c++ - help understanding differences between function and class templates - c++

Question

I'm trying to implement a C++ program that, given a list of elements, prints out unique elements from the list.
I know C a lot better than C++, but I am starting out on C++ practically (coding) only now.
I've only read up on C++ concepts what templates are and I'm comfortable with function templates but I just read up on class templates and I think I'm getting confused as to where to use which one, as applied to the scenario below.
Here's what I've written so far (*note that the function isUnique is supposed to be doing something else but I'm writing just verifiable actions inside it for now):
cppArrays.h
#include <iostream>
#include <cstdlib>
#include <vector>
#include <string>
using namespace std;
template <class T> class cpparray{
private:
int size;
vector<T> elems;
public:
cpparray(int);
~ cpparray();
int isUnique(T arr);
};
template <class T> cpparray<T>::cpparray(int size)
{
vector<T> elems(size);
cout << "Object created with size " << elems.size()<< "\n"<< endl;
}
template <class T> cpparray<T>::~cpparray()
{
cout << "Object del\n" << endl;
}
template <class T> int cpparray<T>::isUnique(T arr)
{
return arr.size();
}
cppArrays.cc
#include "cppArrays.h"
int main()
{
cpparray<int> a(10) ;
//a.push_back(1);
//a.push_back(2);
//cout << a.size() << a.begin() << a.end() << endl;
int b = isUnique(a);
return 0;
}
Details:
[1] I'm trying to use templates, because I want my vector to be able to be instantiated with any data type - char/float/int.
[2] I realize that in using a class template, by calling
cpparray<int>a(10);
I end up declaring an object a of class "cpparray" whose type is Integer. Is this correct?
If yes, then a.push_back(1) wouldn't work because it doesn't refer to the member variable but the object itself and hence it's understandable I'll get a compile time error saying no member named push_back in cpparray.
But this is making it even more difficult for me to
[1] understand when to use class template as opposed to function template,
[2] How do I initialize this vector in the template class and work with it, given my objective?

Answer 1

Your constructor should use initialization list to init members,
like this:
template <class T> array<T>::array(int sz) : size(sz), elems(sz)
{
cout << "Object created with size " << elems.size()<< "\n"<< endl;
}
What you did is to declare the vector as local in the constructor, init it to size, and the local vector was destroyed at the end of block.

Answer 2

Use a class template when you need a generic type that has compile-time variable properties. Template parameters can be types and constants, e.g.
template<typename T, size_t Size>
class MyArray {
T elements_[Size];
public:
MyArray() {}
// ...
};
Use a function template when wish to write a generic function that can be applied to various types/parameters:
#include <cstdio>
#include <iostream>
template<size_t BufSize, typename... Args>
int strprintf(char(&buf)[BufSize], const char* fmt, Args&&... args)
{
static_assert(BufSize > 0, "Buffer too small");
static_assert(BufSize < (1 << 31), "Buffer too large");
return snprintf(buf, BufSize, fmt, std::forward<Args>(args)...);
}
int main() {
char buf[16];
int printed = strprintf(buf, "hello world %s so long", "oversized load");
std::cout << buf << "\n";
}
http://ideone.com/SLUQX3
The above is an example of how you can replace one of those old vsnprintf forwarding printf-type functions; doing all the legwork at compile time makes it drastically more efficient.
BufSize can be deduced by the compiler because the type of buf is char[16]; it can capture the source by reference, the type is char and the array size - the template variable - is 16.
It's also possible to have a templated member function of a templated class:
template<typename T>
class Foo {
T t_;
public:
Foo() : t_() {}
Foo(const T& t) : t_(t) {}
template<typename RhsT>
bool is_same_size(const RhsT& rhs) {
return t_.size() == rhs.size();
}
};
This example will only work for instances when both T and RhsT have a size() member function, leading to the following:
Foo<vector<int>> fvi;
Foo<list<double>> fld;
fvi.is_same_size(fld); // fine
Foo<int> fi;
fvi.is_same_size(fi); // compiler error