[immodule-qt] Requirements about filterEvent() (Re: Completing comments and documents)

[YamaKen]

Hi Cougar,

At Wed, 4 Aug 2004 03:49:17 +0100,
liucougar at gmail.com wrote:
> 
> On Wed, 04 Aug 2004 11:42:00 +0900, YamaKen <yamaken at bp.iij4u.or.jp> wrote:
> > At Wed, 4 Aug 2004 03:06:38 +0100,
> > liucougar at gmail.com wrote:
> > >
> > > On Wed, 04 Aug 2004 10:58:03 +0900, YamaKen <yamaken at bp.iij4u.or.jp> wrote:
> > > > I can understand your thought about simplicity. Yes,
> > > > imtermediate Qt keyevent is less straightforward for SCIM
> > > > because keycodes of SCIM is based on X keycodes. So your
> > > > implementation is reasonable because there is no other
> > > > platforms are available for immodule except for X11.
> > > you are right, but in fact, I just fixed the problem which prevents
> > > scim-qtimm from normal functioning without x11EventFilter.
> > 
> > I've understood that some difficulties are existing without
> > x11EventFilter. I'll look at your code later.
> Yeap, I had some real difficulties if only QEvent are utilized.
> Finally, I got this solution: with some very tricky work around to
> overcome the "unfriendly" QKeyEvent.

I've looked at your code in svn trunk. But I can't found what is
the difficulties you said. Tell me more detail to make our API
more useful.

Because distinguishing right and left modifiers by QKeyEvent is
going to be possible by Lars as you know, ignore that issue during
this discussion.

I think that keyevent_qt_to_scim() is the 'tricky workaround'
that you said. But why don't you use QKeyEvent::state() and
QKeyEvent::stateAfter() to extract modifier state simply? In my
rough understanding about keyevent_qt_to_scim(), it can be
simplified by these methods.

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