Microsoft's <tchar.h> defines _stprintf as 'swprintf' if _UNICODE is defined, and 'sprintf' if not. But these functions take different arguments! In swprintf, the second argument is the buffer size, but sprintf doesn't have this.
Did somebody goof? If so, this is a big one. How can I use _stprintf in my programs, and have them work with and without _UNICODE?
these functions take different arguments!
There are two different versions available with MS compilers. Take a look here. This is in keeping with the ANSI standard. But I think that does not answer your question. I'll skip it for a while and rather tell you how you can have uniformity.
have them work with and without _UNICODE?
You are better off using the 'safe string functions' as per MS recommendations. See this. Use `_stprintf_s' and I think you will get around your problem.
Did somebody goof?
EDITED: I don't think so. I don't have the Rationale handy to give you the answer. I'll post an update when I get my hands on something more concrete. In the meantime look at MSalters' explanation.
A curious thing is MS's C runtime does not claim compatibility with the ISO standard.
Disclaimer: I am not defending the giant of Redmond, only pointing out stuff that strikes me as odd!
You're seeing parallel evolution here. swprintf is a latecomer to standard C, after it was discovered that (A) 8 bits is insufficient for text and (B) you should pass buffer sizes along with buffers. TCHAR is a microsoft idea to unify ASCII and Unicode APIs. They dropped the ball, missing point (B). The proper TCHAR solution should have been to define _stprintf as either swprintf or snprintf.
The solution is then to simply wrap <tchar.h> and do this yourself.