Bug #114586 for Audio-TagLib: Random test failures with perl 5.18 and later (hash randomization?)

Sun May 22 16:26:04 2016 SREZIC [...] cpan.org - Ticket created

Subject:

Random test failures with perl 5.18 and later (hash randomization?)

My smoker machines (linux debian jessie) show random test failures: # Failed test at t/TagLib_ID3v2_FrameList.t line 75. # 'TIT2title' # doesn't match '(?^:^TPE1.*?artist$)' # method sortedInsert(item) and getItem(1) failed # Looks like you failed 1 test of 31. t/TagLib_ID3v2_FrameList.t ....................... Dubious, test returned 1 (wstat 256, 0x100) Failed 1/31 subtests This seems to happen only with perl >= 5.18, so maybe it is related to hash randomization (or some other change introduced in this perl version).

Sun May 22 16:35:19 2016 GLEACH [...] cpan.org - Taken

Sun May 22 16:36:21 2016 GLEACH [...] cpan.org - Status changed from 'new' to 'open'

Sun May 22 16:36:21 2016 GLEACH [...] cpan.org - Severity Important added

Thu Jul 28 17:00:44 2016 GLEACH [...] cpan.org - Status changed from 'open' to 'new'

Sat May 13 05:23:33 2017 SREZIC [...] cpan.org - Broken in 1.67 added

Sat May 13 05:47:13 2017 SREZIC [...] cpan.org - Correspondence added

On 2016-05-22 16:26:04, SREZIC wrote: Show quoted text

> My smoker machines (linux debian jessie) show random test failures: > > # Failed test at t/TagLib_ID3v2_FrameList.t line 75. > # 'TIT2title' > # doesn't match '(?^:^TPE1.*?artist$)' > # method sortedInsert(item) and getItem(1) failed > # Looks like you failed 1 test of 31. > t/TagLib_ID3v2_FrameList.t ....................... > Dubious, test returned 1 (wstat 256, 0x100) > Failed 1/31 subtests > > > This seems to happen only with perl >= 5.18, so maybe it is related to > hash randomization (or some other change introduced in this perl > version).

To reproduce the failure one can use a loop like this: for i in $(seq 1000); do echo $i; env PERL_HASH_SEED=$i /opt/perl-5.26.0-RC1/bin/prove -b t/TagLib_ID3v2_FrameList.t || break; done Usually (for other hash randomization-related problems) such a loop stops always at the same PERL_HASH_SEED number. In this case this number seems to differ from compilation to compilation, and sometimes the loop takes a long time to hit a PERL_HASH_SEED number causing the failure.