Re: Emacs pretest 24.0.90

[Eli Zaretskii <eliz@gnu.org>]

> Date: Wed, 12 Oct 2011 06:04:31 -0700 (PDT)
> From: nabil-82 <sjanjal@web.de>
> 
> After testing the arabic input method I found another problem.  It's not
> emacs-specific. I have it with EVERY bidi-editor in Dos or Unix!  And maybe
> a lot of people will not see any problem and say: "That's what I want!"
> 
> If I type '1' then '2'  in a "L2R paragraph" I get "12" - as expected of
> course.
> 
> But if do this with in a "R2L paragraph" (I've tried it in arabic and
> hebrew) I get on the scren "12". But this is not "normal". Because the
> logical order is 1 then 2. And I'm writing from right to left. So the 1
> should be on the right and 2 the should be on the left. So I think "21"
> should be correct (for me).

What you see is correct behavior according to UBA, the Unicode
Bidirectional Algorithm that Emacs implements (as do other bidi-aware
applications).  1, 2, 3, etc. are the so-called "European Numbers",
and they are always displayed left to right, even in a right-to-left
paragraph.  If you want to have it otherwise, you will have to enclose
the numbers in a "right-to-left override", by typing the RIGHT-TO-LEFT
OVERRIDE character before the group of digits and the POP DIRECTIONAL
FORMATTING character after them.

> As I said you have this problem with every editor. And I see that it is
> easier to make it as I think it should be: ALWAYS from right to left,  also
> for digits and not only for characters - in the R2L-mode. So they must be a
> reason...?

The reason is the UBA, it mandates this behavior.  I don't read Arabic
(unfortunately), so I cannot comment on how this behavior fits or
doesn't fit the accepted practice.