[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [GNU Autoconf 2.59] testsuite: 41 failed
From: |
Stepan Kasal |
Subject: |
Re: [GNU Autoconf 2.59] testsuite: 41 failed |
Date: |
Thu, 18 Aug 2005 10:15:08 +0200 |
User-agent: |
Mutt/1.4.1i |
Hello,
On Wed, Aug 17, 2005 at 10:17:19AM -0700, address@hidden wrote:
> Only this one test failed. Host is a SunBlade 1000 running Solaris 9,
> with fairly recent Sun jumbo patch installed.
>
> SunOS marten.isi.edu 5.9 Generic_118558-10 sun4u sparc SUNW,Sun-Blade-1000
>
> Autoconf seems to be the most recent version and I couldn't find
> any patches that seemed related. Let me know if you need further info.
you can get the development release 2.59b from
ftp://alpha.gnu.org/gnu/autoconf/
Or you can even try the CVS version, see
http://savannah.gnu.org/cvs/?group=autoconf
I'm not able to guess what caused the problem. There is a question whether
the bug is deterministic.
If you still have the tree where you tested, could you try
cd tests ; ./testsuite 41
?
If you want to do further investigation, the following can help:
After unpacking a tarball, do:
./configure; make all check
"make all" is quick, "make check" is very long. But you don't
have to wait for the end. As soon as you see that the slow test suite
runc, you can break it. This is the easiest way to create the testsuite
script and the wrappers.
Then you can run ./testsuite -k -v 41 tu run the test with verbose output
and then cs testsuite.dir/041 to look at the working directory of the test.
The test number changes in releases, for example, in the CVS version it
is 75. Use
./testsuite -l | less
to find out the number of the test.
A guess: perhaps the cause is the ancient aclocal 1.4? Would the problem
disappear if you upgrade to Automake 1.9.6? (aclocal is part of the
Automake package.)
HTH,
Stepan Kasal