I've found that the issue was that Shell::Perl tests were relying in a
behavior of $err output of run() that changed according to this bit of
the diff between 0.41_02 and 0.41_04
(
http://search.cpan.org/diff?from=IPC-Cmd-0.41_02&to=IPC-Cmd-0.41_04):
- my $err = $?;
- if ( $@ and ref $@ and $@->isa( ALARM_CLASS ) ) {
- $ok = 0;
- $err = $@; # the error code is an expired alarm
+ my $err;
+ unless( $ok ) {
+ ### alarm happened
+ if ( $@ and ref $@ and $@->isa( ALARM_CLASS ) ) {
+ $err = $@->(); # the error code is an expired alarm
+
+ ### another error happened, set by the dispatch sub
+ } else {
+ $err = $self->error;
+ }
The test unknowingly did:
for my $switch ( '-v', '--version' ) {
my ( $ok, $err, $full_buf, $out_buf, $err_buf ) = run( command =>
[ @pirl, $switch ] );
ok( $ok, "'pirl $switch' run ok" );
is( $err, 0, 'exited with 0' ); <<<<< HERE
cmp_deeply( $out_buf, [ re(qr/\AThis is pirl/) ], 'printed version info' );
cmp_deeply( $err_buf, [], 'no output to STDERR' );
}
where
is( $err, 0, 'exited with 0' );
went wrong, because now $err returns undef.
According to the docs, $err was previously an error code and now is an
error message (which is supposed to hold something only if $ok is
false). So if that's the correct interpretation, I will be changing
this test to be:
ok( !$err, 'no error' );
Thanks a lot, Andreas, for your precision that made much easier to
find where the issue came from.
Show quoted text> Binary search shows it happened between 0.41_02 and 0.41_04.
Regards,
Adriano Ferreira
On Sun, Oct 19, 2008 at 4:56 AM, Jos I. Boumans via RT
<bug-IPC-Cmd@rt.cpan.org> wrote:
Show quoted text> <URL:
http://rt.cpan.org/Ticket/Display.html?id=40157 >
>
>
> On Oct 19, 2008, at 6:01 AM, Andreas Koenig via RT wrote:
>
>> Sun Oct 19 00:01:17 2008: Request 40157 was acted upon.
>> Transaction: Ticket created by ANDK
>> Queue: IPC-Cmd
>> Subject: IPC-Cmd-0.42 breaks Shell::Perl 0.0017
>> Broken in: 0.42
>> Severity: Important
>> Owner: Nobody
>> Requestors: ANDK@cpan.org
>> Status: new
>> Ticket <URL:
http://rt.cpan.org/Ticket/Display.html?id=40157 >
>>
>>
>> As the subject line says, upgrading from 0.40 to 0.42 breaks the
>> version.t test of Shell-Perl-0.0017.
>>
>> It's a pity this was not detected earlier during the 0.41_* cycle.
>>
>> Binary search shows it happened between 0.41_02 and 0.41_04.
>>
>> I have not tried to determine the cause,
>
> That's an odd bug, but of course i'll look into it.
>
> Thanks for letting me know!
>
> --
> Jos Boumans
>
> "Never ask a man what computer he uses. If it's a Mac, he'll
> tell you. If it's not, why embarrass him?" - Tom Clancy
>
> CPANPLUS
http://cpanplus.sf.net
>
>
>
>
