Skip Menu |

This queue is for tickets about the Dist-Zilla-Plugin-Test-EOL CPAN distribution.

Report information
The Basics
Id: 76038
Status: resolved
Priority: 0/
Queue: Dist-Zilla-Plugin-Test-EOL

People
Owner: Nobody in particular
Requestors: dolmen [...] cpan.org
Cc:
AdminCc:

Bug Information
Severity: Critical
Broken in: 0.04
Fixed in: 0.05



Subject: Tests should not be skipped if Test::EOL is missing
DZP::Test::EOL is a release plugin. What is the point of adding a test to the release if the test is simply skipped if a dependency is missing? That's the current problem of DZPTE. So, instead of conditionally loading Test::EOL with Test::Requires, the test should completely fail if Test::EOL is missing. -- Olivier Mengué - http://perlresume.org/DOLMEN
Subject: Re: [rt.cpan.org #76038] Tests should not be skipped if Test::EOL is missing
Date: Sun, 25 Mar 2012 18:37:20 -0500
To: bug-Dist-Zilla-Plugin-Test-EOL [...] rt.cpan.org
From: Caleb Cushing <xenoterracide [...] gmail.com>
2012/3/25 Olivier Mengué via RT <bug-Dist-Zilla-Plugin-Test-EOL@rt.cpan.org>: Show quoted text
> DZP::Test::EOL is a release plugin. > What is the point of adding a test to the release if the test is simply > skipped if a dependency is missing?
Actually it's an Author test not a release test Show quoted text
> That's the current problem of DZPTE. > > So, instead of conditionally loading Test::EOL with Test::Requires, the > test should completely fail if Test::EOL is missing.
I see your point though, I need to confer with Mike Doherty, because this plugin is a part of TestingMania, and I believe he currently would rather stuff skip instead of fail. I was pretty sure thought that installing this plugin actually requires you to have Test::EOL installed... though it appears that I forgot to add it, I'll see about doing that by the end of the week (my workstation at home currently has network issues). The idea was more that people just running the test suite can have it skipped, though perhaps that's a mistake given it's an author test. -- Caleb Cushing http://xenoterracide.com
Subject: Re: [rt.cpan.org #76038] Tests should not be skipped if Test::EOL is missing
Date: Sun, 25 Mar 2012 18:38:09 -0500
To: bug-Dist-Zilla-Plugin-Test-EOL [...] rt.cpan.org
From: Caleb Cushing <xenoterracide [...] gmail.com>
also , per the pod Please report any bugs or feature requests on the bugtracker website https://github.com/xenoterracide/Dist-Zilla-Plugin-Test-EOL/issues
On Sun Mar 25 19:11:35 2012, DOLMEN wrote: Show quoted text
> DZP::Test::EOL is a release plugin. > What is the point of adding a test to the release if the test is
simply Show quoted text
> skipped if a dependency is missing? > That's the current problem of DZPTE. > > So, instead of conditionally loading Test::EOL with Test::Requires,
the Show quoted text
> test should completely fail if Test::EOL is missing.
This test injects a test into the dist that'll be uploaded to PAUSE. If an end-user runs the test, I don't see why it should fail if they don't already have Test::EOL installed - I think skipping is the right behaviour there. If you're working on the module, it is your responsibility to have Test::EOL installed. To that end, I think the Dist::Zilla plugin should require Test::EOL (because, if you don't have it then how will you run your tests prior to release in a meaningful way?). I take it you disagree with the first case there - you want the end- user's tests to fail, right? My question is why.
Subject: Re: [rt.cpan.org #76038] Tests should not be skipped if Test::EOL is missing
Date: Fri, 30 Mar 2012 17:28:38 -0500
To: bug-Dist-Zilla-Plugin-Test-EOL [...] rt.cpan.org
From: Caleb Cushing <xenoterracide [...] gmail.com>
On Fri, Mar 30, 2012 at 5:14 PM, Mike Doherty via RT <bug-Dist-Zilla-Plugin-Test-EOL@rt.cpan.org> wrote: Show quoted text
> I take it you disagree with the first case there - you want the end- > user's tests to fail, right? My question is why.
I think what he's saying is this test will be skipped if AUTHOR_TESTING is not set, so it should be skipped for users anyways (unless they set that) -- Caleb Cushing http://xenoterracide.com
RT-Send-CC: xenoterracide [...] gmail.com
Le 2012-03-31 00:14:48, DOHERTY a écrit : Show quoted text
> This test injects a test into the dist that'll be uploaded to PAUSE. If > an end-user runs the test, I don't see why it should fail if they don't > already have Test::EOL installed - I think skipping is the right > behaviour there.
EOL problems are not the problem of the final user. They are problems of the author that he has to fix before a release. The test must be skipped on the user machine indepently of wether or not Test::EOL is installed or not. This behavior is correct. Show quoted text
> If you're working on the module, it is your responsibility to have > Test::EOL installed. To that end, I think the Dist::Zilla plugin should > require Test::EOL (because, if you don't have it then how will you run > your tests prior to release in a meaningful way?).
That's my point, and here are the problems: - dependencies are wrong in META.yml: Test::EOL and Test::Requires are missing from the dependencies of DZPTE, so they are not installed on the author machine when he installs DZPTE. The test will fail not because the author has wrong EOL, but because he doesn't have Test::Requires - Test::Requires will skip the test if Test::EOL instead of enforcing its presence. So once the author manually installs Test::Requires (because the problem above), the test succeed, but not because all EOL are fine. Instead it is because Test::EOL is missing so nothing has been tested, but a success is still reported. The fix: - drop Test::Requires usage as Test::EOL is mandatory for the author - add Test::EOL in dependencies
RT-Send-CC: xenoterracide [...] gmail.com