Re: [RFC] Glib::Idle->add() and Glib::Timeout->add() with destroy notification parameter

[muppet <scott asofyet org>]

On Feb 21, 2007, at 11:08 AM, Emmanuele Bassi wrote:

> so I'd like to request the addition of a notify callback to
> Glib::Idle->add() and Glib::Timeout->add(), set to undef by  
> default, to
> be called as the destroy notification function.  I can provide a  
> patch,
> unless someone either shows me how to do it differently with the  
> current
> Glib version, or demonstrates that my idea is pure crack and I  
> should be
> shot down. :-)

*cracks knuckles*  Easily done with available tools....

#!/usr/bin/perl -w

use strict;
use Glib;

my $loop = Glib::MainLoop->new;

Glib::Idle->add (sub {
                my $data = shift;
                print "IDLE $data->{count}\n";
                return --$data->{count};
        },
        # Note: this works because the only reference to this object
        #       is the one stored in the idle event source we're  
creating.
        #       If you create it in two steps, make sure your other
        #       reference goes away!
        GSourceDataWithDestroyNotifier->new (
                count => 10,
                callback => sub {
                        print "destroyed!\n";
                        $loop->quit;
                },
        ));

$loop->run;

print "exiting\n";
exit;

package GSourceDataWithDestroyNotifier;

sub new {
        my $class = shift;
        return bless { @_ }, $class;
}

sub DESTROY {
        my $self = shift;
        $self->{callback}->($self) if exists $self->{callback} and
                                      'CODE' eq ref $self->{callback};
}

__END__


However, i do admit that it can seem like quite a lot of overkill and  
overhead to have to create a new class for each GSource for which you  
want a destruction notification.

Your example application of the destruction notification is a good  
example of why simply cleaning up in the last call to the idle or  
timeout is not sufficient; if the source is removed by  
Glib::Source::remove(), that last call may not happen, and you would  
have to trigger the cleanup by hand.

I think it should apply to Glib::IO::add_watch(), too.

Oh, and we're missing the bindings for the primitives to create your  
own GSource implementations in perl...

So, make a patch, and let's have a look.  :-)


--
I hate to break it to you, but magic data pixies don't exist.
  -- Simon Cozens