Hello Lars and Daniel, thanks much for the links and explanations!
On Thu, Jun 16, 2005 at 01:37:34PM +0300, Lars Wirzenius wrote: > The C standard guarantees (see page 166, 7.1.3, "Reserved identifiers", > if you have a copy) that the standard headers do not define identifiers > that the C standard does not explicitly declare as defined by the > standard, reserved for future versions of the standard, or reserved to > the implementation. "struct timespec" and "nanosleep" are not such > identifiers. Indeed. I've overseen that time.h is also defined by C99. I've changed my opinion regarding whether this is a libc6-dev bug. However, I still have a problem. My intention is to use -std=c99 and define macros like _BSD_SOURCE in order to document all portability issues at the top of the files. After I defined _POSIX_C_SOURCE to 200201L, I'm able to compile the file without problems. However, neither SUSv3, nor Linux man page say anything about it. That is why I used to think that struct timespec and nanosleep MUST be available after a bare #include <time.h>. Does POSIX specify whether the availability can be controlled with a macro? Should Linux man page be updated to mention _POSIX_C_SOURCE? BTW, defining _POSIX_SOURCE, which is described in the glibc documentation, didn't work for me. Is it a bug, or does "POSIX.1" mean POSIX 1990 only? With kind regards, Baurzhan. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
