W3C home > Mailing lists > Public > www-international@w3.org > July to September 2008

Re: Weird bidi behavior?

From: Martin Duerst <duerst@it.aoyama.ac.jp>
Date: Thu, 28 Aug 2008 10:35:20 +0900
Message-Id: <>
To: Matitiahu Allouche <matial@il.ibm.com>
Cc: www-international@w3.org, www-international-request@w3.org

Hello Mati,

Many thanks for your very quick and precise answer.

Thinking over dinner and breakfast (or some other similar
occasions between my two work days), I started to suspect
as much as you confirm below.

At 07:22 08/08/28, Matitiahu Allouche wrote:

>   Hello, Martin! 
>It seems to me that it is working as designed. 

But would we be able to say that it works as intended?

Of course, intent is difficult to capture, but my guess
would be that to a large number of people, a LTR embedding
is something very inherently and strongly LTR, and should
therefore behave as such to the outside.

Also, one could immagine that it would be desirable to
have the following preservation property of the bidi
algorithm: If a sequence of strong LTR characters
(or, a sequence of characters displayed currently as
LTR) is wrappend in an LTR embedding, the display order
doesn't change (and of course, same for RTL). The current
behavior clearly doesn't exhibit this preservation property.

Do you (or anybody else) have any idea why the algorithm is
designed this way? Was this aspect overlooked? Are there other
considerations that justify this behavior? Would it be too
difficult to describe an algorithm that exhibited the above
preservation property? Is this case too rare in practice to
be relevant?

Regards,    Martin.

>Case 1 has a default direction equals to LTR, meaning a paragraph embedding level equal to 0.  The Arabic words implicitly raise the level to 1 at their location.  Simple. 
>Case 2 starts similarly, i.e at embedding level 0.  The span with dir="ltr"  is equivalent to a LRE.  It raises the embedding level to the next even number, which is 2 (rule X3 of UAX #9).  The English words within the span will receive level 2, and the Arabic word will receive level 3 (rule I1). 
>The Arabic word following the span is at level 1, like in case 1 (rule I1). 
>So you have a span at level 2 immediately followed by an Arabic word at level 1.  The regular rules (L2) say that the span will be displayed on the right side of the Arabic word. 
>Is that clear?
>Shalom (Regards),  Mati
>          Bidi Architect
>          Globalization Center Of Competency - Bidirectional Scripts
>          IBM Israel
>          Phone: +972 2 5888802    Fax: +972 2 5870333    Mobile: +972 52 2554160
>Martin Duerst <duerst@it.aoyama.ac.jp> 
>Sent by: www-international-request@w3.org 
>27/08/2008 13:44 
>Weird bidi behavior? 
>Can somebody explain the bidi behavior of Case 2 at
>http://www.sw.it.aoyama.ac.jp/2008/pub/bidi-test.html ?
>The only difference between Case 1 and Case 2 is that
>Case 2 uses dir='ltr' on the green <span>.
>Regards,    Martin.
>#-#-#  Martin J. Du"rst, Assoc. Professor, Aoyama Gakuin University
>#-#-#  http://www.sw.it.aoyama.ac.jp       mailto:duerst@it.aoyama.ac.jp     

#-#-#  Martin J. Du"rst, Assoc. Professor, Aoyama Gakuin University
#-#-#  http://www.sw.it.aoyama.ac.jp       mailto:duerst@it.aoyama.ac.jp     
Received on Thursday, 28 August 2008 02:00:30 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 22:40:56 UTC