Bug #64192 for Text-MicroMason: Safe.pm 2.28 plus bleadperl break T-MM-2.12 (more or less)

Fri Dec 24 08:50:55 2010 ANDK [...] cpan.org - Ticket created

CC:	RGARCIA [...] cpan.org
Subject:	Safe.pm 2.28 plus bleadperl break T-MM-2.12 (more or less)

I just verified on one of my smokers that upgrading Safe from 2.27 to 2.28 broke FERRENCY/Text-MicroMason-2.12.tar.gz. The perl was bleadperl v5.13.4-170-gc99cfaa. Interesting is this table: http://analysis.cpantesters.org/reports_by_field?SUBMIT_xxx=Submit&distv=Text-MicroMason-2.12&field=mod%3ASafe&field=meta%3Aperl&field=mod%3ASafe It shows that there are PASS reports with Safe 2.28. These are all happening with 5.8, 5.10 and 5.12. So there's some additional cause in 5.13 that makes these tests fail. HTH, Cheers,

Fri Dec 24 13:22:40 2010 alan [...] ferrency.com - Correspondence added

Subject:	Re: [rt.cpan.org #64192] Safe.pm 2.28 plus bleadperl break T-MM-2.12 (more or less)
Date:	Fri, 24 Dec 2010 13:22:23 -0500 (EST)
To:	Andreas Koenig via RT <bug-Text-MicroMason [...] rt.cpan.org>
From:	alan [...] ferrency.com

Hello, I am not sure which previous version of Safe had problems; I believe it was 2.27. However, the Safe test suite didn't pass by itself, on the same architectures that MicroMason failed on. I'll take a look at this if I get a chance, but last time I looked at it, everything depended on Safe being fixed in the affected Perls. If they fixed it, MicroMason may not be doing the right thing anymore. Thanks, Alan On Fri, 24 Dec 2010, Andreas Koenig via RT wrote: Show quoted text

> Fri Dec 24 08:50:55 2010: Request 64192 was acted upon. > Transaction: Ticket created by ANDK > Queue: Text-MicroMason > Subject: Safe.pm 2.28 plus bleadperl break T-MM-2.12 (more or less) > Broken in: 2.12 > Severity: Normal > Owner: Nobody > Requestors: ANDK@cpan.org > Status: new > Ticket <URL: https://rt.cpan.org/Ticket/Display.html?id=64192 > > > > I just verified on one of my smokers that upgrading Safe from 2.27 to > 2.28 broke FERRENCY/Text-MicroMason-2.12.tar.gz. > > The perl was bleadperl v5.13.4-170-gc99cfaa. > > Interesting is this table: > > http://analysis.cpantesters.org/reports_by_field?SUBMIT_xxx=Submit&distv=Text-MicroMason-2.12&field=mod%3ASafe&field=meta%3Aperl&field=mod%3ASafe > > It shows that there are PASS reports with Safe 2.28. These are all > happening with 5.8, 5.10 and 5.12. So there's some additional cause in > 5.13 that makes these tests fail. > > HTH, Cheers, > >

Fri Dec 24 13:22:40 2010 The RT System itself - Status changed from 'new' to 'open'

Thu Jan 13 12:20:22 2011 FERRENCY [...] cpan.org - Correspondence added

I looked into this problem, and the basic summary is: Nothing has changed since previous versions of Safe.pm; the failures in Text::MicroMason with Safe 2.28/2.29 and Perl 5.13.* are the same as the failures with Safe 2.27 and Perl 5.13.*. The basic issue in play here is that with Perl 5.13 and later, Safe 2.27 and later, Safe.pm doesn't die the same way it did in earlier Perl versions, in some cases. This was reflected in the fact that Safe.pm 2.27 failed its own test suite on Perl 5.13. Text::MicroMason failed in the exact same way Safe failed. The changes in Safe 2.28 cause it to mark the test that failed in 2.27 as TODO when running under Perl 5.13 or later. However, the behavior it is testing has not changed. MicroMason continues to fail in the same way it always has, because Safe still doesn't behave the same way in Perl versions 5.12 and 5.13. I am inclined to not change any code, but to also mark some tests TODO. I'm not completely convinced MicroMason's Safe mode is particularly safe to begin with; or if it is, it's basically nonfunctional. Alan

Thu Jan 13 13:13:21 2011 FERRENCY [...] cpan.org - Correspondence added

It turns out, I already added SKIP to the Safe-related tests in 2.12, which is why they passed with Safe 2.27. Unfortunately I tied them to Safe version 2.27, and they stopped being SKIPped when 2.28 was released. I've just uploaded Text::MicroMason 2.13, which permanently skips these tests for Perl 5.13.1 and later, independant of the version of Safe in use. Once the changes in Safe's behavior are nailed down, I can update MicroMason's tests to do the right thing. Thanks, Alan

Wed Mar 02 16:02:30 2011 FERRENCY [...] cpan.org - Status changed from 'open' to 'resolved'

Wed Mar 02 16:02:30 2011 FERRENCY [...] cpan.org - Severity Normal changed to Unimportant

Wed Mar 02 16:02:30 2011 FERRENCY [...] cpan.org - Fixed in 2.13 added