[immodule-qt] Should dead key support be builtin feature?

[LiuCougar]

On Tue, 20 Jul 2004 01:55:50 +0900, YamaKen <yamaken at bp.iij4u.or.jp> wrote:
> At Tue, 20 Jul 2004 01:10:49 +0900,
> mover at hct.zaq.ne.jp wrote:
> > I'm now creating the plugin to support dead/multi key, you know.
> > In gtk+, dead/multi key support is provided as gtkimcontextsimple and
> > "uim-gtk" uses this.
> > That is, dead key support is implemented in immodule level.
> 
> There are two orthogonal issues. We should keep them isolated in
> the discussion.
> 
> 1. to provide simple table-based input method as a qt-immodule
>    that supports dead/multi keys (QSimpleInputContext)
> 
> 2. combining QSimpleInputContext into preexisting immodule as
>    slave input context as like as uim-gtk
> 
> Providing 1. is important because some users not need complex
> input method frameworks. Suppose embedded application as an
> example. In the case, Qt rules whole GUI environment and
> QSimpleInputContext satisfies all requirements for input, in
> many countries.
> 
> QSimpleInputContext will also be reference implementation for
> the immodule for Qt.
> 
> > But I think dead key support is indispensable feature, so if we have gtk+ way,
> > the same codes appears in iiimf-qt, scim-qt, uim-qt.
> 
> I think that uim-gtk's way (2. in above) is a temporary
> solution. All of IIIMF, SCIM and uim should support dead/multi
> keys itself and allow combining with other methods according to
> their own framework. I think that the combination is (well?)
> known as 'input method cascading'.
> 
> > To avoid this, it's better to add the support to QInputContext, I think.
> > dead key support is implemented in inputcontext level.
> 
> I disagree with you.
> 
> Keep QInputContext simple. It's not an input method framework
> itself, but an abstractor of complex input method (including
> input method frameworks). Input method cascading is a
> responsibility of concrete input method plugins.
> 
> To actualize your idea, implement QCombiningInputContext as
> shared base class of IIIMF, SCIM and uim is better. But I expect
> that IIIMF and SCIM have their own solution.

Personally, I agree with YamaKen
I think one of the goals of qt-immodule support is that: it is SIMPLE,
yet flexible enough to be extended and compatible with even the most
complicated input method frameworks our there.

Regards,
-- 
"People's characters are strengthened through struggle against
difficulties; they are weakened by comfort."
- Old Chinese adage