This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: GSL: fail in make check
- From: Brian Gough <bjg at network-theory dot co dot uk>
- To: "James W. Haefner" <jhaefner at biology dot usu dot edu>
- Cc: gsl-discuss at sources dot redhat dot com
- Date: Fri, 16 Nov 2001 09:17:46 +0000 (GMT)
- Subject: Re: GSL: fail in make check
- References: <3BF45D9F.6F3B6C6F@biology.usu.edu>
James W. Haefner writes:
> I just compiled 1.0 and ran "make check". I found many errors in herm(4)
> and herm(10). An example is below:
> FAIL: herm(4), eigenvalue(2,1), real, val/asc (9.74766689339348974e+127
> observed vs 0 expected)
> FAIL: herm(10), eigenvalue(1,1), real, unsorted
> (2.16415272160173273e+110 observed vs -8.71992293261537888e-304
> expected)
>
> I have RedHat 7.1 on a 1GHz AMD, 256M RAM.
Thanks for the bug report. What version of gcc do you have? This
looks like a bug that I have seen on the Alpha (see the MACHINES file)
with gcc-2.95/2.96. I came to the conclusion that it was a compiler
optimisation problem in that case -- for example the same code worked
fine without optimisation.
Brian