On Tue, 2012-10-16 at 21:30 +0200, Samuel Thibault wrote: > Svante Signell, le Tue 16 Oct 2012 13:29:58 +0200, a écrit :
> No need to define HAVE_GETCWD_GLIBC to 0, people usually only define it > to 1 when the feature is available, thus allowing to just use #ifdef > HAVE_GETCWD_GLIBC. > > > ]) > > #[AC_MSG_FAILURE([test getcwd(NULL,0) failed])]): Add this for > > cross-compiling? > > Better handle cross-compiling yes. I would however not write that it > failed, but just that it could not be tested. OK now? AC_CHECK_FUNCS([getcwd]) AC_RUN_IFELSE([ AC_LANG_PROGRAM([ #ifdef HAVE_GETCWD #include <unistd.h> #endif ],[ char *result = getcwd(NULL,0); if (result == NULL) return 1; ])],[ AC_MSG_RESULT([checking for getcwd(NULL, 0)... yes]) AC_DEFINE([HAVE_GETCWD_GLIBC], 1, [Define to 1 if you have support for `getcwd(NULL,0)`]) ],[ AC_MSG_RESULT([checking for getcwd(NULL, 0)... no]) ],[ AC_MSG_WARN([getcwd(NULL,0)... cannot run test program while cross compiling]) ]) If so, this code snippet could be added to the GNU/Hurd porting pages. This is a very common case of portability issues. -- To UNSUBSCRIBE, email to debian-hurd-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/1350458839.5747.26.ca...@hp.my.own.domain
