Re: autodie was not installed on http://www.cpantesters.org/cpan/report/6109ee1e-63a7-11e5-80f8-24112cde3c15
On Sat, Sep 26, 2015 at 12:48 PM, Shlomi Fishwrote: > Hi, Leon! > > On Sat, 26 Sep 2015 12:28:20 +0200 > Leon Timmermans wrote: > > > On Sat, Sep 26, 2015 at 10:33 AM, Shlomi Fish > > wrote: > > > > > On > > > > http://www.cpantesters.org/cpan/report/6109ee1e-63a7-11e5-80f8-24112cde3c15 > > > - there's a missing "autodie" despite the fact that it properly > appears in > > > the > > > test_requires in Build.PL and in build_requires in the META.yml. Please > > > investigate why it is happening. > > > > > > > Proper test_requires support in MB requires MB 0.4200 > > > > Well https://metacpan.org/source/SHLOMIF/App-Timestamper-0.0.5/Build.PL > seems to contain the right code to fallback to build_requires (see line > #63), > however with a check for version 0.4004 instead. Should it be 0.4200 ? Is > it a > bug in Dist::Zilla? The latest version of dzil was used. > I had missed that. In theory it should be fine, though I remember there were a few edge-cases (I forgot the details) that would make me recommend 0.4200 instead. I think I figured out what the problem is though: I suspect the version of CPAN.pm is too old to understand META.yml, and the _build/prereqs contains only test_requires (not build_requires), so it all falls on its face. Possibly this is something MB should take into account. > > > I also noticed that the test was done with perl-5.6.2 - is there still > a > > > case > > > for testing with such an ancient version of perl? > > > > > > > Yes there is. > > Until when are we going to do so? And why? > As long as people feel the need to; because they feel the need to. I'm not sure what the disadvantage is to you, if you really mind the messages you can simple disable CPAN Testers reports for 5.6 or declare in advance it wouldn't work. Leon
Re: autodie was not installed on http://www.cpantesters.org/cpan/report/6109ee1e-63a7-11e5-80f8-24112cde3c15
On Sat, Sep 26, 2015 at 10:23:44AM +0300, Shlomi Fish wrote: > I also noticed that the test was done with perl-5.6.2 - is there still a case > for testing with such an ancient version of perl? Better to ask - why not? What's the harm in doing it? -- David Cantrell | semi-evolved ape-thing Vegetarian: n: a person who, due to poor lifestyle choices, is more likely to get arse cancer than a normal person
Re: autodie was not installed on http://www.cpantesters.org/cpan/report/6109ee1e-63a7-11e5-80f8-24112cde3c15
> Proper test_requires support in MB requires MB 0.4200 That shouldn't matter in this case -- dynamic_config is 0, so the prereqs should be read out of META.*, and autodie is in build-requires, so the installer should be satisfying that requirement. Shlomi: please, for the love of kittens, add a META.json to your distributions, and please only use Module::Build when necessary (which, given you're using Dist::Zilla to package your distribution, should be almost never). On Sat, Sep 26, 2015 at 3:48 AM, Shlomi Fishwrote: > Hi, Leon! > > On Sat, 26 Sep 2015 12:28:20 +0200 > Leon Timmermans wrote: > > > On Sat, Sep 26, 2015 at 10:33 AM, Shlomi Fish > > wrote: > > > > > On > > > > http://www.cpantesters.org/cpan/report/6109ee1e-63a7-11e5-80f8-24112cde3c15 > > > - there's a missing "autodie" despite the fact that it properly > appears in > > > the > > > test_requires in Build.PL and in build_requires in the META.yml. Please > > > investigate why it is happening. > > > > > > > Proper test_requires support in MB requires MB 0.4200 > > > > Well https://metacpan.org/source/SHLOMIF/App-Timestamper-0.0.5/Build.PL > seems to contain the right code to fallback to build_requires (see line > #63), > however with a check for version 0.4004 instead. Should it be 0.4200 ? Is > it a > bug in Dist::Zilla? The latest version of dzil was used. > > > > > > I also noticed that the test was done with perl-5.6.2 - is there still > a > > > case > > > for testing with such an ancient version of perl? > > > > > > > Yes there is. > > Until when are we going to do so? And why? > > Regards, > > Shlomi Fish >
Re: autodie was not installed on http://www.cpantesters.org/cpan/report/6109ee1e-63a7-11e5-80f8-24112cde3c15
Hi, Leon! On Sat, 26 Sep 2015 12:28:20 +0200 Leon Timmermanswrote: > On Sat, Sep 26, 2015 at 10:33 AM, Shlomi Fish > wrote: > > > On > > http://www.cpantesters.org/cpan/report/6109ee1e-63a7-11e5-80f8-24112cde3c15 > > - there's a missing "autodie" despite the fact that it properly appears in > > the > > test_requires in Build.PL and in build_requires in the META.yml. Please > > investigate why it is happening. > > > > Proper test_requires support in MB requires MB 0.4200 > Well https://metacpan.org/source/SHLOMIF/App-Timestamper-0.0.5/Build.PL seems to contain the right code to fallback to build_requires (see line #63), however with a check for version 0.4004 instead. Should it be 0.4200 ? Is it a bug in Dist::Zilla? The latest version of dzil was used. > > > I also noticed that the test was done with perl-5.6.2 - is there still a > > case > > for testing with such an ancient version of perl? > > > > Yes there is. Until when are we going to do so? And why? Regards, Shlomi Fish
Re: autodie was not installed on http://www.cpantesters.org/cpan/report/6109ee1e-63a7-11e5-80f8-24112cde3c15
Hi Karen and all, On Sat, 26 Sep 2015 11:00:33 -0700 Karen Etheridgewrote: > > Proper test_requires support in MB requires MB 0.4200 > > That shouldn't matter in this case -- dynamic_config is 0, so the prereqs > should be read out of META.*, and autodie is in build-requires, so the > installer should be satisfying that requirement. > I see. > Shlomi: please, for the love of kittens, add a META.json to your > distributions, This was done for App-Timestamper for v0.0.6, but will take some time to be done for all the other Dist-Zilla-based distributions that I have. Perhaps I should consider creating Dist::Zilla::PluginBundle::Author::SHLOMIF but last time I looked, creating a PluginBundle was quite an ordeal. > and please only use Module::Build when necessary (which, > given you're using Dist::Zilla to package your distribution, should be > almost never). Since I've preferred to use M::B instead of EU::MM for my distributions before I switched some of them to Dist-Zilla (some of them predate Dist-Zilla), I've found it reasonable to use it after the switch to Dist-Zilla as well. I've also been motivated by this old chromatic quote: http://www.shlomifish.org/humour/fortunes/show.cgi?id=chromatic-perl-reliable-state-of-the-art Recently, I attested to it after a EU::MM update which changed the generated syntax broke the build of one of the Perl-based and EU::MM-using Mageia packages. Regards, Shlomi Fish