This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: Possible regressions
- From: Tobias Burnus <burnus at net-b dot de>
- To: Jerry DeLisle <jvdelisle at verizon dot net>
- Cc: Fortran List <fortran at gcc dot gnu dot org>
- Date: Fri, 29 Dec 2006 14:25:49 +0100
- Subject: Re: Possible regressions
- References: <4594512C.5010009@verizon.net>
Hi Jerry,
Jerry DeLisle wrote:
> With latest trunk I am getting about 77 Wrong Results failures with
> the cp2k regression tests. The last good results I had were within a
> few days of my post on 12/16/2006. Compilation seems OK. I have not
> had time to investigate further.
"Wrong results": The purpose of these tests is not to check for compiler
regressions, but for cp2k regressions; there are two possibilities for
checking the result: One can test against known good results or one can
compare against previous results.
The latter is what cp2k does; i.e. the current results don't agree with
the previous run. I think the problem is that gfortran's output is
slightly different than expected by "awk" etc. rather than that the
results are indeed off. (I also saw awk error lines [division by zero],
which should not have happend.)
If your previous run was your first run, it is not surprising that you
didn't get "wrong results".
It would be great, if you could find out why cp2k's regression script
stumbles
Tobias,
which has only a dialup connection at the moment.