[immodule-qt] Re: scim-qtimm 0.7 released

[YamaKen]

Hi Choe,

At Tue, 7 Sep 2004 10:41:22 +0900,
krisna at kldp.org wrote:
> 
> On Tue, Sep 07, 2004 at 01:21:10AM +0900, YamaKen wrote:
> > Hi Mike, nice to meet you, hajimemashite.
> > 
> > At Mon, 06 Sep 2004 15:38:03 +0200,
> > mfabian at suse.de wrote:
> > > 
> > > YamaKen <yamaken at bp.iij4u.or.jp> さんは書きました:
> > > 
> > > > Hi James,
> > > >
> > > > At Mon, 6 Sep 2004 17:04:30 +0800,
> > > > james.su at gmail.com wrote:
> > > >> 
> > > >> Hi,
> > > >>   Changing postIMEvent to:
> > > >> 
> > > >> void QApplication::postIMEvent( QObject *receiver, QIMEvent *event )
> > > >>  {
> > > >>      sendEvent( receiver, event );
> > > >> }
> > > >> 
> > > >> can fix this issue. 
> > > >
> > > > It's exactly same as our first implementation and insufficient
> > > > for some users.
> > > 
> > > What do you mean by insufficient? Does it cause problems or
> > > is it only slow?
> > 
> > Only slow (in some language or hardware). But I can't sense such
> > problem in my environment with 60 chars/sec key repeat with a
> > Japanese input method uim-anthy. So you could use following
> > James's code as temporary solution.
> 
> No, the problem is not the slowenss only. If we call postEvent instead of
> sendEvent, event sequence maybe changed.

Mike and me are discussing about another problem "What is the
problem if QIMEvent compression is not implemented?". It differs
from the problem that you mentioned "What is the problem if we
mixed QIMEvent and QKeyEvent with QIMEvent compression?".

James's code uses only sendEvent(), so we could resolve latter
problem. But it doesn't resolve former problem. Cougar had
implemented the QIMEvent compression feature to resolve the
former problem.

> For example, when user presses enter key, immodule will commit current
> preedit string and forward enter key event. In this case, module code
> will call postEvent and IMEnd event will be pushed into event queue,
> and immodule return false to forward this enter key event.
> Then application will receive enter key event first and IMEnd event
> second.

Thank you for clear explanation. Now I've understood what is the
problem. Although my previous patch had fixed another bug, it
will not resolve the problem.

Cougar, Kazuki, I'm sorry about my previous blaming based on
poor understanding. I'm going to find out best solution.

> I think this problem is quite important to my qt immodule implementation.

I also think so.

Regards,
-------------------------------
YamaKen  yamaken at bp.iij4u.or.jp