Doug MacEachern wrote: > On Thu, 2 Aug 2001, Stas Bekman wrote: > > >>Ok, so with the latest workaround the normal run of 'make test' succeeds >>on my machine. But, if I stress-test, I see segfaults: >> >>% ./t/TEST -order=random -times=5 >> > > i had looked at this a few times before and nothing clicked. then i hit > again today writing the lookup_uri2 test. the problem was triggered by > Apache::Test::init_test_pm() which would tie STDOUT only if it wasn't tied > already. but then never untie it. so the next time around an old > request_rec was being used, which had already been destroyed. this is > fixed now and the stress test above passes. > fun to watch it happen, the -order -times stuff is way cool stas! > >
nice, but apparently it's not the only problem. Run: t/TEST -order=random -times=5 with sid 170482341 and there are loads of failures including a core dump. since -order=random is random, it's possible that the tests will pass one time but not another. Therefore I think to get a better chance of various order coverage, we need to run -times=100 and more. This particular sequence that I've reported before still deterministicly fails with a core dump: t/TEST modperl/exit filter/input_msg modperl/exit _____________________________________________________________________ Stas Bekman JAm_pH -- Just Another mod_perl Hacker http://stason.org/ mod_perl Guide http://perl.apache.org/guide mailto:[EMAIL PROTECTED] http://ticketmaster.com http://apacheweek.com http://singlesheaven.com http://perl.apache.org http://perlmonth.com/ --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
