W dniu 03.01.2012 23:55, Buddy Burden pisze:
> Guys,
>
> Okay, I got it down to this, but I can't seem to narrow it down
> further. Time to call in the big guns.(*)
>
> This code:
>
> #! /usr/bin/perl
>
> use Test::More 0.88;
> use Test::Command;
>
> stderr_like(q{ perl -le 'print STDERR "STDERR points to ", readlink
> "/proc/self/fd/2"' },
> qr/STDERR points to/, 'should always pass');
>
> done_testing;
>
> produces this output (as expected):
>
> [tyr:~] ./test.t
> ok 1 - should always pass
> 1..1
>
> However, _this_ code:
>
> #! /usr/bin/perl
>
> use Test::More 0.88;
> use Test::Command;
>
> use System::Command;
>
> stderr_like(q{ perl -le 'print STDERR "STDERR points to ", readlink
> "/proc/self/fd/2"' },
> qr/STDERR points to/, 'should always pass');
>
> done_testing;
>
> produces this:
>
> [tyr:~] ./test.t
> STDERR points to /dev/pts/0
> not ok 1 - should always pass
> # Failed test 'should always pass'
> # at ./test.t line 8.
> # ''
> # doesn't match '(?-xism:STDERR points to)'
> 1..1
> # Looks like you failed 1 test of 1.
>
> which is entirely unexpected. Can anyone tell me:
>
> 1) What's going on here?
> 2) Is there any way to work around it?
> 3) Is there perhaps another way to do what I'm trying to do that won't
> trigger this?
>
> To clarify #3, I'm not actually using System::Command, I'm using
> Git::Repository (which uses Git::Repository::Command, which uses
> System::Command). I'm using Git::Respository so I can do this test:
>
> lives_ok { Git::Repository->new( work_tree => $packages_dir ) } 'basic
> sanity check: packages dir is in a Git repo';
>
> which works great, but perhaps there's another way to do this without
> using G::R, which is obviously causing me issues.
>
> TIA for any suggestions you guys have! (And happy 2012, everybody!)
I've noticed the same thing, but with STDOUT and with Test::Output
and/or FastCGI.
There is a hack in System::Command for Catalyst which causes this.
See https://rt.cpan.org/Public/Bug/Display.html?id=73614
--
Grzegorz Dzięgielewski
http://jabbas.eu
