learnxinyminutes-docs/c++.html.markdown

---
language: c++
filename: learncpp.cpp
contributors:
    - ["Steven Basart", "http://github.com/xksteven"]
    - ["Matt Kline", "https://github.com/mrkline"]
lang: en
---

C++ is a systems programming language that,
[according to its inventor Bjarne Stroustrup](http://channel9.msdn.com/Events/Lang-NEXT/Lang-NEXT-2014/Keynote),
was designed to

- be a "better C"
- support data abstraction
- support object-oriented programming
- support generic programming

Though its syntax can be more difficult or complex than newer languages,
it is widely used because it compiles to native instructions that can be
directly run by the processor and offers tight control over hardware (like C)
while offering high-level features such as generics, exceptions, and classes.
This combination of speed and functionality makes C++
one of the most widely-used programming languages.

```c++
//////////////////
// Comparison to C
//////////////////

// C++ is _almost_ a superset of C and shares its basic syntax for
// variable declarations, primitive types, and functions.

// Just like in C, your program's entry point is a function called
// main with an integer return type.
// This value serves as the program's exit status.
// See http://en.wikipedia.org/wiki/Exit_status for more information.
int main(int argc, char** argv)
{
    // Command line arguments are passed in by argc and argv in the same way
    // they are in C.
    // argc indicates the number of arguments,
    // and argv is an array of C-style strings (char*)
    // representing the arguments.
    // The first argument is the name by which the program was called.
    // argc and argv can be omitted if you do not care about arguments,
    // giving the function signature of int main()

    // An exit status of 0 indicates success.
    return 0;
}

// However, C++ varies in some of the following ways:

// In C++, character literals are one byte.
sizeof('c') == 1

// In C, character literals are the same size as ints.
sizeof('c') == sizeof(10)


// C++ has strict prototyping
void func(); // function which accepts no arguments

// In C
void func(); // function which may accept any number of arguments

// Use nullptr instead of NULL in C++
int* ip = nullptr;

// C standard headers are available in C++,
// but are prefixed with "c" and have no .h suffix.
#include <cstdio>

int main()
{
    printf("Hello, world!\n");
    return 0;
}

///////////////////////
// Function overloading
///////////////////////

// C++ supports function overloading
// provided each function takes different parameters.

void print(char const* myString)
{
    printf("String %s\n", myString);
}

void print(int myInt)
{
    printf("My int is %d", myInt);
}

int main()
{
    print("Hello"); // Resolves to void print(const char*)
    print(15); // Resolves to void print(int)
}

/////////////////////////////
// Default function arguments
/////////////////////////////

// You can provide default arguments for a function
// if they are not provided by the caller.

void doSomethingWithInts(int a = 1, int b = 4)
{
    // Do something with the ints here
}

int main()
{
    doSomethingWithInts();      // a = 1,  b = 4
    doSomethingWithInts(20);    // a = 20, b = 4
    doSomethingWithInts(20, 5); // a = 20, b = 5
}

// Default arguments must be at the end of the arguments list.

void invalidDeclaration(int a = 1, int b) // Error!
{
}


/////////////
// Namespaces
/////////////

// Namespaces provide separate scopes for variable, function,
// and other declarations.
// Namespaces can be nested.

namespace First {
    namespace Nested {
        void foo()
        {
            printf("This is First::Nested::foo\n");
        }
    } // end namespace Nested
} // end namespace First

namespace Second {
    void foo()
    {
        printf("This is Second::foo\n")
    }
}

void foo()
{
    printf("This is global foo\n");
}

int main()
{
    // Assume everything is from the namespace "Second"
    // unless otherwise specified.
    using namespace Second;

    foo(); // prints "This is Second::foo"
    First::Nested::foo(); // prints "This is First::Nested::foo"
    ::foo(); // prints "This is global foo"
}

///////////////
// Input/Output
///////////////

// C++ input and output uses streams
// cin, cout, and cerr represent stdin, stdout, and stderr.
// << is the insertion operator and >> is the extraction operator.

#include <iostream> // Include for I/O streams

using namespace std; // Streams are in the std namespace (standard library)

int main()
{
   int myInt;

   // Prints to stdout (or terminal/screen)
   cout << "Enter your favorite number:\n";
   // Takes in input
   cin >> myInt;

   // cout can also be formatted
   cout << "Your favorite number is " << myInt << "\n";
   // prints "Your favorite number is <myInt>"

    cerr << "Used for error messages";
}

//////////
// Strings
//////////

// Strings in C++ are objects and have many member functions
#include <string>

using namespace std; // Strings are also in the namespace std (standard library)

string myString = "Hello";
string myOtherString = " World";

// + is used for concatenation.
cout << myString + myOtherString; // "Hello World"

cout << myString + " You"; // "Hello You"

// C++ strings are mutable and have value semantics.
myString.append(" Dog");
cout << myString; // "Hello Dog"


/////////////
// References
/////////////

// In addition to pointers like the ones in C,
// C++ has _references_.
// These are pointer types that cannot be reassigned once set
// and cannot be null.
// They also have the same syntax as the variable itself:
// No * is needed for dereferencing and
// & (address of) is not used for assignment.

using namespace std;

string foo = "I am foo";
string bar = "I am bar";


string& fooRef = foo; // This creates a reference to foo.
fooRef += ". Hi!"; // Modifies foo through the reference
cout << fooRef; // Prints "I am foo. Hi!"

// Doesn't reassign "fooRef". This is the same as "foo = bar", and
//   foo == "I am bar"
// after this line.
fooRef = bar;

const string& barRef = bar; // Create a const reference to bar.
// Like C, const values (and pointers and references) cannot be modified.
barRef += ". Hi!"; // Error, const references cannot be modified.

//////////////////////////////////////////
// Classes and object-oriented programming
//////////////////////////////////////////

// First example of classes
#include <iostream>

// Declare a class.
// Classes are usually declared in header (.h or .hpp) files.
class Dog {
    // Member variables and functions are private by default.
    std::string name;
    int weight;

// All members following this are public
// until "private:" or "protected:" is found.
public:

    // Default constructor
    Dog();

    // Member function declarations (implementations to follow)
    // Note that we use std::string here instead of placing
    // using namespace std;
    // above.
    // Never put a "using namespace" statement in a header.
    void setName(const std::string& dogsName);

    void setWeight(int dogsWeight);

    // Functions that do not modify the state of the object
    // should be marked as const.
    // This allows you to call them if given a const reference to the object.
    // Also note the functions must be explicitly declared as _virtual_
    // in order to be overridden in derived classes.
    // Functions are not virtual by default for performance reasons.
    virtual void print() const;

    // Functions can also be defined inside the class body.
    // Functions defined as such are automatically inlined.
    void bark() const { std::cout << name << " barks!\n" }

    // Along with constructors, C++ provides destructors.
    // These are called when an object is deleted or falls out of scope.
    // This enables powerful paradigms such as RAII
    // (see below)
    // Destructors must be virtual to allow classes to be derived from this one.
    virtual ~Dog();

}; // A semicolon must follow the class definition.

// Class member functions are usually implemented in .cpp files.
void Dog::Dog()
{
    std::cout << "A dog has been constructed\n";
}

// Objects (such as strings) should be passed by reference
// if you are modifying them or const reference if you are not.
void Dog::setName(const std::string& dogsName)
{
    name = dogsName;
}

void Dog::setWeight(int dogsWeight)
{
    weight = dogsWeight;
}

// Notice that "virtual" is only needed in the declaration, not the definition.
void Dog::print() const
{
    std::cout << "Dog is " << name << " and weighs " << weight << "kg\n";
}

void Dog::~Dog()
{
    cout << "Goodbye " << name << "\n";
}

int main() {
    Dog myDog; // prints "A dog has been constructed"
    myDog.setName("Barkley");
    myDog.setWeight(10);
    myDog.printDog(); // prints "Dog is Barkley and weighs 10 kg"
    return 0;
} // prints "Goodbye Barkley"

// Inheritance:

// This class inherits everything public and protected from the Dog class
class OwnedDog : public Dog {

    void setOwner(const std::string& dogsOwner)

    // Override the behavior of the print function for all OwnedDogs. See
    // http://en.wikipedia.org/wiki/Polymorphism_(computer_science)#Subtyping
    // for a more general introduction if you are unfamiliar with
    // subtype polymorphism.
    // The override keyword is optional but makes sure you are actually
    // overriding the method in a base class.
    void print() const override;

private:
    std::string owner;
};

// Meanwhile, in the corresponding .cpp file:

void OwnedDog::setOwner(const std::string& dogsOwner)
{
    owner = dogsOwner;
}

void OwnedDog::print() const
{
    Dog::print(); // Call the print function in the base Dog class
    std::cout << "Dog is owned by " << owner << "\n";
    // Prints "Dog is <name> and weights <weight>"
    //        "Dog is owned by <owner>"
}

//////////////////////////////////////////
// Initialization and Operator Overloading
//////////////////////////////////////////

// In C++ you can overload the behavior of operators such as +, -, *, /, etc.
// This is done by defining a function which is called
// whenever the operator is used.

#include <iostream>
using namespace std;

class Point {
public:
    // Member variables can be given default values in this manner.
    double x = 0;
    double y = 0;

    // Define a default constructor which does nothing
    // but initialize the Point to the default value (0, 0)
    Point() { };

    // The following syntax is known as an initialization list
    // and is the proper way to initialize class member values
    Point (double a, double b) :
        x(a),
        y(b)
    { /* Do nothing except initialize the values */ }

    // Overload the + operator.
    Point operator+(const Point& rhs) const;

    // Overload the += operator
    Point& operator+=(const Point& rhs);

    // It would also make sense to add the - and -= operators,
    // but we will skip those for brevity.
};

Point Point::operator+(const Point& rhs) const
{
    // Create a new point that is the sum of this one and rhs.
    return Point(x + rhs.x, y + rhs.y);
}

Point& Point::operator+=(const Point& rhs)
{
    x += rhs.x;
    y += rhs.y;
    return *this;
}

int main () {
    Point up (0,1);
    Point right (1,0);
    // This calls the Point + operator
    // Point up calls the + (function) with right as its paramater
    Point result = up + right;
    // Prints "Result is upright (1,1)"
    cout << "Result is upright (" << result.x << ',' << result.y << ")\n";
    return 0;
}

/////////////////////
// Templates
/////////////////////

// Templates in C++ are mostly used for generic programming, though they are
// much more powerful than generics constructs in other languages. It also
// supports explicit and partial specialization, functional-style type classes,
// and also it's Turing-complete.

// We start with the kind of generic programming you might be familiar with. To
// define a class or function that takes a type parameter:
template<class T>
class Box {
    // In this class, T can be used as any other type.
    void insert(const T&) { ... }
};

// During compilation, the compiler actually generates copies of each template
// with parameters substituted, and so the full definition of the class must be
// present at each invocation. This is why you will see template classes defined
// entirely in header files.

// To instantiate a template class on the stack:
Box<int> intBox;

// and you can use it as you would expect:
intBox.insert(123);

// You can, of course, nest templates:
Box<Box<int> > boxOfBox;
boxOfBox.insert(intBox);

// Up until C++11, you muse place a space between the two '>'s, otherwise '>>'
// will be parsed as the right shift operator.

// You will sometimes see
//   template<typename T>
// instead. The 'class' keyword and 'typename' keyword are _mostly_
// interchangeable in this case. For full explanation, see
//   http://en.wikipedia.org/wiki/Typename
// (yes, that keyword has its own Wikipedia page).

// Similarly, a template function:
template<class T>
void barkThreeTimes(const T& input)
{
    input.bark();
    input.bark();
    input.bark();
}

// Notice that nothing is specified about the type parameters here. The compiler
// will generate and then type-check every invocation of the template, so the
// above function works with any type 'T' that has a const 'bark' method!

Dog fluffy;
fluffy.setName("Fluffy")
barkThreeTimes(fluffy); // Prints "Fluffy barks" three times.

// Template parameters don't have to be classes:
template<int Y>
void printMessage() {
  cout << "Learn C++ in " << Y << " minutes!" << endl;
}

// And you can explicitly specialize templates for more efficient code (most
// real-world uses of specialization are not as trivial as this):
template<>
void printMessage<10>() {
  cout << "Learn C++ faster in only 10 minutes!" << endl;
}

printMessage<20>();  // Prints "Learn C++ in 20 minutes!"
printMessage<10>();  // Prints "Learn C++ faster in only 10 minutes!"


/////////////////////
// Exception Handling
/////////////////////

// The standard library provides a few exception types
// (see http://en.cppreference.com/w/cpp/error/exception)
// but any type can be thrown an as exception
#include <exception>

// All exceptions thrown inside the _try_ block can be caught by subsequent
// _catch_ handlers.
try {
    // Do not allocate exceptions on the heap using _new_.
    throw std::exception("A problem occurred");
}
// Catch exceptions by const reference if they are objects
catch (const std::exception& ex)
{
  std::cout << ex.what();
// Catches any exception not caught by previous _catch_ blocks
} catch (...)
{
    std::cout << "Unknown exception caught";
    throw; // Re-throws the exception
}

///////
// RAII
///////

// RAII stands for Resource Allocation Is Initialization.
// It is often considered the most powerful paradigm in C++,
// and is the simple concept that a constructor for an object
// acquires that object's resources and the destructor releases them.

// To understand how this is useful,
// consider a function that uses a C file handle:
void doSomethingWithAFile(const char* filename)
{
    // To begin with, assume nothing can fail.

    FILE* fh = fopen(filename, "r"); // Open the file in read mode.

    doSomethingWithTheFile(fh);
    doSomethingElseWithIt(fh);

    fclose(fh); // Close the file handle.
}

// Unfortunately, things are quickly complicated by error handling.
// Suppose fopen can fail, and that doSomethingWithTheFile and
// doSomethingElseWithIt return error codes if they fail.
// (Exceptions are the preferred way of handling failure,
//  but some programmers, especially those with a C background,
//  disagree on the utility of exceptions).
// We now have to check each call for failure and close the file handle
// if a problem occurred.
bool doSomethingWithAFile(const char* filename)
{
    FILE* fh = fopen(filename, "r"); // Open the file in read mode
    if (fh == nullptr) // The returned pointer is null on failure.
        return false; // Report that failure to the caller.

    // Assume each function returns false if it failed
    if (!doSomethingWithTheFile(fh)) {
        fclose(fh); // Close the file handle so it doesn't leak.
        return false; // Propagate the error.
    }
    if (!doSomethingElseWithIt(fh)) {
        fclose(fh); // Close the file handle so it doesn't leak.
        return false; // Propagate the error.
    }

    fclose(fh); // Close the file handle so it doesn't leak.
    return true; // Indicate success
}

// C programmers often clean this up a little bit using goto:
bool doSomethingWithAFile(const char* filename)
{
    FILE* fh = fopen(filename, "r");
    if (fh == nullptr)
        return false;

    if (!doSomethingWithTheFile(fh))
        goto failure;

    if (!doSomethingElseWithIt(fh))
        goto failure;

    fclose(fh); // Close the file
    return true; // Indicate success

failure:
    fclose(fh);
    return false; // Propagate the error
}

// If the functions indicate errors using exceptions,
// things are a little cleaner, but still sub-optimal.
void doSomethingWithAFile(const char* filename)
{
    FILE* fh = fopen(filename, "r"); // Open the file in read mode
    if (fh == nullptr)
        throw std::exception("Could not open the file.");

    try {
        doSomethingWithTheFile(fh);
        doSomethingElseWithIt(fh);
    }
    catch (...) {
        fclose(fh); // Be sure to close the file if an error occurs.
        throw; // Then re-throw the exception.
    }

    fclose(fh); // Close the file
    // Everything succeeded
}

// Compare this to the use of C++'s file stream class (fstream)
// fstream uses its destructor to close the file.
// Recall from above that destructors are automatically called
// whenever an object falls out of scope.
void doSomethingWithAFile(const std::string& filename)
{
    // ifstream is short for input file stream
    std::ifstream fh(filename); // Open the file

    // Do things with the file
    doSomethingWithTheFile(fh);
    doSomethingElseWithIt(fh);

} // The file is automatically closed here by the destructor

// This has _massive_ advantages:
// 1. No matter what happens,
//    the resource (in this case the file handle) will be cleaned up.
//    Once you write the destructor correctly,
//    It is _impossible_ to forget to close the handle and leak the resource.
// 2. Note that the code is much cleaner.
//    The destructor handles closing the file behind the scenes
//    without you having to worry about it.
// 3. The code is exception safe.
//    An exception can be thrown anywhere in the function and cleanup
//    will still occur.

// All idiomatic C++ code uses RAII extensively for all resources.
// Additional examples include
// - Memory using unique_ptr and shared_ptr
// - Containers - the standard library linked list,
//   vector (i.e. self-resizing array), hash maps, and so on
//   all automatically destroy their contents when they fall out of scope.
// - Mutexes using lock_guard and unique_lock


/////////////////////
// Fun stuff
/////////////////////

// Aspects of C++ that may be surprising to newcomers (and even some veterans):

// You can override private methods!
class Foo {
  virtual void bar();
};
class FooSub : public Foo {
  virtual void bar();  // overrides Foo::bar!
};

// 0, false, NULL are all the same thing!
bool* pt = new bool;
*pt = 0;  // Sets the value points by 'pt' to false.
pt = 0;  // Sets 'pt' to the null pointer. Yes both lines compile without warning.

// '=' != '='
Foo f1 = f2;  // Calls Foo::Foo(const Foo&) or some variant copy constructor.

Foo f1;
f1 = f2;  // Calls Foo::operator=(Foo&) or variant.

```
Futher Reading:

An up-to-date language reference can be found at
<http://cppreference.com/w/cpp>

Additional resources may be found at <http://cplusplus.com>