On 03/17/2010 08:24 AM, Ludovic Courtès wrote: >> Yes, you have to unset the cache variable before the second test, >> ac_cv_have_decl_forkpty. > > Then I think this patch should fix it: > > 2010-03-17 Ludovic Courtès <l...@gnu.org> (tiny change) > > * m4/pty.m4: Unset $ac_cv_have_decl_forkpty before the second > `AC_CHECK_DECL' invocation.
Except that you no longer benefit from caching on platforms that really do lack forkpty; if the cache contains ac_cv_have_decl_forkpty=no, then you are forcing the user to waste time repeating the check to reset it back to no. A better fix would be to test for the existence of both headers (<util.h> and <libutil.h>), then write a single AC_CHECK_DECL with a prefix of: #if HAVE_UTIL_H # include <util.h> #endif #if HAVE_LIBUTIL_H # include <libutil.h> #endif I'll try and write a patch along these lines shortly... -- Eric Blake ebl...@redhat.com +1-801-349-2682 Libvirt virtualization library http://libvirt.org
signature.asc
Description: OpenPGP digital signature
