[SCIM] Re: [m17n-lib:00079] Proposal: add minput_reset_ic (MinputContext *ic) method.

[Choe Hwanjin]

On Tue, 21 Dec 2004 14:18:30 +0900 (JST), Kenichi Handa <handa at m17n.org> wrote:
> Choe Hwanjin <choe.hwanjin at gmail.com> writes:
> 
> > I'm very curious about this. the reset method is very essential to
> > korean im. Is there any historical reason about this?
> 
> I don't understand why such an API as minput_reset_ic is so
> essential to Korean IM.  If a user wants to commit the
> current preediting text, he can type some key defined as the
> command for "commit this" in an input method.  For instance,
> currently ko-romaja.mim accepts SPC as a command for "commit
> this and insert space", but it's easy to change it to a
> command for "If there's no preediting text, insert space.
> Otherwise commit the preediting text" if that is more
> preferable.

Korean never use explicit or implicit "commit key" and don't aware of
preedit state.
For example, almost every korean move the focus by clicking mouse on
another widget while the preedit string is still not commited. Because
they don't aware of preedit state. They think the preedit string is
commited, even if it is not really commited. So toolkit should call
reset to commit current preedit string before moving the focus. If
not, the last charactor or word, the preedit string, will be lost.
They move the cursor in the same way.
And there is no explicit commit key in korean input method.