Bug #45074 for PAR-Packer: pp -M should trump the core-module-rejection of -B (Was: PAR doesn't include Time::Local)

Thu Apr 16 03:12:56 2009 MSCHILLI [...] cpan.org - Ticket created

Subject:

PAR doesn't include Time::Local

If you create a PAR archive from a script like # test.pl use Time::Local; print timelocal(0,0,12,01,01,2009), "\n"; via the pp command $ pp -o time.par -p test.pl then the resulting .par file doesn't contain Time::Local: $ unzip -l time.par Archive: time.par Length Date Time Name -------- ---- ---- ---- 0 04-16-09 00:08 lib/ 0 04-16-09 00:08 script/ 428 04-16-09 00:08 MANIFEST 215 04-16-09 00:08 META.yml 262 04-16-09 00:08 script/main.pl 86 04-16-09 00:08 script/test.pl -------- ------- 991 6 files And not even an explicit reference helps: $ pp -o time.par -M Time::Local -p test.pl has the same effect. Is there anything special about Time::Local?

Thu Apr 16 03:29:22 2009 MSCHILLI [...] cpan.org - Correspondence added

Granted, Time::Local is a core module, so -B will fix it, but this even happens if you've installed Time::Local from CPAN. Shouldn't pp archive the CPAN package?

Thu Apr 16 03:29:24 2009 MSCHILLI [...] cpan.org - Status changed from 'new' to 'open'

Thu Apr 16 05:56:44 2009 nj88udd02 [...] sneakemail.com - Correspondence added

Subject:	Re: [rt.cpan.org #45074] PAR doesn't include Time::Local
Date:	Thu, 16 Apr 2009 11:56:21 +0200
To:	bug-PAR [...] rt.cpan.org
From:	Steffen Mueller <nj88udd02 [...] sneakemail.com>

Hi Michael, Michael_Schilli via RT wrote: Show quoted text

> Queue: PAR Ticket <URL: > https://rt.cpan.org/Ticket/Display.html?id=45074 > > > Granted, Time::Local is a core module, so -B will fix it, but this > even happens if you've installed Time::Local from CPAN. Shouldn't pp > archive the CPAN package?

I know that given how it's implemented, it'll simply skip any core modules unless -B is present. Here are some ideas how it could be improved (using Module::CoreList): - The user could be allowed to specify the target perl version (which defaults to the current one). - PAR::Packer then does NOT skip those core modules which have a higher version than what Module::CoreList considers the version that shipped with the target perl. This would generally degrade to the current behaviour *unless* you installed newer versions of core modules from CPAN. I think in case of upgraded core modules, this is actually not what users want most of the time. Most module upgrades aren't necessarily required for proper operation and shipping a load of core modules with EVERY .par archive is almost always a bad idea. Therefore, I'm not sure whether or how this should be changed. Adding more options to pp is generally a bad thing (just look at the synopsis of perldoc pp) unless they're rarely needed and the defaults are good in 95% of all cases. Further opinions on this would be very welcome, but as usual, patches, too! Cheers, Steffen

Thu Apr 16 08:50:34 2009 RSCHUPP [...] cpan.org - Correspondence added

On Thu Apr 16 05:56:44 2009, nj88udd02@sneakemail.com wrote: Show quoted text

> I know that given how it's implemented, it'll simply skip any core

Steffen, you're to close to subject and hence your thinking is overly complex :) I think that all Michael wants is: If I say "pp -M Foo ..." then simply pack Foo.pm, no matter if it's a core module or not. That doesn't need a new option (The Horror! The Horror!). Of course, that means that any module that Foo.pm depends on will also be packed. If someone needs even more fine grained control about what to pack or not to pack he's free to construct the par file by hand (it's simply a zip file after all).

Wed Aug 26 10:03:12 2009 smueller [...] cpan.org - Severity Important changed to Normal

Wed Aug 26 10:03:13 2009 smueller [...] cpan.org - Broken in 0.994 added

Wed Aug 26 10:03:13 2009 smueller [...] cpan.org - Broken in 0.992 deleted

Wed Aug 26 10:41:12 2009 smueller [...] cpan.org - Queue changed from PAR to PAR-Packer

Wed Aug 26 10:42:04 2009 smueller [...] cpan.org - Subject changed from 'PAR doesn't include Time::Local' to 'pp -M should trump the core-module-rejection of -B (Was: PAR doesn't include Time::Local)'