This is the mail archive of the
mailing list for the libstdc++ project.
Re: Current mainline test failures after last generic locale fixupPATCH
- From: Benjamin Kosnik <bkoz at redhat dot com>
- To: rittle at labs dot mot dot com
- Cc: rittle at latour dot rsch dot comm dot mot dot com, libstdc++ at gcc dot gnu dot org
- Date: Sat, 8 Mar 2003 11:29:18 -0600
- Subject: Re: Current mainline test failures after last generic locale fixupPATCH
- References: <200303080951.h289pdYQ086578@latour.rsch.comm.mot.com>
>Ah, that was fun. I will monitor testresults over the weekend.
Thanks again. The gcc-testresults list just got more interesting...
>Current locale failures that were being hidden (i.e. these are the
>only two remaining of ~100 we have been seeing on such ports):
>FAIL: 22_locale/ctype/cons/char/1.cc execution test
>FAIL: 22_locale/ctype/is/char/3.cc execution test
Yeah. These should not fail if possible. I've notice similar fails for
AIX, darwin. I don't know what's up off hand.
Now that this stuff is actually visible, fails like this (the point of
redoing the testsuite) are finally clear....
>Seems to be failing a variety of x86 machines but no other CPUs (thus
>code generation problem?):
>FAIL: 24_iterators/iterator.cc execution test
This indeed is an x86 codegen issue with -O2. It's been happening for
months, see Phil's regression plot.