Re: Tk::Dialog memory leak?

On Saturday 15 July 2006 10:56, ReneeB wrote:
> Hi,
>
> Tk::Tree isn't destroyed because it inherits from Tk::Derived not from
> Tk::Widget.
> I'll try to release a new Tk::Tree version today.
>
> ReneeB

Ok, thanks. I'm still confused:

Tk:Tree has: use base qw(Tk::Derived Tk::HList);
Tk::HList has: use base qw(Tk::Widget);

so it shouldn't really matter, right?

Anyway, removing Tk::Tree from the demo doesn't fix it for other widgets (like 
the default button and all the frames of DialogBox), although it is a widget 
based on Tk::Tree I'm actually having in the real app.

>
> Petr Pajas schrieb:
> > Hi,
> >
> > I have a dialog box which I cannot easily reuse, so it is created many
> > times during the application run. After creating it, I can't find a way
> > to destroy it completely even if I explicitly call $d->destroy                                                                  

oh, *and* was missing here 

> > all variables go out of scope. Here is a demo:

-- Petr


> > #!/usr/bin/perl
> > use strict;
> > $|=1;
> > {
> >   use Tk;
> >   use Tk::Tree;
> >
> >   my $mw = MainWindow->new;
> >
> >   *Tk::Widget::DESTROY = sub {
> >     my $w = shift;
> >     print "DESTROYING $w\n";
> >     $w->destroy if ($w->IsWidget);
> >   };
> >
> >   {
> >     for (1..3) {
> >       print "===================================\n";
> >       print "RUN $_\n";
> >       print "===================================\n";
> >       my $d=$mw->DialogBox(-title=> "test",
> > 			   -buttons=> ["OK", "Cancel"]);
> >       # only the Cancel button will get destroyed
> >
> >       my $edit1 = $d->HList(); # will get destroyed
> >       my $edit2 = $d->Tree();  # won't get destroyed
> >       $d->destroy;
> >       # will destroy only Tk::Button a Tk::HList
> >       sleep 0.1;
> >     }
> >   }
> >   #  MainLoop;
> > }
> >
> > END {
> >   print "===================================\n";
> >   print "STARTING GLOBAL DESTRUCTION\n";
> >   print "===================================\n";
> >   # will destroy everything else
> > }
> >
> > __END__
> >
> > Is there any way to prevent the leak? Thanks,
> > -- Petr
> > --++**==--++**==--++**==--++**==--++**==--++**==--++**==
> > ptk mailing list
> > [email protected]
> > https://mailman.stanford.edu/mailman/listinfo/ptk
--++**==--++**==--++**==--++**==--++**==--++**==--++**==
ptk mailing list
[email protected]
https://mailman.stanford.edu/mailman/listinfo/ptk