On 10/13/2009 09:23 AM, Jim Meyering wrote:
Definitely.
In fact, there should be no uses of AC_CHECK_FUNCS in modules/* files.
Of course, there*could* be justification if someone requires
that a function check be conditional, but if there's enough logic,
then you can argue it belongs in a .m4
Paolo Bonzini wrote:
configure.ac:
+AC_CHECK_FUNCS([usleep])
AC_CHECK_FUNCS_ONCE is better maybe?
Definitely.
In fact, there should be no uses of AC_CHECK_FUNCS in modules/* files.
Of course, there *could* be justification if someone requires
that a function check be conditional, but if
On 10/13/2009 09:47 AM, Bruno Haible wrote:
... or if someone augments LIBS, does an AC_CHECK_FUNCS, and restores LIBS
afterwards.
Better use a .m4 in that case too, I'd say.
Paolo
Jim Meyering wrote:
In fact, there should be no uses of AC_CHECK_FUNCS in modules/* files.
Of course, there *could* be justification if someone requires
that a function check be conditional
... or if someone augments LIBS, does an AC_CHECK_FUNCS, and restores LIBS
afterwards.
Bruno, Paolo,
Eric Blake ebb9 at byu.net writes:
+/* Sleep long enough to notice a timestamp difference on the file
+ system in the current directory. */
+static void
+nap (void)
+{
...
+delay = 200;
+}
+ usleep (delay);
Well I feel stupid. POSIX 2008 withdrew usleep in favor of
Eric Blake ebb9 at byu.net writes:
Date: Tue, 15 Sep 2009 15:54:43 -0600
Subject: [PATCH] fchdir: improve use of replacement functions
* m4/fchdir.m4 (gl_FUNC_FCHDIR): Set appropriate witnesses.
* m4/sys_stat_h.m4 (gl_SYS_STAT_H_DEFAULTS): Add REPLACE_FSTAT.
This patch broke fstat on
Eric Blake ebb9 at byu.net writes:
looking at the assembly, it looks like mingw fails to set errno to
EINVAL, but just blindly returns without sleeping).
Correction - on mingw, usleep(100) returns EINVAL without touching errno,
rather than the POSIX behavior of returning -1 and setting