Bug #40287 for Scriptalicious: 04-fork test unreliable

Thu Oct 23 01:50:34 2008 ANDK [...] cpan.org - Ticket created

Subject:

04-fork test unreliable

I just ran the test suite on a fast and unloaded machine about twenty times in a row with all tests OK, then came this one: t/04-fork............04-fork.t: ERROR: Command `/usr/local/perl-5.10/bin/perl -Mlib=lib t/loopback.pl' killed by signal 13 program output: # Looks like you planned 17 tests but ran 8. # Looks like your test exited with 1 just after 8. Dubious, test returned 1 (wstat 256, 0x100) Failed 9/17 subtests I downloaded all 49 or so failing cpan testers reports and all of them failed on 04-fork. HTH,

Sun Oct 26 01:56:23 2008 ANDK [...] cpan.org - Correspondence added

Looking closer at cpan testers statistics I realize that with Test-Harness-2.64 this failure cannot be reproduced. Indeed I let a loop run for about 12 hours running 'make test' under Test::Harness 2.64 and it did not show up again. At the moment I cannot draw a conclusion from that but I think it's interesting.

Sun Oct 26 01:56:31 2008 ANDK [...] cpan.org - Status changed from 'new' to 'open'

Sun Oct 26 02:28:06 2008 SAMV [...] cpan.org - Correspondence added

Yeah, I've had a few random failures with this and IO::Plumbing where I even went so far as to get system traces from people where it fails. I'm not sure if it's a real failure or just a quirk of the mechanisms that try to print the test run results at END { } time. Sadly I don't really have an answer...