On Sat, Feb 2, 2013 at 6:59 AM, Serguei Trouchelle <s...@cpan.org> wrote: > Andreas Koenig wrote: >>>>> >>>>> If you can point us at a couple of example failure reports, maybe >>>>> someone here can figure out what's going wrong. In my experience it is >>>>> almost always the case that things like this are caused by an error in >>>>> the distribution under test. >>>> >>>> Here's the most recent example of such failures: >>>> >>>> http://www.cpantesters.org/cpan/report/cb01432c-6737-11e2-992e-e7c00a4a7996 >> >> I have contacted Nigel and he sent me his CPAN.pm configuration and I >> cannot see anything wrong in it. Especially my suspect about >> trust_test_report_history was wrong, trust_test_report_history is off. > > At least two of Nigel's smoker setups generate errors because they use NFS, > and everything that uses IPC::Run3 fails because File::Temp cannot guarantee > successful unlinking on NFS. Probably, some modules that use File::Temp, may > fail too. > In my opinion, these results are useless for module authors. I, as an > author, do not benefit from reports like this: > http://www.cpantesters.org/cpan/report/835621d4-6b45-11e2-a82e-ca470f55919d > as it has nothing to do with the module itself. It's not even a problem with > architecture (this module has no problems running on arm), it's a problem > with smoker setup.
If a real world user can also have similar setup using NFS, then this is a legitimate setup, isn't it? Isn't this exactly the same situation as Windows vs Linux, where Windows might not support all the features Linux does? Are those only test failures, or is the module (partially) broken on a system using NFS? If the module does not fully work over NFS then shouldn't it recognize the situation and disallow those features? If only the tests fail on NFS, shouldn't those be skipped? Gabor
