Package: autoconf
Version: 2.60a-1
Severity: serious
Tags: upstream patch
Justification: Is my understanding that this failed test could potentially
break *many* packages
Please include this patch from upstream:
http://www.mail-archive.com/autoconf@gnu.org/msg15087.html
Here is some recent irc mentions of it:
[23:59:05] (des) which severity should a bug in autoconf that breaks other
packages get?
[23:59:26] (des) i.e a basic check which is wrong?
[00:00:32] (vorlon) possibly serious, if you're right
[00:00:50] (des) vorlon:
http://www.mail-archive.com/autoconf@gnu.org/msg15087.html
[00:01:16] (des) that patch is *not* in debian as of today
[00:01:38] (vorlon) only breaks with -Werror?
[00:01:43] (des) and is the reason why I spent my hole day trying to pinpoint a
FTBFS in shadow (not reported
yet)
[00:02:12] (vorlon) new in 2.60a, I guess?
[00:02:32] (vorlon) yeah, please mark that one serious; if it breaks shadow I'm
sure it'll break other packages
[00:02:41] (des) yeah, they rewrote the test in 2.60a apparently
[00:02:53] -*- vorlon nods
I understand that this could potentionally break a lot of packages that
autoreconf at build time, therefore the severity.
Hope to help,
Damián Viano(Des).
P.s.: sorry for the sloppy report, I'm really tired, and been chansing this one
the hole day.
--- lib/autoconf/headers.m4 2006-08-15 16:24:42.000000000 +0000
+++ lib/autoconf/headers.m4.des 2006-09-04 02:52:30.000000000 +0000
@@ -529,7 +529,7 @@
#if defined S_ISSOCK && defined S_IFREG
extern char c4[S_ISSOCK (S_IFREG) ? -1 : 1];
#endif
-]])], ac_cv_header_stat_broken=yes, ac_cv_header_stat_broken=no)])
+]])], ac_cv_header_stat_broken=no, ac_cv_header_stat_broken=yes)])
if test $ac_cv_header_stat_broken = yes; then
AC_DEFINE(STAT_MACROS_BROKEN, 1,
[Define to 1 if the `S_IS*' macros in <sys/stat.h> do not
