This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: crash in 1.0.3a fortran compilation
- To: Joe Buck <jbuck at synopsys dot com>
- Subject: Re: crash in 1.0.3a fortran compilation
- From: Andi Kleen <ak at muc dot de>
- Date: 24 Jul 1998 02:01:15 +0200
- Cc: lindahl at cs dot virginia dot edu (Greg Lindahl), law at cygnus dot com, egcs at cygnus dot com
- References: <199807231448.HAA15033@atrus.synopsys.com>
Joe Buck <jbuck@synopsys.com> writes:
> > That's fine. Is there any reason that there shouldn't be a second
> > test-suite which compiles gobs of code in an attempt to ferret out
> > goofs. There are plenty of people like me with tons of machines who'd
> > be happy to run the "fat suite".
>
> I suggest that the best "fat suite" is production code. For C/C++,
> regularly build a whole Linux distribution (Red Hat/Debian/name your
> favorite). For Fortran, regularly compile the popular public domain
> math libraries (which for some reason all have names ending in PACK
> or PAC).
The problem is that it is hard to regression test a complete distribution
for additional failures caused by miscompiles. It is only useful to detect
bugs that crash the compiler, or break something obvious.
Math code has the advantage that you can at least do simple test runs
and compare the results.
-Andi