Bruno Haible wrote:
> Jim Meyering wrote:
>> Setting the envvars, LIB_CLOCK_GETTIME, LIB_EACCESS or LIB_NANOSLEEP
>> could cause a configure-time and/or build-time malfunction.
>
> Likewise with LIB_GETHRXTIME. How about this proposed fix?
Good catch.
However, please make these two lines adjacent:
(pulling the AC_SUBST "up")
LIB_GETHRXTIME=
AC_SUBST([LIB_GETHRXTIME])
Since AC_SUBST can be used anywhere, it's slightly more
maintainer-friendly to keep these two references to LIB_GETHRXTIME close.
I nearly moved the AC_SUBST lines in the changes I made recently,
but opted not to: I didn't want to mix bug-fix and clean-up fixes.
> 2009-10-17 Bruno Haible <br...@clisp.org>
>
> * m4/gethrxtime.m4 (gl_GETHRXTIME): Initialize LIB_GETHRXTIME here...
> (gl_PREREQ_GETHRXTIME): ... not here.
>
> --- m4/gethrxtime.m4.orig 2009-10-18 15:43:25.000000000 +0200
> +++ m4/gethrxtime.m4 2009-10-18 15:42:56.000000000 +0200
> @@ -1,4 +1,4 @@
> -# gethrxtime.m4 serial 8
> +# gethrxtime.m4 serial 9
> dnl Copyright (C) 2005, 2006, 2008, 2009 Free Software Foundation, Inc.
> dnl This file is free software; the Free Software Foundation
> dnl gives unlimited permission to copy and/or distribute it,
> @@ -12,12 +12,14 @@
> AC_REQUIRE([gl_USE_SYSTEM_EXTENSIONS])
> AC_REQUIRE([gl_XTIME])
> AC_CHECK_DECLS([gethrtime], [], [], [#include <time.h>])
> + LIB_GETHRXTIME=
> case $ac_cv_have_decl_gethrtime,$gl_cv_arithmetic_hrtime_t in
> yes,yes) ;;
> *)
> AC_LIBOBJ([gethrxtime])
> gl_PREREQ_GETHRXTIME;;
> esac
> + AC_SUBST([LIB_GETHRXTIME])
...
