Re: [emacs-bidi] bidi categories, derived from Unicode data

[Eli Zaretskii]

> From: Alex Schroeder <address@hidden>
> Date: Sat, 10 Nov 2001 01:35:27 +0100
> 
> Unfortunately the lisp files required to set this up are rather big,
> we'll have to find a way of dumping the info, later.

Rather, we will have to find a way to load these tables on demand.
Someone who doesn't use Arabic, won't need the tables for Arabic, and
likewise for Hebrew.  And it doesn't make sense to load the huge
tables for CJK, unless the user wants to use those characters.

This has been discussed before, but I don't think there's a decision
how to go about this.

> Attached you will find bidi.el which does the setup.

Thanks.  Please in the future consider posting an abridged version
(no need to send the entire humongous table, for example), and making
the rest available for download by the interested parties.

> Some UCS characters seem not to exist in Emacs; this was surprising.
> An example from the source: (decode-char 'ucs ?\x33FE) -- valid,
> (decode-char 'ucs ?\x3400) -- invalid.

That's because the corresponding mule-unicode-* charset ends at U+33FF
(thus its name, mule-unicode-2500-33ff).  There's a gap in Unicode BMP
coverage between the end of this charset and the beginning of the next
charset, mule-unicode-e000-ffff.

> I currently just ignore the UCS characters where decode-char
> returns nil.

That's The Right Thing to do.