W3C home > Mailing lists > Public > www-international@w3.org > July to September 2005

Re: bidi discussion list was: Bidi Markup vs Unicode control characters

From: Martin Duerst <duerst@it.aoyama.ac.jp>
Date: Mon, 15 Aug 2005 17:09:01 +0900
Message-Id: <6.0.0.20.2.20050815170254.07333ec0@itmail.it.aoyama.ac.jp>
To: Bert Bos <bert@w3.org>, www-international@w3.org

At 08:30 05/08/09, Bert Bos wrote:
 >
 >On Mon, Aug 08, 2005 at 01:13:49PM -0700, Tex Texin wrote:
 >>
 >> Sandra, yes of course. I am not suggesting CSS be made a requirement for
 >> bidi. Only that it should be acceptable and considered best practice in
 >> the right context.
 >>
 >> At the time the standard was written, I took the warning that bidi css
 >> may be ignored by user agents to be a reflection of the state of the art
 >> at the time, since most browsers were not able to support bidi with or
 >> without css.
 >
 >As Fantasai said, it is a design goal of CSS that CSS can be turned
 >off without affecting the meaning of the document (first bullet of
 >section 2.4 http://www.w3.org/TR/CSS21/intro.html#q6).
 >
 >It's sometimes a difficult goal to follow. For example, generic XML
 >doesn't have hyperlinks and it would be quite simple to create a
 >property in CSS to make elements into hyperlinks. Such a property has
 >been tested in practice and it doesn't need to be very complex to be
 >very useful. But so far we are hoping for a solution outside CSS.
 >
 >Bidi properties are similar, but there we have gone the other way:
 >generic XML doesn't have bidi attributes and nobody was even talking
 >about adding them (unlike for links, where XLink has been attempted),
 >so we had no choice but to add bidi properties to CSS.

It may be that this gap is filled in by the ITS WG. Whether their
result will be used as 'generic' bidi attributes (in the sense that
eventually every XML stack should just understand them) or just
as a 'preferred' set of attributes with CSS glue for rendering
remains to be seen. Maybe a bit of both.

 >That's why CSS (in section 9.10
 >http://www.w3.org/TR/CSS21/visuren.html#direction) indeed tells people
 >not to use these properties. In general, creators of an XML-based
 >format should tell how the elements/attributes of their formats map to
 >CSS 'direction' & 'bidi-override' properties and other people should
 >not override them.

Yes, this is a very important point. A stylesheet for an XML document
should just include the stylesheet for the bidi properties of the
relevant document type, and not mess around with them at all, except
for some very special situations (the typical example is to render
Hebrew-encoded Yiddish with a font that substitutes Latin glyphs;
in that case, the directionality needs to be tweaked).


 >> It had not occurred to me that people were interpreting it to mean that
 >> css implementers would intentionally choose not to support it or that
 >> W3C was endorsing such a view. That would be counter to goals for
 >> internationalization and making the web accessible to all.
 >
 >The CSS2 spec used to say that a program was allowed to ignore the
 >properties in the case of HTML documents, but it doesn't say that
 >anymore. There is no difference anymore between HTML and XML in that
 >respect.

When and why has this been changed? To me, the change does not make
sense, and should be reverted.


 >Individual profiles can still say which properties are optional and
 >both the Mobile Profile and the Print Profile say the bidi properties
 >are not required for conformance. I believe that is because both
 >profiles are in practice only used with XHTML1 and thus the properties
 >aren't needed.

If that is the case, this would indeed make sense.


Regards,    Martin. 
Received on Monday, 15 August 2005 10:48:41 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 19:17:05 GMT