Bug #127877 for Plerd: System perl used in tests

Sat Dec 01 14:12:45 2018 SREZIC [...] cpan.org - Ticket created

Subject:

System perl used in tests

It seems that the test suite fails if the first perl in the user's $PATH is not the perl used for running the build (it looks like a dependency error, but it isn't): ... Can't locate namespace/autoclean.pm in @INC (@INC contains: ... .) at /home/cpansand/.cpan/build/2018120104/File-ChangeNotify-0.29-0/blib/lib/File/ChangeNotify.pm line 5. BEGIN failed--compilation aborted at /home/cpansand/.cpan/build/2018120104/File-ChangeNotify-0.29-0/blib/lib/File/ChangeNotify.pm line 5. Compilation failed in require at /home/cpansand/.cpan/build/2018120104/Plerd-1.7-0/t/../bin/plerdwatcher line 11. BEGIN failed--compilation aborted at /home/cpansand/.cpan/build/2018120104/Plerd-1.7-0/t/../bin/plerdwatcher line 11. Tried to launch a plerdwatcher test instance, but no PID file found. at t/daemon.t line 79. t/daemon.t .. Dubious, test returned 2 (wstat 512, 0x200) No subtests run ...

Sat Dec 08 20:45:27 2018 jmac [...] jmac.org - Correspondence added

Hello, Plerd's author here... I'm not sure what to do about this. Using one perl environment to build and then another one to test sounds to me like a fault with the tester, and a pattern that would fail with any candidate Perl module or program with CPAN dependencies. But the implication here is that this is not the case?

Sat Dec 08 20:45:28 2018 The RT System itself - Status changed from 'new' to 'open'

Sun Dec 09 02:04:59 2018 SREZIC [...] cpan.org - Correspondence added

On 2018-12-08 20:45:27, JMAC wrote: Show quoted text

> Hello, Plerd's author here... > > I'm not sure what to do about this. Using one perl environment to > build and then another one to test sounds to me like a fault with the > tester, and a pattern that would fail with any candidate Perl module > or program with CPAN dependencies. But the implication here is that > this is not the case?

I looked into the plerdwatch script and one problem is the usage of the shebang "#!/usr/bin/env perl". Normally ExtUtils::MakeMaker would take care of rewriting the shebang to the correct perl (the one used to build the distribution), but this does not happen in this case. See https://github.com/Perl-Toolchain-Gang/ExtUtils-MakeMaker/issues/58 about this issue. So the solution would be to use something like "#!/usr/bin/perl" or just "#!perl". The other problem is that the rewritten shebang would be only available after "make install". During "make test" time the rewritten version is already available in blib/script, so you could try to call your script from that location. Alternatively you can call the script using something like "$^X bin/..." --- this would be a workaround to get the right perl interpreter.

Sun Dec 09 11:24:55 2018 jmac [...] jmac.org - Correspondence added

On Sun Dec 09 02:04:59 2018, SREZIC wrote: Show quoted text

> I looked into the plerdwatch script and one problem is the usage of > the shebang "#!/usr/bin/env perl". [...]

This is a very clear explanation of the problem. Thank you! Show quoted text

> Alternatively you can call the script > using something like "$^X bin/..." --- this would be a workaround to > get the right perl interpreter.

If I understand you correctly, the specific change you're suggesting is to line 70 of daemon.t, and turning 'perl' into "$^X"... is that right? (Unless I'm mistaken, this is the only time Plerd's tests invoke perl via the system.)

Sun Dec 09 11:38:39 2018 SREZIC [...] cpan.org - Correspondence added

On 2018-12-09 11:24:55, JMAC wrote: Show quoted text

> On Sun Dec 09 02:04:59 2018, SREZIC wrote: >

> > I looked into the plerdwatch script and one problem is the usage of > > the shebang "#!/usr/bin/env perl". [...]

> > This is a very clear explanation of the problem. Thank you! >

> > Alternatively you can call the script > > using something like "$^X bin/..." --- this would be a workaround to > > get the right perl interpreter.

> > If I understand you correctly, the specific change you're suggesting > is to line 70 of daemon.t, and turning 'perl' into "$^X"... is that > right? (Unless I'm mistaken, this is the only time Plerd's tests > invoke perl via the system.)

I tried diff --git a/t/daemon.t b/t/daemon.t index a433350..3d25ae6 100644 --- a/t/daemon.t +++ b/t/daemon.t @@ -67,7 +67,7 @@ END $config_file_path->spew( $config ); system( - 'perl', + $^X, '-I', "$FindBin::Bin/../lib/", $daemon, "--config=$config_file_path", --- and "make test" was then successful.

Sun Dec 09 11:39:32 2018 jmac [...] jmac.org - Correspondence added

Super duper. I'll give it a whirl.

Sun Dec 09 13:34:52 2018 SREZIC [...] cpan.org - Correspondence added

On 2018-12-09 11:39:32, JMAC wrote: Show quoted text

> Super duper. I'll give it a whirl.

Looks much better now. There are still some test failures on some systems, but this is probably due to other reasons. Closing this issue.

Sun Dec 09 13:34:52 2018 SREZIC [...] cpan.org - Status changed from 'open' to 'resolved'