I cast my vote with Andrew, a unified character set such as 10646 seems to me to
be the right approach.
Support of that concept on my platform (Macintosh) is non-trivial, apple
provides no systemic method for mapping its own font encodings to/from 10646,
and recomends applications be coded generically, defering all knowledge of
language-specifics to the OS. Hopefully, Apple (Adobe, IBM...) will provide
mechanisms for Macfont<->10646, untill then I would be on my own.
As to Ohta-san's concerns about misrendered kanji, I am sympathetic, but I am
also realistic, I would expect a UA to employ some itelligent algorithym to
choose Chinese, Japanese, or Korean for rendering of HAN, based on an inspection
of the character stream for nearby hiragana, katakana, or the korean phonetic
alphabet. For homogenous text of any length it will guess corectly. For
fragmented polylingual text, such as lists of mixed oriental personal names (eg,
IGO match results) it may well get mixed up, but would then be expected to allow
the reader to override its decisions.
--
dana s emery <de19(_at_)umail(_dot_)umd(_dot_)edu>