"Seak, Teng-Fong" <[EMAIL PROTECTED]> wrote:
> > TTB_RTL (for Chinese, Japanese, etc.) and TTB_LTR (for Mongolian)
> > primitives.
>
> Actually, there's another possible solution (or even simpler): rotate all
> characters to the left by 90:-) but only if it's supported by LaTeX. If this is
> supported, we just need to type in LTR then TTB and after rotation, the text will be
> TTB then RTL.
LaTeX need not support this at all. You have only to use the rotated
fonts.
CJKvert.sty is designed to use this.
> But I'm not sure if LaTeX supports already double-byte encoding, so I'm not
> going to ask this question in the newsgroup for the moment.
There are essentially two ways to support double-byte encodings. One is
to
use special TeX compilers whose internal string representation is 16
bit.
These compilers have "Translation Process" for reading byte streams of
input
files so that the TeX proper will receive wide character string instead
of
byte streams. pTeX can read Japanese multibyte encodings, hTeX can read
Korean, and omega can read anything as far as you have an appropriate
.ocp
(Omega Compiled translation Process: There already is a Big5 .ocp but
no
one has written Japanese/Korean .ocp yet).
The other way is to use the catcode magic. If you set the multibyte
system
leading bytes as active characters, then TeX considers each multibyte
character a command macro which selects a character in a font (256
each).
So then, if you have divided multibyte character fonts into smaller
pieces
beforehand, the ordinary TeX can process multibyte encoded source files.
This approach is taken by the CJK macro package. You are more likely to
see the TeX stack exhausted errors, though.
Regards,
SMiyata
