| Subject: | t/critic.t, t/pod.t and t/pod_coverage.t should never run at install time |
AUTHOR_TESTING modules and RELEASE_TESTING modules should never run during user installation.
Even if they have the dependencies required for performing tests, it is *STILL* useless to run those tests on a users machine, because
1. It doesn't indicate any useful error
2. And those tests failing do not indicate any useful kind of problem for users
For instance: http://www.cpantesters.org/cpan/report/46504932
Worse, which policies apply the way the tests currently run are system dependent, namely, applying all critic policies a user has installed ( Leading to very unpredictable failures )
And similarly, POD family has been known to change and introduce test failures that a user should be instructed to ignore becuse the failure is completely inconsequential to the usability of the code itself.
And the presence of useless failing tests encourages users not to run useful tests! :)
As per Lancaster Consensus : http://www.dagolden.com/index.php/2098/the-annotated-lancaster-consensus/ , the recommended way to disable these features is by conditionally testing AUTHOR_TESTING or RELEASE_TESTING env variables.
For instance, like this: https://metacpan.org/source/INGY/ByteBeat-0.0.2/t/release-pod-syntax.t
Thanks in advance =).