Re: [emacs-bidi] Arabic implementation

[Alex Schroeder]

"Eli Zaretskii" <address@hidden> writes:

>> What I probably need is a string with all the characters in it which
>> have the right to left property.  Would that be feasible?
> 
> If all the Arabic characters have the same category, you can use the
> feature of char-tables where setting a single entry for the generic
> character has the effect of setting it for the entire charset.  (I
> thought that's why you've chosen category tables as the implementation
> for this; I know that's why I suggested to use char-tables.)

Yes, if you look at the old table, I used this for the private arab
charsets.  For hebrew however, it didn't seem right -- we can't just
assume that all of hebrew-iso8859-8 is left to right.  That's why I
assumed the same is true for arabic-iso8859-6, and it is obviously
true for the unicode charsets.  So in the end, having the generic
characters will not be that useful.  But I think the category
machinery is already there, and I think that bidi categories look very
much like the categories defined for phonetic classifications.  So it
seemed right.

BTW, I like this assignment you gave me.  I've read the MULE paper by
Handa et al., I've read lots of code in lisp/international/, I've read
a lot of stuff on the Unicode web pages.  :)

Alex.
-- 
http://www.emacswiki.org/