Re: Depth First Search through Digraph Results in Memory Leak

Yes, like ysth says, this looks like a bug in perl. A workaround here is to change the sub transitive_closure_DFS like this:

sub transitive_closure_DFS
 {
   my ($self, $node) = @_;
   my $nodes = [];
   my $search = sub { push @$nodes, $_[0] };
   $self->DFS($node, $search);
   my @nodes = @$nodes;
   $nodes = '';
   return [@nodes];
 }
[download]

So it seems that somehow the anonymous array in $nodes is not properly destroyed automatically. But how this is related to the recursiveness dependency ysth mentions escapes me ...

Update
I've written a follow-up with a simplified version of the problem (and an additional segfault :) here.

-- Hofmator

Comment on Re: Depth First Search through Digraph Results in Memory Leak Download Code

Replies are listed 'Best First'.
Re: Re: Depth First Search through Digraph Results in Memory Leak by djantzen (Priest) on Jan 08, 2004 at 20:40 UTC
Well I'll be damned, that fixes it. I've changed `$nodes = ''` to `undef $nodes` with the same results. So for some reason perl needs to be told to destroy the array when it should take a cue from the lexical going out of scope. I'll prepare a bug report after I stop sighing in relief. Thank you Hofmator! "The dead do not recognize context" -- Kai, Lexx	[reply] [d/l] [select]
Re: Re: Re: Depth First Search through Digraph Results in Memory Leak by diotalevi (Canon) on Jan 08, 2004 at 22:49 UTC
Change those print calls to warn(). I'm guessing that those arrays are kept around because the closures were kept around and that both were eventually cleaned up during global destruction. ~~You are always free to clear a variable with prejudice - use undef as a function. `undef $nodes` might give you some different results.~~ (you already did that. Whoops). That may still be a bug though but whether it is in the reference counting of $nodes' array or $sub's clusure I don't know. But then it isn't clear to me that it is a bug.	[reply] [d/l]


Your skill will accomplish what the force of many cannot
	PerlMonks