Re: BIDI question

[Eli Zaretskii <eliz@gnu.org>]

> Date: Mon, 07 Oct 2013 21:25:54 +0400
> From: Dmitry Antipov <dmantipov@yandex.ru>
> CC: Eli Zaretskii <eliz@gnu.org>
> 
> There is a text with a mix of Arabic and ASCII characters (attached), and it is
> displayed correctly (first screenshot a.k.a top.png). But, when I concatenate
> this text with itself (cat /tmp/uthmani.txt /tmp/uthmani.txt > 1.txt) and view
> the result, second Arabic fragment is rendered L2R (second screenshot a.k.a
> next.png). Why it is different from the first Arabic fragment, and isn't it a bug?

It's a feature: a paragraph starts after at least one empty line (or
at point-min).  In this case, there's no empty line between the
English text (which is an L2R paragraph) and the second Arabic
fragment, so the second Arabic fragment is rendered as a continuation
of the L2R paragraph that starts with the English text.

IOW, Emacs does not recognize the second Arabic fragment as a new
paragraph, and does not recompute the base paragraph direction for it,
instead it continues using the base paragraph direction computed for
the English text.