Re: Reverse-wrapping gtkmm widgets to gtk+ (for glade3)

[Murray Cumming <murrayc murrayc com>]

On Mon, 2006-11-20 at 17:28 +0100, Alexander Brinkman wrote:
> On Mon, 2006-11-20 at 16:56 +0100, Murray Cumming wrote:
> > On Mon, 2006-11-20 at 16:49 +0100, Murray Cumming wrote:
> > > 
> > > Maybe we could hook in a GObject init() function to the derived GType,
> > > which would also create the C++ wrapper around it. 
> > 
> > Of course we wouldn't want this to run in addition to the regular
> > constructor, when you are using this class without Glade.
> 
> I'm assuming that a GObject init function will only be called if the
> object is created with a normal C g_object_new call?

Yes, but I think that's how we create the underlying C GObject even when
we use a C++ constructor. It will need some thought/investigation to
figure out how to best solve this.

>  In thoses cases you
> would like to have the C++ wrapper (or else the widget isn't functioning
> at all).
> 
> Also what does the wrap_register function do?

It just registers the GType with the Glib::wrap() system, so that it
knows which wrap_new() function to call when it is given a GObject of
that GType.

>  I was assuming that after
> the call to wrap_register there would be a full C GObject wrapper
> available (which knows how to correctly instantiate a Gtkmm object) for
> that specific GType?
> 
> p.s. I'm afraid my total lack of Gtk+ (and also Gtk+/Gtkmm interaction)
> knowledge isn't helping here :( ...

Yes, it's becoming a bit difficult.

At the very least, if you don't make progress, please mention this in a
bugzilla report, so I (or someone else) can look at it if I have time.

-- 
Murray Cumming
murrayc murrayc com
www.murrayc.com
www.openismus.com