I believe if you write everything in 1 main function, you are eventually going to blow the stack because unused local variables are not popped off.

The final decision to inline is based on the compiler, unless u use __forceinline.

My rule of thumb in inlining are those very small functions that are used very often. (e.g. matrix classes or those small libs).

While we are on the subject, what are the diffs/benefits between _cdecl and _stdcall?

AFAIK, _cdecl results in slightly larger code because the caller cleans up the parameters passed. If we need variable arguments, we must use _cdecl.

So _stdcall is better??

The MSDN docs says _stdcall requires a function prototype. So what happens if I don''t provide one?. ( I looked at the disassembly, nothing unusual there) Anyone knows?