[Uim] new input method 'look'

[Etsushi Kato]

On 2/20/07, IWATA Ray <iwata at quasiquote.org> wrote:
> "Etsushi Kato" <ek.kato at gmail.com> writes:
>
> > Great.  I've just tried look IM and found updating preedit seems
> > needed in some cases.  Also it is needed to modify look.c to compile
> > on Linux (SIZE_T_MAX definition, see attached diff).  In addition, I
> > think use of more specific prefix for the function names in bsdlook.h
> > is preferable as they are exported functions.
>
> OK, look.diff works fine on my environment, too:)

Here is the one more suggestion.  I think it is better to have a
cursor at the position just after the look-context-left.  Or is it
your intention not to set the cursor at the position?

--- look.scm.orig    2007-02-20 16:47:55.000000000 +0900
+++ look.scm     2007-02-20 17:08:55.000000000 +0900
@@ -206,6 +206,8 @@
    lc preedit-none
    (look-context-left lc))
   (im-pushback-preedit
+   lc preedit-cursor "")
+  (im-pushback-preedit
    lc preedit-none
    (look-format-candidates lc))
   (im-update-preedit lc))

Also it wolud be great if there is some kind of learning mechanism
about the committed words.


Anyway I'd like to integrate look IM into uim repository.  YamaKen, is
it OK?  And what is the good way to have bsdlook.c do you think?
Iwata-san's patch integrates bsdlook.c into libuim.  Another candidate
is 1) put these functionality into libuim-skk and libuim-look
separately, or 2) create libuim-look to share it.


> > Thanks for removing popen() in skk.c.  It becomes a lot cleaner :)
>
> I found a bug, on xim/canddisp.cpp:canddisp_singleton.
> If uim_ipc_open_command returns fail, uim-xim will stall.
> In this case, uim-xim should write error-message and exit.

OK.  I'll fix it later.

Cheers,
-- 
Etsushi Kato
ek.kato at gmail.com