Bug #87444 for Unexpected: CPANtesting.pm makes no sense

Tue Jul 30 11:10:00 2013 mst [...] shadowcat.co.uk - Ticket created

Subject:	CPANtesting.pm makes no sense
Date:	Tue, 30 Jul 2013 15:09:44 +0000
To:	bugs-Unexpected [...] rt.cpan.org
From:	Matt S Trout <mst [...] shadowcat.co.uk>

The two referenced reports from BinGOs' smoker are both UNKNOWNs because your module requires a higher version of perl than that smoker's running. Both have an error message in them. UNKNOWN is correct for 'prerequisites not met'. So it seems exceedingly strange to nix that smoker - especially in a way that will do nothing but generate a different sort of UNKNOWN. I don't understand. -- Matt S Trout - Shadowcat Systems - Perl consulting with a commit bit and a clue http://shadowcat.co.uk/blog/matt-s-trout/ http://twitter.com/shadowcat_mst/ Email me now on mst (at) shadowcat.co.uk and let's chat about how our CPAN commercial support, training and consultancy packages could help your team.

Tue Jul 30 11:54:26 2013 Support [...] RoxSoft.co.uk - Correspondence added

On Tue Jul 30 11:10:00 2013, mst@shadowcat.co.uk wrote: Show quoted text

> The two referenced reports from BinGOs' smoker are both UNKNOWNs because > your module requires a higher version of perl than that smoker's running.

True. Bingos smokers used to be misconfigured such that testing with a Perl version below the minimum Perl version produced a bogus pass anyway. Recently he has reconfigured them to produce UNKNOWN reports instead. The advice on the CPAN Testing wiki is that if Build.PL exits 0 without creating a Build file then no report will be filed. It is in effect the CPAN Testing off switch. As currently configured Bingos smokers do no do that. As can be seen from this report df276fba-f57c-11e2-8c80-50d7c5c10595 exiting zero from Build.PL without a Build file still creates an UNKNOWN report, which is contrary to the advice given on the wiki and has nothing to do with dependencies or minimum Perl version. I have already developed a better approach to dealing with this particular issue as can be seen in the current version of CPANTesting.pm in which the sleep 10 while 1 has been removed (Unexpected v0.5.6). This leaves us with the issue of what should a smoker do when it tries to test a distribution with a version of Perl below the minimum Perl version required by that distribution. My opinion is that no report should be filed which is what I was aiming for with the earlier version CPANTesting.pm. Since, given Bingos current configuration, that cannot happen I will settle for a bogus pass instead. Filing an UNKNOWN report makes no sense. There is nothing unknown here. When an UNKNOWN report appears in my RSS feed I click through it to see if it's genuine or not. If genuine then I alter my code accordingly. If bogus then I alter my toolchain to prevent the same bogus report from wasting my time again in the future.

Tue Jul 30 11:54:26 2013 The RT System itself - Status changed from 'new' to 'open'

Tue Jul 30 11:54:26 2013 Support [...] RoxSoft.co.uk - Status changed from 'open' to 'resolved'

Tue Jul 30 11:54:27 2013 Support [...] RoxSoft.co.uk - Taken

Tue Jul 30 15:41:49 2013 ether [...] cpan.org - Correspondence added

On 2013-07-30 08:54:26, PJFL wrote: Show quoted text

> sleep 10 while 1 ...

This, by the way, was exceedingly rude. You should have made some effort to talk to BinGOs or the cpantesters-discuss mailing list, rather than being so deliberately destructive to his infrastructure.

Tue Jul 30 17:43:20 2013 Support [...] RoxSoft.co.uk - Correspondence added

On Tue Jul 30 15:41:49 2013, ETHER wrote: Show quoted text

> You should have made some effort to talk to BinGOs or the > cpantesters-discuss mailing list

I have tried discussing issues about misconfigured smokers on that list before and found it to be frustrating and pointless. Since Bingos deliberately configured his smokers to file an UNKNOWN report when Build.PL exits 0 without creating a build file, see http://www.mail-archive.com/cpan-testers-discuss@perl.org/msg03151.html I saw no point in arguing with him. Usually I can find a solution to these problems without being so aggressive and that is what I have now done, although it took me two attempts and the current solution is sub-optimal. I find it interesting that you show concern for disruption to Bingos infrastructure but none for the accuracy of the CPAN Testers database nor the disruption to module authors workflow caused by the bogus UNKNOWN/FAIL reports.

Tue Jul 30 17:43:20 2013 Support [...] RoxSoft.co.uk - Fixed in v0.5.6 added

Tue Jul 30 20:01:11 2013 mst [...] shadowcat.co.uk - Correspondence added

Subject:	Re: [rt.cpan.org #87444] CPANtesting.pm makes no sense
Date:	Wed, 31 Jul 2013 00:00:51 +0000
To:	Peter Flanigan via RT <bug-Unexpected [...] rt.cpan.org>
From:	Matt S Trout <mst [...] shadowcat.co.uk>

On Tue, Jul 30, 2013 at 05:43:20PM -0400, Peter Flanigan via RT wrote: Show quoted text

> I find it interesting that you show concern for disruption to Bingos > infrastructure but none for the accuracy of the CPAN Testers database > nor the disruption to module authors workflow caused by the bogus > UNKNOWN/FAIL reports.

I don't. He tests a huge number of modules for a huge number of people. This is extremely useful to many members of the perl community. You, on the other hand, have a bunch of CPAN modules that nobody else depends on, none of which I've ever seen in a customer's codebase. This is irrelevant to me, except for the fact that you've used Moo::Role in a way that works for you and I'm glad it helped. Since your code contributes a donation to the community, I thank you for it whether I ever use it or not. However, on a purely utilitarian basis, his smokers are far more important than your inconvenience to those of us who are involved with a substantial percentage of the core "usual" CPAN stack. So others' prioritisation of that should be considered rational and to be expected, whether you agree with them or not. -- Matt S Trout - Shadowcat Systems - Perl consulting with a commit bit and a clue http://shadowcat.co.uk/blog/matt-s-trout/ http://twitter.com/shadowcat_mst/ Email me now on mst (at) shadowcat.co.uk and let's chat about how our CPAN commercial support, training and consultancy packages could help your team.

Wed Jul 31 06:29:23 2013 LEONT [...] cpan.org - Correspondence added

On Tue Jul 30 15:41:49 2013, ETHER wrote: Show quoted text

> On 2013-07-30 08:54:26, PJFL wrote:

> > sleep 10 while 1 ...

> > This, by the way, was exceedingly rude.

IMO it's not just rude, it was outright abusive behavior. Show quoted text

> You should have made some > effort to talk to BinGOs or the cpantesters-discuss mailing list, > rather than being so deliberately destructive to his infrastructure.

Indeed. He was trying to help you, even if you didn't find him helpful. Asking him to stop that would be reasonable, smacking him in the face is not. Leon

Wed Jul 31 06:40:57 2013 Support [...] RoxSoft.co.uk - Correspondence added

BinGOs, I apologise for the tier gruber. It was uncalled for. This one issue where the toolchain indicates to the smoker "file no report please" brings out the arsehole in me. Report pops up in the RSS feed, I click through, scan the report detail, recognise it as one of the ones I've seen before and disagree with, I take it personally and then start thrashing about looking for a technical solution when the problem is my inability to communicate with someone without the snark. Even now one side off my brain is picking Matt's last post apart and constructing a response whilst the other side is telling me to write this and learn to accept the small number incorrect reports as most everyone does.