "Please enter an integer:";
cin >> input;
if (input = 2)
cout << "two" << endl;
cout << "You entered " << input << endl;
}

Listing 5.17 (badequality.cpp) demonstrates a common mistakeusing the assignment operator where
the equality operator is intended. This program, when run, always prints the message two and insists the
user entered 2 regardless of the actual input. Recall from Section 4.3 that the assignment expression has a
value. The value of an assignment expression is same as the value that is assigned; thus, the expression
input = 2

has the value 2. When you consider also that every integer can be treated as a Boolean value (see Sec-
tion 5.1) and any non-zero value is interpreted as true, you can see that the condition of if statement
if (input = 2)
cout << "two" << endl;

is always true. Additionally, the variable input is always assigned the value 2.
Since it is such a common coding error, most C++ compilers can check for such misuse of assignment. At
warning Level 4, for example, Visual C++ will issue a warning when assignment appears where a conditional
expression is expected:

warning C4706: assignment within conditional expression

Occasionally the use of assignment within a conditional expression is warranted, so the compiler does not
perform this check by default. For our purposes it is good idea to direct the compiler to perform this extra
check.
Carefully consider each compound conditional used, such as
value > 0 && value <= 10

found in Listing 5.7 (newcheckrange.cpp). Confusing logical and and logical or is a common programming
error. If you substitute || for &&, the expression
x > 0 || x <= 10

is always true, no matter what value is assigned to the variable x. A Boolean expression that is always true
is known as a tautology. Think about it. If x is an int, what value could the variable x assume that would
make
x > 0 || x <= 10

false? Regardless of its value, one or both of the subexpressions will be true, so this compound logical or
expression is always true. This particular or expression is just a complicated way of expressing the value
true.
Another common error is contriving compound Boolean expressions that are always false, known as
contradictions. Suppose you wish to exclude values from a given range; for example, reject values in the

2015 Richard L. Halterman Draft date: January 18, 2015

5.10. SUMMARY 121

range 0...10 and accept all other numbers. Is the Boolean expression in the following code fragment up to
the task?
// I want to use all but 0, 1, 2, ..., 10
if (value < 0 && value > 10)
/* Code to execute goes here . . . */

A closer look at the condition reveals it can never be true. What number can be both less than zero and
greater than ten at the same time? None can, of course, so the expression is a contradiction and a compli-
cated way of expressing false. To correct this code fragment, replace the && operator with ||.

5.10 Summary

The bool data type represents the values true and false.

The name bool comes from Boolean algebra, the mathematical study of operations on truth values.

In C++ the value true is represented by the integer one, and false is represented by zero.

Any integer value except zero is treated as true.

Integers and bools are interchangeable and can be assigned to each other.

Expressions involving the relational operators (==, !=, <, >, <=, and >=) evaluate to Boolean values.

! is the unary not operator.

Boolean expressions can be combined via && (logical AND) and || (logical OR).

The if statement can be used to optionally execute statements.

A compound statement is a sequence of statements within a pair of curly braces.

The if statement has an optional else clause to require the selection between two alternate paths
of execution.

The if/else statements can be nested to achieve arbitrary complexity.

The bodies of if/else statements should be indented to aid human readers. Indentation doe not
affect the logic of the program; when multiple statements are to be part of the body of an if or else,
the statements must be part of a compound statement.

Beware placing a semicolon immediately after the close parenthesis of an if statements condition.

Complex Boolean expressions require special attention, as they are easy to get wrong.

5.11 Exercises

1. What values can a variable of type bool assume?

2. Where does the term bool originate?

3. What is the integer equivalent to true in C++?

2015 Richard L. Halterman Draft date: January 18, 2015

5.11. EXERCISES 122

4. What is the integer equivalent to false in C++?

5. Is the value -16 interpreted as true or false?

6. May an integer value be assigned to a bool variable?

7. Can true be assigned to an int variable?

8. Given the following declarations:

int x = 3, y = 5, z = 7;
bool b1 = true, b2 = false, b3 = x == 3, b4 = y < 3;

evaluate the following Boolean expressions:

(a) x == 3
(b) x < y
(c) x >= y
(d) x <= y
(e) x != y - 2
(f) x < 10
(g) x >= 0 && x < 10
(h) x < 0 && x < 10
(i) x >= 0 && x < 2
(j) x < 0 || x < 10
(k) x > 0 || x < 10
(l) x < 0 || x > 10
(m) b1
(n) !b1
(o) !b2
(p) b1 && b2

9. Express the following Boolean expressions in simpler form; that is, use fewer operators. x is an int.

(a) !(x == 2)
(b) x < 2 || x == 2
(c) !(x < y)
(d) !(x <= y)
(e) x < 10 && x > 20
(f) x > 10 || x < 20
(g) x != 0
(h) x == 0

10. What is the simplest tautology?

11. What is the simplest contradiction?

2015 Richard L. Halterman Draft date: January 18, 2015

5.11. EXERCISES 123

12. Write a C++ program that requests an integer value from the user. If the value is between 1 and 100
inclusive, print OK; otherwise, do not print anything.

13. Write a C++ program that requests an integer value from the user. If the value is between 1 and 100
inclusive, print OK; otherwise, print Out of range.

14. Write a C++ program that requests an integer value from the user. If the value is between 1 and 100
inclusive,

15. The following program attempts to print a message containing the the English word correspond-
ing to a given integer input. For example, if the user enters the value 3, the program should print
"You entered a three". In its current state, the program contains logic errors. Locate the
problems and repair them so the program will work as expected.
#include <iostream>

using namespace std;

int main() {
cout << "Please in value in the range 1...5: ";
int value;
cin >> value;
// Translate number into its English word
if (month == 1)
cout << "You entered a";
cout << "one";
cout << endl;
else if (month == 2)
cout << "You entered a";
cout << "two";
cout << endl;
else if (month == 3)
cout << "You entered a";
cout << "three";
cout << endl;
else if (month == 4)
cout << "You entered a";
cout << "four";
cout << endl;
else if (month == 5)
cout << "You entered a";
cout << "five";
cout << endl;
else // Value out of range
cout << "You entered a";
cout << "value out of range";
cout << endl;
}

16. Consider the following section of C++ code:

// i, j, and k are ints

2015 Richard L. Halterman Draft date: January 18, 2015

5.11. EXERCISES 124

if (i < j) {
if (j < k)
i = j;
else
j = k;
}
else {
if (j > k)
j = i;
else
i = k;
}
cout << "i = " << i << " j = " << j << " k = " << k << endl;

What will the code print if the variables i, j, and k have the following values?

(a) i is 3, j is 5, and k is 7
(b) i is 3, j is 7, and k is 5
(c) i is 5, j is 3, and k is 7
(d) i is 5, j is 7, and k is 3
(e) i is 7, j is 3, and k is 5
(f) i is 7, j is 5, and k is 3

17. Consider the following C++ program that prints one line of text:
#include <iostream>
using namespace std;

int main() {
int input;
cin >> input;
if (input < 10) {
if (input != 5)
cout << "wow ";
else
input++;
}
else {
if (input == 17)
input += 10;
else
cout << "whoa ";
}
cout << input << endl;
}

What will the program print if the user provides the following input?

(a) 3
(b) 21

2015 Richard L. Halterman Draft date: January 18, 2015

5.11. EXERCISES 125

(c) 5
(d) 17
(e) -5

18. Why does the following section of code always print "ByeHi"?
int x;
cin >> x;
if (x < 0);
cout << "Bye";
cout << "Hi" << endl;

19. Write a C++ program that requests five integer values from the user. It then prints the maximum and
minimum values entered. If the user enters the values 3, 2, 5, 0, and 1, the program would indicate
that 5 is the maximum and 0 is the minimum. Your program should handle ties properly; for example,
if the user enters 2, 4 2, 3 and 3, the program should report 2 as the minimum and 4 as maximum.

20. Write a C++ program that requests five integer values from the user. It then prints one of two
things: if any of the values entered are duplicates, it prints "DUPLICATES"; otherwise, it prints
"ALL UNIQUE".

2015 Richard L. Halterman Draft date: January 18, 2015

5.11. EXERCISES 126

2015 Richard L. Halterman Draft date: January 18, 2015

 127

Chapter 6

Iteration

Iteration repeats the execution of a sequence of code. Iteration is useful for solving many programming
problems. Iteration and conditional execution are key components of algorithm construction.

6.1 The while Statement

Listing 6.1 (counttofive.cpp) counts to five by printing a number on each output line.

Listing 6.1: counttofive.cpp

#include <iostream>

using namespace std;

int main() {
cout << 1 << endl;
cout << 2 << endl;
cout << 3 << endl;
cout << 4 << endl;
cout << 5 << endl;
}

When compiled and run, this program displays

1
2
3
4
5

How would you write the code to count to 10,000? Would you copy, paste, and modify 10,000 printing
statements? You could, but that would be impractical! Counting is such a common activity, and computers

2015 Richard L. Halterman Draft date: January 18, 2015

6.1. THE WHILE STATEMENT 128

routinely count up to very large values, so there must be a better way. What we really would like to do
is print the value of a variable (call it count), then increment the variable (count++), and repeat this
process until the variable is large enough (count == 5 or perhaps count == 10000). This process
of executing the same section of code over and over is known as iteration, or looping, and in C++ we can
implement loops in several different ways.
Listing 6.2 (iterativecounttofive.cpp) uses a while statement to count to five:

Listing 6.2: iterativecounttofive.cpp

#include <iostream>

using namespace std;

int main() {
int count = 1; // Initialize counter
while (count <= 5) {
cout << count << endl; // Display counter, then
count++; // Increment counter
}
}

Listing 6.2 (iterativecounttofive.cpp) uses a while statement to display a variable that is counting
up to five. Unlike the approach taken in Listing 6.1 (counttofive.cpp), it is trivial to modify Listing 6.2
(iterativecounttofive.cpp) to count up to 10,000just change the literal value 5 to 10000.
The line
while (count <= 5)

begins the while statement. The expression within the parentheses must be a Boolean expression. If the
Boolean expression is true when the the programs execution reaches the while statement, the program
executes the body of the while statement and then checks the condition again. The program repeatedly
executes the statement(s) within the body of the while as long as the Boolean expression remains true.
If the Boolean expression is true when the while statement is executed, the body of the while state-
ment is executed, and the body is executed repeatedly as long as the Boolean expression remains true.
The statements
cout << count << endl;
count++;

constitute the body of the while statement. The curly braces are necessary since more than one statement
makes up the body.
The while statement has the general form:

while ( condition )

statement

2015 Richard L. Halterman Draft date: January 18, 2015

6.1. THE WHILE STATEMENT 129

The reserved word while begins the while statement.

The Boolean expression condition determines whether the body will be (or will continue to be) exe-
cuted. The expression must be enclosed within parentheses as shown.
The statement is the statement to be executed while the Boolean expression is true. The statement
makes up the body of the while statement. The statement may be a compound statement (multiple
statements enclosed within curly braces, see Section 5.4).

Except for using the reserved word while instead of if, a while statement looks identical to an
if statement. Sometimes beginning programmers confuse the two or accidentally type if when they
mean while or vice-versa. Usually the very different behavior of the two statements reveals the problem
immediately; however, sometimes, especially in nested complex logic, this mistake can be hard to detect.
Figure 6.1 shows how program execution flows through Listing 6.2 (iterativecounttofive.cpp).

Figure 6.1: while flowchart for Listing 6.2 (iterativecounttofive.cpp)

The program checks the whiles condition before executing the body, and then re-checks the condition
each time after it executes the body. If the condition is initially false the programs execution skips the

2015 Richard L. Halterman Draft date: January 18, 2015

6.1. THE WHILE STATEMENT 130

body completely and continues executing the statements that follow the whiles body. If the condition is
initially true, the program repeatedly executes the body until the condition becomes false, at which point
the loop terminates. Program execution then continues with the statements that follow the loops body, if
any. Observe that the body may never be executed if the Boolean expression in the condition is initially
false.
Listing 6.3 (countup.cpp) counts up from zero as long as the user wishes to do so.

Listing 6.3: countup.cpp

/*
* Counts up from zero. The user continues the count by entering
* 'Y'. The user discontinues the count by entering 'N'.
*/

#include <iostream>

using namespace std;

int main() {
char input; // The users choice
int count = 0; // The current count
bool done = false; // We are not done

while (!done) {

// Print the current value of count

cout << count << endl;
cout << "Please enter \"Y\" to continue or \"N\" to quit: ";
cin >> input;
// Check for "bad" input
if (input != 'Y' && input != 'y' && input != 'N' && input != 'n')
cout << "\"" << input << "\""
<< " is not a valid choice" << endl;
else if (input == 'Y' || input == 'y')
count++; // Keep counting
else if (input == 'N' || input == 'n')
done = true; // Quit the loop
}
}

A sample run of Listing 6.3 (countup.cpp) produces

2015 Richard L. Halterman Draft date: January 18, 2015

6.1. THE WHILE STATEMENT 131

0
Please enter "Y" to continue or "N" to quit: y
1
Please enter "Y" to continue or "N" to quit: y
2
Please enter "Y" to continue or "N" to quit: y
3
Please enter "Y" to continue or "N" to quit: q
"q" is not a valid choice
3
Please enter "Y" to continue or "N" to quit: r
"r" is not a valid choice
3
Please enter "Y" to continue or "N" to quit: W
"W" is not a valid choice
3
Please enter "Y" to continue or "N" to quit: Y
4
Please enter "Y" to continue or "N" to quit: y
5
Please enter "Y" to continue or "N" to quit: n

In Listing 6.3 (countup.cpp) the expression

input != 'Y' && input != 'y' && input != 'N' && input != 'n'

is true if the character variable input is not equal to one of the listed character literals. The Boolean
variable done controls the loops execution. It is important to note that the expression
!done

inside the whiles condition evaluates to the opposite truth value of the variable done; the expression
does not affect the value of done. In other words, the ! operator applied to a variable does not modify the
variables value. In order to actually change the variable done, you would need to reassign it, as in
done = !done; // Invert the truth value

For Listing 6.3 (countup.cpp) we have no need to invert its value. We ensure that its value is false initially
and then make it true when the user enters a capital or lower-case N.
Listing 6.4 (addnonnegatives.cpp) is a program that allows a user to enter any number of non-negative
integers. When the user enters a negative value, the program no longer accepts input, and it displays the
sum of all the non-negative values. If a negative number is the first entry, the sum is zero.

Listing 6.4: addnonnegatives.cpp

/*
* Allow the user to enter a sequence of non-negative
* integers. The user ends the list with a negative
* integer. At the end the sum of the non-negative
* integers entered is displayed. The program prints
* zero if the users no non-negative integers.

2015 Richard L. Halterman Draft date: January 18, 2015

6.1. THE WHILE STATEMENT 132

*/

#include <iostream>

using namespace std;

int main() {
int input = 0, // Ensure the loop is entered
sum = 0; // Initialize sum

// Request input from the user

cout << "Enter numbers to sum, negative number ends list:";

while (input >= 0) { // A negative number exits the loop

cin >> input; // Get the value
if (input >= 0)
sum += input; // Only add it if it is non-negative
}
cout << "Sum = " << sum << endl; // Display the sum
}

The initialization of input to zero coupled with the condition input >= 0 of the while guarantees
that program will execute the body of the while loop at least once. The if statement ensures that a
negative entry will not be added to sum. (Could the condition have used > instead of >= and achieved the
same results?) When the user enters a negative integer the program will not update sum, and the condition
of the while will no longer be true. The programs execution then leaves the loop and executes the print
statement at the end.
Listing 6.4 (addnonnegatives.cpp) shows that a while loop can be used for more than simple counting.
The program does not keep track of the number of values entered. The program simply accumulates the
entered values in the variable named sum.
It is a little awkward in Listing 6.4 (addnonnegatives.cpp) that the same condition appears twice, once
in the while and again in the if. Furthermore, what if the user wishes to enter negative values along
with non-negative values? The code can be simplified with a common C++ idiom that uses cin and the
extraction operator as a condition within an while statement.
If x is an integer, the expression
cin >> x

evaluates to false if the user does not enter a valid integer literal. Armed with this knowledge we can
simplify and enhance Listing 6.4 (addnonnegatives.cpp) as shown in Listing 6.5 (addnumbers.cpp).

Listing 6.5: addnumbers.cpp

#include <iostream>

using namespace std;

int main() {
int input, sum = 0;
cout << "Enter numbers to sum, type 'q' to end the list:";
while (cin >> input)
sum += input;
cout << "Sum = " << sum << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

6.1. THE WHILE STATEMENT 133

The condition reads a value from the input stream and, if it is successful, it is interpreted as true.
When the user enters 'q', the loop is terminated. If the user types 'q' at the beginning, the loop is not
entered. The if statement is no longer necessary, since the statement

sum += input;

can be executed only if input has been legitimately assigned. Also, the variable input no longer needs
to initialized with a value simply so the loop is entered the first time; now it is assigned and then checked
within the condition of the while.
In Listing 6.5 (addnumbers.cpp), the programs execution will terminate with any letter the user types;
an entry of 'x' or Ctrl-Z will terminate the sequence just as well as 'q'.

If you use the

while (cin >> x) {
// Do something . . .
}
idiom, be aware that when the loop is exited due to the input stream being unable
to read a value of the type of variable x, the cin input stream object is left in an
error state and cannot be used for the rest of the programs execution. It you wish
to use this technique and reuse cin later, you must reset cin and extract and
discard keystrokes entered since the last valid use of the extractor operator. This
recovery process is covered in Section 13.2, but, for now, use this idiom to control
a loop only if the program does not require additional user input later during its
execution.

Listing 6.6 (powersof10.cpp) prints the powers of 10 from 1 to 1,000,000,000 (the next power of ten,
10,000,000,000, is outside the range of the int type).

Listing 6.6: powersof10.cpp

#include <iostream>

using namespace std;

int main() {
int power = 1;
while (power <= 1000000000) {
cout << power << endl;
power *= 10;
}
}

Listing 6.6 (powersof10.cpp) produces

2015 Richard L. Halterman Draft date: January 18, 2015

6.1. THE WHILE STATEMENT 134

1
10
100
1000
10000
100000
1000000
10000000
100000000
1000000000

It is customary to right justify a column of numbers, but Listing 6.6 (powersof10.cpp) prints the powers
of ten with their most-significant digit left aligned. We can right align the numbers using a stream object
called a steam manipulator. The specific stream manipulator we need is named setw. setw means set
width. It can be used as

cout << setw(3) << x << endl;

This statement prints the value of x right justified within a three character horizontal space on the screen.
Listing 6.7 (powersof10justified.cpp) shows the affects of setw.

Listing 6.7: powersof10justified.cpp

#include <iostream>
#include <iomanip>

using namespace std;

// Print the powers of 10 from 1 to 1,000,000,000

int main() {
int power = 1;
while (power <= 1000000000) {
// Right justify each number in a field 10 wide
cout << setw(10) << power << endl;
power *= 10;
}
}

Listing 6.7 (powersof10justified.cpp) prints

2015 Richard L. Halterman Draft date: January 18, 2015

6.1. THE WHILE STATEMENT 135

1
10
100
1000
10000
100000
1000000
10000000
100000000
1000000000

Observe that in order to use setw the compiler needs to be made aware of it. The needed information about
setw is not found in the iomanip header file, so an additional preprocessor include directive is required:
#include <iomanip>

The setw manipulator conditions the output stream for the next item to be printed. The values passed
to the conditioned stream are all right justified within the number of spaces specified by setw.
We can use a while statement to make Listing 5.11 (troubleshoot.cpp) more convenient for the user.
Recall that the computer troubleshooting program forces the user to rerun the program once a potential
program has been detected (for example, turn on the power switch, then run the program again to see what
else might be wrong). A more desirable decision logic is shown in Figure 6.2.

Figure 6.2: Decision tree for troubleshooting a computer system Listing 6.8 (troubleshootloop.cpp)

Listing 6.8 (troubleshootloop.cpp) incorporates a while statement so that the programs execution
continues until the problem is resolved or its resolution is beyond the capabilities of the program.

2015 Richard L. Halterman Draft date: January 18, 2015

6.1. THE WHILE STATEMENT 136

Listing 6.8: troubleshootloop.cpp

#include <iostream>

using namespace std;

int main() {
cout << "Help! My computer doesn't work!" << endl;
char choice;
bool done = false; // Initially, we are not done

while (!done) { // Continue until we are done

cout << "Does the computer make any sounds "
<< "(fans, etc.) or show any lights? (y/n):";
cin >> choice;
// The troubleshooting control logic
if (choice == 'n') { // The computer does not have power
cout << "Is it plugged in? (y/n):";
cin >> choice;
if (choice == 'n') { // It is not plugged in, plug it in
cout << "Plug it in." << endl;
}
else { // It is plugged in
cout << "Is the switch in the \"on\" position? (y/n):";
cin >> choice;
if (choice == 'n') { // The switch is off, turn it on!
cout << "Turn it on." << endl;
}
else { // The switch is on
cout << "Does the computer have a fuse? (y/n):";
cin >> choice;
if (choice == 'n') { // No fuse
cout << "Is the outlet OK? (y/n):";
if (choice == 'n') { // Fix outlet
cout << "Check the outlet's circuit "
<< "breaker or fuse. Move to a "
<< "new outlet, if necessary. "
<< endl;
}
else { // Beats me!
cout << "Please consult a service "
<< "technician." << endl;
done = true; // Exhausted simple fixes
}
}
else { // Check fuse
cout << "Check the fuse. Replace if "
<< "necessary." << endl;
}
}
}
}
else { // The computer has power
cout << "Please consult a service technician." << endl;
done = true; // Only troubleshoots power issues

2015 Richard L. Halterman Draft date: January 18, 2015

6.2. NESTED LOOPS 137

}
}
}

The bulk of the body of the Listing 6.8 (troubleshootloop.cpp)is wrapped by a while statement. The
Boolean variable done is often called a flag. You can think of the flag being down when the value is
false and raised when it is true. In this case, when the flag is raised, it is a signal that the program should
terminate.
Notice the last 11 lines of Listing 6.8 (troubleshootloop.cpp):
}
}
}
}
else { // The computer has power
cout << "Please consult a service technician." << endl;
done = true; // Only troubleshoots power issues
}
}
}

In the way this code is organized, the matching opening curly brace of a particular closing curly brace can be
found by scanning upward in the source code until the closest opening curly brace at the same indentation
level is found. Our programming logic is now getting complex enough that the proper placement of curly
braces is crucial for human readers to more quickly decipher how the program should work. See Section 4.5
for guidelines on indentation and curly brace placement to improve code readability.

6.2 Nested Loops

Just like in if statements, while bodies can contain arbitrary C++ statements, including other while
statements. A loop can therefore be nested within another loop. To see how nested loops work, consider a
program that prints out a multiplication table. Elementary school students use multiplication tables, or times
tables, as they learn the products of integers up to 10 or even 12. Figure 6.3 shows a 10 10 multiplication
table. We want our multiplication table program to be flexible and allow the user to specify the tables
size. We will begin our development work with a simple program and add features as we go. First, we will
not worry about printing the tables row and column titles, nor will we print the lines separating the titles
from the contents of the table. Initially we will print only the contents of the table. We will see we need
a nested loop to print the tables contents, but that still is too much to manage in our first attempt. In our
first attempt we will print the rows of the table in a very rudimentary manner. Once we are satisfied that
our simple program works we can add more features. Listing 6.9 (timestable-1st-try.cpp) shows our first
attempt at a muliplication table.

Listing 6.9: timestable-1st-try.cpp

#include <iostream>

using namespace std;

int main() {
int size; // The number of rows and columns in the table

2015 Richard L. Halterman Draft date: January 18, 2015

6.2. NESTED LOOPS 138

Figure 6.3: A 10 10 multiplication table

cout << "Please enter the table size: ";

cin >> size;
// Print a size x size multiplication table
int row = 1;
while (row <= size) { // Table has 10 rows.
cout << "Row #" << row << endl;
row++; // Next row
}
}

The output of Listing 6.9 (timestable-1st-try.cpp) is somewhat underwhelming:

Please enter the table size: 10

Row #1
Row #2
Row #3
Row #4
Row #5
Row #6
Row #7
Row #8
Row #9
Row #10

Listing 6.9 (timestable-1st-try.cpp) does indeed print each row in its proper placeit just does not supply
the needed detail for each row. Our next step is to refine the way the program prints each row. Each row
should contain size numbers. Each number within each row represents the product of the current row and
current column; for example, the number in row 2, column 5 should be 2 5 = 10. In each row, therefore,
we must vary the column number from from 1 to size. Listing 6.10 (timestable-2nd-try.cpp) contains the
needed refinement.

2015 Richard L. Halterman Draft date: January 18, 2015

6.2. NESTED LOOPS 139

Listing 6.10: timestable-2nd-try.cpp

#include <iostream>

using namespace std;

int main() {
int size; // The number of rows and columns in the table
cout << "Please enter the table size: ";
cin >> size;
// Print a size x size multiplication table
int row = 1;
while (row <= size) { // Table has size rows.
int column = 1; // Reset column for each row.
while (column <= size) { // Table has size columns.
int product = row*column; // Compute product
cout << product << " "; // Display product
column++; // Next element
}
cout << endl; // Move cursor to next row
row++; // Next row
}
}

We use a loop to print the contents of each row. The outer loop controls how many total rows the program
prints, and the inner loop, executed in its entirity each time the program prints a row, prints the individual
elements that make up a row.
The result of Listing 6.10 (timestable-2nd-try.cpp) is

Please enter the table size: 10

1 2 3 4 5 6 7 8 9 10
2 4 6 8 10 12 14 16 18 20
3 6 9 12 15 18 21 24 27 30
4 8 12 16 20 24 28 32 36 40
5 10 15 20 25 30 35 40 45 50
6 12 18 24 30 36 42 48 54 60
7 14 21 28 35 42 49 56 63 70
8 16 24 32 40 48 56 64 72 80
9 18 27 36 45 54 63 72 81 90
10 20 30 40 50 60 70 80 90 100

The numbers within each column are not lined up nicely, but the numbers are in their correct positions rela-
tive to each other. We can use the setw stream manipulator introduced in Listing 6.7 (powersof10justified.cpp)
to right justify the numbers within a four-digit area. Listing 6.11 (timestable-3rd-try.cpp) contains this align-
ment adjustment.

Listing 6.11: timestable-3rd-try.cpp

#include <iostream>
#include <iomanip>

2015 Richard L. Halterman Draft date: January 18, 2015

6.2. NESTED LOOPS 140

using namespace std;

int main() {
int size; // The number of rows and columns in the table
cout << "Please enter the table size: ";
cin >> size;
// Print a size x size multiplication table
int row = 1;
while (row <= size) { // Table has size rows.
int column = 1; // Reset column for each row.
while (column <= size) { // Table has size columns.
int product = row*column; // Compute product
cout << setw(4) << product; // Display product
column++; // Next element
}
cout << endl; // Move cursor to next row
row++; // Next row
}
}

Listing 6.11 (timestable-3rd-try.cpp) produces the tables contents in an attractive form:

Please enter the table size: 10

1 2 3 4 5 6 7 8 9 10
2 4 6 8 10 12 14 16 18 20
3 6 9 12 15 18 21 24 27 30
4 8 12 16 20 24 28 32 36 40
5 10 15 20 25 30 35 40 45 50
6 12 18 24 30 36 42 48 54 60
7 14 21 28 35 42 49 56 63 70
8 16 24 32 40 48 56 64 72 80
9 18 27 36 45 54 63 72 81 90
10 20 30 40 50 60 70 80 90 100

Input values of 5:

Please enter the table size: 5

1 2 3 4 5
2 4 6 8 10
3 6 9 12 15
4 8 12 16 20
5 10 15 20 25

and 15:

2015 Richard L. Halterman Draft date: January 18, 2015

6.2. NESTED LOOPS 141

Please enter the table size: 15

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
3 6 9 12 15 18 21 24 27 30 33 36 39 42 45
4 8 12 16 20 24 28 32 36 40 44 48 52 56 60
5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
6 12 18 24 30 36 42 48 54 60 66 72 78 84 90
7 14 21 28 35 42 49 56 63 70 77 84 91 98 105
8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
9 18 27 36 45 54 63 72 81 90 99 108 117 126 135
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
11 22 33 44 55 66 77 88 99 110 121 132 143 154 165
12 24 36 48 60 72 84 96 108 120 132 144 156 168 180
13 26 39 52 65 78 91 104 117 130 143 156 169 182 195
14 28 42 56 70 84 98 112 126 140 154 168 182 196 210
15 30 45 60 75 90 105 120 135 150 165 180 195 210 225

also give good results.

All that is left is to add the row and column titles and the lines that bound the edges of the table.
Listing 6.12 (timestable.cpp) adds the necessary code.

Listing 6.12: timestable.cpp

#include <iostream>
#include <iomanip>

using namespace std;

int main() {
int size; // The number of rows and columns in the table
cout << "Please enter the table size: ";
cin >> size;
// Print a size x size multiplication table

// First, print heading: 1 2 3 4 5 etc.

cout << " ";
// Print column heading
int column = 1;
while (column <= size) {
cout << setw(4) << column; // Print heading for this column.
column++;
}
cout << endl;

// Print line separator: +------------------

cout << " +";
column = 1;
while (column <= size) {
cout << "----"; // Print line for this column.
column++;
}

2015 Richard L. Halterman Draft date: January 18, 2015

6.2. NESTED LOOPS 142

cout << endl;

// Print table contents

int row = 1;
while (row <= size) { // Table has size rows.
cout << setw(2) << row << " |"; // Print heading for this row.
int column = 1; // Reset column for each row.
while (column <= size) { // Table has size columns.
int product = row*column; // Compute product
cout << setw(4) << product; // Display product
column++; // Next element
}
row++; // Next row
cout << endl; // Move cursor to next row
}
}

When the user supplies the value 10, Listing 6.12 (timestable.cpp) produces

Please enter the table size: 10

1 2 3 4 5 6 7 8 9 10
+----------------------------------------
1 | 1 2 3 4 5 6 7 8 9 10
2 | 2 4 6 8 10 12 14 16 18 20
3 | 3 6 9 12 15 18 21 24 27 30
4 | 4 8 12 16 20 24 28 32 36 40
5 | 5 10 15 20 25 30 35 40 45 50
6 | 6 12 18 24 30 36 42 48 54 60
7 | 7 14 21 28 35 42 49 56 63 70
8 | 8 16 24 32 40 48 56 64 72 80
9 | 9 18 27 36 45 54 63 72 81 90
10 | 10 20 30 40 50 60 70 80 90 100

An input of 15 yields

2015 Richard L. Halterman Draft date: January 18, 2015

6.2. NESTED LOOPS 143

Please enter the table size: 15

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
+------------------------------------------------------------
1 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
2 | 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
3 | 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45
4 | 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60
5 | 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75
6 | 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90
7 | 7 14 21 28 35 42 49 56 63 70 77 84 91 98 105
8 | 8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
9 | 9 18 27 36 45 54 63 72 81 90 99 108 117 126 135
10 | 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
11 | 11 22 33 44 55 66 77 88 99 110 121 132 143 154 165
12 | 12 24 36 48 60 72 84 96 108 120 132 144 156 168 180
13 | 13 26 39 52 65 78 91 104 117 130 143 156 169 182 195
14 | 14 28 42 56 70 84 98 112 126 140 154 168 182 196 210
15 | 15 30 45 60 75 90 105 120 135 150 165 180 195 210 225

If the user enters 7, the program prints

Please enter the table size: 7

1 2 3 4 5 6 7
+----------------------------
1 | 1 2 3 4 5 6 7
2 | 2 4 6 8 10 12 14
3 | 3 6 9 12 15 18 21
4 | 4 8 12 16 20 24 28
5 | 5 10 15 20 25 30 35
6 | 6 12 18 24 30 36 42
7 | 7 14 21 28 35 42 49

The user even can enter a 1:

Please enter the table size: 1

1
+----
1 | 1

As we can see, the table automatically adjusts to the size and spacing required by the users input.
This is how Listing 6.12 (timestable.cpp) works:

It is important to distinguish what is done only once (outside all loops) from that which is done

2015 Richard L. Halterman Draft date: January 18, 2015

6.2. NESTED LOOPS 144

repeatedly. The column heading across the top of the table is outside of all the loops; therefore, it is
printed all at once.

The work to print the heading for the rows is distributed throughout the execution of the outer loop.
This is because the heading for a given row cannot be printed until all the results for the previous row
have been printed.

A code fragment like

if (x < 10)
cout << " ";
cout << x;

prints x in one of two ways: if x is a one-digit number, it prints a space before it; otherwise, it does
not print the extra space. The net effect is to right justify one and two digit numbers within a two
character space printing area. This technique allows the columns within the times table to be properly
right aligned.

In the nested loop, row is the control variable for the outer loop; column controls the inner loop.

The inner loop executes size times on every single iteration of the outer loop. How many times is
the statement
cout << product << " "; // Display product

executed? size size times, one time for every product in the table.

A newline is printed after the contents of each row is displayed; thus, all the values printed in the
inner (column) loop appear on the same line.

Nested loops are used when an iterative process itself must be repeated. In our times table example, a
while loop is used to print the contents of each row, but multiple rows must be printed. The inner loop
prints the contents of each row, while the outer is responsible for printing all the rows.
Listing 6.13 (permuteabc.cpp) uses a triply-nested loop to print all the different arrangements of the
letters A, B, and C. Each string printed is a permutation of ABC.

Listing 6.13: permuteabc.cpp

// File permuteabc.cpp

#include <iostream>

using namespace std;

int main() {
char first = 'A'; // The first letter varies from A to C
while (first <= 'C') {
char second = 'A';
while (second <= 'C') { // The second varies from A to C
if (second != first) { // No duplicate letters
char third = 'A';
while (third <= 'C') { // The third varies from A to C
// Don't duplicate first or second letter
if (third != first && third != second)
cout << first << second << third << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

6.3. ABNORMAL LOOP TERMINATION 145

third++;
}
}
second++;
}
first++;
}
}

Notice how the if statements are used to prevent duplicate letters within a given string. The output of
Listing 6.13 (permuteabc.cpp) is all six permutations of ABC:

ABC
ACB
BAC
BCA
CAB
CBA

6.3 Abnormal Loop Termination

By default, a while statement executes until its condition becomes false. The executing program checks
this condition only at the top of the loop. This means that even if the Boolean expression that makes
up the condition becomes false before the program completes executing all the statements within the body
of the loop, all the remaining statements in the loops body must complete before the loop can once again
check its condition. In other words, the while statement in and of itself cannot exit its loop somewhere in
the middle of its body.
Ordinarily this behavior is not a problem. Usually the intention is to execute all the statements within
the body as an indivisible unit. Sometimes, however, it is desirable to immediately exit the body or recheck
the condition from the middle of the loop instead. C++ provides the break and continue statements to
to give programmers more flexibility designing the control logic of loops.

6.3.1 The break statement

C++ provides the break statement to implement middle-exiting control logic. The break statement causes
the immediate exit from the body of the loop. Listing 6.14 (addmiddleexit.cpp) is a variation of Listing 6.4
(addnonnegatives.cpp) that illustrates the use of break.

Listing 6.14: addmiddleexit.cpp

#include <iostream>

using namespace std;

int main() {
int input, sum = 0;
cout << "Enter numbers to sum, negative number ends list:";

2015 Richard L. Halterman Draft date: January 18, 2015

6.3. ABNORMAL LOOP TERMINATION 146

while (true) {
cin >> input;
if (input < 0)
break; // Exit loop immediately
sum += input;
}
cout << "Sum = " << sum << endl;
}

The condition of the while in Listing 6.14 (addmiddleexit.cpp) is a tautology. This means the condition
is true and can never be false. When the programs execution reaches the while statement it is guaranteed
to enter the loops body and the while loop itself does not provide a way of escape. The if statement in
the loops body:
if (input < 0) // Is input negative
break; // If so, exit the loop immediately

provides the necessary exit. In this case the break statement, executed conditionally based on the value of
the variable input, exits the loop. In Listing 6.14 (addmiddleexit.cpp) the break statement executes only
when the user enters a negative number. When the programs execution encounters the break statement,
it immediately jumps out of the loop. It skips any statements following the break within the loops body.
Since the statement
sum += input; // Accumulate user input

appears after the break, it is not possible for the program to add a negative number to the sum variable.
Some software designers believe that programmers should use the break statement sparingly because
it deviates from the normal loop control logic. Ideally, every loop should have a single entry point and
single exit point. While Listing 6.14 (addmiddleexit.cpp) has a single exit point (the break statement),
some programmers commonly use break statements within while statements in the which the condition
for the while is not a tautology. Adding a break statement to such a loop adds an extra exit point (the
top of the loop where the condition is checked is one point, and the break statement is another). Using
multiple break statements within a single loop is particularly dubious and you should avoid that practice.
Why have the break statement at all if its use is questionable and it is dispensable? The logic in
Listing 6.4 (addnonnegatives.cpp) is fairly simple, so the restructuring of Listing 6.14 (addmiddleexit.cpp)
is straightforward; in general, the effort to restructure code to avoid a break statement may complicate
the logic a bit and require the introduction of an additional Boolean variable. As shown in Figure 6.4, any
program that uses a break statement can be rewritten so that the break statement is not used. The no-
break version introduces a Boolean variable, and the loop control logic is a little more complicated. The
no-break version uses more memory (an extra variable) and more time to execute (requires an extra check
in the loop condition during every iteration of the loop). This extra memory is insignificant, and except for
rare, specialized applications, the extra execution time is imperceptible. In most cases, the more important
issue is that the more complicated the control logic for a given section of code, the more difficult the code
is to write correctly. In some situations, even though it violates the single entry point, single exit point
principle, a simple break statement is an acceptable loop control option.

6.3.2 The goto Statement

The break statement exits the single loop in which it is located. A break statement is insufficient to jump
completely out of the middle of a nested loop. The goto statement allows the programs execution flow

2015 Richard L. Halterman Draft date: January 18, 2015

6.3. ABNORMAL LOOP TERMINATION 147

bool looping = true;

while ( Condition 1 ) { while ( looping && Condition 1 ) {

Part A Part A

if ( Condition 2 ) { if ( Condition 2 ) {

Part B Eliminate Part B

the
break; looping = false;
break
} }
statement
else {
Part C
Part C
}
}
}

Figure 6.4: The code on the left generically represents any loop that uses a break statement. It is possible to transform
the code on the left to eliminate the break statement, as the code on the right shows.

to jump to a specified location within the function. Listing 6.15 (exitnested.cpp) uses a goto statement to
jump out from the middle of a nested loop.

Listing 6.15: exitnested.cpp

#include <iostream>

using namespace std;

int main() {
// Compute some products
int op1 = 2;
while (op1 < 100) {
int op2 = 2;
while (op2 < 100) {
if (op1 * op2 == 3731)
goto end;
cout << "Product is " << (op1 * op2) << endl;
op2++;
}
op1++;
}
end:
cout << "The end" << endl;
}

When op1 * op2 is 3731, program flow will jump to the specified label within the program. In this
example, the label is named end, but this name is arbitrary. Like variable names, label names should be

2015 Richard L. Halterman Draft date: January 18, 2015

6.3. ABNORMAL LOOP TERMINATION 148

chosen to indicate their intended purpose. The label here named end comes after and outside the nested
while loops.
A labels name is an identifier (see Section 3.3), and a label is distinguished by the colon that immedi-
ately follows its name. A label represents a target to which a goto can jump. A goto label must appear
before a statement within a function.
With the goto statement, the while is superfluous; for example, Listing 6.2 (iterativecounttofive.cpp)
could be rewritten without the while statement as shown in Listing 6.16 (gotoloop.cpp).

Listing 6.16: gotoloop.cpp

#include <iostream>

using namespace std;

int main() {
int count = 1; // Initialize counter
top:
if (count > 5)
goto end;
cout << count << endl; // Display counter, then
count++; // Increment counter
goto top;
end:
; // Target is an empty statement
}

Early programming languages like FORTRAN and early versions of BASIC did not have structured
statements like while, so programmers were forced to use goto statements to write loops. The problem
with using goto statements is that it is easy to develop program logic that is very difficult to under-
stand, even for the original author of the code. See the Wikipedia article about spaghetti code (http:
//en.wikipedia.org/wiki/Spaghetti_code). The structured programming revolution of the 1960s in-
troduced constructs such as the while statement and resulted in the disappearance of the use of goto in
most situations. All modern programming languages have a form of the while statement, so the goto
statement in C++ is largely ignored except for the case of breaking out of a nested loop. You similarly
should restrict your use of the goto statement to the abnormal exit of nested loops.

6.3.3 The continue Statement

When a programs execution encounters a break statement inside a loop, it skips the rest of the body
of the loop and exits the loop. The continue statement is similar to the break statement, except the
continue statement does not necessarily exit the loop. The continue statement skips the rest of the
body of the loop and immediately checks the loops condition. If the loops condition remains true, the
loops execution resumes at the top of the loop. Listing 6.17 (continueexample.cpp) shows the continue
statement in action.

Listing 6.17: continueexample.cpp

#include <iostream>

using namespace std;

int main() {

2015 Richard L. Halterman Draft date: January 18, 2015

6.3. ABNORMAL LOOP TERMINATION 149

int input, sum = 0;

bool done = false;
while (!done) {
cout << "Enter positive integer (999 quits): ";
cin >> input;
if (input < 0) {
cout << "Negative value " << input << " ignored"
<< endl;
continue; // Skip rest of body for this iteration
}
if (input != 999) {
cout << "Tallying " << input << endl;
sum += input;
}
else
done = (input == 999); // 999 entry exits loop
}
cout << "sum = " << sum << endl;
}

Programmers do not use the continue statement as frequently as the break statement since it
is easy to transform the code the would use contunue into an equivalent form that does not. List-
ing 6.18 (nocontinueexample.cpp) works exactly like Listing 6.17 (continueexample.cpp), but it avoids
the continue statement.
Listing 6.18: nocontinueexample.cpp
#include <iostream>

using namespace std;

int main() {
int input, sum = 0;
bool done = false;
while (!done) {
cout << "Enter positive integer (999 quits): ";
cin >> input;
if (input < 0) {
cout << "Negative value " << input << " ignored"
<< endl;
}
else
if (input != 999) {
cout << "Tallying " << input << endl;
sum += input;
}
else
done = (input == 999); // 999 entry exits loop
}
cout << "sum = " << sum << endl;
}

Figure 6.5 shows how we can rewrite any program that uses a continue statement into an equivalent
form that does not use continue. The transformation is simpler than for break elimination (see Fig-
ure 6.4), since the loops condition remains the same, and no additional variable is needed. The version

2015 Richard L. Halterman Draft date: January 18, 2015

6.4. INFINITE LOOPS 150

while ( Condition 1 ) { while ( Condition 1 ) {

Part A Part A

if ( Condition 2 ) { if ( Condition 2 ) {

Part B Eliminate Part B

the
continue; }
continue
} else {
statement
Part C
Part C
}
}
}

Figure 6.5: The code on the left generically represents any loop that uses a continue statement. It is possible to
transform the code on the left to eliminate the continue statement, as the code on the right shows.

that uses continue is no more efficient than the version that uses else; in fact, the Visual C++ and
GNU C++ compilers generate the same machine language code for Listing 6.17 (continueexample.cpp) and
Listing 6.18 (nocontinueexample.cpp). Also, the logic of the else version is no more complex than the
continue version. Therefore, unlike the break statement above, there is no compelling reason to use
the continue statement. Sometimes a programmer may add continue statement at the last minute to
an existing loop body to handle an exceptional condition (like ignoring negative numbers in the example
above) that initially went unnoticed. If the body of the loop is lengthy, a conditional statement with a
continue can be added easily near the top of the loop body without touching the logic of the rest of the
loop. The continue statement thus merely provides a convenient alternative for the programmer. The
else version is preferred.

6.4 Infinite Loops

An infinite loop is a loop without an exit. Once the program flow enters an infinite loops body it cannot
escape. Some infinite loops are by design; for example, a long-running server application, like a Web
server, may need to continuously check for incoming connections. This server application can perform this
checking within a loop that runs indefinitely. All too often, however, beginning programmers create infinite
loops by accident, and these infinite loops represent logic errors in their programs.
Intentional infinite loops should be made obvious. For example,
while (true) {
/* Do something forever . . . */
}

2015 Richard L. Halterman Draft date: January 18, 2015

6.4. INFINITE LOOPS 151

The Boolean literal true is always true, so it is impossible for the loops condition to be false. The only
ways to exit the loop is via a break statement, return statement (see Chapter 9), or an exit call (see
Section 8.1) embedded somewhere within its body.
It is easy to write an intentional infinite loop. Accidental infinite loops are quite common, but can be
puzzling for beginning programmers to diagnose and repair. Consider Listing 6.19 (findfactors.cpp) that
attempts to print all the integers with their associated factors from 1 to 20.

Listing 6.19: findfactors.cpp

#include <iostream>

using namespace std;

int main() {
// List of the factors of the numbers up to 20
int n = 1;
const int MAX = 20;
while (n <= MAX) {
int factor = 1;
cout << n << ": ";
while (factor <= n)
if (n % factor == 0) {
cout << factor << " ";
factor++;
}
cout << endl; // Go to next line for next n
n++;
}
}

It displays

1: 1
2: 1 2
3: 1

and then freezes up or hangs, ignoring any user input (except the key sequence Ctrl C on most
systems which interrupts and terminates the running program). This type of behavior is a frequent symptom
of an unintentional infinite loop. The factors of 1 display properly, as do the factors of 2. The first factor
of 3 is properly displayed and then the program hangs. Since the program is short, the problem may be
easy to locate. In some programs, though, the error may be challenging to find. Even in Listing 6.19
(findfactors.cpp) the debugging task is nontrivial since it involves nested loops. (Can you find and fix the
problem in Listing 6.19 (findfactors.cpp) before reading further?)
In order to avoid infinite loops, we must ensure that the loop exhibits certain properties:

The loops condition must not be a tautology (a Boolean expression that can never be false). For
example,
while (i >= 1 || i <= 10) {
/* Body omitted */

2015 Richard L. Halterman Draft date: January 18, 2015

6.4. INFINITE LOOPS 152

is an infinite loop since any value chosen for i will satisfy one or both of the two subconditions.
Perhaps the programmer intended to use a && instead of || to stay in the loop as long as i remains
in the range 1...10.
In Listing 6.19 (findfactors.cpp) the outer loop condition is
n <= MAX

If n is 21 and MAX is 20, then the condition is false, so this is not a tautology. Checking the inner
loop condition:
factor <= n

we see that if factor is 3 and n is 2, then the expression is false; therefore, it also is not a tautology.
The condition of a while must be true initially to gain access to its body. The code within the body
must modify the state of the program in some way so as to influence the outcome of the condition
that is checked at each iteration. This usually means code within the body of the loop modifies one of
the variables used in the condition. Eventually the variable assumes a value that makes the condition
false, and the loop terminates.
In Listing 6.19 (findfactors.cpp) the outer loops condition involves the variable n and constant MAX.
MAX cannot change, so to avoid an infinite loop it is essential that n be modified within the loop.
Fortunately, the last statement in the body of the outer loop increments n. n is initially 1 and MAX is
20, so unless the circumstances arise to make the inner loop infinite, the outer loop should eventually
terminate.
The inner loops condition involves the variables n and factor. No statement in the inner loop
modifies n, so it is imperative that factor be modified in the loop. The good news is factor is
incremented in the body of the inner loop, but the bad news is the increment operation is protected
within the body of the if statement. The inner loop contains one statement, the if statement. That
if statement in turn has two statements in its body:
while (factor <= n)
if (n % factor == 0) {
cout << factor << " ";
factor++;
}

If the condition of the if is ever false, the state of the program will not change when the body of the
inner loop is executed. This effectively creates an infinite loop. The statement that modifies factor
must be moved outside of the if statements body:
while (factor <= n) {
if (n % factor == 0)
cout << factor << " ";
factor++;
}

Note that the curly braces are necessary for the statement incrementing factor to be part of the body
of the while. This new version runs correctly.

Programmers can use a debugger to step through a program to see where and why an infinite loop
arises. Another common technique is to put print statements in strategic places to examine the values of the
variables involved in the loops control. The original inner loop can be so augmented:

2015 Richard L. Halterman Draft date: January 18, 2015

6.5. ITERATION EXAMPLES 153

while (factor <= n) {

cout << "factor = " << factor
<< " n = " << n << endl;
if (n % factor == 0) {
cout << factor << " ";
factor++;
}
}

It produces the following output:

1: factor = 1 n = 1
1
2: factor = 1 n = 2
1 factor = 2 n = 2
2
3: factor = 1 n = 3
1 factor = 2 n = 3
factor = 2 n = 3
factor = 2 n = 3
factor = 2 n = 3
factor = 2 n = 3
factor = 2 n = 3
.
.
.

The program continues to print the same line until the user interrupts its execution. The output demonstrates
that once factor becomes equal to 2 and n becomes equal to 3 the programs execution becomes trapped
in the inner loop. Under these conditions:

1. 2 < 3 is true, so the loop continues and

2. 3 % 2 is equal to 1, so the if statement will not increment factor.

It is imperative that factor be incremented each time through the inner loop; therefore, the statement
incrementing factor must be moved outside of the ifs guarded body.

6.5 Iteration Examples

We can implement some sophisticated algorithms in C++ now that we are armed with if and while
statements. This section provides several examples that show off the power of conditional execution and
iteration.

2015 Richard L. Halterman Draft date: January 18, 2015

6.5. ITERATION EXAMPLES 154

6.5.1 Drawing a Tree

Suppose we must write a program that draws a triangular tree, and the user provides the trees height. A
tree that is five levels tall would look like

*
***
*****
*******
*********

whereas a three-level tree would look like

*
***
*****

If the height of the tree is fixed, we can write the program as a simple variation of Listing 2.4 (arrow.cpp)
which uses only printing statements and no loops. Our program, however, must vary its height and width
based on input from the user.
Listing 6.20 (startree.cpp) provides the necessary functionality.

Listing 6.20: startree.cpp

#include <iostream>

using namespace std;

int main() {
int height; // Height of tree
cout << "Enter height of tree: ";
cin >> height; // Get height from user
int row = 0; // First row, from the top, to draw
while (row < height) { // Draw one row for every unit of height
// Print leading spaces
int count = 0;
while (count < height - row) {
cout << " ";
count++;
}
// Print out stars, twice the current row plus one:
// 1. number of stars on left side of tree
// = current row value
// 2. exactly one star in the center of tree
// 3. number of stars on right side of tree
// = current row value
count = 0;
while (count < 2*row + 1) {

2015 Richard L. Halterman Draft date: January 18, 2015

6.5. ITERATION EXAMPLES 155

cout << "*";

count++;
}
// Move cursor down to next line
cout << endl;
// Change to the next row
row++;
}
}

When a user runs Listing 6.20 (startree.cpp) and enters 7, the program displays

Enter height of tree: 7

*
***
*****
*******
*********
***********
*************

Listing 6.20 (startree.cpp) uses two sequential while loops both nested within a while loop. The
outer while loop is responsible for drawing one row of the tree each time its body is executed:

The program will execute the outer while loops body as long as the user enters a value greater than
zero; if the user enters zero or less, the program terminates and does nothing. This is the expected
behavior.
The last statement in the body of the outer while:
row++;

ensures that the variable row increases by one each time through the loop; therefore, it eventually
will equal height (since it initially had to be less than height to enter the loop), and the loop will
terminate. There is no possibility of an infinite loop here.
The body of the outer loop consists of more than one statement; therefore, the body must be enclosed
within curly braces. Whenever a group of statements is enclosed within curly braces a block is
formed. Any variable declared within a block is local to that block. A variables scope (the section of
the source code in which the variable exists and can be used) is from its point of declaration to the end
of the block in which it is declared. For example, the variables height and row are declared in the
block that is mains body; thus, they are local to main. The variable count is declared within the
block that is the body of the outer while statement; therefore, count is local to the outer while
statement. An attempt to use count outside the body of the outer while statement would be an
error.

What does it mean for a variable x to be local to a particular section of code? It means x does not exist
outside its scope. There may be other variables in the program named x, but they are different variables. If
it seems odd that you can have two different variables in the same program with the same name, consider
the fact that there can be two people in the same room with the same name. They are different people, but

2015 Richard L. Halterman Draft date: January 18, 2015

6.5. ITERATION EXAMPLES 156

they have the same name. Similarly, the meaning of a variable depends on its context, and its name is not
necessarily unique.
The two inner loops play distinct roles:

The first inner loop prints spaces. The number of spaces printed is equal to the height of the tree the
first time through the outer loop and decreases each iteration. This is the correct behavior since each
succeeding row moving down contains fewer leading spaces but more asterisks.

The second inner loop prints the row of asterisks that make up the tree. The first time through the
outer loop, row is zero, so no left side asterisks are printed, one central asterisk is printed (the top of
the tree), and no right side asterisks are printed. Each time through the loop the number of left-hand
and right-hand stars to print both increase by one and the same central asterisk is printed; therefore,
the tree grows one wider on each side each line moving down. Observe how the 2*row + 1 value
expresses the needed number of asterisks perfectly.

While it seems asymmetrical, note that no third inner loop is required to print trailing spaces on the
line after the asterisks are printed. The spaces would be invisible, so there is no reason to print them!

6.5.2 Printing Prime Numbers

A prime number is an integer greater than one whose only factors (also called divisors) are one and itself.
For example, 29 is a prime number (only 1 and 29 divide into it with no remainder), but 28 is not (2, 4, 7,
and 14 are factors of 28). Prime numbers were once merely an intellectual curiosity of mathematicians, but
now they play an important role in cryptography and computer security.
The task is to write a program that displays all the prime numbers up to a value entered by the user.
Listing 6.21 (printprimes.cpp) provides one solution.

Listing 6.21: printprimes.cpp

#include <iostream>

using namespace std;

int main() {
int max_value;
cout << "Display primes up to what value? ";
cin >> max_value;
int value = 2; // Smallest prime number
while (value <= max_value) {
// See if value is prime
bool is_prime = true; // Provisionally, value is prime
// Try all possible factors from 2 to value - 1
int trial_factor = 2;
while (trial_factor < value) {
if (value % trial_factor == 0) {
is_prime = false; // Found a factor
break; // No need to continue; it is NOT prime
}
trial_factor++;
}
if (is_prime)
cout << value << " "; // Display the prime number

2015 Richard L. Halterman Draft date: January 18, 2015

6.5. ITERATION EXAMPLES 157

value++; // Try the next potential prime number

}
cout << endl; // Move cursor down to next line
}

Listing 6.21 (printprimes.cpp), with an input of 90, produces:

Display primes up to what value? 90

2 3 5 7 11 13 17 19 23 29 31 37 41 43 47 53 59 61 67 71 73 79 83 89

The logic of Listing 6.21 (printprimes.cpp) is a little more complex than that of Listing 6.20 (startree.cpp).
The user provides a value for max_value. The main loop (outer while iterates over all the values from
two to max_value:

Two new variables, local to the body of the outer loop, are introduced: trial_factor and
is_prime. is_prime is initialized to true, meaning value is assumed to be prime unless our
tests prove otherwise. trial_factor takes on all the values from two to value - 1 in the inner
loop:
int trial_factor = 2;
while (trial_factor < value) {
if (value % trial_factor == 0) {
is_prime = false; // Found a factor
break; // No need to continue; it is NOT prime
}
trial_factor++;
}

The expression value % trial_factor is zero when trial_factor divides into value
with no remainderexactly when trial_factor is a factor of value. If the executing program
determines that any of the values of trial_factor is a factor of value, then it sets is_prime
to false and exits the loop via the break. If the loop continues to completion, the program never sets
is_prime to false, which means it found no factors and value is indeed prime.

The if statement after the inner loop:

if (is_prime)
cout << value << " "; // Display the prime number

simply checks the status of is_prime. If is_prime is true, then value must be prime, so it
prints value along with an extra space for separation from output it may produce during subsequent
iterations.

Some important questions we can ask include:

1. If the user enters a 2, will it be printed?

In this case max_value = value = 2, so the condition of the outer loop
value <= max_value

2015 Richard L. Halterman Draft date: January 18, 2015

6.5. ITERATION EXAMPLES 158

is true, since 2 <= 2. is_prime is set to true, but the condition of the inner loop
trial_factor < value

is not true (2 is not less than 2). Thus, the inner loop is skipped, is_prime is not changed from
true, and 2 is printed. This behavior is correct because 2 is the smallest prime number (and the only
even prime).

2. if the user enters a number less than 2, is anything printed?

The while condition ensures that values less than two are not considered. The body of the while
will never be entered. Only the newline is printed, and no numbers are displayed. This behavior is
correct.

3. Is the inner loop guaranteed to always terminate?

In order to enter the body of the inner loop, trial_factor must be less than value. value
does not change anywhere in the loop. trial_factor is not modified anywhere in the if
statement within the loop, and it is incremented within the loop immediately after the if state-
ment. trial_factor is, therefore, incremented during each iteration of the loop. Eventually,
trial_factor will equal value, and the loop will terminate.

4. Is the outer loop guaranteed to always terminate?

In order to enter the body of the outer loop, value must be less than or equal to max_value.
max_value does not change anywhere in the loop. value is increased in the last statement within
the body of the outer loop, and value is not modified anywhere else. Since the inner loop is guaran-
teed to terminate as shown in the previous answer, eventually value will exceed max_value and
the loop will end.

We can rearrange the logic of the inner while to avoid the break statement. The current version is:
while (trial_factor < value) {
if (value % trial_factor == 0) {
is_prime = false; // Found a factor
break; // No need to continue; it is NOT prime
}
trial_factor++;
}

We can rewrite it as:

while (is_prime && trial_factor < value) {
is_prime = (value % trial_factor != 0);
trial_factor++; // Try next factor
}

This version without the break introduces a slightly more complicated condition for the while but re-
moves the if statement within its body. is_prime is initialized to true before the loop. Each time through
the loop it is reassigned. trial_factor will become false if at any time value % trial_factor
is zero. This is exactly when trial_factor is a factor of value. If is_prime becomes false, the
loop cannot continue, and if is_prime never becomes false, the loop ends when trial_factor be-
comes equal to value. Due to operator precedence, the parentheses are not necessary. The parentheses do
improve readability, since an expression including both == and != is awkward for humans to parse. When
parentheses are placed where they are not needed, as in

2015 Richard L. Halterman Draft date: January 18, 2015

6.6. SUMMARY 159

x = (y + 2);

the compiler simply ignores them, so there is no efficiency penalty in the compiled code.
We can shorten the loop even further:
while (is_prime && trial_factor < value)
is_prime = (value % trial_factor++ != 0);

This version uses the post-increment operator within the test expression (see Section 4.9). Recall that with
the post-increment operator the value of the variable is used in the surrounding expression (if any), and then
the variable is incremented. Since the whiles body now contains only one statement, the curly braces are
not needed.

6.6 Summary

The while statement allows the execution of code sections to be repeated multiple times.

The condition of the while controls the execution of statements within the whiles body.

The statements within the body of a while are executed over and over until the condition of the
while is false.

If the whiles condition is initially false, the body is not executed at all.

In an infinite loop, the whiles condition never becomes false.

The statements within the whiles body must eventually lead to the condition being false; otherwise,
the loop will be infinite.

Do not confuse while statements with if statements; their structure is very similar (while re-
served word instead of the if word), but they behave differently.

Infinite loops are rarely intentional and are usually accidental.

An infinite loop can be diagnosed by putting a printing statement inside its body.

An assignment expression has a value; the expressions value is the same as the expression on the
right side of the assignment operator. This fact can be used to streamline the control of a loop that
repeats based on user input.

A loop contained within another loop is called a nested loop.

Iteration is a powerful mechanism and can be used to solve many interesting problems.

A block is any section of source code enclosed within curly braces. A compound statement is one
example of a block.

A variable declared within a block is local to that block.

Complex iteration using nested loops mixed with conditional statements can be difficult to do cor-
rectly.

Sometimes simple optimizations can speed up considerably the execution of loops.

2015 Richard L. Halterman Draft date: January 18, 2015

6.7. EXERCISES 160

The break statement immediately exits a loop, skipping the rest of the loops body, without checking
to see if the condition is true or false. Execution continues with the statement immediately following
the body of the loop.
In a nested loop, the break statement exits only the loop in which the break is found.
The goto statement directs the programs execution to a labeled statement within the function. The
goto statement is legitimately used only to exit completely from the depths of a nested loop.
The continue statement immediately checks the loops condition, skipping the rest of the loops
body. If the condition is true, the execution continues at the top of the loop as usual; otherwise, the
loop is terminated and execution continues with the statement immediately following the loops body.
false.
In a nested loop, the continue statement affects only the loop in which the continue is found.

6.7 Exercises

1. In Listing 6.4 (addnonnegatives.cpp) could the condition of the if statement have used > instead of
>= and achieved the same results? Why?
2. In Listing 6.4 (addnonnegatives.cpp) could the condition of the while statement have used > instead
of >= and achieved the same results? Why?
3. Use a loop to rewrite the following code fragment so that it uses just one cout and one endl.
cout << 2 << endl;
cout << 4 << endl;
cout << 6 << endl;
cout << 8 << endl;
cout << 10 << endl;
cout << 12 << endl;
cout << 14 << endl;
cout << 16 << endl;

4. In Listing 6.4 (addnonnegatives.cpp) what would happen if the statement containing cin is moved
out of the loop? Is moving the assignment out of the loop a good or bad thing to do? Why?
5. How many asterisks does the following code fragment print?
int a = 0;
while (a < 100) {
cout << "*";
a++;
}
cout << endl;

6. How many asterisks does the following code fragment print?

int a = 0;
while (a < 100)
cout << "*";
cout << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

6.7. EXERCISES 161

7. How many asterisks does the following code fragment print?

int a = 0;
while (a > 100) {
cout << "*";
a++;
}
cout << endl;

8. How many asterisks does the following code fragment print?

int a = 0;
while (a < 100) {
int b = 0;
while (b < 55) {
cout << "*";
b++;
}
cout << endl;
}

9. How many asterisks does the following code fragment print?

int a = 0;
while (a < 100) {
if (a % 5 == 0)
cout << "*";
a++;
}
cout << endl;

10. How many asterisks does the following code fragment print?
int a = 0;
while (a < 100) {
int b = 0;
while (b < 40) {
if ((a + b) % 2 == 0)
cout << "*";
b++;
}
cout << endl;
a++;
}

11. How many asterisks does the following code fragment print?
int a = 0;
while (a < 100) {
int b = 0;
while (b < 100) {
int c = 0;
while (c < 100) {

2015 Richard L. Halterman Draft date: January 18, 2015

6.7. EXERCISES 162

cout << "*";

c++;
}
b++;
}
a++;
}
cout << endl;

12. What is printed by the following code fragment?

int a = 0;
while (a < 100)
cout << a++;
cout << endl;

13. What is printed by the following code fragment?

int a = 0;
while (a > 100)
cout << a++;
cout << endl;

14. Rewrite the following code fragment using a break statement and eliminating the done variable.
Your code should behave identically to this code fragment.
bool done = false;
int n = 0, m = 100;
while (!done && n != m) {
cin >> n;
if (n < 0)
done = true;
cout << "n = " << endl;
}

15. Rewrite the following code fragment so it does not use a break statement. Your code should behave
identically to this code fragment.
// Code with break ...

16. Rewrite the following code fragment so it eliminates the continue statement. Your new codes
logic should be simpler than the logic of this fragment.
int x = 100, y;
while (x > 0) {
cin >> y;
if (y == 25) {
x--;
continue;
}
cin >> x;
cout << "x = " << x << endl;
}

2015 Richard L. Halterman Draft date: January 18, 2015

6.7. EXERCISES 163

17. Suppose you were given some code from the 1960s in a language that did not support structured
statements like while. Your task is to modernize it and adapt it to C++. The following code fragment
has been adapted to C++ already, but you must now structure it with a while statement to replace
the gotos. Your code should be goto free and still behave identically to this code fragment.
int i = 0;
top: if (i >= 10)
goto end;
cout << i << endl;
i++;
goto top;
end:

18. Simplify with goto...

19. What is printed by the following code fragment?
int a = 0;
while (a < 100);
cout << a++;
cout << endl;

20. Write a C++ program that accepts a single integer value entered by the user. If the value entered is
less than one, the program prints nothing. If the user enters a positive integer, n, the program prints
an n n box drawn with * characters. If the users enters 1, for example, the program prints

If the user enters a 2, it prints

**
**

An entry of three yields

***
***
***

and so forth. If the user enters 7, it prints

*******
*******
*******
*******
*******
*******
*******

2015 Richard L. Halterman Draft date: January 18, 2015

6.7. EXERCISES 164

that is, a 7 7 box of * symbols.

21. Write a C++ program that allows the user to enter exactly twenty double-precision floating-point
values. The program then prints the sum, average (arithmetic mean), maximum, and minimum of the
values entered.

22. Write a C++ program that allows the user to enter any number of non-negative double-precision
floating-point values. The user terminates the input list with any negative value. The program then
prints the sum, average (arithmetic mean), maximum, and minimum of the values entered. The
terminating negative value is not used in the computations.

23. Redesign Listing 6.20 (startree.cpp) so that it draws a sideways tree pointing right; for example, if
the user enters 7, the program would print

*
**
***
****
*****
******
*******
******
*****
****
***
**
*

24. Redesign Listing 6.20 (startree.cpp) so that it draws a sideways tree pointing left; for example, if the
user enters 7, the program would print

*
**
***
****
*****
******
*******
******
*****
****
***
**
*

2015 Richard L. Halterman Draft date: January 18, 2015

 165

Chapter 7

Other Conditional and Iterative

Statements

The if/else and while statements are sufficient to implement any algorithms that involve conditional
execution and looping. The break and continue statements are convenient but are not necessary. C++
provides some additional conditional and iterative statements that are more convenient to use in some
circumstances. These additional statements include

switch: an alternative to some multi-way if/else statements

the conditional operator: an expression that exhibits the behavior of an if/else statement

do/while: a loop that checks its condition after its body is executed

for: a loop convenient for counting

These alternate constructs allow certain parts of algorithms to expressed more clearly and succinctly.
This chapter explores these other forms of expressing conditional execution and iteration.

7.1 The switch Statement

The switch statement provides a convenient alternative for some multi-way if/else statements like the
one in Listing 5.15 (restyleddigittoword.cpp). Listing 7.1 (switchdigittoword.cpp) is a new implementation
of Listing 5.15 (restyleddigittoword.cpp) that uses a switch statement instead of a multi-way if/else
statement.

Listing 7.1: switchdigittoword.cpp

#include <iostream>

using namespace std;

int main() {
int value;
cout << "Please enter an integer in the range 0...5: ";

2015 Richard L. Halterman Draft date: January 18, 2015

7.1. THE SWITCH STATEMENT 166

cin >> value;

switch (value) {
case 0:
cout << "zero";
break;
case 1:
cout << "one";
break;
case 2:
cout << "two";
break;
case 3:
cout << "three";
break;
case 4:
cout << "four";
break;
case 5:
cout << "five";
break;
default:
if (value < 0)
cout << "Too small";
else
cout << "Too large";
break;
}
cout << endl;
}

The general form of a switch is:

2015 Richard L. Halterman Draft date: January 18, 2015

7.1. THE SWITCH STATEMENT 167

switch ( integral expression ) {

case integral constant 1 :

statement sequence 1
break;

case integral constant 2 :

statement sequence 2
break;

case integral constant 3 :

statement sequence 3
break;

case integral constant n :

statement sequence n
break;

default:
default statement sequence
}

In a switch statement

The reserved word switch identifies a switch statement.

The required parenthesized expression that follows the word switch must evaluate to an integral
value. Any integer type, characters, and Boolean expressions are acceptable. Floating point expres-
sions and other non-integer types are forbidden.

2015 Richard L. Halterman Draft date: January 18, 2015

7.1. THE SWITCH STATEMENT 168

The body of the switch is enclosed by required curly braces.

Each occurrence of the word case is followed by an integral constant and a colon (:). We call the
integral constant a case label. This label can be either a literal value or a const symbolic value
(see Section 3.6). In particular, non-const variables and other expressions are expressly forbidden.
The case label defines a position within the code; it is not an executable statement. A case label
represents a target to which the programs execution flow can jump.
If the case label matches the switchs expression, then the statements that follow that label are
executed up until the break statement is encountered. The statements and break statement that
follow each case label are optional. One way to execute one set of statements for more than one
case label is to provide empty statements for one or more of the labels, as in:
cin >> key; // get key from user
switch (key) {
case 'p':
case 'P':
cout << "You choose \"P\"" << endl;
break;
case 'q':
case 'Q':
done = true;
break;
}

Here either an upper- or lowercase P result in the same action You chose P is printed. If the user
enters either an upper- or lowercase Q, the done Boolean variable is set to true. If the user enters
neither P nor Q, none of the statements in the switch is executed.
The break statement is optional. When a case label is matched, the statements that follow are
executed until a break statement is encountered. The control flow then transfers out of the body of
the switch. In this way, the break within a switch works just like a break within a loop: the
rest of the body of the statement is skipped and program execution resumes at the next statement fol-
lowing the body. A missing break statement, a common error, when its omission is not intentional,
causes the statements of the succeeding case label to be executed. The process continues until a
break is encountered or the end of the switch body is reached.

The default label is matched if none of the case labels match. It serves as a catch all option like
the final else in a multi-way if/else statement. The default label is optional. If it is missing
and none of the case labels match the expression, then no statement within the switchs body is
executed.

The switch statement has two restrictions that make it less general than the multi-way if/else:

The switch argument must be an integral expression.

Case labels must be constant integral values. Integral literals and constants are acceptable. Variables
or expressions are not allowed.

To illustrate these restrictions, consider the following if/else statement that translates easily to an equiv-
alent switch statement:

2015 Richard L. Halterman Draft date: January 18, 2015

7.1. THE SWITCH STATEMENT 169

if (x == 1) {
// Do 1 stuff here . . .
}
else if (x == 2) {
// Do 2 stuff here . . .
}
else if (x == 3) {
// Do 3 stuff here . . .
}

The corresponding switch statement is:

switch (x) {
case 1:
// Do 1 stuff here . . .
break;
case 2:
// Do 2 stuff here . . .
break;
case 3:
// Do 3 stuff here . . .
break;
}

Now consider the following if/else:

if (x == y) {
// Do "y" stuff here . . .
}
else if (x > 2) {
// Do "> 2" stuff here . . .
}
else if (z == 3) {
// Do 3 stuff here . . .
}

This code cannot be easily translated into a switch statement. The variable y cannot be used as a case
label. The second choice checks for an inequality instead of an exact match, so direct translation to a case
label is impossible. In the last condition, a different variable is checked, z instead of x. The control flow
of a switch statement is determined by a single value (for example, the value of x), but a multi-way
if/else statement is not so constrained.
Where applicable, a switch statement allows programmers to compactly express multi-way selection
logic. Most programmers find a switch statement easier to read than an equivalent multi-way if/else
construct.
A positive consequence of the switch statements restrictions is that it allows the compiler to produce
more efficient code for a switch than for an equivalent if/else. If a choice must be made from one of
several or more options, and the switch statement can be used, then the switch statement will likely be
faster than the corresponding multi-way if/else.

2015 Richard L. Halterman Draft date: January 18, 2015

7.2. THE CONDITIONAL OPERATOR 170

7.2 The Conditional Operator

As purely a syntactical convenience, C++ provides an alternative to the if/else construct called the
conditional operator. It has limited application but is convenient nonetheless. The following code fragment
assigns one of two things to x:

the result of y/z, if z is nonzero, or

zero, if z is zero; we wish to avoid the run-time error of division by zero.

// Assign a value to x:
if (z != 0)
x = y/z; // Division is possible
else
x = 0; // Assign a default value instead

This code has two assignment statements, but only one is executed at any given time. The conditional
operator makes for a more compact statement:
// Assign a value to x:
x = (z != 0) ? y/z : 0;

The general form of a conditional expression is:

( condition )? expression 1 : expression 2

condition is a normal Boolean expression that might appear in an if statement. Parentheses around
the condition are not required but should be used to improve the readability.
expression 1 is the overall value of the conditional expression if condition is true.
expression 2 is the overall value of the conditional expression if condition is false.

The conditional operator uses two symbols (? and :) and three operands. Since it has three operands it
is classified as a ternary operator (C++s only one). The overall type of a conditional expression is the more
dominant of exp1 and exp2 The conditional expression can be used anywhere an expression can be used. It
is not a statement itself; it is used within a statement.
As another example, the absolute value of a number is defined in mathematics by the following formula:

n, when n 0
|n| =
n, when n < 0

In other words, the absolute value of a positive number or zero is the same as that number; the abso-
lute value of a negative number is the additive inverse (negative of) of that number. The following C++
expression represents the absolute value of the variable n:

2015 Richard L. Halterman Draft date: January 18, 2015

7.3. THE DO/WHILE STATEMENT 171

(n < 0) ? -n : n

Some argue that the conditional operator is cryptic, and thus its use reduces a programs readability. To
seasoned C++ programmers it is quite understandable, but it is used sparingly because of its very specific
nature.

7.3 The do/while Statement

An executing program checks the condition of a while statement (Section 6.1) before executing any of
the statements in its body; thus, we say a while loop is a top-checking loop. Sometimes this sequence of
checking then executing is inconvenient; for example, consider
The while statement (Section 6.1) checks its condition before its body is executed; thus, it is a top-
checking loop. Sometimes this sequence of checking the condition first then executing the body is inconve-
nient; for example, consider Listing 7.2 (goodinputonly.cpp).

Listing 7.2: goodinputonly.cpp

#include <iostream>

using namespace std;

int main() {
int in_value = -1;
cout << "Please enter an integer in the range 0-10: ";
// Insist on values in the range 0...10
while (in_value < 0 || in_value > 10)
cin >> in_value;
// in_value at this point is guaranteed to be within range
cout << "Legal value entered was " << in_value << endl;
}

The loop in Listing 7.2 (goodinputonly.cpp) traps the user in the while until the user provides a number
in the desired range. Heres how it works:

The condition of the while specifies a set that includes all values that are not in the desired range.
The initialization of in_value to 1 ensures the condition of the while will be true initially, and,
thus, the program always will execute the loops body at least one time.
The user does not get a chance to enter a value until programs execution is inside the loop.
The only way the user can escape the loop is to enter a value that violates the conditionprecisely a
value in the desired range.

The initialization of in_value before the loop check is somewhat artificial. It is there only to ensure
entry into the loops body. It seems unnatural to check for a valid value before the user gets a chance to
enter it. A loop that checks its condition after its body is executed at least once would be more appropriate.
The do/while statement is a bottom-checking loop that behaves exactly in this manner. Listing 7.3
(betterinputonly.cpp) uses a do/while statement to check for valid input.

Listing 7.3: betterinputonly.cpp

2015 Richard L. Halterman Draft date: January 18, 2015

7.3. THE DO/WHILE STATEMENT 172

#include <iostream>

using namespace std;

int main() {
int in_value;
cout << "Please enter an integer in the range 0-10: ";
// Insist on values in the range 0...10
do
cin >> in_value;
while (in_value < 0 || in_value > 10);
// in_value at this point is guaranteed to be within range
cout << "Legal value entered was " << in_value << endl;
}

Notice that there is no need to initialize in_value since its value is not used until after it is assigned
through the input stream cin. Figure 7.1 compares the flowcharts of a while and do/while loop.

Figure 7.1: The flowcharts for while and do/while loops

The do/while statement has the general form:

2015 Richard L. Halterman Draft date: January 18, 2015

7.4. THE FOR STATEMENT 173

do

statement

while ( condition );

The reserved words do and while identify a do/while statement. The do and while keywords
delimit the loops body, but curly braces are still required if the body consists of more than one
statement.
The condition is associated with the while at the end of the loop. The condition is a Boolean
expression and must be enclosed within parentheses.
The statement is exactly like the statement in the general form of the while loop (see Section 6.1).
It can be a compound statement enclosed within curly braces.

The body of a do/while statement, unlike the while statement, is guaranteed to execute at least
once.
The do/while loop is a convenience to the programmer and is not an essential programming construct.
It is easy to transform any code that uses a do/while statement into code that behaves identically that
uses a while statement instead. In practice, programmers use while loops much more frequently than
do/while loops because more algorithms require top-checking loops than bottom-checking loops. The
do/while statement is included in C++ for a reason, however. Transforming an algorithm that can be
expressed more naturally with a bottom-checking loop into one the uses a top-checking loop can lead to
awkward code. Use do/while when appropriate.

7.4 The for Statement

Recall Listing 6.2 (iterativecounttofive.cpp). It simply counts from one to five. Counting is a frequent
activity performed by computer programs. Certain program elements are required in order for any program
to count:

A variable must be used to keep track of the count; in Listing 6.2 (iterativecounttofive.cpp), count
is the aptly named counter variable.
The counter variable must be given an initial value. In the case of Listing 6.2 (iterativecounttofive.cpp),
the initial value is 1.
The variable must be modified (usually incremented) as the program counts. The statement
count++;

increments count in Listing 6.2 (iterativecounttofive.cpp).

A way must be provided to determine if the counting has completed. In Listing 6.2 (iterativecounttofive.cpp),
the condition of the while statement determines if the counting is complete or must contine.

2015 Richard L. Halterman Draft date: January 18, 2015

7.4. THE FOR STATEMENT 174

C++ provides a specialized loop that packages these four programming elements into one convenient
statement. Called the for statement, its general form is

for ( initialization ; condition ; modification )

statement

The reserved word for identifies a for statement.

The loop is controlled by a special variable called the loop variable.

The header, contained in parentheses, contains three parts, each separated by semicolons:

Initialization. The initialization part assigns an initial value to the loop variable. The loop
variable may be declared here as well; if it is declared here, then its scope is limited to the for
statement. This means you may use that loop variable only within the loop. It also means you
are free to reuse that variables name outside the loop to declare a different variable with the
same name as the loop variable.
The initialization part is performed one time.

Condition. The condition part is a Boolean expression, just like the condition of a while
statement. The condition is checked each time before the body is executed.

Modification. The modification part generally changes the loop variable. The change should be
such that the condition will eventually become false so the loop will terminate. The modification
is performed during each iteration after the body is executed.
Notice that the last part (modification) is not following by a semicolon; semicolons are used
strictly to separate the three parts.

The statement is like the body of any other loop. It may be a compound statement within curly braces.

Any for loop can be rewritten as a while loop. The general form of the for loop given above can be
written equivalently as

2015 Richard L. Halterman Draft date: January 18, 2015

7.4. THE FOR STATEMENT 175

initialization

while ( condition ) {

statement

modification

Listing 7.4 (forcounttofive.cpp) uses a for statement to count to five.

Listing 7.4: forcounttofive.cpp

#include <iostream>

using namespace std;

int main() {
for (int count = 1; count <= 5; count++)
cout << count << endl; // Display counter
}

With a while loop, the four counting components (variable declaration, initialization, condition, and
modification can be scattered throughout the code. With a for loop, a programmer should be able to
determine all the important information about the loops control by looking at one statement.
Recall Listing 6.12 (timestable.cpp) that prints a multiplication table on the screen. We can organize
its code better by converting all the while statements to for statements. The result uses far less code, as
shown in Listing 7.5 (bettertimestable.cpp).

Listing 7.5: bettertimestable.cpp

#include <iostream>
#include <iomanip>

using namespace std;

int main() {
int size; // The number of rows and columns in the table
cout << "Please enter the table size: ";
cin >> size;
// Print a size x size multiplication table

// First, print heading

cout << " ";
for (int column = 1; column <= size; column++)

2015 Richard L. Halterman Draft date: January 18, 2015

7.4. THE FOR STATEMENT 176

cout << setw(4) << column; // Print heading for this column.
cout << endl;
// Print line separator
cout << " +";
for (int column = 1; column <= size; column++)
cout << "----"; // Print separator for this column.
cout << endl;
// Print table contents
for (int row = 1; row <= size; row++) {
cout << setw(4) << row << " |"; // Print row label.
for (int column = 1; column <= size; column++)
cout << setw(4) << row*column; // Display product
cout << endl; // Move cursor to next row
}
}

A for loop is ideal for stepping through the rows and columns. The information about the control of
both loops is now packaged in the respective for statements instead of being spread out in various places
in main. In the while version, it is easy for the programmer to forget to update one or both of the counter
variables (row and/or column). The for makes it harder for the programmer to forget the loop variable
update, since it is done right up front in the for statement header.
It is considered bad programming practice to do either of the following in a for statement:

Modify the loop control variable within the body of the loopif the loop variable is modified
within the body, then the logic of the loops control is no longer completely isolated to the for state-
ments header. The programmer must look elsewhere within the statement to understand completely
how the loop works.
Prematurely exit the loop with a breakthis action also violates the concept of keeping all the
loop control logic in one place (the fors header).

The language allows both of these practices, but experience shows that it is best to avoid them. If it seems
necessary to violate this advice, consider using a different kind of loop. The while and do/while loops
do not imply the same degree of control regularity expected in a for loop.
Listing 7.6 (permuteabcd.cpp) is a rewrite of Listing 6.13 (permuteabc.cpp) that replaces its while
loops with for loops and adds an additional character.

Listing 7.6: permuteabcd.cpp

// File permuteabcd.cpp

#include <iostream>

using namespace std;

int main() {
for (char first = 'A'; first <= 'D'; first ++)
for (char second = 'A'; second <= 'D'; second++)
if (second != first) // No duplicate letters
for (char third = 'A'; third <= 'D'; third++)
if (third != first && third != second)
for (char fourth = 'A'; fourth <= 'D'; fourth++)
if (fourth != first && fourth != second && fourth != third)

2015 Richard L. Halterman Draft date: January 18, 2015

7.4. THE FOR STATEMENT 177

cout << first << second << third << fourth << endl;
}

Notice that since all the variable initialization and incrementing is taken care of in the for statement
headers, we no longer need compound statements in the loop bodies, so the curly braces are unnecessary.
Listing 7.6 (permuteabcd.cpp) prints all 24 permutations of ABCD:

ABCD
ABDC
ACBD
ACDB
ADBC
ADCB
BACD
BADC
BCAD
BCDA
BDAC
BDCA
CABD
CADB
CBAD
CBDA
CDAB
CDBA
DABC
DACB
DBAC
DBCA
DCAB
DCBA

Listing 7.7 (forprintprimes.cpp) is a rewrite of Listing 6.21 (printprimes.cpp) that replaces its while
loops with for loops.

Listing 7.7: forprintprimes.cpp

#include <iostream>

using namespace std;

int main() {

int max_value;
cout << "Display primes up to what value? ";
cin >> max_value;
for (int value = 2; value <= max_value; value++) {
// See if value is prime
bool is_prime = true; // Provisionally, value is prime
// Try all possible factors from 2 to value - 1

2015 Richard L. Halterman Draft date: January 18, 2015

7.4. THE FOR STATEMENT 178

for (int trial_factor = 2;

is_prime && trial_factor < value;
trial_factor++)
is_prime = (value % trial_factor != 0);
if (is_prime)
cout << value << " "; // Display the prime number
}
cout << endl; // Move cursor down to next line
}

As shown in Listing 7.7 (forprintprimes.cpp), the conditional expression in the for loop is not limited
to a simple test of the loop control variable; it can be any legal Boolean expression. Programmers can use
logical the and (&&), or (||), and not (!) operators to create complex Boolean expressions, if necessary.
The modification part of the for loop is not limited to simple arithmetic and can be quite elaborate. For
example:
for (double d = 1000; d >= 1; cin >> d) {
/* Body goes here */
}

Here d is reassigned from the input stream. If necessary, multiple variables can be initialized in the initial-
ization part:
for (int i = 0, j = 100; i < j; i++) {
/* Body goes here */
}

While the for statement supports such complex headers, simpler is usually better. Ordinarily the for
loop should manage just one control variable, and the initialization, condition, and modification parts should
be straightforward. If a particular programming situation warrants an overly complicated for construction,
consider using another kind of loop.
Any or all of the parts of the for statement (initialization, condition, modification, and body) may be
omitted:

Initialization. If the initialization is missing, as in

for (; i < 10; i++)
/* Body goes here */

then no initialization is performed by the for loop, and it must be done elsewhere.

Condition. If the condition is missing, as in

for (int i = 0; ; i++)
/* Body goes here */

then the condition is true by default. A break or goto must appear in the body unless an infinite
loop is intended.

Modification. If the modification is missing, as in

for (int i = 0; i < 10; )
/* Body goes here */

2015 Richard L. Halterman Draft date: January 18, 2015

7.4. THE FOR STATEMENT 179

then the for performs no automatic modification; the modification must be done by a statement in
the body to avoid an infinite loop.
Body. An empty body, as in
for (int i = 0; i < 10; i++) {}

or
for (int i = 0; i < 10; i++);

results in an empty loop. Some programmers use an empty loop to produce a non-portable delay in
the programs execution. A programmer may, for example, need to slow down a graphical animation.
Such an attempt using an empty loop is non-portable for several reasons. If the program actually
executes the loop, slower computers will delay longer than faster computers. The timing of the
programs delay will differ from one computer to another. Worse yet, some compilers may detect
that such code has no functional effect and optimize away the empty loop. This means the compiler
will ignore the for statement altogether.

As mentioned in Section 5.3, be careful about accidentally putting a semicolon at

the end of the for header, as in
for (int i = 0; i < 10; i++);
/* Intended body goes here */
The semicolon terminates the for statement, and the intended body that follows
is not the body, even though it may be properly indented.

One common C/C++ idiom to make an intentional infinite loop is to use a for statement with all control
information missing:
for ( ;; )
/* Body goes here */

Omitting all the parts of the for header is a statement from the programmer that says I know what I am
doingI really want an infinite loop here. In reality the loop may not be infinite at all; its body could
contain a break or goto statement.
While the for statement supports the omission of parts of its header, such constructs should be avoided.
The intention of the for loop is to allow the programmer to see all the aspects of the loops control in one
place. If some of these control responsibilities are to be handled elsewhere (not in the fors header) then
consider using another kind of loop.
Programmers usually select a simple name for the control variable of a for statement. Recall that
variable names should be well chosen to reflect the meaning of their use within the program. It may come
as a surprise that i is probably the most common name used for an integer control variable in a for loop.
This practice has its roots in mathematics where variables such as i, j, and k are commonly used to index
vectors and matrices. Such mathematical structures have programming analogs in arrays and vectors,
which we explore in Chapter 11. Computer programmers make considerable use of for loops in array
and vector processing, so programmers have universally adopted this convention of short control variable
names. Thus, it generally is acceptable to use simple identifiers like i as loop control variables.
C++ allows the break, continue, and goto statements to be used in the body of a for statement.
Like with the while and do/while statements, break causes immediate loop termination, continue

2015 Richard L. Halterman Draft date: January 18, 2015

7.5. SUMMARY 180

causes the condition to be immediately checked to determine if the iteration should continue, and goto
jumps to a label somewhere in the function. As previously mentioned, however, for loop control should be
restricted to its header, and the use of break, continue, and goto within for loops should be avoided.
Any for loop can be rewritten with a while loop and behave identically. For example, consider the
for loop
for (int i = 1; i <= 10; i++)
cout << i << endl;

and next consider the while loop that behaves exactly the same way:
int i = 1;
while (i <= 10) {
cout << i << endl;
i++;
}

Which is better? The for loop conveniently packages the loop control information in its header, but in
the while loop this information is distributed throughout the small section of code. The for loop thus
provides a better organization of the loop control code. Does one loop outperform the other? No, most
compilers produce essentially the same code for both constructs. Thus, the for loop is preferred in this
example.

7.5 Summary

The switch statement is an alternative to some multi-way if/else statements.

Not all multi-way if/else statements can be converted directly to a switch statement.

The expression to evaluate within the switch statement must evaluate to an integral value (int,
short, long, char, or bool).

The case labels within a switch statement must be integral literals or constants. Specifically,
case labels may not be variables or other expressions. The types int, short, long, char, and
bool qualify as permissible integral types.

Program execution jumps to the case label with the value that matches the switch expression.

Once a case label is matched, program execution continues within the switch statement until a
break statement is encountered.

If no case labels match the switch expression, the program execution jumps to the default
label, if it is present. If no case labels match and no default label is present, no part of the
switch body is executed.

The conditional operator is an expression that evaluates to one of two values depending on a given
condition.

The do/while is a bottom-checking loop, unlike the while loop, which is a top-checking loop.

The body of a do/while loop is always executed at least once, regardless of the condition. The
body of a while loop is not executed if the condition is initially false.

2015 Richard L. Halterman Draft date: January 18, 2015

7.6. EXERCISES 181

The for loop is a top-checking loop that, when used properly, concentrates all the information about
its control in one convenient location.
The three parts of the for loop control are initialization, condition, and modification.
The for statement is best used for a loop that can be controlled by a single variable with a definite
starting value, a definite ending value, and a regular way to update the variables value.
A for loop is ideal for counting.
Any or all of the three parts in the for loop header can be omitted; however, if you feel the need to
omit one or more of the parts, the while statement may be a better choice.
Best practice avoids modifying the control variable of a for statement within the loops body; the
modification should be limited to the third part of the for header.
The <iomanip> library provides the setw object that provides special formatting for output sent
to cout.

7.6 Exercises

1. Consider the following code fragment.

int x;
cin >> x;
switch (x + 3) {
case 5:
cout << x << endl;
break;
case 10:
cout << x - 3 << endl;
break;
case 20:
cout << x + 3 << endl;
break;
}

(a) What is printed when the user enters 2?

(b) What is printed when the user enters 5?
(c) What is printed when the user enters 7?
(d) What is printed when the user enters 17?
(e) What is printed when the user enters 20?

2. Consider the following code fragment.

char ch;
cin >> ch;
switch (ch) {
case 'a':
cout << "*" << endl;
break;

2015 Richard L. Halterman Draft date: January 18, 2015

7.6. EXERCISES 182

case 'A':
cout << "**" << endl;
break;
case 'B':
case 'b':
cout << "***" << endl;
case 'C':
case 'c':
cout << "****" << endl;
break;
default:
cout << "*****" << endl;
}

(a) What is printed when the user enters a?

(b) What is printed when the user enters A?
(c) What is printed when the user enters b?
(d) What is printed when the user enters B?
(e) What is printed when the user enters C?
(f) What is printed when the user enters c?
(g) What is printed when the user enters t?

3. What is printed by the following code fragment?

int x = 0;
do {
cout << x << " ";
x++;
} while (x < 10);
cout << endl;

4. What is printed by the following code fragment?

int x = 20;
do {
cout << x << " ";
x++;
} while (x < 10);
cout << endl;

5. What is printed by the following code fragment?

for (int x = 0; x < 10; x++)
cout << "*";
cout << endl;

6. Rewrite the following code fragment so that a switch is used instead of the if/else statements.

2015 Richard L. Halterman Draft date: January 18, 2015

7.6. EXERCISES 183

int value;
char ch;
cin >> ch;
if (ch == 'A')
value = 10;
else if (ch == 'P')
value = 20;
else if (ch == 'T')
value = 30;
else if (ch == 'V')
value = 40;
else
value = 50;
cout << value << endl;

7. Rewrite the following code fragment so that a multi-way if/else is used instead of the switch
statement.
int value;
char ch;
cin >> ch;
switch( ch) {
case 'A':
value = 10;
break;
case 'P':
cin >> value;
break;
case 'T':
value = ch;
break;
case 'V':
value = ch + 1000;
break;
default:
value = 50;
}
cout << value << endl;

8. Rewrite the following code fragment so that a multi-way if/else is used instead of the switch
statement.
int value;
char ch;
cin >> ch;
switch (ch) {
case 'A':
cout << ch << endl;
value = 10;
break;
case 'P':

2015 Richard L. Halterman Draft date: January 18, 2015

7.6. EXERCISES 184

case 'E':
cin >> value;
break;
case 'T':
cin >> ch;
value = ch;
case 'C':
value = ch;
cout << "value=" << value << ", ch=" << ch << endl;
break;
case 'V':
value = ch + 1000;
break;
}
cout << value << endl;

9. Rewrite the following code fragment so a while loop is used instead of the for statement.
for (int i = 100; i > 0; i--)
cout << i << endl;

10. Rewrite the following code fragment so that it uses the conditional operator instead of an if state-
ment:
if (value % 2 != 0) // Is value even?
value = value + 1; // If not, make it even.

11. Rewrite the following code fragment so that it uses the conditional operator instead of an if/else
statement:
if (value % 2 == 0) // Is value even?
value = 0; // If so, make it zero.
else
value = value + 1; // Otherwise, make it even.

12. Would the following multi-way if/else be a good candidate to rewrite as a switch statement?
If so, rewrite the code using a switch; otherwise, explain why it is impractical to do so.
int x, y;
cin >> x >> y;
if (x < 10)
y = 10;
else if (x == 5)
y = 5;
else if (x == y)
y = 0;
else if (y > 10)
x = 10;
else
x = y;

2015 Richard L. Halterman Draft date: January 18, 2015

 185

Chapter 8

Using Functions

Suppose you must write a C++ program that computes the square root of a number supplied by the user.
Listing 8.1 (computesquareroot.cpp) provides a simple implementation.

Listing 8.1: computesquareroot.cpp

// File squareroot.cpp

#include <iostream>

using namespace std;

int main() {
double input;

// Get value from the user

cout << "Enter number: ";
cin >> input;
double diff;
// Compute a provisional square root
double root = 1.0;

do { // Loop until the provisional root

// is close enough to the actual root
root = (root + input/root) / 2.0;
cout << "root is " << root << endl;
// How bad is the approximation?
diff = root * root - input;
}
while (diff > 0.0001 || diff < -0.0001);

// Report approximate square root

cout << "Square root of " << input << " = " << root << endl;
}

The program is based on a simple algorithm, Newtons Method, that uses successive approximations to
zero in on an answer that is within 0.0001 of the true answer.

2015 Richard L. Halterman Draft date: January 18, 2015

 186

One sample run is

Enter number: 2
root is 1.5
root is 1.41667
root is 1.41422
Square root of 2 = 1.41422

The actual square root is approximately 1.4142135623730951 and so the result is within our accepted
tolerance (0.0001). Another run is

Enter number: 100

root is 50.5
root is 26.2401
root is 15.0255
root is 10.8404
root is 10.0326
root is 10.0001
root is 10
Square root of 100 = 10

which is, of course, the exact answer.

While this code may be acceptable for many applications, better algorithms exist that work faster and
produce more precise answers. Another problem with the code is this: What if you are working on a
significant scientific or engineering application and must use different formulas in various parts of the
source code, and each of these formulas involve square roots in some way? In mathematics, for example,
you use square root to compute the distance between two geometric points (x1 , y1 ) and (x2 , y2 ) as
q
(x2 x1 )2 + (y2 y1 )2

and, using the quadratic formula, the solution to the equation ax2 + bx + c = 0 is

b b2 4ac
2a
In electrical engineering and physics, the root mean square of a set of values {a1 , a2 , a3 , . . . , an } is
s
a21 + a22 + a23 + . . . + a2n
n
Suppose we are writing one big program that, among many other things, needs to compute distances and
solve quadratic equations. Must we copy and paste the relevant portions of the square root code in List-
ing 8.1 (computesquareroot.cpp) to each location in our source code that requires a square root compu-
tation? Also, what if we develop another program that requires computing a root mean square? Will we
need to copy the code from Listing 8.1 (computesquareroot.cpp) into every program that needs to compute
square roots, or is there a better way to package the square root code and reuse it?

2015 Richard L. Halterman Draft date: January 18, 2015

8.1. INTRODUCTION TO USING FUNCTIONS 187

Code is made reusable by packaging it in functions. A function is a unit of reusable code. In Chapter 9
we will write our own reusable functions, but in this chapter we examine some of the functions available
in the C++ standard library. C++ provides a collection of standard precompiled C and C++ code stored
in libraries. Programmers can use parts of this library code within their own code to build sophisticated
programs.

8.1 Introduction to Using Functions

In mathematics, a function computes a result from a given value; for example, from the function definition
f (x) = 2x + 3, we can compute f (5) = 13 and f (0) = 3. A function in C++ works like a mathematical
function. To introduce the function concept, we will look at the standard C++ function that implements
mathematical square root.
In C++, a function is a named sequence of code that performs a specific task. A program itself consists
of a collection of functions. One example of a function is the mathematical square root function. Such
a function, named sqrt, is available to C and C++ programs (see Section 8.2). The square root function
accepts one numeric value and produces a double value as a result; for example, the square root of 16 is 4,
so when presented with 16.0, sqrt responds with 4.0. Figure 8.1 visualizes the square root function. For

Figure 8.1: Conceptual view of the square root function

the programmer using the sqrt function within a program, the function is a black box; the programmer is
concerned more about what the function does, not how it does it.
This sqrt function is exactly what we need for our square root program, Listing 8.1 (computesquareroot.cpp).
The new version, Listing 8.2 (standardsquareroot.cpp), uses the library function sqrt and eliminates the
complex logic of the original code.

Listing 8.2: standardsquareroot.cpp

#include <iostream>
#include <cmath>

using namespace std;

int main() {
double input;

// Get value from the user

cout << "Enter number: ";
cin >> input;

2015 Richard L. Halterman Draft date: January 18, 2015

8.1. INTRODUCTION TO USING FUNCTIONS 188

// Compute the square root

double root = sqrt(input);

// Report result
cout << "Square root of " << input << " = " << root << endl;
}

The line
#include <cmath>

directs the preprocessor to augment our source code with the declarations of a collection of mathematical
functions in the cmath library. The sqrt function is among them. Table 8.1 lists some of the other
commonly used mathematical functions available in the cmath library. The compiler needs this augmented
code so it can check to see if we are using the sqrt function properly.
The expression
sqrt(input)

is a function invocation, also known as a function call. A function provides a service to the code that uses
it. Here, our main function is the caller1 that uses the service provided by the sqrt function. We say
main calls, or invokes, sqrt passing it the value of input. The expression sqrt(input) evaluates to
the square root of the value of the variable input. Behind the scenesinside the black box as it were
precompiled C code uses the value of the input variable to compute its square root. There is nothing
special about this precompiled C code that constitutes the sqrt function; it was written by a programmer
or team of programmers working for the library vendor using the same tools we have at our disposal. In
In Chapter 9 we will write our own functions, but for now we will enjoy the functions that others have
provided for us.
When calling a function, a pair of parentheses follow the functions name. Information that the function
requires to perform its task must appear within these parentheses. In the expression
sqrt(input)

input is the information the function needs to do its work. We say input is the argument, or parameter,
passed to the function. We also can say we are passing input to the sqrt function.
While we might say we are passing input to the sqrt function, the program really is not giving the
function access to mains input variable. The sqrt function itself cannot change the value of mains
input variable, it simply uses the variables value to perform the computation.
The following simple analogy may help explain how the communication works between main and
sqrt, The main function has work to do, but instead of doing all the work itself, it delegates some of
the work (in this case the hard part) to sqrt. When main needs to compute the square root of input,
it writes down the value of its input variable on a piece of paper and hands it to sqrt. main then
sits idly until sqrt finishes its work. The sqrt function accepts mains note and begins working on
the task (computing the square root of the number on the note main gave it). When it is finished, sqrt
does two things: sqrt hands back to main a different piece of paper with the answer, and sqrt throws
away the piece of paper main originally passed to it. When main receives the note from sqrt it uses the
information on the note and then discards the note. The main function then can continue with its other
business.
1 The term client can be used as well, although we reserve the term client for code that interacts with objects (see Chapter 13).

2015 Richard L. Halterman Draft date: January 18, 2015

8.1. INTRODUCTION TO USING FUNCTIONS 189

The sqrt function thus has no access to mains original input variable; it has only a copy of input,
as if written on a piece of paper. (Similarly, if the sqrt function uses any variables to do its work, main
is oblivious to them and has no way to access them.) After sqrt is finished and returns to main its
computed answer, sqrt discards its copy of input (by analogy, the function throws away the paper
with the copy of input that main gave it). Thus, during a function call the parameter is a temporary,
transitory value used only to communicate information to the function. The parameter lives only as long as
the function is executing.
Figure 8.2 illustrates a programs execution involving simple function calls. Figure 8.2 shows that a

int main() {
double value; Program Execution

// Assign variable Time

value = 16;
main
// Compute s square root
double root = sqrt(value); 16 100
4 10
// Compute another
root = sqrt(100); sqrt sqrt
}

Figure 8.2: The diagram on the right visualizes the execution of the program on the left. Time flows from left to right. A
rectangular bar represents the time that a function is active. A C++ programs execution begins with its main function.
Here, main calls the sqrt function twice. The shaded parts of mains bar shows the times main has to wait for
sqrt to complete.

programs execution begins in its main function. Here main calls the sqrt function twice. A vertical
bar represents the time that a function is active, or alive. A functions variables exist while a function is
active. Observe that the main function is active for the duration of the programs execution. The sqrt
becomes active twice, exactly the two times main calls it.
The sqrt function can be called in other ways, as Listing 8.3 (usingsqrt.cpp) illustrates:

Listing 8.3: usingsqrt.cpp

/*
* This program shows the various ways the
* sqrt function can be used.
*/

#include <iostream>
#include <cmath>

using namespace std;

int main() {
double x = 16.0;
// Pass a literal value and display the result
cout << sqrt(16.0) << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

8.1. INTRODUCTION TO USING FUNCTIONS 190

// Pass a variable and display the result

cout << sqrt(x) << endl;
// Pass an expression
cout << sqrt(2 * x - 5) << endl;
// Assign result to variable
double y = sqrt(x);
// Use result in an expression
y = 2 * sqrt(x + 16) - 4;
// Use result as argument to a function call
y = sqrt(sqrt(256.0));
cout << y << endl;
}

The sqrt function accepts a single numeric argument. The parameter that a caller can pass to sqrt
can be a literal number, a numeric variable, an arithmetic expression, or even a function invocation that
produces an acceptable numeric result.
Some C++ functions, like sqrt, compute a value and return it to the caller. The caller can use this result
in various ways, as shown in Listing 8.3 (usingsqrt.cpp
p ). The next to the last statement passes the result of
calling sqrt to sqrt, thereby computing 256, which is 4.
If the caller code attempts to pass a parameter to the function that is incompatible with the type expected
by the function, the compiler will issue an error.
cout << sqrt("16") << endl; // Illegal, a string is not a number

The compiler is able to determine that the above statement is illegal based on the additional information the
preprocessor added to the code via the
#include<cmath>

directive.
Listing 8.3 (usingsqrt.cpp) shows that a program can call the sqrt function as many times and in as
many places as needed. As noted in Figure 8.1, to the caller of the square root function, the function is a
black box; the caller is concerned strictly about what the function does, not how the function accomplishes
its task.
We safely can treat all functions as black boxes. We can use the service that a function provides without
being concerned about its internal details. We are guaranteed that we can influence the functions behavior
only via the parameters that we pass, and that nothing else we do can affect what the function does or how
it does it. Furthermore, the function cannot affect any of our code, apart from what we do with the value it
computes.
Some functions take more than one parameter; for example, the C++ max function requires two argu-
ments in order to produce a result. The max function selects and returns the larger of the two parameters.
The max function is visualized in Figure 8.3. The max function could be used as
cout << "The larger of " << 4 << " and " << 7
<< " is " << max(4, 7) << endl;

Notice that the parameters are contained in parentheses following the functions name, and the parameters
are separated by commas.
From the callers perspective a function has three important parts:

2015 Richard L. Halterman Draft date: January 18, 2015

8.1. INTRODUCTION TO USING FUNCTIONS 191

Figure 8.3: Conceptual view of the maximum function

Name. Every function has a name that identifies the location of the code to be executed. Function
names follow the same rules as variable names; a function name is another example of an identifier
(see Section 3.3).

Parameter type(s). A caller must provide the exact number and types of parameters that a function
expects. If a caller attempts to call a function with too many or too few parameters, the compiler will
issue an error message and not compile the code. Similarly, if the caller passes parameters that are
not compatible with the types specified for the function, the compiler will report appropriate error
messages.

Result type. A function can compute a result and return this value to the caller. The callers use of
this result must be compatible with the functions specified result type. The result type returned to
the caller and the parameter types passed in by the caller can be completely unrelated.

These three crucial pieces of information are formally described for each function in a specification
known as a function prototype. The prototype for the sqrt function is
double sqrt(double)

and the max functions prototype can be expressed as 2

int max(int, int)

In a function prototype, the return type is listed first, followed by the functions name, and then the param-
eter types appear in parentheses. Sometimes it is useful to list parameter names in the functions prototype,
as in
double sqrt(double n)

or
int max(int a, int b)

The specific parameter names are irrelevant. The names make it easier to describe what the function does;
for example, sqrtcomputes the square root of n and max determines the larger of a and b.
When using a library function the programmer must include the appropriate #include directive in
the source code. The file specified in an #include directive contains prototypes for library functions. In
order to use the sqrt function, a program must include the
2 The prototype for the actual library function max uses generic types; generic types are beyond the scope of this introductory

book, so the prototype provided here is strictly for illustrating a point.

2015 Richard L. Halterman Draft date: January 18, 2015

8.2. STANDARD MATH FUNCTIONS 192

#include <cmath>

preprocessor directive. For the max function, include

#include <algorithm>

although under Visual C++, includeing <iostream> is sufficient.

Armed with the function prototypes, the compiler can check to see if the calling code is using the library
functions correctly. An attempt to use the sqrt function as

cout << sqrt(4.0, 7.0) << endl; // Error

will result in an error because the prototype for sqrt specifies only one numeric parameter, not two.
Some functions do not accept parameters; for example, the C++ function to generate a pseudorandom
number, rand, is called with no arguments:

cout << rand() << endl;

The rand function returns an int value, but the caller does not pass the function any information to do its
task. The rand prototype is

int rand()

Notice the empty parentheses that indicate this function does not accept any parameters.
Unlike mathematical functions that must produce a result, C++ does not require a function to return a
value to its caller. The C++ function exit expects an integer value from the caller, but it does not return a
result back to the caller. A prototype for a function that returns nothing uses void as the return type, as in:

void exit(int);

The exit function immediately terminates the programs execution. The integer argument passed to exit
is returned to the operating system which can use the value to determine if the program terminated normally
or due to an error. C++ programs automatically return zero when main finishes executingno exit call
is necessary.
Note that since exit does not return a value to the caller, code such as

cout << exit(8) << endl; // Illegal!

will not compile since the expression exit(8) evaluates to nothing, and the cout stream object requires
an actual value of some kind to print. A void function is useful for the side effects it produces instead
a value it computes. Example side effects include printing something on the console, sending data over a
network, or animating a graphical image.

8.2 Standard Math Functions

The cmath library provides much of the functionality of a scientific calculator. Table 8.1 lists only a few
of the available functions.

2015 Richard L. Halterman Draft date: January 18, 2015

8.2. STANDARD MATH FUNCTIONS 193

mathfunctions Module
double sqrt(double x)
Computes the square root of a number: sqrt(x) = x
double exp(double x)
Computes e raised a power: exp(x) = ex
double log(double x)
Computes the natural logarithm of a number: log(x) = loge x = ln x
double log10(double x)
Computes the common logarithm of a number: log(x) = log10 x
double cos(double)
Computes the cosine of a value specified in radians: cos(x) = cos x; other trigonometric
functions include sine, tangent, arc cosine, arc sine, arc tangent, hyperbolic cosine, hyper-
bolic sine, and hyperbolic tangent
double pow(double x, double y)
Raises one number to a power of another: pow(x, y) = xy
double fabs(double x)
Computes the absolute value of a number: fabs(x) = |x|

Table 8.1: A few of the functions from the cmath library

The cmath library also defines a constant named HUGE_VAL. Programmers can use this constant to
represent infinity or an undefined value such the slope of a vertical line or a fraction with a zero denominator.
A complete list of the numeric functions available to C++ can be found at http://www.cplusplus.com/
reference/clibrary/cmath/.

Be careful to put the functions arguments in the proper order when calling a
function; for example, the call pow(10,2) computes 102 = 100, but the call
pow(2,10) computes 210 = 1, 024.

A C++ program that uses any of the functions from the cmath library must use the following prepro-
cessor #include directive:
#include <cmath>

Functions in the cmath library are ideal for solving problems like the one shown in Figure 8.4. Suppose
a spacecraft is at a fixed location in space relative to some planet. The spacecrafts distance to the planet,
therefore, also is fixed. A satellite is orbiting the planet in a circular orbit. We wish to compute how much
farther away the satellite will be from the spacecraft when it has progressed 10 degrees along its orbital
path.
We will let the origin of our coordinate system (0,0) be located at the center of the planet which cor-
responds also to the center of the circular orbital path. The satellite is initially at point (x1 , y1 ) and the
spacecraft is stationary at point (px , py ). The spacecraft is located in the same plane as the satellites orbit.
We need to compute the difference in the distances between the moving point (satellite) and the fixed point
(spacecraft) at two different times during the satellites orbit.
Facts from mathematics provide solutions to the following two problems:

2015 Richard L. Halterman Draft date: January 18, 2015

8.2. STANDARD MATH FUNCTIONS 194

(x2 ,y2 )

d2

(x1 ,y1 ) d1
(0,0) (px ,py )

Figure 8.4: Orbital distance problem

1. Problem: We must recompute the location of the moving point as it moves along the circle.
Solution: Given an initial position (x1 , y1 ) of the moving point, a rotation of degrees around the
origin will yield a new point at (x2 , y2 ), where

x2 = x1 cos y1 sin
y2 = x1 sin + y1 cos

2. Problem: The distance between the moving point and the fixed point must be recalculated as the
moving point moves to a new position.
Solution: The distance d1 in Figure 8.4 between two points (px , py ) and (x1 , y1 ) is given by the
formula q
d1 = (x1 px )2 + (y1 py )2

Similarly, the distance d2 in Figure 8.4 is

q
d2 = (x2 px )2 + (y2 py )2

Listing 8.4 (orbitdist.cpp) uses these mathematical results to compute the difference in the distances.

Listing 8.4: orbitdist.cpp

#include <iostream>
#include <cmath>

using namespace std;

int main() {
// Location of orbiting point is (x,y)
double x; // These values change as the
double y; // satellite moves
const double PI = 3.14159;

// Location of fixed point is always (100, 0),

// AKA (p_x, p_y). Change these as necessary.
const double p_x = 100;

2015 Richard L. Halterman Draft date: January 18, 2015

8.2. STANDARD MATH FUNCTIONS 195

const double p_y = 0;

// Radians in 10 degrees
const double radians = 10 * PI/180;

// Precompute the cosine and sine of 10 degrees

const double COS10 = cos(radians);
const double SIN10 = sin(radians);

// Get starting point from user

cout << "Enter initial satellite coordinates (x,y):";
cin >> x >> y;

// Compute the initial distance

double d1 = sqrt((p_x - x)*(p_x - x) + (p_y - y)*(p_y - y));

// Let the satellite orbit 10 degrees

double x_old = x; // Remember x's original value
x = x*COS10 - y*SIN10; // Compute new x value
// x's value has changed, but y's calculate depends on
// x's original value, so use x_old instead of x.
y = x_old*SIN10 + y*COS10;

// Compute the new distance

double d2 = sqrt((p_x - x)*(p_x - x) + (p_y - y)*(p_y - y));

// Print the difference in the distances

cout << "Difference in distances: " << d2 - d1 << endl;
}

We can use the square root function to improve the efficiency of our primes program. Instead of trying
all the factors of n up to n 1, we need only try potential factors up to the square root of n. Listing 8.5
(moreefficientprimes.cpp) uses the sqrt function to reduce the number of factors that need be considered.

Listing 8.5: moreefficientprimes.cpp

#include <iostream>
#include <cmath>

using namespace std;

int main() {

int max_value;
cout << "Display primes up to what value? ";
cin >> max_value;
for (int value = 2; value <= max_value; value++) {
// See if value is prime
bool is_prime = true; // Provisionally, value is prime
double r = value, root = sqrt(r);
// Try all possible factors from 2 to the square
// root of value
for (int trial_factor = 2;
is_prime && trial_factor <= root; trial_factor++)
is_prime = (value % trial_factor != 0);

2015 Richard L. Halterman Draft date: January 18, 2015

8.3. MAXIMUM AND MINIMUM 196

if (is_prime)
cout << value << " "; // Display the prime number
}
cout << endl; // Move cursor down to next line
}

The sqrt function comes in three forms:

double sqrt(double)
float sqrt(float)
long double sqrt(long double)

The function names are the same, but the parameter types differ. We say that the sqrt function is over-
loaded. (Overloaded functions are covered in more detail in Section 10.3.) When a caller invokes the sqrt
function, the compiler matches the call to the closest matching prototype. If the caller passes a double
parameter, the compiler generates code to call the double version. If the caller instead passes a float
variable, the compiler selects the float version of sqrt. When an int is passed to sqrt, the compiler
cannot decide which version to use, because an int can be converted automatically to either a float,
double, or long double. The compiler thus needs some help to resolve the ambiguity, so we intro-
duced an additional variable of type double so the compiler will use the double version of the sqrt
function. Another option is to use a type cast to convert the integer value into one of the types acceptable
to the sqrt function.

8.3 Maximum and Minimum

C++ provides standard functions for determining the maximum and minimum of two numbers. Listing 8.6
(maxmin.cpp) exercises the standard min and max functions.

Listing 8.6: maxmin.cpp

#include <iostream>
#include <algorithm>

using namespace std;

int main() {
int value1, value2;
cout << "Please enter two integer values: ";
cin >> value1 >> value2;
cout << "max = " << max(value1, value2)
<< ", min = " << min(value1, value2) << endl;
}

To use the standard max and min functions in in program you must include the <algorithm> header.

8.4 clock Function

The clock function from the <ctime> library requests from the operating system the amount of time an
executing program has been running. The units returned by the call clock() is system dependent, but it

2015 Richard L. Halterman Draft date: January 18, 2015

8.4. CLOCK FUNCTION 197

can be converted into seconds with the constant CLOCKS_PER_SEC, also defined in the ctime library.
Under Visual C++, the CLOCKS_PER_SEC constant is 1,000, which means the call clock() returns the
number of milliseconds that the program has been running.
Using two calls to the clock function you can measure elapsed time. Listing 8.7 (timeit.cpp) measures
how long it takes a user to enter a character from the keyboard.

Listing 8.7: timeit.cpp

#include <iostream>
#include <ctime>

using namespace std;

int main() {
char letter;
cout << "Enter a character: ";
clock_t seconds = clock(); // Record starting time
cin >> letter;
clock_t other = clock(); // Record ending time
cout << static_cast<double>(other - seconds)/CLOCKS_PER_SEC
<< " seconds" << endl;
}

The type clock_t is a type defined in the <ctime> header file. clock_t is equivalent to an
unsigned long, and you can perform arithmetic on clock_t values and variables just as if they are
unsigned longs. In the expression
static_cast<double>(other - seconds)/CLOCKS_PER_SEC

the cast is required to force floating-point division; otherwise, the result is truncated to an integer value.
Listing 8.8 (measureprimespeed.cpp) measures how long it takes a program to display all the prime
numbers up to half a million using the algorithm from Listing 7.7 (forprintprimes.cpp).

Listing 8.8: measureprimespeed.cpp

#include <iostream>
#include <ctime>
#include <cmath>

using namespace std;

// Display the prime numbers between 2 and 500,000 and

// time how long it takes

int main() {
clock_t start_time = clock(), // Record start time
end_time;
for (int value = 2; value <= 500000; value++) {
// See if value is prime
bool is_prime = true; // Provisionally, value is prime
// Try all possible factors from 2 to n - 1
for (int trial_factor = 2;
is_prime && trial_factor < value;
trial_factor++)

2015 Richard L. Halterman Draft date: January 18, 2015

8.5. CHARACTER FUNCTIONS 198

is_prime = (value % trial_factor != 0);

if (is_prime)
cout << value << " "; // Display the prime number
}
cout << endl; // Move cursor down to next line
end_time = clock();
// Print the elapsed time
cout << "Elapsed time: "
<< static_cast<double>(end_time - start_time)/CLOCKS_PER_SEC
<< " sec." << endl;
}

On one system, the program took 93 seconds, on average, to print all the prime numbers up to 500,000.
By comparison, the newer, more efficient version, Listing 8.5 (moreefficientprimes.cpp), which uses the
square root optimization takes only 15 seconds to display all the primes up to 500,000. Exact times will
vary depending on the speed of the computer.
As it turns out, much of the programs execution time is taken up printing the output, not computing the
prime numbers to print. We can compare the algorithms better by redirecting the programs output to a file.
If the executable program is named primes.exe, you can redirect its output at the command line by issuing
the command

primes > run1.out

This creates a text file named run1.out that can be viewed with any text editor. Its contents are exactly what
would have been printed to the screen if the redirection is not used.
When run using redirection, the time difference is even more dramatic: The unoptimized version gen-
erates the prime numbers up to 500,000 in 77 seconds, while the optimized square root version requires
only 2 seconds to generate the same number of primes! An even faster prime generator can be found in
Listing 11.12 (fasterprimes.cpp); it uses a completely different algorithm to generate prime numbers.
The ctime header must be #included to use the standard time function in a program.

8.5 Character Functions

The C library provides a number of character functions that are useful to C++ programmers. Listing 8.9
(touppercase.cpp) converts lowercase letters to uppercase letters.

Listing 8.9: touppercase.cpp

#include <iostream>
#include <cctype>

using namespace std;

int main() {
for (char lower = 'a'; lower <= 'z'; lower++) {
char upper = toupper(lower);
cout << lower << " => " << upper << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

8.5. CHARACTER FUNCTIONS 199

}
}

The first lines printed by Listing 8.9 (touppercase.cpp) are

a => A
b => B
c => C
d => D

Interestingly, the toupper function returns an int, not a char. At the enhanced warning level 4 for
Visual C++ a cast is required to assign the result to the variable upper:

char upper = static_cast<char>(toupper(lower));

Some of the more useful character functions are described in Table 8.2.

charfunctions Module
int toupper(int ch)
Returns the uppercase version of the given character; returns the original character if no
uppercase version exists (such as for punctuation or digits)
int tolower(int ch)
Returns the lowercase version of the given character; returns the original character if no
lowercase version exists (such as for punctuation or digits)
int isupper(int ch)
Returns a nonzero value (true) if ch is an uppercase letter (AZ); otherwise, it returns
0 (false)
int islower(int ch)
Returns a nonzero value (true) if ch is an lowercase letter (az); otherwise, it returns
0 (false)
int isalpha(int ch)
Returns a nonzero value (true) if ch is a letter from the alphabet (AZ or az);
otherwise, it returns 0 (false)
int isdigit(int ch)
Returns a nonzero value (true) if ch is a digit (09); otherwise, it returns 0 (false)

Table 8.2: A few of the functions from the cctype library

Other functions exist to determine if a character is a punctuation character like a comma or semicolon
(ispunct), a space, tab, or newline character (isspace).
To use the standard C character functions in your C++ program, you must include the <cctype> header
file.

2015 Richard L. Halterman Draft date: January 18, 2015

8.6. RANDOM NUMBERS 200

8.6 Random Numbers

Some applications require behavior that appears random. Random numbers are useful particularly in games
and simulations. For example, many board games use a die (one of a pair of dice) to determine how many
places a player is to advance. (See Figure 8.5.) A die or pair of dice are used in other games of chance. A
die is a cube containing spots on each of its six faces. The number of spots range from one to six. A player
rolls a die or sometimes a pair of dice, and the side(s) that face up have meaning in the game being played.
The value of a face after a roll is determined at random by the complex tumbling of the die. A software
adaptation of a game that involves dice would need a way to simulate the random roll of a die.

Figure 8.5: A pair of dice

All algorithmic random number generators actually produce pseudorandom numbers, not true random
numbers. A pseudorandom number generator has a particular period, based on the nature of the algorithm
used. If the generator is used long enough, the pattern of numbers produced repeats itself exactly. A
sequence of true random numbers would not contain such a repeating subsequence. The good news is
that all practical algorithmic pseudorandom number generators have periods that are large enough for most
applications.
C++ programmers can use two standard C functions for generating pseudorandom numbers: srand and
rand:
void srand(unsigned)
int rand()

srand establishes the first value in the sequence of pseudorandom integer values. Each call to rand
returns the next value in the sequence of pseudorandom values. Listing 8.10 (simplerandom.cpp) shows
how a sequence of 100 pseudorandom numbers can be printed.

Listing 8.10: simplerandom.cpp

#include <iostream>
#include <cstdlib>

using namespace std;

int main() {
srand(23);
for (int i = 0; i < 100; i++) {
int r = rand();

2015 Richard L. Halterman Draft date: January 18, 2015

8.6. RANDOM NUMBERS 201

cout << r << " ";

}
cout << endl;
}

The numbers printed by the program appear to be random. The algorithm is given a seed value to begin,
and a formula is used to produce the next value. The seed value determines the sequence of numbers gener-
ated; identical seed values generate identical sequences. If you run the program again, the same sequence is
displayed because the same seed value, 23, is used. In order to allow each program run to display different
sequences, the seed value must be different for each run. How can we establish a different seed value for
each run? The best way to make up a random seed at run time is to use the time function which is found
in the ctime library. The call time(0) returns the number of seconds since midnight January 1, 1970.
This value obviously differs between program runs, so each execution will use a different seed value, and the
generated pseudorandom number sequences will be different. Listing 8.11 (betterrandom.cpp) incorporates
the time function to improve its randomness over multiple executions.

Listing 8.11: betterrandom.cpp

#include <iostream>
#include <cstdlib>
#include <ctime>

using namespace std;

int main() {
srand(static_cast<unsigned>(time(0)));
for (int i = 0; i < 100; i++) {
int r = rand();
cout << r << " ";

}
cout << endl;
}

Each execution of Listing 8.11 (betterrandom.cpp) produces a different pseudorandom number se-
quence. The actual type of value that time returns is time_t, so the result from a call to time must
be cast to unsigned int before being used with srand.
Notice that the numbers returned by rand can be rather large. The pseudorandom values range from 0
to a maximum value that is implementation dependent. The maximum value for Visual C++s rand function
is 32,767, which corresponds to the largest 16-bit signed int value. The cstdlib header defines the
constant RAND_MAX that represents the largest value in the range. The following statement
cout << RAND_MAX << endl;

would print the value of RAND_MAX for a particular system.

Ordinarily we need values in a more limited range, like 1...100. Simple arithmetic with the modulus
operator can produce the result we need. If n is any non-negative integer and m is any positive integer, the
expression

n%m

2015 Richard L. Halterman Draft date: January 18, 2015

8.6. RANDOM NUMBERS 202

produces a value in the range 0 . . . m 1.

This means the statement
int r = rand() % 100;

can assign only values in the range 0...99 to r. If we really want values in the range 1...100, what can we
do? We simply need only add one to the result:
int r = rand() % 100 + 1;

This statement produces pseudorandom numbers in the range 1...100.

We now have all we need to write a program that simulates the rolling of a die.
Listing 8.12 (die.cpp) simulates rolling die.

Listing 8.12: die.cpp

#include <iostream>
#include <cstdlib>
#include <ctime>

using namespace std;

int main() {
// Set the random seed value
srand(static_cast<unsigned>(time(0)));

// Roll the die three times

for (int i = 0; i < 3; i++) {
// Generate random number in the range 1...6
int value = rand() % 6 + 1;

// Show the die

cout << "+-------+" << endl;
switch (value) {
case 1:
cout << "| |" << endl;
cout << "| * |" << endl;
cout << "| |" << endl;
break;
case 2:
cout << "| * |" << endl;
cout << "| |" << endl;
cout << "| * |" << endl;
break;
case 3:
cout << "| * |" << endl;
cout << "| * |" << endl;
cout << "| * |" << endl;
break;
case 4:
cout << "| * * |" << endl;
cout << "| |" << endl;
cout << "| * * |" << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

8.7. SUMMARY 203

break;
case 5:
cout << "| * * |" << endl;
cout << "| * |" << endl;
cout << "| * * |" << endl;
break;
case 6:
cout << "| * * * |" << endl;
cout << "| |" << endl;
cout << "| * * * |" << endl;
break;
default:
cout << " *** Error: illegal die value ***" << endl;
break;
}
cout << "+-------+" << endl;
}
}

The output of one run of Listing 8.12 (die.cpp) is

+-------+
| * * |
| |
| * * |
+-------+
+-------+
| * * * |
| |
| * * * |
+-------+
+-------+
| |
| * |
| |
+-------+

Since the values are pseudorandomly generated, actual output will vary from one run to the next.

8.7 Summary

The C and C++ standard library provides a collection of routines that can be incorporated into code
that you write.
When faced with the choice of using a standard library function or writing your own code to solve
the same problem, choose the library function. The standard function will be tested thoroughly, well
documented, and likely more efficient than the code you would write.
The function is a standard unit of reuse in C++.

2015 Richard L. Halterman Draft date: January 18, 2015

8.8. EXERCISES 204

Code that uses a function is known as calling code.

A function has a name, a list of parameters (which may be empty), and a result type (which may be
void). A function performs some computation or action that is useful to callers. Typically a function
produces a result based on the parameters passed to it.

Callers communicate information to a function via its parameters (also known as arguments).

In order to use a standard function, callers must #include the proper header file.

The arguments passed to a function by a caller consist of a comma-separated list enclosed by paren-
theses.

Calling code (code outside the functions definition that invoke the function) must pass the correct
number and types of parameters that the function expects.

The C standard library (<cmath>) includes a variety of mathematical functions.

The time function from the <ctime> library returns the number of seconds since midnight, January
1, 1970.

The C standard library (<cctype>) includes a variety of character manipulation functions.

The rand function (from <cstdlib>) returns a pseudorandom integer. The first pseudorandom
number can be set with the srand function. The constant RAND_MAX is largest possible value
returned by rand.

8.8 Exercises

1. Suppose you need to compute the square root of a number in a C++ program. Would it be a good idea
to write the code to perform the square root calculation? Why or why not?

2. In C++ source code what is one way to help you distinguish a variable name from a function name?

3. Which one of the following values could be computed by the rand function?
4.5 34 -1 @RAND_MAX@ + 1

4. What does clock_t represent?

5. What does CLOCKS_PER_SEC represent?

6. Ordinarily how often should a program call the srand function?

7. In Listing 8.2 (standardsquareroot.cpp), what does the main function do while the sqrt function
is computing the square root of the argument that main provides?

8. Consider each of the following code fragments below that could be part of a C++ program. Each
fragment contains a call to a standard C/C++ library function. Answer each question in one of the
following three ways:

If the code fragment contains a compile-time error, write the word error for the answer.
If the code fragment contains no compile-time errors and you can determine its output at
compile-time, provide the fragments literal output.

2015 Richard L. Halterman Draft date: January 18, 2015

8.8. EXERCISES 205

If the code fragment contains no compile-time errors but you cannot determine its exact output
at compile-time, provide one possible evaluation and write the word example for the answer and
provide one possible literal output that the code fragment could produce.

(a) cout << sqrt(4.5) << endl;

(b) cout << sqrt(4.5, 3.1) << endl;

(c) cout << rand(4) << endl;

(d) double d = 16.0;

cout << sqrt(d) << endl;

(e) cout << srand() << endl;

(f) cout << rand() << endl;

(g) int i = 16;

cout << sqrt(i) << endl;

(h) cout << srand(55) << endl;

(i) cout << tolower('A') << endl;

(j) cout << exp() << endl;

(k) cout << sqrt() << endl;

(l) cout << toupper('E') << endl;

(m) cout << toupper('e') << endl;

(n) cout << toupper("e") << endl;

(o) cout << exp(4.5) << endl;

(p) cout << toupper('h', 5) << endl;

(q) cout << ispunct('!') << endl;

(r) cout << tolower("F") << endl;

(s) char ch = 'D';

cout << tolower(ch) << endl;

(t) cout << exp(4.5, 3) << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

8.8. EXERCISES 206

(u) cout << toupper('7') << endl;

(v) double a = 5, b = 3;
cout << exp(a, b) << endl;

(w) cout << exp(3, 5, 2) << endl;

(x) cout << tolower(70) << endl;

(y) double a = 5;
cout << exp(a, 3) << endl;

(z) double a = 5;
cout << exp(3, a) << endl;

9. From geometry: Write a computer program that given the lengths of the two sides of a right triangle
adjacent to the right angle computes the length of the hypotenuse of the triangle. (See Figure 8.6.)
If you are unsure how to solve the problem mathematically, do a web search for the Pythagorean
theorem.

Hy
po
ten
Side 2

us
e

Side 1

Figure 8.6: Right triangle

10. From trigonometry: Write a computer program that given the lengths of the two sides of a right
triangle adjacent to the right angle computes the length of the hypotenuse of the triangle. If you are
unsure how to solve the problem mathematically, do a web search for the Pythagorean theorem.

2015 Richard L. Halterman Draft date: January 18, 2015

 207

Chapter 9

Writing Functions

As programs become more complex, programmers must structure their programs in such a way as to ef-
fectively manage their complexity. Most humans have a difficult time keeping track of too many pieces of
information at one time. It is easy to become bogged down in the details of a complex problem. The trick
to managing complexity is to break down the problem into more manageable pieces. Each piece has its
own details that must be addressed, but these details are hidden as much as possible within that piece. The
problem is ultimately solved by putting these pieces together to form the complete solution.
So far all of our programs have been written within one functionmain. As the number of statements
within a function increases, the function can become unwieldy. The code within such a function that does
all the work by itself is called monolithic code. Monolithic code that is long and complex is undesirable for
several reasons:

It is difficult to write correctly. All the details in the entire piece of code must be considered when
writing any statement within that code.

It is difficult to debug. If the sequence of code does not work correctly, it is often difficult to find
the source of the error. The effects of an erroneous statement that appears earlier in the code may not
become apparent until a correct statement later uses the erroneous statements incorrect result.

It is difficult to extend. All the details in the entire sequence of code must be well understood before
it can be modified. If the code is complex, this may be a formidable task.

Using a divide and conquer strategy, a programmer can decompose a complicated function (like main)
into several simpler functions. The original function can then do its job by delegating the work to these
other functions. In this way the original function can be thought of as a work coordinator.
Besides their code organization aspects, functions allow us to bundle functionality into reusable parts. In
Chapter 8 we saw how library functions can dramatically increase the capabilities of our programs. While
we should capitalize on library functions as much as possible, sometimes we need a function exhibiting
custom behavior that is not provided by any standard function. Fortunately we can create our own functions,
and the same function may be used (called) in numerous places within a program. If the functions purpose
is general enough and we write the function properly, we may be able to reuse the function in other programs
as well.

2015 Richard L. Halterman Draft date: January 18, 2015

9.1. FUNCTION BASICS 208

9.1 Function Basics

Recall the handwritten square root code we saw in Listing 8.1 (computesquareroot.cpp). We know that
the better option is the standard library function sqrt; however, we will illustrate custom function devel-
opment by writing our own square root function based on the code in Listing 8.1 (computesquareroot.cpp).
In Listing 9.1 (customsquareroot.cpp) we see the definition for the square_root function.

Listing 9.1: customsquareroot.cpp

#include <iostream>
#include <iomanip>
#include <cmath>

using namespace std;

// Compute an approximation of
// the square root of x
double square_root(double x) {
double diff;
// Compute a provisional square root
double root = 1.0;

do { // Loop until the provisional root

// is close enough to the actual root
root = (root + x/root) / 2.0;
//cout << "root is " << root << endl;
// How bad is the approximation?
diff = root * root - x;
} while (diff > 0.0001 || diff < -0.0001);
return root;
}

int main() {
// Compare the two ways of computing the square root
for (double d = 1.0; d <= 10.0; d += 0.5)
cout << setw(7) << square_root(d) << " : " << sqrt(d) << endl;
}

The main function in Listing 9.1 (customsquareroot.cpp) compares the behavior of our custom square_root
function to the sqrt library function. Its output:

2015 Richard L. Halterman Draft date: January 18, 2015

9.1. FUNCTION BASICS 209

1 : 1
1.22474 : 1.22474
1.41422 : 1.41421
1.58116 : 1.58114
1.73205 : 1.73205
1.87083 : 1.87083
2 : 2
2.12132 : 2.12132
2.23607 : 2.23607
2.34521 : 2.34521
2.44949 : 2.44949
2.54952 : 2.54951
2.64577 : 2.64575
2.73861 : 2.73861
2.82843 : 2.82843
2.91548 : 2.91548
3 : 3
3.08221 : 3.08221
3.16228 : 3.16228

shows a few small differences in the results. Clearly we should use the standard sqrt function instead of
ours.
There are two aspects to every C++ function:

Function definition. The definition of a function specifies the functions return type and parameter
types, and it provides the code that determines the functions behavior. In Listing 9.1 (customsquareroot.cpp)
the definition of the square_root function appears above the main function.
Function invocation. A programmer uses a function via a function invocation. The main function
invokes both our square_root function and the sqrt function. Every function has exactly one
definition but may have many invocations.

A function definition consists of four parts:

Nameevery function in C++ has a name. The name is an identifier (see Section 3.3). As with
variable names, the name chosen for a function should accurately portray its intended purpose or
describe its functionality.
Typeevery function has a return type. If the function returns a value to its caller, its type corre-
sponds to the type of the value it returns. The special type void signifies that the function does not
return a value.
Parametersevery function must specify the types of parameters that it accepts from callers. The
parameters appear in a parenthesized comma-separated list like in a function prototype (see Sec-
tion 8.1). Unlike function prototypes, however, parameters usually have names associated with each
type.
Bodyevery function definition has a body enclosed by curly braces. The body contains the code to
be executed when the function is invoked.

2015 Richard L. Halterman Draft date: January 18, 2015

9.1. FUNCTION BASICS 210

Type of value the Name of Type of value the

function computes function function requires the
caller to provide

double square_root(double x) {
double diff;
// Compute a provisional square root
double root = 1.0; The name the function
uses for the value
Body of do { // Loop until the provisional root
// is close enough to provided by the
the actual rootcaller
function
root = (root + x/root) / 2.0;
// How bad is the approximation?
diff = root * root - x;
} while (diff > 0.0001 || diff < -0.0001);
return root;
}

Figure 9.1: Function definition dissection

Figure 9.1 dissects a our square_root function definition.

The simplest function accepts no parameters and returns no value to the caller. Listing 9.2 (simplefunction.cpp)
is a variation of Listing 4.1 (adder.cpp) that contains such a simple function definition.

Listing 9.2: simplefunction.cpp

#include <iostream>

using namespace std;

// Definition of the prompt function

void prompt() {
cout << "Please enter an integer value: ";
}

int main() {
int value1, value2, sum;
cout << "This program adds together two integers." << endl;
prompt(); // Call the function
cin >> value1;
prompt(); // Call the function again
cin >> value2;
sum = value1 + value2;
cout << value1 << " + " << value2 << " = " << sum << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

9.1. FUNCTION BASICS 211

The prompt function simply prints a message. The program runs as follows:

1. The programs execution, like in all C++ programs, begins with the first executable statement in the
function named main. The first line in the main function simply declares some variables needed for
compiler housekeeping, so the next line actually begins the executable code.
2. The first executable statement prints the message of the programs intent.
3. The next statement is a call of the prompt function. At this point the programs execution transfers
to the body of the prompt function. The code within prompt is executed until the end of its body
or until a return statement is encountered. Since prompt contains no return statement, all of
prompts body (the one print statement) will be executed.
4. When prompt is finished, control is passed back to the point in main immediately after the call of
prompt.
5. The next action after prompt call reads the value of value1 from the keyboard.
6. A second call to prompt transfers control back to the code within the prompt function. It again
prints its message.
7. When the second call to prompt is finished, control passes back to main at the point of the second
input statement that assigns value2 from the keyboard.
8. The remaining two statements in main are executed, and then the programs execution terminates.

As another simple example, consider Listing 9.3 (countto10.cpp).

Listing 9.3: countto10.cpp

#include <iostream>

using namespace std;

int main() {
for (int i = 1; i <= 10; i++)
cout << i << endl;
}

which simply counts to ten:

1
2
3
4
5
6
7
8
9
10

2015 Richard L. Halterman Draft date: January 18, 2015

9.1. FUNCTION BASICS 212

If counting to ten in this way is something we want to do frequently within a program, we can write a
function as shown in Listing 9.4 (countto10func.cpp) and call it as many times as necessary.

Listing 9.4: countto10func.cpp

#include <iostream>

using namespace std;

// Count to ten and print each number on its own line

void count_to_10() {
for (int i = 1; i <= 10; i++)
cout << i << endl;
}

int main() {
cout << "Going to count to ten . . .";
count_to_10();
cout << "Going to count to ten again. . .";
count_to_10();
}

Our prompt and countto10 functions are a bit underwhelming. The prompt function could be
eliminated, and each call to prompt could be replaced with the statement in its body. The same could be
said for the countto10 function, although it is convenient to have the simple one-line statement that hides
the complexity of the loop. Using the prompt function does have one advantage, though. If prompt is
removed and the two calls to prompt are replaced with the print statement within prompt, we have to make
sure that the two messages printed are identical. If we simply call prompt, we know the two messages
printed will be identical because only one possible message can be printed (the one in the body of prompt).
We can alter the behavior of a function through a mechanism called parameter passing. If a function
is written to accept information from the caller, the caller must supply the information in order to use the
function. The caller communicates the information via one or more parameters as required by the function.
The countto10 function does us little good if we sometimes want to count up to a different number.
Listing 9.5 (countton.cpp) generalizes Listing 9.4 (countto10func.cpp) to count as high as the caller needs.

Listing 9.5: countton.cpp

#include <iostream>

using namespace std;

// Count to n and print each number on its own line

void count_to_n(int n) {
for (int i = 1; i <= n; i++)
cout << i << endl;
}

int main() {
cout << "Going to count to ten . . .";
count_to_n(10);
cout << "Going to count to five . . .";
count_to_n(5);
}

2015 Richard L. Halterman Draft date: January 18, 2015

9.1. FUNCTION BASICS 213

When the caller, in this case main, issues the call

count_to_n(10);

the argument 10 is assigned to n before the functions statements begin executing.

A caller must pass exactly one integer parameter (or other type that is assignment-compatible with
integers) to countton during a call. An attempt to do otherwise one will result in a compiler error or
warning:
count_to_n(); // Error, missing parameter during the call
count_to_n(3, 5); // Error, too many parameters during the call
count_to_n(3.2); // Warning, possible loss of data (double to int)

We can enhance the prompt functions capabilities as shown in Listing 9.6 (betterprompt.cpp)

Listing 9.6: betterprompt.cpp

#include <iostream>

using namespace std;

// Definition of the prompt function

int prompt() {
int result;
cout << "Please enter an integer value: ";
cin >> result;
return result;
}

int main() {
int value1, value2, sum;
cout << "This program adds together two integers." << endl;
value1 = prompt(); // Call the function
value2 = prompt(); // Call the function again
sum = value1 + value2;
cout << value1 << " + " << value2 << " = " << sum << endl;
}

In this version, prompt takes care of the input, so main does not have to use any input statements. The
assignment statement within main:
value1 = prompt();

implies prompt is no longer a void function; it must return a value that can be assigned to the variable
value1. Furthermore, the value that prompt returns must be assignment compatible with an int be-
cause value1s declared type is int. A quick look at the first line of prompts definition confirms our
assumption:
int prompt()

This indicates that prompt returns an int value.

Because prompt is declared to return an int value, it must contain a return statement. A return
statement specifies the exact value to return to the caller. When a return is encountered during a func-
tions execution, control immediately passes back to the caller. The value of the function call is the value
specified by the return statement, so the statement

2015 Richard L. Halterman Draft date: January 18, 2015

9.1. FUNCTION BASICS 214

value1 = prompt();

assigns to the variable value1 the value indicated when the return statement executs.
Note that in Listing 9.6 (betterprompt.cpp), we declared a variable named result inside the prompt
function. This variable is local to the function, meaning we cannot use this particular variable outside of
prompt. It also means we are free to use that same name outside of the prompt function in a different
context, and that use will not interfere with the result variable within prompt. We say that result is
a local variable.
We can further enhance our prompt function. Currently prompt always prints the same message.
Using parameters, we can customize the message that prompt prints. Listing 9.7 (evenbetterprompt.cpp)
shows how parameters are used to provide a customized message within prompt.

Listing 9.7: evenbetterprompt.cpp

#include <iostream>

using namespace std;

// Definition of the prompt function

int prompt(int n) {
int result;
cout << "Please enter integer #" << n << ": ";
cin >> result;
return result;
}

int main() {
int value1, value2, sum;
cout << "This program adds together two integers." << endl;
value1 = prompt(1); // Call the function
value2 = prompt(2); // Call the function again
sum = value1 + value2;
cout << value1 << " + " << value2 << " = " << sum << endl;
}

In Listing 9.7 (evenbetterprompt.cpp), the parameter influences the message that it printed. The user is
now prompted to enter value #1 or value #2. The call
value1 = prompt(1);

passes the value 1 to the prompt function. Since prompts parameter is named n, the process works as
if the assignment statement
n = 1;

were executed as the first action within prompt.

In the first line of the function definition:
int prompt(int n)

n is called the formal parameter. A formal parameter is used like a variable within the functions body, but
it is declared in the functions parameter list; it is not declared in the functions body. A formal parameter
is a parameter as used in the formal definition of the function.

2015 Richard L. Halterman Draft date: January 18, 2015

9.1. FUNCTION BASICS 215

At the point of the function call:

value1 = prompt(1);

the parameter (or argument) passed into the function, 1, is called the actual parameter. An actual parameter
is the parameter actually used during a call of the function. When a function is called, any actual parameters
are assigned to their corresponding formal parameters, and the function begin executing. Another way to say
it is that during a function call, the actual parameters are bound to their corresponding formal parameters.

The parameters used within a function definition are called formal parameters.
Formal parameters behave as local variables within the functions body; as such,
the name of a formal parameter will not conflict with any local variable or for-
mal parameter names from other functions. This means as a function developer
you may choose a parameter name that best represents the parameters role in the
function.

If you are writing a function, you cannot predict the callers actual parameters.
You must be able to handle any value the caller sends. The compiler will ensure
that the types of the callers parameters are compatible with the declared types of
your formal parameters.

To remember the difference between formal and actual parameters, remember this:

A formal parameter is a parameter declared and used in a functions formal

definition.
An actual parameter is a parameter supplied by the caller when the caller
actually uses (invokes or calls) the function.

When the call

value1 = prompt(1);

is executed in main, and the statement

cout << "Please enter integer #" << n << ": ";

within the body of prompt is executed, n will have the value 1. Similarly, when the call
value2 = prompt(2);

is executed in main, and the statement

cout << "Please enter integer #" << n << ": ";

within the body of prompt is executed, n will have the value 2. In the case of
value1 = prompt(1);

n within prompt is bound to 1, and in the case of

value2 = prompt(2);

2015 Richard L. Halterman Draft date: January 18, 2015

9.2. USING FUNCTIONS 216

n within prompt is bound to 2.

A functions definition requires that all formal parameters be declared in the paren-
theses following the functions name. A caller does not provide actual parameter
type declarations when calling the function. Given the square_root function
defined in Listing 9.1 (customsquareroot.cpp), the following caller code fragment
is illegal:
double number = 25.0;
// Legal, pass the variable's value to the function
cout << square_root(number) << endl;
// Illegal, do not declare the parameter during the call
cout << square_root(double number) << endl;
The function definition is responsible for declaring the types of its parameters, not
the caller.

9.2 Using Functions

The general form of a function definition is

type name( parameterlist )

{
body
}
The type of the function indicates the type of value the function returns. Often a function will perform
a calculation and the result of the calculation must be communicated back to the place where the
function was invoked. The special type void indicates that the function does not return a value.
The name of the function is an identifier (see Section 3.3). The functions name should indicate the
purpose of the function.
The parameterlist is a comma separated list of pairs of the form
type name
where type is a C++ type and name is an identifier representing a parameter. The caller of the func-
tion communicates information into the function via parameters. The parameters specified in the
parameter list of a function definition are called formal parameters. A parameter is also known as an
argument. The parameter list may be empty; an empty parameter list indicates that no information
may be passed into the function by the caller.
The body is the sequence of statements, enclosed within curly braces, that define the actions that the
function is to perform. The statements may include variable declarations, and any variables declared
within the body are local to that function.
The body may contain only one statement, many statements, or no statements at all; regardless, the
curly braces always are required.

2015 Richard L. Halterman Draft date: January 18, 2015

9.2. USING FUNCTIONS 217

Observe that multiple pieces of information can be passed into a function via multiple parameters, but
only one piece of information can be passed out of the function via the return value. Recall the greatest
common divisor (also called greatest common factor) from elementary mathematics. To determine the GCD
of 24 and 18 we list all of their common factors and select the largest one:
24: 1, 2, 3, 4, 6 , 8, 12, 24
The greatest common divisor function is useful when reducing fractions
18: 1, 2, 3, 6 , 9, 18
18
to lowest terms; for example, consider the fraction . The greatest common divisor of 18 and 24 is 6, and
24
18 6 3
so we divide the numerator and the denominator of the fraction by 6: = . The GCD function has
24 6 4
applications in other areas besides reducing fractions to lowest terms. Consider the problem of dividing a
piece of plywood 24 inches long by 18 inches wide into square pieces of maximum size without wasting
any material. Since the GCD(24, 18) = 6, we can cut the plywood into twelve 6 inch 6 inch square
pieces as shown in Figure 9.2. If we cut the plywood into squares of any other size without wasting the

18 inches

24 inches

Figure 9.2: Cutting plywood

any of the material, the squares would have to be smaller than 6 inches 6 inches; for example, we could
make forty-eight 3 inch 3 inch squares as shown in pieces as shown in Figure 9.3. If we cut squares

3
3

18 inches

24 inches

Figure 9.3: Squares too small

2015 Richard L. Halterman Draft date: January 18, 2015

9.2. USING FUNCTIONS 218

larger than 6 inches 6 inches, not all the plywood can be used to make the squares. Figure 9.4. shows
how some larger squares would fare. The GCD function plays a vital role in cryptography, enabling secure

9 in.

9 in.

Waste
18 inches

24 inches

8 in.

8 in.

18 inches

Waste

24 inches

Figure 9.4: Squares too large

communication accross an insecure network.

The following code defines a function that computes the greatest common divisor of two integers. It
determines largest factor (divisor) common to its parameters:

int gcd(int num1, int num2) {

// Determine the smaller of num1 and num2
int min = (num1 < num2) ? num1 : num2;
// 1 is definitely a common factor to all ints
int largestFactor = 1;
for (int i = 1; i <= min; i++)
if (num1 % i == 0 && num2 % i == 0)
largestFactor = i; // Found larger factor
return largestFactor;
}

2015 Richard L. Halterman Draft date: January 18, 2015

9.2. USING FUNCTIONS 219

This function is named gcd and expects two integer arguments. Its formal parameters are named num1 and
num2. It returns an integer result. Its body declares three local variables: min, largestFactor, and i
(i is local to the for statement). The last line in its body is a return statement. A return statement is
required for functions that return a value. A void function is not required to have a return statement. If a
void function does have a return statement, it must simply consist of return followed by a semicolon
(in other words, it cannot return a value, like gcds return statement does). A void function that does
not contain a return statement simply returns at the end of its body.
Recall from Section 6.5 that local variables have meaning only within their scope. This means that
when you write a function you can name a local variable without fear that its name may be used already
in another part of the program. Two different functions can use local variables named x, and these are two
different variables that have no influence on each other. Anything local to a function definition is hidden to
all code outside that function definition.
Since a formal parameter is a local variable, you can reuse the names of formal parameters in different
functions without a problem.
Another advantage of local variables is that they occupy space in the computers memory only when
the function is executing. Space is allocated for local variables and parameters when the function begins
executing. When the function is finished and control returns to the caller, the variables and parameters go
out of scope, and the memory they held is freed up for other purposes within the running program. This
process of local variable allocation and deallocation happens each time a caller invokes the function. More
information about how C++ handles memory management during a programs execution can be found in
Section D.11.
Once we have written a complete function definition we can use the function within our program. We
invoke a programmer-defined function in exactly the same way as a standard library function like sqrt
(Section 8.2) or rand (Section 8.6). If the function returns a value (that is, it is not declared void), then we
can use its invocation anywhere an expression of that type is allowed. The parameters used for the function
call are known as actual parameters. The function gcd can be called as part of an assignment statement:
int factor = gcd(val, 24);

This call uses the variable val as its first actual parameter and the literal value 24 as its second actual
parameter. Variables, expressions, and literals can be freely used as actual parameters. The function then
computes and returns its result. This result is assigned to the variable factor.
How does the function call and parameter mechanism work? Its actually quite simple. The actual
parameters, in order, are assigned (bound) to each of the formal parameters in the function definition, then
control is passed to the body of the function. When the functions body is finished executing, control passes
back to the point in the program where the function was called. The value returned by the function, if any,
replaces the function call expression. In the statement
int factor = gcd(val, 24);

an integer value is assigned to factor. The expression on the right is a function call, so the function is
invoked to determine what to assign. The value of the variable val is assigned to the formal parameter
num1, and the literal value 24 is assigned to the formal parameter num2. The body of the gcd function is
then executed. When the return statement in the body is encountered, program execution returns back to
where the function was called. The argument of the return statement becomes the value that is assigned to
factor. This process of copying actual parameters to formal parameters works exactly like assignment,
so widening and narrowing (see Section 4.2 is performed automatically as needed, For example, if val is
declared to a char, its value would automatically be copied to a temporary location and converted to an
int. This temporary value would then be bound to the formal parameter num1. Note that gcd could be

2015 Richard L. Halterman Draft date: January 18, 2015

9.2. USING FUNCTIONS 220

called from many different places within the same program, and, since different parameter values could be
passed at each of these different invocations, gcd could compute a different result at each invocation.
Other invocation examples include:

cout << gcd(36, 24);

This example simply prints the result of the invocation. The value 36 is bound to num1
and 24 is bound to num2 for the purpose of the function call. The value 12 will be printed,
since 12 is the greatest common divisor of 36 and 24..

x = gcd(x - 2, 24);

The execution of this statement would evaluate x - 2 and bind its value to num1. num2
would be assigned 24. The result of the call is then assigned to x. Since the right side of
the assignment statement is evaluated before being assigned to the left side, the original
value of x is used when calculating x - 2, and the function return value then updates x.

x = gcd(x - 2, gcd(10, 8));

This example shows two invocations in one statement. Since the function returns an inte-
ger value its result can itself be used as an actual parameter in a function call. Passing the
result of one function call as an actual parameter to another function call is called function
composition.

The compiler will report an error if a function call does not agree with the functions definition. Possible
problems include:

Number of actual parameters do not agree with the number of formal parameters. The number
of parameters must agree exactly. For example, the statement
int factor = gcd(24); // Error: too few parameters

is illegal given the above definition of gcd, since only one actual parameter is provided when two
are required.

Passing an actual parameter that is not assignment compatible with the formal parameter. For
example, passing the cout object when an int has been defined, as in
int factor = gcd(36, cout); // Error: second parameter is wrong type

The compiler will detect that cout is not a valid int and report an error.

Using the result in a context where an expression of that type is not allowed. For example, a
function that returns void cannot be used where an int is expected:
cout << srand(2); // Error: srand does not return anything

The compiler will disallow this code.

2015 Richard L. Halterman Draft date: January 18, 2015

9.3. PASS BY VALUE 221

9.3 Pass by Value

The default parameter passing mechanism in C++ is classified as pass by value, also known as call by value.
This means the value of the actual parameter is copied to the formal parameter for the purpose of executing
the functions code. Since it is working on a copy of the actual parameter, the functions execution cannot
affect the value of the actual parameter owned by the caller.
Listing 9.8 (passbyvalue.cpp) illustrates the consequences of pass by value.

Listing 9.8: passbyvalue.cpp

#include <iostream>

using namespace std;

/*
* increment(x)
* Illustrates pass by value protocol.
*/
void increment(int x) {
cout << "Beginning execution of increment, x = "
<< x << endl;
x++; // Increment x
cout << "Ending execution of increment, x = "
<< x << endl;
}

int main() {
int x = 5;
cout << "Before increment, x = " << x << endl;
increment(x);
cout << "After increment, x = " << x << endl;
}

For additional drama we chose to name the actual parameter the same as the formal parameter. Since the
actual parameter and formal parameter are declared and used in different contexts and represent completely
different memory locations, their names can be the same without any problems.
Listing 9.8 (passbyvalue.cpp) produces

Before increment, x = 5
Beginning execution of increment, x = 5
Ending execution of increment, x = 6
After increment, x = 5

The memory for the variable x in main is unaffected since increment works on a copy of the actual
parameter.
C++ supports another way of passing parameters called pass by reference. Pass by reference is intro-
duced in Section 10.8.
A function communicates its return value to the caller in the same way that the caller might pass a
parameter by value. In the prompt function we saw earlier:

2015 Richard L. Halterman Draft date: January 18, 2015

9.4. FUNCTION EXAMPLES 222

int prompt(int n) {
int result;
cout << "Please enter integer #" << n << ": ";
cin >> result;
return result;
}

the return statement is

return result;

The variable result is local to prompt. We informally may say we are returning the result variable,
but, in fact, we really are returning only the value of the result variable. The caller has no access to
the local variables declared within any function it calls. In fact, the local variables for a function exist
only when the function is active (that is, executing). When the function returns to its caller all of its local
variables disappear from memory. During subsequent invocations, the functions local variables reappear
when the function becomes active and disappear again when it finishes.

9.4 Function Examples

This section contains a number of examples of how we can use functions to organize a programs code.

9.4.1 Better Organized Prime Generator

Listing 9.9 (primefunc.cpp) is a simple enhancement of Listing 8.5 (moreefficientprimes.cpp). It uses the
square root optimization and adds a separate is_prime function.

Listing 9.9: primefunc.cpp

#include <iostream>
#include <cmath>

using namespace std;

/*
* is_prime(n)
* Determines the primality of a given value
* n an integer to test for primality
* Returns true if n is prime; otherwise, returns false
*/
bool is_prime(int n) {
bool result = true; // Provisionally, n is prime
double r = n, root = sqrt(r);
// Try all possible factors from 2 to the square
// root of n
for (int trial_factor = 2;
result && trial_factor <= root; trial_factor++)
result = (n % trial_factor != 0);
return result;
}

2015 Richard L. Halterman Draft date: January 18, 2015

9.4. FUNCTION EXAMPLES 223

/*
* main
* Tests for primality each integer from 2
* up to a value provided by the user.
* If an integer is prime, it prints it;
* otherwise, the number is not printed.
*/
int main() {
int max_value;
cout << "Display primes up to what value? ";
cin >> max_value;
for (int value = 2; value <= max_value; value++)
if (is_prime(value)) // See if value is prime
cout << value << " "; // Display the prime number
cout << endl; // Move cursor down to next line
}

Listing 9.9 (primefunc.cpp) illustrates several important points about well-organized programs:

The complete work of the program is no longer limited to the main function. The effort to test for
primality is delegated to a separate function. main is focused on a simpler task: generating all the
numbers to be considered and using another function (is_prime) to do the hard work of determin-
ing if a given number is prime. main is now simpler and more logically coherent. A function is
coherent when it is focused on a single task. Coherence is a desirable property of functions. If a
function becomes too complex by trying to do too many different things, it can be more difficult to
write correctly and debug when problems are detected. A complex function should be decomposed
into several, smaller, more coherent functions. The original function would then call these new sim-
pler functions to accomplish its task. Here, main is not concerned about how to determine if a given
number is prime; main simply delegates the work to is_prime and makes use of the is_prime
functions findings.
Each function is preceded by a thorough comment that describes the nature of the function. It explains
the meaning of each parameter, and it indicates what the function should return. The comment for
main may not be as thorough as for other functions; this is because main usually has no parameters,
and it always returns a code to the operating system upon the programs termination.
While the exterior comment indicates what the function is to do, comments within each function
explain in more detail how the function accomplishes its task.

The call to is_prime return true or false depending on the value passed to it. The means a condition
like
if (is_prime(value) == true) . . .

can be expressed more compactly as

if (is_prime(value)) . . .

because if is_prime(value) is true, true == true is true, and if is_prime(value) is false,

false == true is false. The expression is_prime(value) suffices.
Just as it is better for a loop to have exactly one entry point and exactly one exit point, preferably a
function will have a single return statement. Simple functions with a small number of returns are
generally tolerable, however. Consider the following version of is_prime:

2015 Richard L. Halterman Draft date: January 18, 2015

9.4. FUNCTION EXAMPLES 224

bool is_prime(int n) {
for (int trialFactor = 2;
trialFactor <= sqrt(static_cast<double>(n));
trialFactor++)
if (n % trialFactor == 0) // Is trialFactor a factor?
return false; // Yes, return right away
return true; // Tried them all, must be prime
}

This version uses two return statements, but eliminates the need for a local variable (result). Because
a return statement exits the function immediately, no break statement is necessary. The two return
statements are close enough textually in source code that the logic is fairly transparent.

9.4.2 Command Interpreter

Some functions are useful even if they accept no information from the caller and return no result. List-
ing 9.10 (calculator.cpp) uses such a function.

Listing 9.10: calculator.cpp

#include <iostream>
#include <cmath>

using namespace std;

/*
* help_screen
* Displays information about how the program works
* Accepts no parameters
* Returns nothing
*/
void help_screen() {
cout << "Add: Adds two numbers" << endl;
cout << " Example: a 2.5 8.0" << endl;
cout << "Subtract: Subtracts two numbers" << endl;
cout << " Example: s 10.5 8.0" << endl;
cout << "Print: Displays the result of the latest operation"
<< endl;
cout << " Example: p" << endl;
cout << "Help: Displays this help screen" << endl;
cout << " Example: h" << endl;
cout << "Quit: Exits the program" << endl;
cout << " Example: q" << endl;
}

/*
* menu
* Display a menu
* Accepts no parameters
* Returns the character entered by the user.
*/
char menu() {
// Display a menu

2015 Richard L. Halterman Draft date: January 18, 2015

9.4. FUNCTION EXAMPLES 225

cout << "=== A)dd S)ubtract P)rint H)elp Q)uit ===" << endl;
// Return the char entered by user
char ch;
cin >> ch;
return ch;
}

/*
* main
* Runs a command loop that allows users to
* perform simple arithmetic.
*/
int main() {
double result = 0.0, arg1, arg2;
bool done = false; // Initially not done
do {
switch (menu()) {
case 'A': // Addition
case 'a':
cin >> arg1 >> arg2;
result = arg1 + arg2;
cout << result << endl;
break;
case 'S': // Subtraction
case 's':
cin >> arg1 >> arg2;
result = arg1 - arg2;
// Fall through, so it prints the result
case 'P': // Print result
case 'p':
cout << result << endl;
break;
case 'H': // Display help screen
case 'h':
help_screen();
break;
case 'Q': // Quit the program
case 'q':
done = true;
break;
}
}
while (!done);
}

The help_screen function needs no information from main, nor does it return a result. It behaves
exactly the same way each time it is called. The menu function returns the character entered by the user.

9.4.3 Restricted Input

Listing 7.3 (betterinputonly.cpp) forces the user to enter a value within a specified range. We now can easily
adapt that concept to a function. Listing 9.11 (betterinputfunc.cpp) uses a function named get_int_range
that does not return until the user supplies a proper value.

2015 Richard L. Halterman Draft date: January 18, 2015

9.4. FUNCTION EXAMPLES 226

Listing 9.11: betterinputfunc.cpp

#include <iostream>

using namespace std;

/*
* get_int_range(first, last)
* Forces the user to enter an integer within a
* specified range
* first is either a minimum or maximum acceptable value
* last is the corresponding other end of the range,
* either a maximum or minimum * value
* Returns an acceptable value from the user
*/
int get_int_range(int first, int last) {
// If the larger number is provided first,
// switch the parameters
if (first > last) {
int temp = first;
first = last;
last = temp;
}
// Insist on values in the range first...last
cout << "Please enter a value in the range "
<< first << "..." << last << ": ";
int in_value; // User input value
bool bad_entry;
do {
cin >> in_value;
bad_entry = (in_value < first || in_value > last);
if (bad_entry) {
cout << in_value << " is not in the range "
<< first << "..." << last << endl;
cout << "Please try again: ";
}
}
while (bad_entry);
// in_value at this point is guaranteed to be within range
return in_value;
}

/*
* main
* Tests the get_int_range function
*/
int main() {
cout << get_int_range(10, 20) << endl;
cout << get_int_range(20, 10) << endl;
cout << get_int_range(5, 5) << endl;
cout << get_int_range(-100, 100) << endl;
}

Listing 9.11 (betterinputfunc.cpp) forces the user to enter a value within a specified range. This func-
tionality could be useful in many programs.

2015 Richard L. Halterman Draft date: January 18, 2015

9.4. FUNCTION EXAMPLES 227

In Listing 9.11 (betterinputfunc.cpp)

The high and low values are specified by parameters. This makes the function more flexible since
it could be used elsewhere in the program with a completely different range specified and still work
correctly.
The function is supposed to be called with the lower number passed as the first parameter and the
higher number passed as the second parameter. The function will also accept the parameters out of
order and automatically swap them to work as expected; thus,
num = get_int_range(20, 50);

will work exactly like

num = get_int_range(50, 20);

The Boolean variable bad_entry is used to avoid evaluating the Boolean expression twice (once
to see if the bad entry message should be printed and again to see if the loop should continue).

9.4.4 Better Die Rolling Simulator

Listing 9.12: betterdie.cpp

#include <iostream>
#include <cstdlib>
#include <ctime>

using namespace std;

/*
* initialize_die
* Initializes the randomness of the die
*/
void initialize_die() {
// Set the random seed value
srand(static_cast<unsigned>(time(0)));
}

/*
* show_die(spots)
* Draws a picture of a die with number of spots
* indicated
* spots is the number of spots on the top face
*/
void show_die(int spots) {
cout << "+-------+" << endl;
switch (spots) {
case 1:
cout << "| |" << endl;
cout << "| * |" << endl;
cout << "| |" << endl;
break;
case 2:
cout << "| * |" << endl;

2015 Richard L. Halterman Draft date: January 18, 2015

9.4. FUNCTION EXAMPLES 228

cout << "| |" << endl;

cout << "| * |" << endl;
break;
case 3:
cout << "| * |" << endl;
cout << "| * |" << endl;
cout << "| * |" << endl;
break;
case 4:
cout << "| * * |" << endl;
cout << "| |" << endl;
cout << "| * * |" << endl;
break;
case 5:
cout << "| * * |" << endl;
cout << "| * |" << endl;
cout << "| * * |" << endl;
break;
case 6:
cout << "| * * * |" << endl;
cout << "| |" << endl;
cout << "| * * * |" << endl;
break;
default:
cout << " *** Error: illegal die value ***" << endl;
break;
}
cout << "+-------+" << endl;
}

/*
* roll
* Returns a pseudorandom number in the range 1...6
*/
int roll() {
return rand() % 6 + 1;
}

/*
* main
* Simulates the roll of a die three times
*/
int main() {

// Initialize the die

initialize_die();

// Roll the die three times

for (int i = 0; i < 3; i++)
show_die(roll());
}

In Listing 9.12 (betterdie.cpp), main is no longer concerned with the details of pseudorandom number
generation, nor is it responsible for drawing the die. These important components of the program are now

2015 Richard L. Halterman Draft date: January 18, 2015

9.4. FUNCTION EXAMPLES 229

in functions,