Re: Glibc-2.6 (was Re: Error from coreutils CVS version)

[James Youngman]

On 5/18/07, Greg Schafer <address@hidden> wrote:
> Alexander Kahl wrote:
>
> > gcc -std=gnu99  -I.      -march=pentium4 -O2 -pipe -fomit-frame-pointer -s 
> -fgnu89-inline -c utimecmp.c
> > In file included from utimecmp.c:33:
> > utimens.h:2: error: conflicting types for 'futimens'
> > /tools/include/sys/stat.h:370: error: previous declaration of 'futimens' was 
> here
> > make[2]: *** [utimecmp.o] Error 1
> > make[2]: Leaving directory `/mnt/lfs/coreutils-build/coreutils/lib'
> > make[1]: *** [all] Error 2
> > make[1]: Leaving directory `/mnt/lfs/coreutils-build/coreutils/lib'
> > make: *** [all-recursive] Error 1
> >
> > using newest glibc from cvs
>
> Hmmmm, Glibc-2.6 is now released and coreutils-6.9 doesn't compile
> against it because of this error. It seems Glibc has now gained
> futimens() and it disagrees with the Coreutils implementation.

In fact this is probably a gnulib issue.   The prototypes for the
relevant functions in the current Austin Group draft document are:-

int futimens(int fd, const struct timespec times[2]);

int utimensat(int fd, const char *path, const struct timespec
times[2], int flag);
/* flag may contain AT_SYMLINK_NOFOLLOW */


The current gnulib implementation of futimens() is more like
utimensat() than the futimens() above, or so a cursory examination
leads me to believe.

The underlying problem here I suppose is that both gnulib and glibc
are tracking an as-yet-not-final standard document.

James.