Front page | perl.perl5.porters |
Postings from July 2013
On Tue Jul 30 09:43:27 2013, demerphq wrote:
> On 30 July 2013 16:32, Dave Mitchell <davem@iabyn.com> wrote:
> > On Tue, Jul 30, 2013 at 01:31:30PM +0200, demerphq wrote:
> >> While I agree with your analysis I thought that PADMY vars werent
> >> freed until global destruction.
> >
> > PADMY vars are freed when the CV is freed whose pad they reside in
> > (as well as any other CVs which might also have captured them).
> >
> > For named CVs this is typically during global destruction; for anon CVs
> > this is typically when the ref to the anon CV is freed.
>
> This was in response to:
>
> > I do not consider this a bug, because the closure does not close over
> > $var1 or $var3. Those variables may already have been freed by the time
> > you call peek_my.
>
> Given what you say I dont understand how $var1 or $var3 could ever
> freed by the time we call peek_my().
Not in the example given in the original report, but it can happen
easily other ways:
use PadWalker 'peek_my';
use Data::Dumper;
use Scalar::Util 'weaken';
my @vars;
sub outer {
my ($var1, $var2, $var3);
push @vars, \$var1, \$var2, \$var3;
weaken $_ for @vars;
return sub { $var2; @vars; warn Dumper peek_my(0); }
}
outer->()
__END__
$VAR1 = {
'$var2' => \undef,
'@vars' => [
undef,
$VAR1->{'$var2'},
undef
]
};
You see that two of the weak references in @vars have gone stale by the
time the closure is called.
--
Father Chrysostomos
---
via perlbug: queue: perl5 status: open
https://rt.perl.org:443/rt3/Ticket/Display.html?id=119049
Thread Previous
|
Thread Next