Okay here's the deal. In our application, I implement various different libraries, each which like to use their own datatypes. (SDL, OpenGL, and OpenAL are the big three). In addition to that, our application defines its own types, which are nothing more than #define wrappers around the SDL types (we didn't like the naming scheme of the SDL types). So my question here is how do we make all these types friendly with each other?


For example, I want to assign a float to a ALfloat (OpenAL's float) and vice-versa. Do I need to do explicit static_cast calls for each type conversion to eliminate warnings, or is there another trick that we could use with the C preprocessor? (Because there would be A LOT of casting that needs to be done otherwise!).


Also are there any issues we should worry about with word size? All of the standard integer types we use in our app are suffixed with 8, 16, or 32 to define the bit-size, but if I want to assign a Alint (OpenAL's integer) to a uint8, int16 (signed), or uint32, couldn't that potentially be a provlem? Especially considering that this application needs to be cross-platform across 32 and 64 bit architectures, from Linux to MacOS X to (yes, sadly) Windows.


I'm just curious about these issues and how serious they might be. So far everything works across all three systems though. Thanks in advance.

I've never used openAL as well i've read its not the finished item and i question wether you need to use it as SDL has sound capabilties. As for opengl/openAl datatypes, I use gl and never ever use GLint Glfloat etc etc i just dont see the point; if you look at the source you should just see:
typedef float GLfloat;
typedef int GLint;
therefore they are just aliases for built in types

the problem is just that int on machine A might not be the same as on machine B
there exist 16Bit ints, but I dont't know if any non-freak machine uses them. I have a similar problem ( writing code, that runs evrywhere, talks with evry one by udp, even the apple machines which tend to use the wrong byteorder, as we know ( well as wrong as the side of the street, the British folks use to drive, it's all relative ) ) so made up my mind to use apache APR, and the apr_int32_t etc.
But what about new data types which use 64 bit ? Is there any C/C++ standard yet for those long longs and extra doubles? or does the 64 in Athlon64 just stand for the bigger adress space?