Bug #48570 for threads: Incorrectly skips tests (UNKNOWN) when it should do an NA

Mon Aug 10 01:14:50 2009 adamk [...] cpan.org - Ticket created

Subject:

Incorrectly skips tests (UNKNOWN) when it should do an NA

The threads module is invalid if the Perl it is installed into does not have thread support. And yet, it will still build and test even when not on a threading Perl. This results in a large number of false-UNKNOWN (which basically says "Testing was unreliable for this, or crashed") results for this module. What it probably should do instead is to check for thread support during the Makefile.PL and return NA if the Perl does not support threads. (This is usually done by returning success exit(0) without generating a Makefile)

Mon Aug 10 08:15:19 2009 jdhedden [...] cpan.org - Correspondence added

RT-Send-CC:

perl5-porters [...] perl.org

On Mon Aug 10 01:14:50 2009, ADAMK wrote: Show quoted text

> The threads module is invalid if the Perl it is installed into does not > have thread support. > > And yet, it will still build and test even when not on a threading Perl. > > This results in a large number of false-UNKNOWN (which basically says > "Testing was unreliable for this, or crashed") results for this module. > > What it probably should do instead is to check for thread support during > the Makefile.PL and return NA if the Perl does not support threads. > > (This is usually done by returning success exit(0) without generating a > Makefile)

The original philosophy for this module is that 'threads' and its ilk are allowed to be installed in an unthreaded Perl. threads.pm checks if the Perl is threaded, and 'die's if it is not. This allows apps to be written such that they can be supported on both threaded and unthreaded Perls using a construct similar to this: my $i_am_threaded; BEGIN { $i_am_threaded = eval 'use threads; 1'; } Backward compatibility concerns also dictate that this behavior not be modified.

Mon Aug 10 08:15:21 2009 The RT System itself - Status changed from 'new' to 'open'

Mon Aug 10 08:15:25 2009 jdhedden [...] cpan.org - Status changed from 'open' to 'rejected'

Mon Aug 10 14:52:22 2009 jdhedden [...] cpan.org - Correspondence added

RT-Send-CC:

perl5-porters [...] perl.org

On Mon Aug 10 01:14:50 2009, ADAMK wrote: Show quoted text

> The threads module is invalid if the Perl it is installed into does not > have thread support. > > And yet, it will still build and test even when not on a threading Perl. > > This results in a large number of false-UNKNOWN (which basically says > "Testing was unreliable for this, or crashed") results for this module.

Of further consideration, I believe it would be appropriate for me to add at least one test that will pass on a non-threaded Perl. This will allow the module to be installed on non-threaded Perls (which is required for backward compatibility), and to 'pass' at the same time. This should eliminate those ambiguous UNKNOWN results.

Mon Aug 10 14:52:23 2009 The RT System itself - Status changed from 'rejected' to 'open'

Mon Aug 10 15:01:05 2009 adam [...] ali.as - Correspondence added

CC:	adamk [...] cpan.org
Subject:	Re: [rt.cpan.org #48570] Incorrectly skips tests (UNKNOWN) when it should do an NA
Date:	Tue, 11 Aug 2009 05:00:41 +1000
To:	bug-threads [...] rt.cpan.org
From:	Adam Kennedy <adamkennedybackup [...] gmail.com>

Removing spurious UNKNOWN results via a single passing test would be the next best way to fix this, yes. The Padre "Percentage likely to install" dependency graph appreciates this. :) Adam K 2009/8/11 Jerry D. Hedden via RT <bug-threads@rt.cpan.org> Show quoted text

> <URL: https://rt.cpan.org/Ticket/Display.html?id=48570 > > > On Mon Aug 10 01:14:50 2009, ADAMK wrote:

> > The threads module is invalid if the Perl it is installed into does not > > have thread support. > > > > And yet, it will still build and test even when not on a threading Perl. > > > > This results in a large number of false-UNKNOWN (which basically says > > "Testing was unreliable for this, or crashed") results for this module.

> > Of further consideration, I believe it would be appropriate for me to add > at least one test that will pass on a non-threaded Perl. This will > allow the module to be installed on non-threaded Perls (which is > required for backward compatibility), and to 'pass' at the same time. > This should eliminate those ambiguous UNKNOWN results. >

Tue Jan 19 14:24:59 2010 jdhedden [...] cpan.org - Status changed from 'open' to 'resolved'

Tue Jan 19 14:25:03 2010 jdhedden [...] cpan.org - Fixed in 1.74 added

Tue Jan 19 14:25:03 2010 jdhedden [...] cpan.org - Fixed in 1.73 deleted