Reasons of Using Templates 

Software developers find that many existing codes have to be re-implemented merely because they were implemented for a particular data type. Take the function swap(...) as an example. Basically, it is to swap the value of two variables. In practice, however, a swap function for int, implemented in a strong-type programming language, cannot be re-used to correctly swap two double variables. The same problem appears for data structure. A linked list basically is a list of linked item; but at the programming level, the type of items to be linked does matter.

One convenient solution is to duplicate the original code and modify the parameter type (e.g. replace int to double). This works. But it takes time to make these tedious changes. It is also easy to make mistakes. Later, if the implementation needs to be modified (e.g. fixing a bug), then you have to modify n versions, where n is the number of "clones". Another approach is to have a typedef to specify the data type. The disadvantages of this approach are: 1) you need to edit the header file for each type change. 2) it is not possible to have one typedef to represent two different types simultaneously.

C++ Template is a better solution for the above problems. The swap function can now be implemented independent of a specific data type. This is same as for pair. The implementation is ready for any type, and modification is localized.

Standard Template Library (STL)

STL is a standard set of common data structures and algorithms written in templates. We introduce how to use vector because it is easy to understand and is sometimes more convenient to use than an array.

Reference

Stephen Prata, C++ Primer Plus, Waite Group Press.