[patch] generate marshallers and metadata from dbus-glib-tool

[Maciej Katafiasz]

Dnia 06-09-2004, pon o godzinie 18:56 -0500, Paul Kuliniewicz napisał:
> I guess it depends on what kinds of objects you're "supposed" to expose
> using D-BUS.  If you're supposed to create a separate object just for
> D-BUS stuff, you could get away with just exporting all signals (and
> methods, if GLib had a way to get that information [0]).  If you're
> supposed to take existing objects and also expose them to D-BUS, there
> could be lots of methods and signals that either don't make sense to
> expose or don't have D-BUS-friendly interfaces.

IMHO we should go for dedicated DBUS objects. I don't think allowing for
exposure of arbitrary GObjects is going to do any good, both in terms of
autogenerating DBUS wrappers and quality of client code written this way

> [0] I had toyed with the idea of adding method metadata to GObjects in
> general via a hypothetical GIntrospectable interface, but that approach
> started looking messier and messier.

http://bugzilla.gnome.org/show_bug.cgi?id=139486
Rob Melby is currently giving a spin to first-cut implementation, but I
have no clue what are real chances of it going in in Glib 2.8 [0]

Cheers,
Maciej

[0] I'd absolutely, totally LOVE seeing it in 2.8 (and in that case, Rob
would get lots of free beers from me ;). But I'm not the one who has the
say here

-- 
"Tautologizm to coś tautologicznego"
   Maciej Katafiasz <mnews2 at wp.pl>
       http://mathrick.blog.pl