How to avoid problems with size_t and int types in 64bit C++ builds?

Question

Today I made a 64bit build of my project for the first time. Basically it compiled, linked and ran ok, except for warnings complaining about incompatibility between the new, 64bit size_t type and the simple int type. This mostly occurs in situations like this in my code:

void func(std::vector<Something> &vec)
{
    int n = vec.size();
    for (int i=0; i < n; i++)
    {
        ....vec[i]....
    }
}

This is quite easy to fix, and I read an article saying one should rather use size_t or ptrdif_t as loop indices. But what can I do in a situation like this?

void outsideLibraryFunc(int n);

void func(std::vector<Something> &vec)
{
    int n = vec.size();
    outsideLibraryFunc(n);
}

I can't change the outside library's function declaration, which expects an argument of type int, and I need to pass it the number of the vector elements. What can I do other than disabling the compiler warnings?

Answer 1

Do an explicit cast to int, e.g.

void outsideLibraryFunc(int n);

void func(std::vector<Something> &vec)
{
    outsideLibraryFunc(static_cast<int>(vec.size()));
}

It doesn't eliminate any of the potential problems with converting size_t to int, but it does tell the compiler that you're doing the conversion on purpose, and it won't warn you about it.

Answer 2

Cast it? Seriously if you can't change the external library, there is not much you can do. To be extra safe check for overflow.