On Fri, Jul 12, 2013 at 11:59:14PM +0100, Dominic Hargreaves wrote: > On Wed, Jan 27, 2010 at 10:06:39AM -0800, Robert Stone wrote: > > On Wed, Jan 27, 2010 at 12:31:44PM +0200, Niko Tyni wrote: > > > > > > Sorry for the lack of action on this. > > > > No problem at all, thanks for looking into this. > > > > > > What's wrong with Devel::Peek:SvREFCNT(\%m) ? > > > > "\%m" is a reference constructor, so SvREFCNT() ends up > > reporting the reference count of that newly constructed SV. The > > situation can be illustrated with this small script: > > > > #!/usr/bin/perl > > use strict; > > use warnings; > > use Devel::Peek; > > my %m = (a => 'b'); > > my $mr = \%m; > > Devel::Peek::Dump($mr); > > print Devel::Peek::SvREFCNT(\%m), "\n"; > > > > From the Dump you can see that %m (the PVHV) has a refcount of > > 2, but "Devel::Peek::SvREFCNT(\%m)" reports 1. The reason for this can > > be illustrated by breaking that call up a little bit. One would > > normally expect that: > > > > Devel::Peek::SvREFCNT(\%m) > > > > should be equivalent to > > > > my $tmp = \%m; > > Devel::Peek::SvREFCNT($tmp) > > > > which unsurprisingly reports a refcount of 1 for $tmp. Nothing changes > > when $tmp is an immediate, unlabeled value on the argument stack to > > SvREFCNT() and I think that's expected and correct behavior. > > > > The proposed patch solves this difficulty by always passing the > > argument from Perl by reference and then unconditionally dereferencing > > to access the target value. > > Hi Robert, > > Sorry for the delay in pointing this, but I forwarded this to the > upstream bug tracker and various people are discussing a fix. > > https://rt.perl.org/rt3//Ticket/Display.html?id=117793
A fix for this has now been applied upstream: http://perl5.git.perl.org/perl.git/commit/da1929e7566d4399b20541f8609bd48b40663034 As the bug is old and the severity is minor, I don't plan any backporting work, but hopefully this will meet your needs when it makes it into Debian at some point? Cheers, Dominic. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
