c++ template

c++ template

c++ template allows you to declare/implement a generic function/class that has a type parameter T instead of declaring/implementing all functions/classes that have the same function for each data type. And compiler generates functions/classs of corresponding type if necessary according to its actual code. There are 2 kind of c++ template that one is function template and another is class template. When use template to function, it is a function template, and when use template to class, it is a class template. There is no conceptual difference.

function template

Source code

Either ‘typename’ or ‘class’ can be used in the angle bracksets. they are same meaning.
ex) cpp_template_function.cpp

#include <iostream>

using namespace std; 

template <typename T>
T add(T a1, T a2)
{
	return a1 + a2;
}

template <typename t1, typename t2>
t2 sub(t1 a1, t2 a2)
{
	return a1 - a2;
}

int main(void)
{
	cout<<"template int type add 1 + 2 = "<<add<int>(1, 2)<<endl;
	cout<<"template float type add 1.0 + 2.0 = "<<add<float>(1.0, 2.0)<<endl;
	cout<<"template float type sub 1.0 - 2.0 = "<<sub(1, 2.0)<<endl;

	return 0;
}

Output

$ ./cpp_template_function
template int type add 1 + 2 = 3
template float type add 1.0 + 2.0 = 3
template float type sub 1.0 - 2.0 = -1

First output, type T is replaced with int. Second output, type T is replaced with float. Third output, type t1 is replaced with int, type t2 is replaced with float. Angle brackets(<>) and data type between angle brackets can be omitted if the data type to replace type T is clear. For ex). there is no angle brackets at third output. But if you want to call a function which is of specific data type, you need to specify it explicitly. For ex). add(1, 2) will call add function of float data type.

class template

In case of class template, the angle brackets and data type cannot be omitted.

Source code

ex) cpp_template_class.cpp

#include <iostream>

using namespace std; 

template <typename T>
class divide
{
	public :
		T result;

		divide(T a, T b) {
			result = a / b;
		}

		void showResult() {
			cout<<"result = "<<result<<endl;
		}
};


int main(void)
{
	divide<int> aClass(4,3);
	divide<float> bClass(4.0,3.0);

	aClass.showResult();
	bClass.showResult();

	return 0;
}

Output

$ ./cpp_template_class 
result = 1
result = 1.33333 

First output, it’s division operation of int types, so the result is 1 Second output, it’s division operation of float type, so the result is 1.33333

template specialization

Basically, code is generated according to a function or class declared as a template, But “template sepcialization” means not to use the function or class that are implemented as template, instead it is works as you implemented it. Here is an example code about function template specialization.

Source code

ex) cpp_template_function_specialization.cpp

#include <iostream>

using namespace std; 

template <typename T1>
void operation(T1 a1, T1 a2)
{
	cout<<"operation template function is called" <<endl;
	cout<<a1<<" + "<<a2<<" = "<<a1 + a2<<endl;
}

template <>
void operation(char a1, char a2)
{
	cout<<"operation template specialization function is called" <<endl;
	cout<<int(a1)<<" + "<<int(a2)<<" = "<<a1 + a2<<endl;
}


int main(void)
{
	operation<int>(1, 2.5);
	operation<float>(1.0, 2.5);
	operation<char>('1', 2);

	return 0;
}

Output

$ ./cpp_template_function_specialization 
operation template function is called
1 + 2 = 3
operation template function is called
1 + 2.5 = 3.5
operation template specialization function is called
49 + 2 = 51

Third output, template specialization is applied to the ‘char’ type

Side note.

If you use template, source code would be more simple because you don’t need to declare/implement for each type, but this does not mean that actual executable file size is also smaller than declaring/implementing for each type.