[developers] [afuchs at franz.com: Re: [spr30362] bug using climxm]

[Francis Bond]

G'day,

With some help from Ben I have finally tested these patches for both
Korean and Japanese.

I proceeded as follows:
 - added the files to the code sub-directory in the lisp directory
 - rebuilt base.dxl (and clim and bclim etc)
 - started lisp (with Norwegain utf-8)
 -- I reset it to the grammar encoding when I load the grammar
 - recompiled the logon tree: (lmt t)

After doing these four steps, everything except for direct input into
the parse window appears to work.  Using clim windows we could display
Korean and Japanese, including the parse windows, compare window
(necessary for treebanking) and orthography within avms.

I will try to get input working (XMODIFIERS=@im=uim didn't work,
although it has in the past) but thought I should report on what I had
done so far.

> Prof. Yang and I jointly tried out the bug fix for the Korean grammar
> (on cypriot) and so far we only have partial success.  We can now paste
> Korean characters into the LKB Parse input window, and see them there,
> and then parse successfully.  But unfortunately the leaf nodes of the
> parse trees and the discriminants in the Compare window are still
> mangled.  It is of course possible that we did not quite get the
> locale settings properly set, but we think we did, so we would
> appreciate some counsel (perhaps from Francis) on how to proceed, or
> what in particular we should check.

Did you recompile everything that you could (base/clim/blicm) and the
logon/delphin code?  Things were very wierd for me before I did all of
them.

> One thing that confuses me
> is how the parse-tree drawing engine chooses the font that it will
> use to print the nodes of the parse tree.  It looks to me as if it
> is hard-wired into the code that the font family and face will be
> :sans-serif and :roman, respectively, and I don't know if these are
> sensible for Korean.

As Ann said, I think it looks through a font-set for characters, so it
will include various fonts.  This can be problematic for Chinese
characters, as they glyphs are often very different in the different
fonts which order you chose the fonts in should be different for
mainly Japanese or mainly Chinese use.  I am sure there is some way to
set the font-sets if necessary, but I hope  we can ignore this for the
moment.


-- 
Francis Bond <http://www2.nict.go.jp/x/x161/en/member/bond/>
NICT Computational Linguistics Group

P.S. I also CCed Andreas --- thanks for the, hopefully, full solution at last.