Question

I’ve been learning C++ and one thing that I’m not really comfortable with is the fact that datatype sizes are not consistent. Depending on what system something is deployed on an int could be 16 bits or 32 bits, etc.

So I was thinking it might be a good idea to make my own header file with data types like byte, word, etc. that are defined to be a specific size and will maintain that size on any platform.

Two questions. First is this a good idea? Or is it going to create other problems I’m not aware of? Second, how do you define a type as being, say, 8 bits? I can’t just say #define BYTE char, cause char would vary across platforms.

Answer 1

Fortunately, other people have noticed this same problem. In C99 and C++11 (so set your compiler to compatibility with one of those two modes, there should be a switch in your compiler settings), they added the header stdint.h (for C) and cstdint (for C++). If you #include <cstdint>, you get the types int8_t, int16_t, int32_t, int64_t, and the same prefixed with a u for unsigned versions. If your platform supports those types, they will be defined in the header, along with several others.

If your compiler does not yet support that standard (or you are forced by reasons out of your control to remain on C++03), then there is also Boost.

However, you should only use this if you care exactly about the size of the type. int and unsigned are fine for throw-away variables in most cases. size_t should be used for indexing std::vector, etc.