[Python-apps-team] Bug#939756: mercurial: FTBFS on hppa - test-hghave.t timed out

[Julien Cristau]

Control: tag -1 - patch
Control: notforwarded -1

On Fri, Feb 07, 2020 at 04:32:08PM +0100, Julien Cristau wrote:
> Control: tag -1 upstream patch
> Control: forwarded -1 https://phab.mercurial-scm.org/D8092
> 
> On Sun, Sep 08, 2019 at 09:51:42AM -0400, John David Anglin wrote:
> > See for example this build log:
> > https://buildd.debian.org/status/fetch.php?pkg=mercurial&arch=hppa&ver=5.1.1-1&stamp=1567811172&raw=0
> > 
> > I had a successful build on hppa by increasing the timeout settings in
> > debian/rules:
> > override_dh_auto_test:
> > ifeq (,$(filter nocheck,$(DEB_BUILD_OPTIONS)))
> > # test-pull-pull-corruption tries to execute two hg commands at the
> > # same time to test for corruption.  It does so by running an hg
> > # command in the background, sleeping one second and running the
> > # second command.  In architectures with slow buildds, one second is
> > # not enough and the second test starts before the first one has
> > # started printing.  This is hacky and we just enhance the hack by
> > # incrementing the sleep time in those architectures.
> > ifneq ($(or $(ARCH_IS_MIPS), $(ARCH_IS_KFREEBSD-i386), $(ARCH_IS_ARMEL), $(ARCH_
> > IS_SPARC)),)
> >                 sed -i.deb-backup -e 's/sleep 1/sleep 2/' $(CURDIR)/tests/test-p
> > ull-pull-corruption.t
> > endif
> > 
> >         http_proxy='' dh_auto_test -- TESTFLAGS="--verbose --timeout 12000 --slowtimeout 36000 $(PARALLEL_TEST_JOBS) --blacklist $(CURDIR)/debian/mercurial.test_blacklist"
> >         file-rename 's/\.deb-backup$$//' $(CURDIR)/tests/*
> >         # run blacklisted tests but ignore their results
> >         -cd tests && python run-tests.py --verbose `grep ^test ../debian/mercurial.test_blacklist`
> > endif
> > 
> > Here is full log:
> > https://buildd.debian.org/status/fetch.php?pkg=mercurial&arch=hppa&ver=5.1.1-1&stamp=1567919941&raw=0
> > 
> Thanks for reporting this.  Rather than increasing the timeout I'm
> hoping we can just make the test faster.  Sent a patch upstream for
> feedback.
> 
So my change (https://www.mercurial-scm.org/repo/hg/rev/e48a996d12bc) got into
5.4, but we're still seeing test-hghave.t time out:
https://buildd.debian.org/status/fetch.php?pkg=mercurial&arch=hppa&ver=5.5-1&stamp=1597658853&raw=0

I don't know what would make it so much slower on hppa than everywhere else...

The latest log at
https://buildd.debian.org/status/fetch.php?pkg=mercurial&arch=hppa&ver=5.5-2&stamp=1597702396&raw=0
also shows some other test failures:
> Failed test-bisect.t: timed out
> Failed test-copies-chain-merge.t#compatibility: timed out
> Failed test-copytrace-heuristics.t: timed out
> Failed test-glog-beautifygraph.t: timed out
> Failed test-graft.t: timed out
> Failed test-hghave.t: timed out
> Failed test-largefiles-update.t: timed out
> Failed test-merge-combination.t: timed out
> Failed test-mv-cp-st-diff.t: timed out
> Failed test-obsmarker-template.t: timed out
> Failed test-obsolete-bundle-strip.t: timed out
> Failed test-obsolete.t: timed out
> Failed test-push-race.t#strict: output changed
> Failed test-push-race.t#unrelated: output changed
> Failed test-shelve.t#phasebased: output changed
> Failed test-shelve2.t#phasebased#abortcommand#continuecommand: timed out
> Failed test-shelve2.t#phasebased#abortcommand#continueflag: timed out
> Failed test-shelve2.t#phasebased#abortflag#continuecommand: timed out
> Failed test-shelve2.t#phasebased#abortflag#continueflag: timed out
> Failed test-shelve2.t#stripbased#abortcommand#continuecommand: timed out
> Failed test-shelve2.t#stripbased#abortcommand#continueflag: timed out
> Failed test-shelve2.t#stripbased#abortflag#continuecommand: timed out

Cheers,
Julien