Bug #61241 for Module-Build: Generated MANIFEST.SKIP contains duplicates, provoking Test::DistManifest warnings

Fri Sep 10 08:41:46 2010 SGRAHAM [...] cpan.org - Ticket created

Subject:

Generated MANIFEST.SKIP contains duplicates, provoking Test::DistManifest warnings

The MANIFEST.SKIP generated if you do not have one, contains duplicate lines, which provokes warnings from Test::DistManifest: t/95-manifest.t ........ 1/4 Duplicate file or mask ^MYMETA\. at /usr/local/share/perl/5.10.1/Test/DistManifest.pm line 211 Duplicate file or mask \bBuild$ at /usr/local/share/perl/5.10.1/Test/DistManifest.pm line 211 Duplicate file or mask \bBuild.bat$ at /usr/local/share/perl/5.10.1/Test/DistManifest.pm line 211 Duplicate file or mask \bBuild.COM$ at /usr/local/share/perl/5.10.1/Test/DistManifest.pm line 211 Duplicate file or mask \bBUILD.COM$ at /usr/local/share/perl/5.10.1/Test/DistManifest.pm line 211 Duplicate file or mask \bbuild.com$ at /usr/local/share/perl/5.10.1/Test/DistManifest.pm line 211 Whether the fault lies in Module::Build or Test::DistManifest I couldn't say, however patching Module::Build seems likely to provide a more effective way to fix the issue since that module version is under the control of the distribution author rather than the end-user. Flagged as unimportant since you can manually edit the MANIFEST.SKIP to remove the duplicates, and it's only warnings rather than test failures.

Mon Jan 24 21:05:19 2011 dagolden [...] cpan.org - Correspondence added

(apologies for a very tardy response) It seems like the examples are all case variations -- I don't think MANIFEST.SKIP is applied case-insensitively, so I think that's a bug in Test::DistManifest not to allow 'BUILD' vs 'build'. I'm going to close this ticket. -- David

Mon Jan 24 21:05:20 2011 The RT System itself - Status changed from 'new' to 'open'

Mon Jan 24 21:05:29 2011 dagolden [...] cpan.org - Status changed from 'open' to 'rejected'

Tue Jan 25 10:56:15 2011 SGRAHAM [...] cpan.org - Correspondence added

Thanks for the response, I've reopened because the previous close was a misunderstanding based on my poor description of the problem. The problem isn't the multiple case checks, rather there's exact duplicate lines: # Avoid Module::Build generated and utility files. \bBuild$ \b_build/ \bBuild.bat$ \bBuild.COM$ \bBUILD.COM$ \bbuild.com$ then further down in the MANIFEST.SKIP there is a repeat: #!end included /usr/local/share/perl/5.10.1/ExtUtils/MANIFEST.SKIP # Avoid configuration metadata file ^MYMETA\. # Avoid Module::Build generated and utility files. \bBuild$ \bBuild.bat$ \b_build \bBuild.COM$ \bBUILD.COM$ \bbuild.com$ ^MANIFEST\.SKIP It's the exact duplicates that cause the issue, rather than the Build vs BUILD regexps. I've attached a generated MANIFEST.SKIP that shows the problem and apologies for the poor description in the first place.

Subject:

MANIFEST.SKIP

Download MANIFEST.SKIP
application/octet-stream 1.1k

Message body not shown because it is not plain text.

Tue Jan 25 10:56:16 2011 The RT System itself - Status changed from 'rejected' to 'open'

Tue Jan 25 11:49:48 2011 dagolden [...] cpan.org - Correspondence added

Subject:	Re: [rt.cpan.org #61241] Generated MANIFEST.SKIP contains duplicates, provoking Test::DistManifest warnings
Date:	Tue, 25 Jan 2011 11:49:04 -0500
To:	bug-Module-Build [...] rt.cpan.org
From:	David Golden <dagolden [...] cpan.org>

On Tue, Jan 25, 2011 at 10:56 AM, Sam Graham via RT <bug-Module-Build@rt.cpan.org> wrote: Show quoted text

> Thanks for the response, I've reopened because the previous close was a > misunderstanding based on my poor description of the problem. > > The problem isn't the multiple case checks, rather there's exact > duplicate lines:

I see. I've reviewed the code and see that the issue is that M::B includes whatever defaults are provided by ExtUtils::Manifest, but on a sufficiently old EU::Manifest, it needs to include the M::B entries, whereas on a newer EU::Manifest, it doesn't. Ultimately, I think that Test::Distmanifest is being unnecessarily strict about duplicate entries in MANIFEST.SKIP. I'll leave this ticket open, but low priority and someone may get around to patching it in the future. (Note to future patcher: in Module::Build::Base::_write_default_maniskip(), filter any duplicate lines before writing the file, ideally, preserving comments.) -- David