[Uim] Re: [SCIM] Re: Design philosophy and strength of uim

[Ken Deeter]

On Sat, 26 Jun 2004 11:20:51 +0800
James Su <suzhe at tsinghua.org.cn> wrote:

> Hi,
>   First, though there is no simple API (like uim) for client usage, 
> SCIM's IMEngines can be loaded into client's process directly by using
> 
> libscim. We are planning a C binding of SCIM's IMEngine interface,
> just like uim.
>   The socket protocol of SCIM is also implemented as an IMEngine
>   module, 
> if a client want to connect to the SCIM socket daemon, it can just
> load the socket IMEngine module and use it just like other normal
> IMEngines, without re-implement the socket interface.
> 

Ok. that sounds pretty similar then. I will leave the exact details to
the real developers ;-)

>   The position of the preedit string window is hardcoded into source 
> code. Most Chinese users like the current style. If you think putting
> it right after the cursor is better, I can change it as you wish.
> 

Perhaps if it could be made as an option that you could set in the GUI,
that would be good. Also I'm not sure if the X11 frontend can get info
about the font size, but it seems useful if hte font size used for the
pre-edit window could be the same as the text in the widget.

Another thing I notice was that when I set the lookup table to be
vertical, the pre-edit string shifts upward when the new table is
displayed.. I suspect this may be somewhat confusing to users who ahve
been using Japanese IM's in which the pre-edit string always is in the
same place, and the conversion candidate list is always displayed right
udner (or right above if there is not enough screen space).

>   And You must use a locale with UTF-8 encoding in order to use all 
> input methods in one client. Because typing characters outside a 
> specific charset makes no sense.
> 

I agree that it makes less sense, but it seems to me that if you are
accessing from a gtk2 immodule, you should be able to safely assume that
the application side will only take utf-8 strings, so the current locale
shouldn't matter. Could it not be set up so that if things are accessed
through gtk2, all languages are available, while if they are accessed
thru XIM, it is limited by the current locale/charset?

gan xie ni de cheng jiu. ;-)

Cheers,
Ken