[XeTeX] The future of XeTeX

[mskala at ansuz.sooke.bc.ca]

On Tue, 31 Jul 2012, Jonathan Kew wrote:
> On 31/7/12 13:26, mskala at ansuz.sooke.bc.ca wrote:
> > There's the rub.  Non-Latin scripts are a big part of the constituency of
> > XeTeX.  I routinely have to manually activate Korean-specific OpenType
> > features that are specified to be default but that XeTeX/fontspec doesn't
> > activate by default, just to get acceptable output in Korean at all.
>
> Which specific features are you referring to here? Maybe we can get this
> improved...

ccmp - glyph (de)composition; and ljmo, vjmo, and tjmo - lead, vowel, and
tail jamo shaping.  It's possible that ccmp may already be default, and my
own application doesn't actually need tjmo, but both should be turned on
for Korean when available.  Ligatures (liga) are also needed but
presumably already default.  There's some Microsoft documentation of these
linked from http://www.microsoft.com/typography/otspec/featurelist.htm ;
technical description of how my own Jieubsida fonts use them is included
in the Tsukurimashou package user manual, available from
  http://sourceforge.jp/projects/tsukurimashou/releases/56223

Notwithstanding the slightly confusing description in the Microsoft
documents (which describes more than one feature as "overriding all
others"), these should be applied in the order ccmp, ljmo, vjmo, tjmo,
liga.  XeTeX does do the ordering correctly already; it only needs to have
the features manually turned on.

Honestly, the glyph-pointer bug we've already discussed on this list,
   http://tug.org/pipermail/xetex/2011-November/022298.html
   http://bugs.icu-project.org/trac/ticket/7753
is a bigger problem for Korean and I think a higher priority for fixing.
But it will probably be harder to fix because it's internal to ICU.

-- 
Matthew Skala
mskala at ansuz.sooke.bc.ca                 People before principles.
http://ansuz.sooke.bc.ca/