W3C home > Mailing lists > Public > public-iri@w3.org > May 2003

Re: Closing issue [arabicnum-03]

From: Matitiahu Allouche <matial@il.ibm.com>
Date: Thu, 8 May 2003 18:01:23 +0300
To: momoi@netscape.com (Katsuhiko Momoi)
Cc: Martin Duerst <duerst@w3.org>, "Rasha Morgan" <rasham@eg.ibm.com>, public-iri@w3.org, "Ahmed Talaat" <AHMEDT@eg.ibm.com>, "Gilan Felfela" <gilanf@eg.ibm.com>, "Tarek Abou Aly" <Tarek_Abou-Ali@eg.ibm.com>, "Mark Davis" <mark.davis@us.ibm.com>, Mark Davis <mark@macchiato.com>, Simon Montagu <smontagu@netscape.com>, Yung-Fong Tang <ytang0648@aol.com>
Message-ID: <OFAF0BC4C5.502CA995-ONC2256D20.005266D7-C2256D20.0052C3A1@telaviv.ibm.com>

If I remember well, the preference for number shaping appears (in Windows 
Regional Options) only when the selected locale is an Arabic one.  This 
preference determines whether digits (hexa 30 to 39) should be presented 
always as European numbers, always as Hindi numbers or depending on 
context.  Depending on context means that hexa 30 to 39 are presented as 
European numbers when the last previous letter was not an Arabic one, and 
as Hindi numbers if it was an Arabic letter.

This is not undefined, nor unpredictable and varying, to use Martin's 

Section 13.3 in TUS 3.0 only deprecates using Numeric Shape selectors 
(U+206E and U+206F) to control the glyphs used for presenting numbers. The 
last sentence says "This state (nominal) is the default state in the 
absence of any numeric shape selector or a higher-level protocol", which 
leaves the door open to selection of digit glyphs based on a higher-level 
protocol.   Preferences at the Operating System level or at the browser 
level are definitely a higher-level protocol, and they should be 
considered good Unicode citizens, IMHO.

So it seems to me that Netscape is performing correctly (I don't remember 
how it sets its default handling of numeric shaping, probably not based on 
Windows preferences, since it tries to be platform independent), and IE 
behaves differently because it relies on the Windows preferences, which on 
a Japanese-locale will always select the European glyphs for numbers.

Shalom (Regards),  Mati
           Bidi Architect
           Globalization Center Of Competency - Bidirectional Scripts
           IBM Israel
           Phone: +972 2 5888802    Fax: +972 2 5870333    Mobile: +972 52 

To:        Martin Duerst <duerst@w3.org>
cc:        Rasha Morgan/Egypt/IBM@IBMEG, public-iri@w3.org, Matitiahu 
Allouche/Israel/IBM@IBMIL, Ahmed Talaat/Egypt/IBM@IBMEG, Gilan 
Felfela/Egypt/IBM@IBMEG, Tarek Abou Aly/Egypt/IBM@IBMEG, Mark 
Davis/Cupertino/IBM@IBMUS, Mark Davis <mark@macchiato.com>, Simon Montagu 
<smontagu@netscape.com>, Yung-Fong Tang <ytang0648@aol.com> 
Subject:        Re: Closing issue [arabicnum-03]

Please refer to the following bug for this issue:


Rasha, I think you should argue your position in the above bug if you
feel strongly but Simon indicates in his latest comments that Gecko
should follow Section 13.3 of Unicode 3.0 as referred to by Martin.

- Kat

Martin Duerst wrote:

> Hello Rasha,
> Many thanks for your comments.
> At 14:09 03/05/06 +0300, Rasha Morgan wrote:
>> Hello Martin,
>> Regarding numeric shapes in the browser, they should follow the system
>> settings,
> I have a Japanese system, and don't remember having made such
> a system setting. And you seem to imply that if such a system
> setting is present, all the digits would appear as Arabic
> digits. Fortunately for me, that's not the case. Even on the page
> in question (http://www.w3.org/International/iri-edit/BidiExamples),
> only digits in vicinity of Arabic characters change shape.
> I'm not sure what the rule/algorithm is that Netscape uses for this.
> I don't like undefined (and therefore unpredictable and varying)
> behavior.
>> so if it is contextual or national they should appear in Arabic
>> digits. Otherwise they should appear in European digits.
> Who made up this rule? Unicode clearly states otherwise, as I already
> have explained (see below). And HTML and XML are based on Unicode.
>> So Netscape is working properly, please don't ask to disable this
>> feature.
>> also IE behaves the same way.
> In my case, IE doesn't, as I have explained before (see below).
>> Numeric shapes are part of the country culture and should not be
>> deprecated.
> I haven't said anything that would deprecate Arabic(-Indic) digits.
> Unicode has them, and anybody who wants to use them can use them.
> However, I wanted the digits in my page, which are European digits,
> to be displayed as European digits, and I think any browser (or other
> application, for that matter) that tries to change this behind my
> back is doing something wrong.
> Regards, Martin.
>> Have a nice day
>> Rasha Morgan
>> Advisory Globalization Specialist, team leader
>> IBM Egypt Branch,
>> ----- Forwarded by Rasha Morgan/Egypt/IBM on 06/05/2003 01:18 -----
>> ----- Forwarded by Ahmed Talaat/Egypt/IBM on 04/05/2003 12:47 -----
>> ---------------------- Forwarded by Matitiahu Allouche/Israel/IBM on
>> 04/05/2003 09:50 ---------------------------
>> Martin Duerst <duerst@w3.org>@w3.org on 02/05/2003 23:57:44
>> Sent by: public-iri-request@w3.org
>> To: public-iri@w3.org
>> cc:
>> Subject: Closing issue [arabicnum-03]
>> This serves to close
>> http://www.w3.org/International/iri-edit/#arabicnum-03.
>> I had put this issue on the issues list because in examples
>> 7-9 of the BidiExamples page
>> (http://www.w3.org/International/iri-edit/BidiExamples),
>> I saw Arabic-Indic digits in the Arabic examples rather
>> than the European digits that I was expecting.
>> Further examination has shown that this is a browser issue.
>> Netscape 7 (/Mozilla) is the only browser that I have found
>> that converts digit shapes when displaying them. Tango,
>> IE6, Safari, and Opera don't change digit shapes.
>> Section 13.3 (http://www.unicode.org/book/ch13.pdf, p. 320)
>> of Unicode 3.0 is clear that nominal behavior (not changing
>> the glyphs used to display the digits) is correct, and
>> using national digit shaping would only be acceptable if
>> using the deprecated and strongly discouraged numeric
>> shape selectors (which I have of course not used).
>> So Netscape/Mozilla is wrong here, and should be fixed.
>> I have told somebody in the i18n team at Netscape.
>> I have added a note at the top of the BidiExamples page
>> saying that a browser doing digit shaping correctly should
>> be used. I have closed this issue.
>> Regards, Martin.
Katsuhiko Momoi <momoi@netscape.com>
Senior International Manager, Web Standards/Embedding
Netscape Technology Evangelism/Developer Support
Received on Thursday, 8 May 2003 11:06:00 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:39:37 UTC