Re: the Korean in Unicode (Re: GScript is no longer ...)

[Owen Taylor <otaylor redhat com>]

Changwoo Ryu <cwryu@debian.org> writes:

> I didn't study the GScript code well, but after seeing the example, I
> guess GScript does:
> 
> 1. If some Unicode characters belong to Korean
> 2. convert them to KS C 5601 with iconv(). 
> 3. display them with the ordinary gdk function.

Yes, that's the way the "basic" rendering module works.
(It actually goes straight to X instead of going through GDK)
 
> The problem is, KS C 5601 only includes 2,350 (pre-composed) syllables
> while Unicode 2.0 includes 11,172 syllables (all in modern Korean).
> And "Hangul Jamo" code (0x1100~0x11FF) provides more times of
> syllables (ancient Hangul) by composing the Jamos.
> 
> In short, we could not see some valid (in Unicode) Korean characters
> in the above senario.
> 
> What could I do, if I want to make GScript display all the possible
> Korean characters?  Of course, I am not questioning about Korean
> specific things.  I just want to know how to stop the above senario,
> and run another better (and complicated) algorithm.  Do I need to
> write a separate rendering module?

Exactly.

A rendering module takes a string of Unicode characters
and converts it into a GlyphString. 

A GlyphString is a list of:

 Font
 glyph index
 (x_offset, y_offset)
 advance width

I think this should be quite sufficient for doing
Jamo composition. 

 It doesn't look like ksc5601 includes the necessary
glyphs for forming the characters, (it has all the syllable
components, but not scaled to the right sizes), so there
might be some trouble in making sure people had appropriate
fonts. I'm not really sure, though, what fonts would
typically be available on a Korean-speakers machine.

Regards,
                                        Owen