Re: component architecture

[Tom Tromey <tromey creche cygnus com>]

Aadi> As far as GUI framework is concerned, I think that Owen Taylor
Aadi> is working on the C bindings for MICO so that GTK+ and the
Aadi> supporting GNOME widgets (like XmHTML) can be CORBAized. I don't
Aadi> know what stages this is in. Can someone update on this?

Do we really want to CORBA-ize all the widgets?  This sounds like a
lot of effort for little payoff to me.

My recollection is that Elliot is working on C bindings for MICO.  Is
this wrong?

Aadi> I believe that the idea is to show (either as proof-of-concept
Aadi> or something more) that GNOME is a very feasible and usable idea
Aadi> that's got the app support built-in and then move from there.
Aadi> Could one of the principals please
Aadi> veryify/refute/flame/otherwise on this?

Am I a principal?  I don't know.

The plan, as I understand it, is to do things incrementally, roughly
like this:

1. Write core code and apps
2. Use CORBA to let apps talk to each other in a coarse-grained way
   (e.g., have an address book service)
3. Work on a component/plugin model
4. Work on a compound document model (see all the old OpenDoc
   discussions)

Of course, the actual order depends on interest and hacker support.

Aadi> Though i'm not fully comfortable with the idea (and like it
Aadi> matters if i am!), the drive is probably to get GNOME up to par
Aadi> with KDE as soon as possible.

Are you uncomfortable with the idea of incremental change, or the idea
of getting up to par with KDE?  It isn't clear (to me) from the
context.

The incremental model makes sense to me, because I think it is
important to have something working.  I think that, generally
speaking, it is better to have something imperfect working now,
instead of waiting for the perfect thing to crystallize all at once.

Tom