[emacs-bidi] Arabic in Unicode

[TAKAHASHI Naoto]

Alex Schroeder writes:

> I don't know how Arab is represented in Unicode.  Picking up my book,
> however, I see that there are often three ways of writing a letter --
> one for the beginning of words, one for within words, one for the end
> of words.

Not three but four, actually; the fourth is called "isolated form".

> Are these represented as different characters in Unicode
> (or Latin 6)?

In Latin-6, no.
In Unicode, yes and no.

Latin-6 defines only one codepoint for each Arabic letter; glyph
variations (word beginning, middle, end or isolated) are not
represented.

Unicode defines one basic codepoint plus up to four
representation-codepoints for each Arabic letter.  Representation
forms are prepared for compatibility purpose and their usage is
discouraged.

Alex Schroeder writes:

> When Emacs uses Unicode internally, then we can move the table to
> Unicode.  At the moment, however, I think we should use Mule because
> of the following reasons:

> 1. Nobody knows when Unicode will happen.

Just for your info: here, Handa and I have already succeeded in
displaying Arabic letters using Emacs-21.0.104.  They are represented
in the mule-unicode-0100-24ff charset.  The buffer contains only basic
forms; appropriate glyph variations are selected on the fly, using a
mechanism similar to font-lock, and displayed on the screen.  Bi-di
support is minimum; it uses the same algorithm as in Mule-2.3.

To release the modification, we need to get an official permission
from the responsible section of our institute.  A necessary procedure
has just begun, but nobody knows how long it will take.

-- 
TAKAHASHI Naoto
address@hidden
http://www.m17n.org/ntakahas/