autoconf-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [GNU Autoconf 2.60] testsuite: 95 failed


From: Paul Eggert
Subject: Re: [GNU Autoconf 2.60] testsuite: 95 failed
Date: Fri, 07 Jul 2006 11:00:02 -0700
User-agent: Gnus/5.1008 (Gnus v5.10.8) Emacs/21.4 (gnu/linux)

Ralf Wildenhues <address@hidden> writes:

>       * tests/torture.at (Configuring subdirectories): Another
>       instance of configure that needs CONFIG_SITE set.

How about this more-general patch instead?  It tries to kill off
CONFIG_SITE for the entire test.  I'd rather not track down the
CONFIG_SITE usages one by one.

Perhaps we should even turn off CONFIG_SITE altogether for tests.

Anyway, I installed this (before reading your email).  This
patch assumes you didn't apply my earlier patch.

2006-07-07  Paul Eggert  <address@hidden>

        * tests/torture.at (Configuring subdirectories): Set CONFIG_SITE
        more globally, since the 2006-06-30 patch didn't suffice.  Problem
        reported by Keith Marshall.  Also, don't bother with builddir2,
        since it shouldn't be needed any more.

--- torture.at  23 Jun 2006 14:56:30 -0000      1.65
+++ torture.at  7 Jul 2006 17:56:16 -0000       1.67
@@ -794,6 +794,21 @@ AT_CHECK([aclocal --version || exit 77],
 # It should understand configure.ac.
 AT_CHECK([[grep '1.[01234]' stdout && exit 77]], [1], [ignore])
 
+# We don't want to run this test if this shell doesn't support
+# `unset'.
+AT_CHECK([
+if (FOO=FOO; unset FOO) >/dev/null 2>&1; then
+  exit 0
+else
+  exit 77
+fi
+])
+
+# Set CONFIG_SITE to a nonexistent file, so that there are
+# no worries about nonstandard values for 'prefix'.
+CONFIG_SITE=no-such-file
+export CONFIG_SITE
+
 # The contents of `inner/', and `inner/innermost/'.
 AS_MKDIR_P([inner/innermost])
 
@@ -862,12 +877,9 @@ prefix=/usr/local
 ])
 
 # Make sure precious variables and command line options are properly
-# passed, even when there are duplicates.  Run another directory so
-# that if the user has set config.site to have a config.cache, then
-# the two runs don't have colliding caches.
-AS_MKDIR_P([builddir2])
-AT_CHECK([cd builddir2 && ../configure --prefix /bad --prefix /good INNER=bad 
INNER=good], 0, [ignore])
-AT_CHECK([cat builddir2/inner/innermost/config], 0,
+# passed, even when there are duplicates.
+AT_CHECK([cd builddir && ../configure --prefix /bad --prefix /good INNER=bad 
INNER=good], 0, [ignore])
+AT_CHECK([cat builddir/inner/innermost/config], 0,
 [INNER=good
 srcdir=../../../inner/innermost
 top_srcdir=../../../inner
@@ -882,6 +894,8 @@ AT_CHECK([autoreconf inner], [], [], [ig
 AT_CHECK([cd inner && autoreconf configure.in], [], [], [ignore])
 AT_CHECK([autoreconf inner/configure.in], [], [], [ignore])
 
+unset CONFIG_SITE
+
 AT_CLEANUP
 
 








reply via email to

[Prev in Thread] Current Thread [Next in Thread]