Nicholas Clark wrote: > This little run doesn't make sense: > > On Tue, Oct 30, 2007 at 11:26:00PM +0000, Steve Hay wrote: >> Automated smoke report for 5.8.8 patch 32025 >> Summary: PASS > > On Wed, Oct 31, 2007 at 11:25:00PM +0000, Steve Hay wrote: >> Automated smoke report for 5.8.8 patch 31996 >> Summary: PASS > > So why the earlier source? > > On Thu, Nov 01, 2007 at 11:25:00PM +0000, Steve Hay wrote: >> Automated smoke report for 5.8.8 patch 32025 >> Summary: FAIL(F) > > And then why does it fail once it gets back to 32025?
It's the same reason that all my other smokes have been useless lately, e.g. my most recent blead smoke was only patch 32072 instead of more like 32225: http://www.nntp.perl.org/group/perl.daily-build.reports/2007/11/msg51322 .html Basically, my rsync has somehow restored itself to the broken version 2.6.9 despite my having carefully downgraded it to 2.6.6, so none of the rsync's having been working. See: http://www.nntp.perl.org/group/perl.perl5.porters/2007/09/msg129036.html It's therefore been smoking whatever source trees happen to be in place, and a different tree is used on Wednesday because it does a PERL_OLD_COPY_ON_WRITE smoke that evening. Not sure why the second attempt at 32025 failed when the first one had passed. Looks like I'll have to upgrade my Test-Smoke after all to get Abe's fix for this rsync madness. (I chickened out before because some tests failed.) Thanks for pointing this out.
