Skip Menu |

This queue is for tickets about the Parse-HTTP-UserAgent CPAN distribution.

Report information
The Basics
Id: 112106
Status: resolved
Priority: 0/
Queue: Parse-HTTP-UserAgent

People
Owner: Nobody in particular
Requestors: SREZIC [...] cpan.org
Cc: gregoa [...] cpan.org
AdminCc:

Bug Information
Severity: (no value)
Broken in:
  • 0.39
  • 0.40_01
Fixed in: (no value)



Subject: Tests fail with version 0.9913
t/02-parser.t fails, probably only with newest version.pm: # Failed test 'Frozen data matches parse result for 'Java/1.6.0_21' -> generic_name_version -> 233' # at t/02-parser.t line 73. # Structures begin differing at: # $got->{version} = '1.006021' # $expected->{version} = '1.006000_021' # Looks like you failed 1 test of 754. t/02-parser.t ... Dubious, test returned 1 (wstat 256, 0x100) Failed 1/754 subtests Here's the statistical analysis of test reports on my smoker machines --- negative theta is bad: **************************************************************** Regression 'mod:version' **************************************************************** Name Theta StdErr T-stat [0='const'] 1.0000 0.0000 9893305126815986.00 [1='eq_0.9902'] 0.0000 0.0000 0.45 [2='eq_0.9904'] 0.0000 0.0000 0.25 [3='eq_0.9907'] 0.0000 0.0000 0.00 [4='eq_0.9908'] 0.0000 0.0000 0.25 [5='eq_0.9909'] 0.0000 0.0000 0.53 [6='eq_0.9912'] 0.0000 0.0000 1.08 [7='eq_0.9913'] -1.0000 0.0000 -9432896322410662.00 R^2= 1.000, N= 74, K= 8 ****************************************************************
On 2016-02-17 15:59:53, SREZIC wrote: Show quoted text
> t/02-parser.t fails, probably only with newest version.pm: > > # Failed test 'Frozen data matches parse result for 'Java/1.6.0_21' > -> generic_name_version -> 233' > # at t/02-parser.t line 73. > # Structures begin differing at: > # $got->{version} = '1.006021' > # $expected->{version} = '1.006000_021' > # Looks like you failed 1 test of 754. > t/02-parser.t ... > Dubious, test returned 1 (wstat 256, 0x100) > Failed 1/754 subtests > > > Here's the statistical analysis of test reports on my smoker machines > --- negative theta is bad: > > **************************************************************** > Regression 'mod:version' > **************************************************************** > Name Theta StdErr T-stat > [0='const'] 1.0000 0.0000 9893305126815986.00 > [1='eq_0.9902'] 0.0000 0.0000 0.45 > [2='eq_0.9904'] 0.0000 0.0000 0.25 > [3='eq_0.9907'] 0.0000 0.0000 0.00 > [4='eq_0.9908'] 0.0000 0.0000 0.25 > [5='eq_0.9909'] 0.0000 0.0000 0.53 > [6='eq_0.9912'] 0.0000 0.0000 1.08 > [7='eq_0.9913'] -1.0000 0.0000 -9432896322410662.00 > > R^2= 1.000, N= 74, K= 8 > ****************************************************************
Possibly relevant explanation about the version.pm change: https://gist.github.com/dagolden/9559280
Here's the patch we're applying in Debian currently. Cheers, gregor, Debian Perl Group
Subject: new-version.pm.patch
Description: replace _ with . version->numify() doesn't treat underscores as a tuple separator anymore but simply ignores them. . - raw version: 1.6.0_21 - old version->numify: 1.006000_021 - new version->numify: 1.006021 . By converting underscores to dots we get 1.006021 consistently independent of the version of version.pm. Origin: vendor Bug: https://rt.cpan.org/Public/Bug/Display.html?id=112106 Bug-Debian: https://bugs.debian.org/825611 Forwarded: https://rt.cpan.org/Public/Bug/Display.html?id=112106 Author: gregor herrmann <gregoa@debian.org> Last-Update: 2016-12-22 --- a/t/data/robot/1 +++ b/t/data/robot/1 @@ -177,7 +177,7 @@ parser => 'generic_name_version', version_raw => '1.6.0_21', -version => '1.006000_021', +version => '1.006000021', name => 'Java', generic => 1, --- a/lib/Parse/HTTP/UserAgent.pm +++ b/lib/Parse/HTTP/UserAgent.pm @@ -265,6 +265,13 @@ push @removed, '-' x $rc if INSIDE_VERBOSE_TEST; } + # convert _ to . + # version.pm has changed its interpretation of versions with underlines + # cf. https://bugs.debian.org/825611 + if ( my $rc = $v =~ tr/_/./ ) { + push @removed, '-' x $rc if INSIDE_VERBOSE_TEST; + } + # Finally, be aggressive to prevent dying on bogus stuff. # It's interesting how people provide highly stupid version "numbers". # Version parameters are probably more stupid than the UA string itself.
Path released with v0.40_02