Tom Lane wrote: > Bruce Momjian <[EMAIL PROTECTED]> writes: > > Andrew Francis wrote: > >> How about avoiding #define altogether, and: > >> - Always use pgrename/pgunlink instead of rename/unlink > > > We could do it but we have avoided that for cases where Unix would just > > be a pass-through. > > To put that in a more positive light: we like to think that our code is > Posix-compliant and runs in a Posix-compliant environment. We're not > thrilled about introducing non-Posix-isms for the convenience of one > platform ... especially if there's no easy way to enforce that the > nonstandard coding convention be used. > > Back on track: if rename() does exist under Windows then my idea is > unreliable. Any other thoughts? How about #including <io.h> in port.h > (for Windows only of course) before we #define these things?
Sure, should we do that? I see 12 mentions of io.h in the code, and we already include some win32 includes in port.h. -- Bruce Momjian | http://candle.pha.pa.us [EMAIL PROTECTED] | (610) 359-1001 + If your life is a hard drive, | 13 Roberts Road + Christ can be your backup. | Newtown Square, Pennsylvania 19073 ---------------------------(end of broadcast)--------------------------- TIP 9: the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match