Re: idn.el and confusables.txt

[Eli Zaretskii <eliz@gnu.org>]

[Resending, as the original message didn't get through.]

> From: Kenichi Handa <handa@m17n.org>
> Cc: emacs-devel@gnu.org
> Date: Fri, 17 Jun 2011 17:15:23 +0900
> 
> In article <tl7r57wo2wc.fsf@m17n.org>, handa <handa@m17n.org> writes:
> 
> > But, at first, I'll work on improving C interface to
> > uni-*.el.
> 
> Attached is the first version for that.  It provides two C
> functions (excerpt from chartab.c).

Thanks!

> 	Lisp_Object bidi_class_table = uniprop_table (intern ("bidi-class"));
> 
>    and get a property value for character CH as this:
> 
> 	Lisp_Object bidi_class = uniprop_table_lookup (CH, bidi_class_table);
> 
>    In this case, what you actually get is an index number to the
>    vector of property values (symbols nil, L, R, etc).  See the
>    comment of uniprop_table_lookup for the detail.  */
> 
> Eli, I arranged that bidi_class got as above is an integer
> value that is the same as enum bidi_type_t defined in
> dispextern.h.

This sounds good, but I think we should add a comment to both
dispextern.h and where you create the properties regarding the fact
that the order must be in sync if any changes are made in any of these
two places.

> And, updata_table (intern "mirroring") gives a char-table for
> mirroring character.

And what is the entry in that table for characters that don't have a
mirrored character defined by BidiMirroring.txt?  Is it the character
itself or nil?

Also, uni-mirroring.el will clash with uni-mirrorred.el on 8+3
filesystems.  Can't we use the same uni-mirrorred.el file for both
tables?

> Could you check if the attached patch gives sufficient
> facility to be used in the bidi code?

I will do that, thanks.  But please wait for a while, as I'd like to
finish the changes needed for string reordering first.

Thanks again for working on this.