You know how to use built-in types, like ints, doubles, chars, etc. You should know some types that are part of the STL, like vector, etc.
Those types that come in the STL are just C++; you can create your own types just like those! Non built-in types are referred to as classes. To create a class, you just use the keyword class, the name of the class, and curly brackets.
CODE
//A class named MyClass
class MyClass {
};
class MyClass {
};
In fact, that is all we need to create variables, pointers, or references to that class.
CODE
MyClass a;
MyClass *b = &a;
MyClass &c = a;
MyClass *b = &a;
MyClass &c = a;
The variable "a" is an object of type MyClass.
Of course, something like that hardly does anything. We need to create members of that class - ways it can be created (called constructors), operations that we can perform on it (called member functions), and data it can hold (called data members).
Data Members
I'll start with data members, because you often need them to use constructors and member functions.
A data member is a piece of data, the same as you use normally. It can be a built-in type, or another class type (i.e. vector). A class can have as many data members as you want or need. To declare a data member, simply put the declaration in the body of the class as follows:
CODE
#include <vector>
class MyClass {
int i;
std::vector<int> v;
};
class MyClass {
int i;
std::vector<int> v;
};
So MyClass has two data members: i and v. Normal rules apply. You can also use pointers and references as usual.
- An important thing to note is that i and v are uninitialized. We'll talk about initializing data members later.
- Also, each object of type MyClass has its own different data. The data is referred to by the same name, but each object of type MyClass has its own value for i and v.
Access Control
Before we move on, I would like to discuss access control. There are three main types of access control in a class.
public: Any part of the program can access it
private: Only members of the class can access it
protected: Private except inherited classes can access it NOT COVERED
To change access control, just use the name followed by a colon, and that will be the access control from then on. The default access control is private. You can bypass this by using the struct keyword instead of class - the default access control is then public.
CODE
#include <vector>
class MyClass {
int i; //Default is private
public:
int integer; //Public
private:
vector v; //Private
protected:
vector vec; //Protected
};
struct MyClass2 {
int i; //Public, as default
private:
int i2; //Private
};
class MyClass {
int i; //Default is private
public:
int integer; //Public
private:
vector v; //Private
protected:
vector vec; //Protected
};
struct MyClass2 {
int i; //Public, as default
private:
int i2; //Private
};
- A common mistake is to assume that struct has a deeper meaning. It doesn't. It's the exact same as using class and following it with "public:".
- To access a public member of a class, use ObjectName.MemberName. So, using the above example:
CODE
MyClass c;
c.integer = 10; //Value of integer is changed
c.i = 10; //Error: private member cannot be accessed
c.integer = 10; //Value of integer is changed
c.i = 10; //Error: private member cannot be accessed
Constructors
Usually, we need to do something when a class is created. We also often need to have different ways to create objects of the same type. That is why we have constructors.
A constructor looks a bit like a function, with a few key differences. For one, it has no return value. Its name must be the same as the class name. For example:
CODE
#include <vector>
class MyClass {
public:
MyClass() {
//Constructor
i = 10;
v.push_back(i);
}
int i;
std::vector<int> v;
};
class MyClass {
public:
MyClass() {
//Constructor
i = 10;
v.push_back(i);
}
int i;
std::vector<int> v;
};
As you can see, you can do anything you want with your data members in the constructor. So now, when we create an object like the following, the operations in the default constructor are called. The default constructor is a constructor with no arguments, and it is called whenever an object of that type is created.
CODE
MyClass my;
Since my is initialized with the default constructor, i should now have a value of 10 and v should have one element with a value of 10.
Also, constructors can take values like a normal function.
CODE
#include <vector>
class MyClass {
public:
//Default constructor
MyClass() {
i = 10;
v.push_back(i);
}
//Constructor that takes an argument
MyClass(int i2) {
i = i2;
v.push_back(i);
}
int i;
std::vector<int> v;
};
//...
MyClass a; //Initializes with default constructor
MyClass b(); //Initializes with default constructor
MyClass c(4); //Initializes with second constructor
class MyClass {
public:
//Default constructor
MyClass() {
i = 10;
v.push_back(i);
}
//Constructor that takes an argument
MyClass(int i2) {
i = i2;
v.push_back(i);
}
int i;
std::vector<int> v;
};
//...
MyClass a; //Initializes with default constructor
MyClass b(); //Initializes with default constructor
MyClass c(4); //Initializes with second constructor
Another important concept is an initializer list. With the above method, your compiler automatically initializes all the data members to their default value, then executes the body of the constructor. This is normally not a big deal, but if you have big classes where initializing them takes a lot of resources, it can be important to do it only once. That is why initializer lists are provided. You can change the default initialization, with the following as an example.
CODE
#include <vector>
class MyClass {
public:
//Default constructor
MyClass() : i(10), v(7,2) { //Initialize i with a value of 10; v is initialized with 2 elements with a value of 7
v.push_back(i);
}
//Constructor that takes an argument
MyClass(int i2) : i(i2) { //Initialize i with the argument i2
v.push_back(i);
}
int i;
std::vector<int> v;
};
class MyClass {
public:
//Default constructor
MyClass() : i(10), v(7,2) { //Initialize i with a value of 10; v is initialized with 2 elements with a value of 7
v.push_back(i);
}
//Constructor that takes an argument
MyClass(int i2) : i(i2) { //Initialize i with the argument i2
v.push_back(i);
}
int i;
std::vector<int> v;
};
To create the initializer list, follow the argument list with a colon, then initialize each data member with whatever values. Multiple initializations are seperated by commas. When you use the initializer lists, you can ensure that the data members are initialized only once.
- If you do not provide a default constructor, your compiler will synthesize one for you. This constructor does nothing except initialize all the data members to their default values.
- If your default constructor is private, objects must be created with another constructor - if there are no public constructors, objects of that type cannot be created. Note that the synthesized constructor is public.
Member Functions
A member function is an operation that can be performed on an object. You can declare member functions in the same way as normal functions, except within the body of the class. Member functions can use data members and can call other member functions. They cannot call constructors, but constructors can call them.
You can use access control for member functions. Public functions can be performed as Object.Function(args). Private (or protected) member functions can only be called by other members.
CODE
#include <vector>
class MyClass {
public:
void setI(int i2) { i = i2; }
int getI() { return i; }
void multiplyBy6() { doubleI(); tripleI(); }
private:
void doubleI() { i *= 2; }
void tripleI() { i *= 3; }
int i;
};
//...
MyClass m;
m.setI(10);
int i = m.getI(); //i==10
m.multiplyBy6(); //i==60
m.doubleI(); //Error: can't access private member function
class MyClass {
public:
void setI(int i2) { i = i2; }
int getI() { return i; }
void multiplyBy6() { doubleI(); tripleI(); }
private:
void doubleI() { i *= 2; }
void tripleI() { i *= 3; }
int i;
};
//...
MyClass m;
m.setI(10);
int i = m.getI(); //i==10
m.multiplyBy6(); //i==60
m.doubleI(); //Error: can't access private member function
The body of a member function does not need to be defined inside the class. It can be defined outside, by declaring it inside and using the scope operator (double colon).
CODE
class MyClass {
public:
int getI();
int i;
};
int MyClass::getI() {
return i;
}
public:
int getI();
int i;
};
int MyClass::getI() {
return i;
}
This operates the same as getI in the previous example, with one difference.
Inline Functions
An inline function is a function that is actually changed into the appropriate code by the compiler. With short functions, this can increase run-time efficiency. Member functions that are defined within a class are inline by default. Member functions defined outside the class and normal functions need the inline keyword.
CODE
inline int sum(int i, int i2) { //Inline function
return i1+i2;
}
class MyClass {
public:
int getI();
int setI(int i2) { i = i2; } //Inline by default
int i;
};
inline int MyClass::getI() { //Inline by keyword
return i;
}
return i1+i2;
}
class MyClass {
public:
int getI();
int setI(int i2) { i = i2; } //Inline by default
int i;
};
inline int MyClass::getI() { //Inline by keyword
return i;
}
- Note that calling a function inline is a request, not a guarantee. Functions that are expanded inline may actually be slower if they are large or if there are certain other conditions. It is up to the compiler to determine whether to expand a function inline or not. It is not a good idea to call all your functions inline; it should only be used on short, simple functions called many times.
That is a basic introduction to classes. Reply if you have questions, comments, etc.
