This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: GCC build of HEAD failed for native with your patch on 2004-09-04T15:32:13Z.
James E Wilson <wilson@specifixinc.com> writes:
> On Mon, 2004-09-06 at 19:22, Daniel Berlin wrote:
> > On a completely unrelated note, i'm not sure what speed ppc you have,
>
> I have a single processor 1.8GHz PowerMac G5, which isn't too bad of a
> machine, but I find it a little slow compared to the Pentium4/Athlon64
> machines I do most of my gcc work on. I suspect it might be faster for
> gcc work if I installed linux on a spare partition. I've been meaning
> to try that for a while, but I haven't gotten around to it yet.
>
> Part of the problem here is that since this is a secondary machine for
> me, I usually have no baseline test results, which means I have to do
> two bootstraps and two make checks when I want to test a patch, and that
> really slows things down. That turns testing into an overnight affair.
You may find the attached shell script useful. It compares .sum files
in the current build tree with versions that the automated tester has
created. It's somewhat unpolished, but it does solve the
two-bootstrap problem.
#!/bin/sh
for i in gcc g++ objc ; do
if [ -f gcc/testsuite/$i.sum ] ; then
wget http://gcc.gnu.org/regtest/native-logsum/gcc/testsuite/$i.sum.gz - \
| gunzip | grep '^X\?FAIL' > /tmp/$$.$i
grep '^X\?FAIL' gcc/testsuite/$i.sum | diff -u /tmp/$$.$i - | grep '^[+-]'
rm /tmp/$$.$i
fi
done
for i in libstdc++-v3/testsuite/libstdc++.sum libjava/testsuite/libjava.sum
do
if [ -f powerpc-*/$i ] ; then
NM=`basename $i`
wget \
http://gcc.gnu.org/regtest/native-logsum/powerpc-apple-darwin7.4.0/$i.gz - \
| gunzip | grep '^X\?FAIL' > /tmp/$$.$NM
grep '^X\?FAIL' powerpc-*/$i | diff -u /tmp/$$.$NM - | grep '^[+-]'
rm /tmp/$$.$NM
fi
done