So, where's the user pulling '%s' from then? What if the user gives '%q'? (Not defined for strftime at all) Or '-%s-'? (Also has %s in it). Or '%k' (Which comes from the same package as '%s' comes from)? Or '%P' (A GNU extension, likely to work on Linux, but not a real Unix)?

I think you should either say that the user gets whatever his current platform support (meaning that a given format will be the same for different applications on the same platform that use strftime), or you should write your own 'strftime' implementation (meaning that your application behaves the same on different platforms). Both solutions have different advantages when it comes to interoperatbility. But writing your own strftime implementation takes a lot more time, and is unlikely to be as fast as the one from libc. Whether or not that's an issue, I cannot judge.