Re: Gtk2::SimpleList

[Ross McFarland <rwmcfa1 neces com>]

On Fri, 2003-07-18 at 09:54, muppet wrote:
> Gavin Brown said:
> > Basically this is an RFC - I'd like to get an idea of how people want to
> > use/are using Gtk2 lists and how closely this might meet their needs.

i'll look at this and my stuff when i have more time.

> i would use them more if they weren't such a pain, so i am excited about your
> effort.

i've got some code that's the beginnings of a possible tree helper. i
decided that a good way to learn about how the whole MVC stuff works
would be to look at that.

> how about Tie'ing the perl representation (an array for a GtkListStore, maybe
> a hash for a GtkTreeStore) to the underlying GtkTreeModel, so that you don't
> maintain two structures?  the perl variable would always be up to date,
> because FETCH would call $treemodel->get and STORE would call $treemodel->set.

> well, the tree thing might be problematic because of the undefined order for
> hash elements, even though the underlying structure would preserve the order. 
> just an idea.

my stuff was an array of anon hashes so as to preserve order :)

my @tree = (
        {
                name => 'p1',
                number => 42,
                children => (
                        {
                                name => 'c1p1',
                                number => 42.1
                        },
                        {
                                name => 'c2p1',
                                number => 42.2
                        },
                ),
        },
        {
                name => 'p2',
                number => 1,
                children => (
                        {
                                name => 'c1p2',
                                number => 1.1
                        },
                        {
                                name => 'c2p2',
                                number => 1.2
                        },
                ),
        },
        ...
)

and a function that recursively descended the tree, handling the
appropriate parent iter usage.

> this may make more sense implemented as another type of model...  i dunno.

that would be an interesting experiment.

-rm