Re: Pango Multilingual Input Reading

[hiura openi18n org (Hideki Hiura)]

> From: Noah Levitt <nlevitt columbia edu>
> Gtk+ isnʼt the right place for input methods for every writing
> system. Xkb and XIM should be able to handle everything in X, and
> win32 has its own input system.

Well, I actually want XIM to taper out as soon as possible ;-).

XIM is designed as mono-lingual IM system, and unnecessarily carrying
heavily dependency on X itself.

Alternative approach I took several years after XIM is to separate the
target platform/windowing system part from IM's system part as client
side bindings, as well as aiming to be natively multilingual. In another
word, the approach of having central repository of language engines
with multiple lightweight client side bindings. 

You are right that Gtk+ isn't the right place for input methods *logics*
to be implemented for every writing system, since it can only serve Gtk+
apps.

With the approach of central repository of language engines with
multiple bindings would resolve the issues here, as same language
engine(and actually same instance) can serve to X apps. Gtk+ apps,
SO/OOo, Mozilla, Java, emacs and even MS-Windows simultaneously and
directly(without going through XIM layer).

For example, the Gtk+ immodule in IIIMF(im-iiim.so) does not carry any
language knowledge or logics. It is just a binding module and actual
logics/engines reside outside, which would be shared among all of X,
Gtk+, Java, emacs, SO/OOo, Windows etc.

--
hiura {freestandards org,OpenI18N.org,li18nux.org,unicode.org,sun.com} 
Chair, OpenI18N.org/The Free Standards Group          http://www.OpenI18N.org
Architect/Sr. Staff Engineer, Sun Microsystems, Inc, USA   eFAX: 509-693-8356