Re: Exculding Hangul compatibility jamo (U+3130 - U+318f) from the basic shaper?

[Owen Taylor <otaylor redhat com>]

Changwoo Ryu <cwryu debian org> writes:

> > > The patch was generated half-automatically..  I marked every char as
> > > ksc5601-printable if iconv(from WCHAR_T to EUC-KR) successes on the
> > > char.  Then I replaced all the marked chars with the char_mask_map[]
> > > indices, which I added by hand.
> > > 
> > > Where is the program you used to generate the table?  I guess it could
> > > be easily modified.
> > 
> > The table is generated by:
> > 
> >  pango/tools/make-table.sh
> >  pango/tools/compress-table.pl
> > 
> > The source tables are described in:
> > 
> >  http://mail.gnome.org/archives/gtk-i18n-list/2000-August/msg00016.html
> > 
> > And, with further additions in:
> > 
> >  http://bugzilla.gnome.org/show_bug.cgi?id=50633
> > 
> > (See comment from 2001-08-06)
> > 
> > Since the EASTASIAN tables on unicode.org are considered obsolete, it
> > might be best to switch make-table.sh to be able to read Unihan.txt
> > in addition. Or, maybe, we should add another script that takes Unihan.txt
> > and creates tables in the format make-table.sh expects.
> 
> I just wrote the below script which can generate make-table.sh's input
> from Unihan.txt.  It seemed to work (the local encoding values it
> produces are not correct but anyway they will be ignored by
> compress-table.pl).  But Unihan.txt does NOT seem to have all the
> informations the obsolete mapping files have.
> 
> It doesn't have jis-0201 mappings.  furthermore the ksc-5601 mapping
> only includes the Hanja area, with no Hangul compatibility jamos or
> Hangul syllables.

It seems to me to be most important to handle the Hanja area, since this
is the area we can't handle with the Hangul shaper. (And it's important
to get the glyphs from the KSC5601 font for these so they match the 
Hangul in style.)

For the Hangul compatiblity jamos, we can probably just go with your
original suggestion and mark them as with a "" language list rather
than with a "*" language list in the coverage information that the
basic shaper reports.

It's fine if the generation process for tables-big.i involves some
combination of tables we supply (like the TIS table) tables generated
from Unihan.txt and tables that the person generating the table needs
to download from ftp.unicode.org (ISO-8859-* tables and JIS-0201 table)

Regards,
                                        Owen