Re: Input Method of Gnome

[NIIBE Yutaka <gniibe chroot org>]

Hello Akira!

Akira Higuchi writes:
 > No. Current code in gtk+ is based on the backend model. 

I'm glad to hear that.  Good.  That's the right thing.  I'll check it.
I've never known the code which use XIM in backend style.

 > Why do you think XIM is useless for Japanese Input based on the
 > backend model? It seems to be working fine for me.

Works? Good.  I don't argue if it works or not.  I want good one,
better one.  "It works" is just a first step, not goal.  Would we
have better user interface?

IMNSHO, XIM is totally useless in backend model, now we have CORBA,
why don't you use it then?  It's powerful scheme, you don't need to 
depend upon poor implementation of communication around 80s.

 > P.S. I use an XIM server called 'kinput2' with XEmacs. Kinput2
 > works fine with gtk+ also.

Please don't get me wrong, but I don't like it, really.  Why you need
kinput2?  Here, X sends to GTK+, and GTK+ sends to Kinput2, and
Kinput2 sends to Wnn/Canna/SJ3 or whatever conversion server.  Kinput2
does nothing but just delegates packets.  I think it's useless.
Well, it not quite true.  Kinput does translation... too little job.

My idea is that X sends to EGGnome and EGGnome sends to Wnn/Canna/SJ3
via CORBA.  Here, I assume that translation can be done within the
widget.  Sounds interesting?

One more reason to support backend model with no XIM.  I usually use
EGG V4 in modeless style.  Input goes to as usual text, then convert
it later on demand, sort of lazy conversion.

Thanks for letting me know the fact,
-- 
Niibe Yutaka