Bug #84146 for DBI: DBI broken on perl 5.17.10

Fri Mar 22 13:29:04 2013 zefram [...] fysh.org - Ticket created

Subject:	DBI broken on perl 5.17.10
Date:	Fri, 22 Mar 2013 17:28:50 +0000
To:	bug-DBI [...] rt.cpan.org
From:	Zefram <zefram [...] fysh.org>

DBI fails some Gofer cache tests on perl 5.17.10. Failure allegedly arrived about the same time as the hash randomisation core change (each hash object now has its own iteration order). Several modules require fixes of the form s/keys/sort keys/ for this change. I tried adding a bunch of key sorting operations of this nature to the DBI code, but the test failures remained. -zefram

Fri Mar 22 14:32:51 2013 TIMB [...] cpan.org - Correspondence added

Thanks Zefram. I'd actually just noticed the DBI was mentioned in that context in p5p and I'm building a blead perl now...

Fri Mar 22 14:32:53 2013 The RT System itself - Status changed from 'new' to 'open'

Fri Mar 22 16:33:35 2013 TIMB [...] cpan.org - Correspondence added

Fixed in r15585. Release to follow. FYI the diff was: --- lib/DBI/Gofer/Serializer/Storable.pm (revision 15575) +++ lib/DBI/Gofer/Serializer/Storable.pm (working copy) @@ -46,6 +46,7 @@ sub serialize { my $self = shift; local $Storable::forgive_me = 1; # for CODE refs etc + local $Storable::canonical = 1; # for go_cache my $frozen = nfreeze(shift); return $frozen unless wantarray; return ($frozen, $self->{deserializer_class}); I imagine that, since it's "just a cache", I reasoned that the cost of canonicalization wasn't worth it. Up until now I was right.

Sun Mar 24 09:13:08 2013 TIMB [...] cpan.org - Correspondence added

Released. Thanks.

Sun Mar 24 09:13:10 2013 TIMB [...] cpan.org - Status changed from 'open' to 'resolved'