> -----Original Message-----
> From: Tomohiro KUBOTA [mailto:[EMAIL PROTECTED]]
...
> > > Chinese text _must_ be displayed with Chinese font. Japanese text
> > > _must_ be displayed with Japanese font.
> >
> > This is not undisputed. Quotes of Chinese text in an otherwise
> > Japanese text are often, I'm told, all set in a Japanese font.
>
> There are about 120,000,000 Japanese people. It is natural a few of
> them have a bad habit to do so.
Fair enough.
> At least, such a habit is
> never allowed
> for textbook on Chinese language for Japanese students. Or,
> do you think
> it is natural that Unicode cannot be used for such purposes?
You would not use so-called plain text for a textbook (of any kind,
on any subject, at least not these days; assuming that the source is
electronic). Then, in the mark-up (whether "binary" or text based) you
would select different fonts for the different languages in the text.
For HTML, as I said, both IE (not too old, I don't know the exact
version where this was introduced) and for Netscape 6, they take
the HTLM 4 language tags into account when choosing 'default' fonts
for a piece of text. So if ' lang="zh" ' the browser would
choose among "Chinese" fonts, and if ' lang="jp" ' the browser
would choose among "Japanese" fonts (assuming such fonts are
installed on the system where the web page is displayed). You
can set the language differently for different pieces of text
on the same web page. I haven't experimented with this myself,
so I'm not sure how good it is in practice. (And you're right,
I can't read either Japanese or Chinese myself.)
> > > which should be the
> > > default font, Chinese or Japanese, for "world version" OSes?
> >
> > This is like asking "Which should be the default font: one
> that covers
> > MES-2 only" (MES-2 is a subset of 10646, covering many, but not all,
> > Latin letters), "or should we use a Devanagari one?".
>
> A Unicode font which include both of Latin and Devanagari.
There will be just a few fonts that have that kind of coverage
in a single font. And it is not recommendable to cover completely
different scripts in a single font, though possible.
> It is possible because codepoints for Latin and Devanagari don't
> conflict each other. On the other hand, CJK variants conflict.
Well, conflict and conflict. Since they are the same characters
it's not really a conflict, but glyph related preference. So either
one uses different fonts, or use a font that is internally sensitive
to "language" (really printing house tradition, but there are no
standard codes for that) variants. Opentype has such a language
glyph variant feature, which will be used among other things for
avoiding automatic fi-ligatures for Turkish, different inclination
of acute accent for Polish, different italic shapes for some
Cyrillic letters depending on "language" or rather language code,
and, of course, different glyphic variants for some (just a few,
I guess, otherwise one might as well use different fonts) CJK
ideographs depending on language code. But then you need to be
able to use Opentype...
(See in particular the 'locl' feature tag in
http://partners.adobe.com/asn/developer/opentype/appendices/feattags.html.)
Kind regards
/kent k
-
Linux-UTF8: i18n of Linux on all levels
Archive: http://mail.nl.linux.org/lists/
