Re: Threaded posts with a single table

[prev] [thread] [next] [Date index for 2004/08/12]

jinx@xxxxxxx.xxx wrote:
> # sorry, Perrin, didn't reply all the first time (don't use this client
> usually, sigh)
> 
>>On Wed, 2004-08-04 at 17:10, jinx@xxxxxxx.xxx wrote:
>>
>>>__PACKAGE__->might_have(parent => __PACKAGE__);
>>>__PACKAGE__->has_many(replies => __PACKAGE__);
>>
>>I would change that might_have to this:
>>__PACKAGE__->has_a(parent => __PACKAGE__);
> 
> 
> Sure thing. I had it that way first but wasn't sure since a post could be
> original (parentless).
> 
> 
>>I also like the explicit form of has_many instead of the magical form
>>you're using there:
>>__PACKAGE__->has_many(replies => __PACKAGE__, 'parent');
>>

This snippet might also be useful to you.. It pulls the entire tree both 
ways
#
# this application's parents
sub parents($) {
     my $self=shift;
     my @application=@_;
     my @parents;
     my @parentspassed;
     my %parents;

     push @parents, $self;
     while (@parents) {
         my $pid=pop @parents;
         my $parent = $pid->application_parent;
         if (defined($parent)) {
             next if ($parents{$parent->id}) ;

             $parents{$parent->id} = $parent;
             push @parents, $parent;
             push @parentspassed, $parent;
         }
     }
     return reverse @parentspassed;
}
#
# this applications children
# (so you can figure out what needs updating)
sub children($) {
     my $self=shift;
     my @application=@_;
     my @children;
     my @childrenpassed;
     my %children;

     push @children , $self;
     while (@children) {
         my $pid=pop @children;
         my @childs = 
Nagios::Application->search({application_parent=>$pid});
         foreach my $child (@childs) {
             next if ($children{$child->id});
             push @children, $child;
             push @childrenpassed, $child;
             $children{$child->id} = $child;
         }
     }
     return @childrenpassed;
}
#
>>
>>>So, the "parent" I can see working fine. It's just pointing to a
>>>"foreign" id in its own table. It's the "replies" that obviously won't
>>>work that way.
>>
>>What makes you think that?
> 
> 
> The relative density of my gray matter, I fear. CDBI relations can be a
> bit magical/confusing to me. That looks great if it does what I think it
> does. I also noticed that the POD for CDBI is much fuller now than it was
> a few versions back (or else I'm just understanding more now) so I have
> some reading to catch up on.
> 
> Thanks for the help. And to Simon, of course!
> http://search.cpan.org/search?query=Email%3A%3AStore%3A%3AThread&mode=all
> is very interesting code.
> 
> -Ashley
> 
>