Introduction to C++

Day 1

Marco Frailis
INAF – Osservatorio Astronomico di Trieste

Marco Frailis Introduction to C++ 26 Feb 2015 1/45

The C++ language 1/2

C++ is designed to be a statically typed, general purpose language, to

be as compatible with the C language as possible
It is designed to support multiple programming styles:
Procedural programming
Data abstraction
Object-oriented programming (OOP)
Generic programming (with templates)
Functional programming (to some extent, e.g. lambda functions)

C++ is a compiled language

Source text files processed by a compiler producing object files
Object files combined by a linker yielding an executable program
The executable program is created for a specific hardware/system
combination

Marco Frailis Introduction to C++ 26 Feb 2015 2/45

The C++ language 2/2

It provides a standard library, extending the core language,

including:
Input/output classes
Data structures: dynamic arrays, linked lists, binary
trees, hash tables
Several algorithms: searching, sorting, counting,
manipulating
Memory management
Low level: pointers and raw arrays, allocation,
deallocation
Higher level: allocation and smart pointers

Marco Frailis Introduction to C++ 26 Feb 2015 3/45

Object-oriented programming

An object packages both data and procedures that operate on the data
Such procedures are called member functions or methods or
operations
The implementation of an object is specified in its class, which defines
the internal data (or attributes) of the object and the operations that
the object can perform
New classes can be defined as subclasses of a parent class,
inheriting its attributes and operations
An abstract class is one whose main purpose is to define a common
interface for its subclasses
When inheriting from an abstract class, we speak about polimorphism,
since the same interface is associated to different implementations
through its subclasses.

Marco Frailis Introduction to C++ 26 Feb 2015 4/45

Generic programming

Generic programming parameterizes algorithms so that they work for a

variety of types and data structures
For instance, data structures holding a collection of elements of some
kind, such as vectors, lists, queues, associative containers, are
general concepts and should be independent of the type of elements
Also, algorithms to sort, copy or search a sequence of elements
should be independent of the particular type of the elements or type of
the container
They can be defined as templates, parameterized by the types to
which type are applied
In C++, templates are a compile time mechanism, avoiding any run-
time overhead

Marco Frailis Introduction to C++ 26 Feb 2015 5/45

C++ standard evolution 1/2

First ISO standard in 1998 (C++98)

New revision of the standard in 2003 (C++03)
No new language features, just a bug fix release for
compiler writers
C++ Technical Report 1 (TR1) in 2005
A document proposing additions to the standard library

Marco Frailis Introduction to C++ 26 Feb 2015 6/45

C++ standard evolution 2/2

New C++ ISO standard in 2011 (C++11)

Additions to the core language, including:
Deducing the type of an object from its initializer (auto)
Lambda expressions
Move semantics
The range-for statement
Type aliases
Additions to the standard library, including
Hashed containers
Basic concurrency library (threads)
Regular expression library
tuple library
Unique and shared pointers

Last C++ standard revision in 2014 (C++14)

Minor revision with small improvements, e.g. function return type deduction

Marco Frailis Introduction to C++ 26 Feb 2015 7/45

C++ compilers (with C++11 conformance)

GCC (GNU Compiler Collection), version 4.8 or greater

Linux, Mac OS X, Windows (with MinGW or Cygwin), iOS, Android
ISO support status:
https://gcc.gnu.org/onlinedocs/libstdc+/manual/status.html#status.iso.2014

Clang, version 3.3 or greater

Linux, Mac OS X, Windows (with MinGW or Cygwin), iOS, Android

Intel(R) C++ Compiler (commercial), version 13.0 or greater

Linux, Mac OS X, Windows, Android

Microsoft Visual C++, version in Visual Studio 2013 or greater

Free version (with license limitations): Visual Studio Community 2013
Windows, Android, iOS

Marco Frailis Introduction to C++ 26 Feb 2015 8/45

C++ in Astronomy
Planck
Most of the processing pipelines developed in C++ (e.g. telemetry processing,
calibration)

LSST
Core processing developed in C++

ALMA
ACS software: C++ implementation for the control system

MAGIC telescopes
MARS, the analysis and reconstruction software is based on ROOT and written in C++

Euclid
C++ and Python selected for developing all the processing levels

FITS standard
cfitsio (C)
CCfits (C++)

Marco Frailis Introduction to C++ 26 Feb 2015 9/45

References
On-line references and FAQ:
http://www.cplusplus.com

http://en.cppreference.com/w/
https://isocpp.org/faq

http://www.stroustrup.com/C++11FAQ.html

On-line tutorials
http://www.tutorialspoint.com/cplusplus/index.htm
http://www.cplusplus.com/doc/tutorial/

http://www.learncpp.com/

http://www.cprogramming.com/tutorial/c++-tutorial.html

Books (in suggested order)

Accelerated C++: Practical Programming by Example, by A. Koenig and B. E. Moo
The C++ Standard Library: A Tutorial and Reference, 2nd edition, by N. M. Josuttis
C++ Primer (5th edition), by S. B. Lippman, J. Lajoie and B. E. Moo
The C++ Programming Language, 4th edition, B. Stroustrup

More advanced books

Effective Modern C++: 42 Specific Ways to Improve Your Use of C++11 and C++14, S. Meyers
C++ Templates: The Complete Guide, by D. Vandevoorde and N. M. Josuttis
Exceptional C++: 47 Engineering Puzzles, Programming Problems, and Solutions, by E. Sutter
More Exceptional C++: 40 New Engineering Puzzles, Programming Problems, and Solutions, by E. Sutter

Marco Frailis Introduction to C++ 26 Feb 2015 10/45

Integrated Development Environments (IDEs)

Code::Blocks (
http://www.codeblocks.org/)
A lightweight IDE (it is provided in
the virtual machine created for
this course)

Eclipse CDT
(https://eclipse.org/cdt/)

NetBeans IDE (https://netbeans.org/features/cpp/)

Emacs (http://www.gnu.org/software/emacs/)

Coding Ground (http://www.tutorialspoint.com/compile_cpp11_online.php)

An on-line web IDE, for short tests and examples

Others:
QT Creator, Xcode (Mac OS X), Visual Studio (Windows)
Marco Frailis Introduction to C++ 26 Feb 2015 11/45
C++ program: main function example
#include <cmath> Including standard library
#include <iostream>
functions and classes
using namespace std;

const double PI = 3.1415926535897931;

Constants definitions
const double RAD2DEG = 180.0/PI;

// Convert Cartesian coordinates to spherical coordinates

int main()
{ Main function definition
// Coordinates to be converted for the executable
double x=10.3, y=-5.2, z=36;

// Norm
double r = sqrt(x*x + y*y + z*z);

// Longitude statements for

double phi = atan2(y, x);
phi = phi < 0 ? phi + 2*PI : phi;
coordinates
conversion
// Colatitude
double theta = acos(z/r);

cout << "Longitude = " << phi*RAD2DEG << " deg, Colatitude = "
<< theta*RAD2DEG << " deg, Radius = " << r << endl; Printing output

return 0;
}

Marco Frailis Introduction to C++ 26 Feb 2015 12/45

Compiling with GNU g++

C++ source code should be given one of the valid C++ file extensions:
.cpp, .cc, .cxx or .C, while the .c extension is reserved to C programs
Saving the previous source code in file “coord.cpp”, in order to compile
it and create an executable, we can use the GNU g++ compiler:

$ g++ -std=c++11 coord.cpp -o coord

This command compiles the source code to machine code, saving it in

the executable file “coord”
The option “-std=c++11” enables compiler support for the 2011 ISO
C++ standard

Marco Frailis Introduction to C++ 26 Feb 2015 13/45

The #include preprocessor directive 1

Many C++ fundamental facilities, such as I/O, are not part of the core
language
They are part of the standard library. These facilities are requested
using the #include directive
Example: requesting the I/O standard library facilities and standard
strings
#include <iostream>
#include <string>

where iostream and string are called a standard headers

Including a header file produces the same results as copying the
header file into each source file that needs it
Normally the include directives are inserted at the beginning of a
source file
1
The preprocessor is a macro processor, called by the compiler to transform the program before
compilation
Marco Frailis Introduction to C++ 26 Feb 2015 14/45
Variable definition and initialization

A variable (also called object in C++) provides a named storage (a

region of memory) that our program can manipulate
Each variable definition starts with a type specifier, followed by a
comma separated list of one or more names (identifiers)
int day, month, year;
double salary;

A definition may also provide an initial value for the object (initialization)
C++ supports four forms of variable initialization: copy-initialization,
direct-initialization, list-copy-initialization, list-direct-initialization
int month = 9, year = 2012, day; // copy-initialization OK for built-in
double salary(1250.23); // direct-initialization types

C++11
std::complex<float> c = {-1.0, 0.0}; // list-copy-initialization Recommended
long int count{35600}; // list-direct-initialization for class types

It is safer to initialize every object of built-in type

Marco Frailis Introduction to C++ 26 Feb 2015 15/45

C++ built-in types
// Boolean type
bool empty = true;

// Character types
char single_letter = 'D';

// Integer types
char tiny_value = 127;
short small_value = 32767;
int value = 2147483647;
long large_value = 2147483647;
long long huge_value = -9.22E18LL;

// Unsigned integer types

unsigned char r = 255;
unsigned short small_pvalue = 65535;
// ...etc.

//Floating point types

float temperature = 2.323787;
double theta = 2.663832728147556;
long double ltheta = 2.6638327281475567373L;

Additional character types: wchar_t and char16_t and char32_t,

for Unicode characters (UTF-16 and UTF-32)

Marco Frailis Introduction to C++ 26 Feb 2015 16/45

Literals
Literal integer constants
Notations:

242 /* decimal */ 0362 /* octal */ 0xF2 /* hexadecimal */

Integer literal types: int or long by default, depending on value.

Use of suffix to force type

128U /* unsigned int */ 12ULL /* unsigned long long */

Floating point literals: double by default

2.323787F /* float */ 2.6638327281475567373L /* long double */

Character literals (char type, ASCII): 'a' '3'

Nonprintable characters: use of escape sequence

'\n' /* newline */ '\t' /* horizontal tab */ '\?' /* question mark */
'\'' /* single quote */ '\"' /* double quote */ '\0' /* null character */

String literals are arrays of characters

"\tHello World\n" "" /* empty string */

Marco Frailis Introduction to C++ 26 Feb 2015 17/45

Expression: Arithmetic, relational and logical
operators
An expression is composed of one or more operands and (usually) an
operator; it produces a result
Arithmetic operators:
a + b a - b Reminder after
a * b a / b
division a / b
a % b

Relational operators:
a < b a <= b
a > b a >= b
a == b a != b

Logical operators

Operator Meaning Example

&& and (a >= 0) && (a < 5)

|| or (a < 0) || (a >= 5)

! not !((a >= 0) && (a < 5))

Marco Frailis Introduction to C++ 26 Feb 2015 18/45

Bitwise operators

Bitwise operators:
Operator Meaning Example
<< left shift a << 3
>> right shift a >> 2
& bitwise and a & mask
| bitwise or a | mask
^ bitwise xor a ^ mask
~ bitwise not ~mask

unsigned char a = 20, mask = 1; // i.e. a = 00010100, mask = 00000001

a >> 2 // result = 00000101

mask << 3 // result = 00001000
~mask // result = 11111110

a & (mask << 4) // result = 00010000 (check if bit 4 is set)

a | (mask << 1) // result = 00010110 (set bit 1)
a ^ (mask << 3) // result = 00011100 (flip bit 3)

Marco Frailis Introduction to C++ 26 Feb 2015 19/45

Assignment and compound assignment
operators
// definition and initialization
int index = 0 , step = 5;
double deltaT = 0;

// assignment
deltaT = 0.3;
index = step * deltaT ; // here an implicit truncation is performed

The result of an assignment is the left-hand operand

We often apply an operator to an object and reassign the result to the
same object:

index = index + step * deltaT;

C++ provides compound-assignment operators as a shorthand, for

arithmetic and bitwise operators

index += step * deltaT; // equivalent to the previous statement

Marco Frailis Introduction to C++ 26 Feb 2015 20/45

Increment and decrement operators

Increment (++) and decrement (--) operators can be used as a

shorthand for adding or subtracting 1 from an object
There are two forms of these operators: prefix and postifix
The prefix form increment (decrement) its operand and yelds the
changed value as its result
The postfix form increment (decrement) its operand but yelds a copy of
the original, unchanged value as its result

int i = 0, j;

j = ++i; // j = 1 , i = 1;
j = i++; // j = 1 , i = 2;

Marco Frailis Introduction to C++ 26 Feb 2015 21/45

Implicit type conversions

In expressions with operands of mixed types, the types are converted

to a common type by the compiler
For arithmetic conversions, the rules define a hierarchy of conversions
in which operands are converted to the widest type in the expression
The conversion rules are defined so as to preserve the precision of the
values involved
Conversion to/from bool:
Integral and floating-point values can be converted to bool. If the value is
0, the resulting bool is false, otherwise it is true.
bool can be converted to other types: true is converted to 1, false to 0

Marco Frailis Introduction to C++ 26 Feb 2015 22/45

Explicit conversions

An explicit conversion is spoken of as a cast

C old-style cast:

int x = 37, y = 6;
float result = (float) x / y; // or float(x) / y

Casts should be discouraged: they turn off normal type-checking

C++ tries to make casts more visible providing named cast operators:

result = static_cast<float>(x) / y;

Additional C++ named cast operators:

dynamic_cast, const_cast, reinterpret_cast

Marco Frailis Introduction to C++ 26 Feb 2015 23/45

Strings in C++
The Standard Library provides the string type to support variable length
character strings
The library takes care of managing the memory associated with storing the
characters
It provides several ways to initialize a string variable:
#include <string>

int main()
{
std::string s1; // s1 is an empty string
std::string s2(s1); // s2 is initialized as a copy of s1
std::string s3("swing") // s3 is a copy of the string literal
std::string s4(5,'-'); // s4 = "-----"
}

string is part of the std namespace

A namespace is a collection of related names: the standard library uses std to
contain all the names that it defines
std::string is a qualified name, using the scope operator ( ::)

Marco Frailis Introduction to C++ 26 Feb 2015 24/45

String operations
#include <string>

using namespace std;

int main()
{
string s1;
string s2 = "Hell0";
string s3 = s2 + " World"; // string + literal concatenation

bool isEmpty = s1.empty(); // check if s1 is an empty string

char first = s3[0]; // get the first character
s3[4] = 'o'; // modify the 5th character in s3
bool isLess = s2 < s1; // comparison operator between strings
// (lexicographic order)
s1 = s2; // copy s2 in s1, replacing s1 characters
string s4 = s3.substr(6, 5) // get the substring starting at index 6
// and taking 5 chars (s4 = "World")
string::size_type size = s4.size(); // get string size in bytes
}
Type defined in the string class:
synonym of an unsigned int, but more portable

Marco Frailis Introduction to C++ 26 Feb 2015 25/45

C++ Input/Output

The standard library defines a family of types that support I/O to and from
devices such as files and console windows. Additional types allow strings to
act like I/O streams
The I/O types are defined in three separate headers:
<iostream> defines the types used to read from and write to a console
window
<fstream> defines the types used to read from and write to named files

<sstream> defines the types used to read from a and write to in-
memory strings
The iostream header includes the definition of three I/O objects:
The istream object named cin, also referred to as standard input
The ostream object named cout, also referred to as standard output
The ostream object named cerr, also referred to as standard error

Marco Frailis Introduction to C++ 26 Feb 2015 26/45

Standard I/O example
#include <iostream>
using namespace std;

int main()
{
cout << "Please, insert two numbers:" << endl;
float v1, v2;

cin >> v1 >> v2;

cout << "The sum of " << v1 << " and " << v2
<< " is " << v1 + v2 << endl;
return 0;
}

The stream operators (<< and >>) are left associative and return the
left operand, i.e. the stream object)
So we can chain the stream operations

Marco Frailis Introduction to C++ 26 Feb 2015 27/45

Formatting floating point values
#include <iostream> I/O format state (parametric)
#include <iomanip> manipulators
using namespace std;

int main()
{
const double PI = 3.1415926535897931;

cout << PI << endl;

cout << setprecision(15) << PI << endl;
cout << scientific << PI << endl;
cout << fixed << setw(20) << PI << endl; // setw sets the field width
cout << resetiosflags(ostream::floatfield); // revert to default notation

cout << PI << endl;

}

Output:
3.14159
3.14159265358979
3.141592653589793e+00
3.141592653589793
3.14159265358979

Marco Frailis Introduction to C++ 26 Feb 2015 28/45

Statements

Most statements in C++ end with a semicolon

We have already seen some expression statements

i.e. an expression followed by a semicolon
Commonly expression statements affect the program's state: assignment,
increment, input/output operators, declaration statements

A compound statement, or block, is a (possibly empty) sequence of

statements surrounded by a pair of curly braces

Compound statements can be used where the rules of the language

require a single statement

Marco Frailis Introduction to C++ 26 Feb 2015 29/45

Scope of a name

Every name in a C++ program must refer to a unique entity (e.g. a variable,
function, type, etc.)
A name can be reused, also with different meanings, as long as it is used in
different context
Such context is the scope, i.e. a region of the program
There are different kinds of scope:
The statements between a pair of matching braces form a scope. The body of
the main or the body of every function and the scopes nested inside a function
(such as a block) form local scopes
Names defined outside any function have global scope. They are accessible
from anywhere in the program
Namespaces partition the global namespace. A namespace is a scope, as we
have already seen with the standard library namespace std
Every class defines its own new scope. An examples is given by the type
size_type of the string class

Marco Frailis Introduction to C++ 26 Feb 2015 30/45

Scope example
#include <iostream>

using namespace std;

Global const variable
const string s = "---------";

int main() s is a variable in the main

{
unsigned int s = 0; function local scope

{
unsigned int s = 1; In this block, s definition
cout << "Inner s: " << s << endl; hides the outer one
}

cout << "Outer s: " << s << endl;

cout << "Global s: " << ::s << endl;

return 0;
}

Output:
Inner s: 1
Outer s: 0
Global s: ---------

Marco Frailis Introduction to C++ 26 Feb 2015 31/45

if statement
Condition
if (rawVal > 90) {
cout << "Warning: hard limit exceeded" << endl;
Statement 1
++hardLimit ;
}
else if (rawVal > 70)
++softLimit;
else { Statement 2
sum += rawVal;
++numSamples;
}

Second if form: without the else branch

if (i < j)
++i;

Marco Frailis Introduction to C++ 26 Feb 2015 32/45

while and do while statements
while statement
Condition
while (i > 0 && v[i] < key) {
v[i+1] = v[i];
Statement
--i;
}

do-while statement
double threshold = -1;
do {
cout << "Please enter a threshold level in the range [0, 1]" Statement executed
<< endl; at least once
cin >> threshold;
} while (threshold < 0 || threshold > 1);

int i = 0;
while (i < size) { Recurrent pattern:
// operations that don't
// change the value of i
for loop
++i;
}

Marco Frailis Introduction to C++ 26 Feb 2015 33/45

for and range-based for loops
for statement

init-statement condition expression statement

for (int i = 1, j = 0; i < v.size(); i += 2, ++j) {
s[j] = (v[i-1] + v[i])/2;
d[j] = v[i-1] - s[j]; Executed after body
} of for loop

range-for statement
for (int x : v) {
sum += x; for each value x in
sum2 += x*x; range v
} C++11

Infinite loop until event

for (;;) { for (int i = 0; i < v.size(); ++i)

cout << "Processing next event" << endl: {
checkEventQueue(); if (v[i] < 0 || v[i] > 5)
if (stopExecution) continue;
break; sum += v[i];
skip rest of
} } iteration

ends the nearest enclosing loop

Marco Frailis Introduction to C++ 26 Feb 2015 34/45
switch-case statement

int apId = packet.apId();

switch (apId) { expression with integral result

case 1536:
case 1538:
++lfiHkCounter; // LFI housekeeping telemetry found
break;
case 1540:
++lfiSciCounter; // LFI scientific telemetry found
break;
default:
cout << "Unknown telemetry packet" << endl;
}

It provides a more convenient way to write deeply nested if/else logic

The result of the expression is compared with the value associated with
each case
Execution starts with the first statement following the matching label, till a
break is found

Marco Frailis Introduction to C++ 26 Feb 2015 35/45

Standard vector 1/3
The standard library provides a type, named vector, that holds a sequence of values of a
given type and grows as needed
It is defined using a language feature called template classes
#include <iostream>
#include <vector>
#include <complex>

using namespace std;

int main() Template

{ parameter

int n = 10;
vector<float> a(n, 5.0); // a has n elements that are copies of 5.0
C++11 vector<int> b = {3,2,7,11,23}; // b contains the elements provided in the list

cout << "First and last elements of b: "

<< b[0] << " " << b[b.size()-1] << endl;

// vector from a list of numbers gathered from the standard input

double value;
vector<double> c;
cout << "Please insert some values (use Ctrl-D to end): " << endl;

// The user ends inputting numbers with the EOF character

while (cin >> value)
c.push_back(value); // appends a new element at the end of vector c Continue ...

Marco Frailis Introduction to C++ 26 Feb 2015 36/45

 Standard vector 2/3
// Print vector elements
cout << "c elements: ";
for (double x : c)
cout << x << " ";
cout << endl;

vector<double> d(c); // defines d as a copy of c

// e holds a copy of the first half of d

vector<double> e(d.begin(), d.begin() + d.size()/2);

c.clear(); // removes all elements in c; c.size() == 0

// erase elements of first half of d

d.erase(d.begin(), d.begin() + d.size()/2);

c = e; // c is now a copy of e

// prepend elements of a in e
e.insert(e.begin(), a.begin(), a.end());

vector<complex<double>> f;
f.push_back({1, -1}); // analogous to push_back(complex<double>(1,-1)), but shorter
// thanks to initializer list
C++11 f.emplace_back(1, -1); // emplace_back is analogous to push_back but it just needs the
// arguments to construct the value, which is created in-place
}

Marco Frailis Introduction to C++ 26 Feb 2015 37/45

Standard vector 3/3
vector is one of the sequential containers provided by the standard
library
To support fast random access to the elements, vector elements are
stored contiguously
Vectors grow dynamically: they allocate capacity beyond what is
immediately needed, as a reserve for new elements
When the capacity is exceeded, a new block of contiguous memory is
automatically allocated (e.g. by a factor of 2 larger than the previous one)
Vector class provides two member functions concerning its memory block:
vector::capacity() and vector::reserve(n)
a.size() == 33;
vector<int> a;
vector<int> a;
for (int i = 0; i < 33; ++i) {
a.push_back(i); 0 1 2 3 31 32
cout << "a capacity: "
<< a.capacity()
<< endl; a.capacity() == 64
}

a.reserve(n) lets us set the initial vector capacity

Marco Frailis Introduction to C++ 26 Feb 2015 38/45

The auto type specifier and type aliases
When a definition of a variable has an initializer, we can let the variable
have the type of the initializer using the auto type “placeholder”

vector<complex<double>> v = {{1,0}, {1,-1}, {0,1}, {0,-1}};

C++11 auto size = v.size(); The size variable has type

vector<complex<double>>::size_type
cout << "v elements: ";
for (auto x : v)
cout << x << " "; x has type
cout << endl; complex<double>

When defining a variable with auto, prefer the “=” syntax

Sometime we need a new name for a type, e.g. because it is too long
or complex or because we need different types to have same name

Older syntax using typedef

C++11
using vec_compd = vector<complex<double>>;

vec_compd v = {{1,0}, {1,-1}, {0,1}, {0,-1}}; typedef vector<complex<double>> vec_compd;

Marco Frailis Introduction to C++ 26 Feb 2015 39/45

Functions

Functions are named computations. We can identify important parts of

our problems and create named computations corresponding to those
parts
A function is defined by specifying its return type, followed by the
function name and then by a parameter list enclosed in () and finally
the function body, which is enclosed in {}
A function is uniquely represented by its name and the list of
parameter types. They form the so called function signature
int gcd(int v1, int v2) // return the greatest common divisor
{
while (v2) {
int temp = v2;
v2 = v1 % v2;
v1 = temp;
}
return v1;
}

Marco Frailis Introduction to C++ 26 Feb 2015 40/45

Argument passing

In the previous example, parameters are passed by value, i.e. they

are initialized by copying the corresponding argument

When we pass built-in types by value, the cost of copying them into the
function parameters is negligible

However, if we need to pass a long vector or string, copying them can

be time consuming

Often, we also need to define functions that modify the arguments

passed, or we need them to return additional information to the caller

Marco Frailis Introduction to C++ 26 Feb 2015 41/45

References and const references

A reference to an object is another name, or synonym for that object

vector<double> samples;

vector<double>& s = samples; // s is a synonym for samples

In the example above, variable s is defined as a reference (synonym) to

samples
Any operation performed on s is equivalent to doing the same to samples,
and vice versa
A reference always refers to the object to which it was initialized (no null
references)
The main use of references is for specifying arguments and return values
for functions in general
When we add a const to a reference, we can only use the reference to
read values from the original object, but we cannot change its values (no
write access)

Marco Frailis Introduction to C++ 26 Feb 2015 42/45

Argument passing: const references

double mean(const vector<double>& samples) samples is a synonym for the

{ actual argument (a vector)
auto size = samples.size(); passed in the function call
if (size == 0)
throw domain_error("mean of an empty vector");

double sum = 0;
for (const auto& x : samples) each element in samples is
sum += x; passed to x as const reference
return sum/size;
}

The type that we specify for the function argument is called “reference
to const vector of double”
No copy of the actual argument passed with the function call (efficient)
The actual argument is passed as a read-only object

In the range-for within the function body, x is a const reference to each

element in samples (no copy performed)

Marco Frailis Introduction to C++ 26 Feb 2015 43/45

Argument passing: non-const reference
There are situations where passing parameters by value or using const
references don't work. For instance, in a simple function that swaps two
values we need to use non-const references:
void swap(int& v1, int& v2)
{
int tmp = v2;
v2 = v1;
v1 = tmp:
}

int main()
{
int i = 5, j = 20;
swap(i,j); // After this call: i = 20, j = 5

This function has no return value. Its return type is void, a C/C++ built-
in type used in a few restricted ways, e.g. to name a return type
The purpose of the swap function is to change its argument values. So
its parameters are declared as references

Marco Frailis Introduction to C++ 26 Feb 2015 44/45

Argument passing: summary

When function parameters are built-in types, they can be passed by

value if the function does not need to change their value

Whenever the function does not need to change the parameter's value,
and the parameter has a type that can be time consuming to copy,
then the parameter should be a const reference

When the function intends to change the arguments values, the

corresponding parameters must be declared as non-const references
Arguments passed as non-const reference parameters must be non-
temporary objects (lvalues)

Marco Frailis Introduction to C++ 26 Feb 2015 45/45