- From: Stephen Deach <sdeach@adobe.com>
- Date: Mon, 15 Aug 2005 07:46:04 -0700
- To: Martin Duerst <duerst@it.aoyama.ac.jp>, Ognyan Kulev <ogi@fmi.uni-sofia.bg>, Stephen Deach <sdeach@adobe.com>
- Cc: Tex Texin <tex@xencraft.com>, Addison Phillips <addison.phillips@quest.com>, www-international@w3.org, Richard Ishida <ishida@w3.org>, Bert Bos <bert@w3.org>
I would like to see your list of languages using RTL scripts. The only scripts identified as RTL in Unicode are Arabic and Hebrew. (Then there is the strange case of Mongolian which is marked as LTR but I think should be treated as "RTL rotated to read top-down".) As you & I both said, nearly all of the cases where you would need to mark directionality are either detected by the Unicode BiDi Algorithm or coincide with semantic markup, I would need to question the need to do anything further, especially if preferred directionality (the embed cases) can be tied to xml:lang (which I think will cause the override cases to disappear). Maybe the language/country/script codes should be fixed/completed rather than adding another "layer" of markup or a new xml:dir attribute, I think these are tightly coupled and need to be identifiable in general XML markup. I question the value of marking them independently (it's hard enough to get people to mark language at all, let alone correctly). I would prefer to see a markup/attribute approach in XML rather than character-level Unicode markers, since it is too easy to get the markup and the markers out of sync. (Except for people on these working groups, most people tend to understand "incremental state" or "hierarchical markup", but not both. -- Mixing the two is asking for trouble.) At 2005.08.15-16:54(+0900), Martin Duerst wrote: >At 15:32 05/08/10, Ognyan Kulev wrote: > > >Isn't direction implied in xml:lang which is part of the core XML spec? > >No. This was considered for html, but the problem is that there are too >many languages written with Arabic, Hebrew, and other RTL scripts, and >some of them (the rarer ones) may not yet have a code, so it would be >difficult to implement this in a browser. That's why it was rejected. > >Regards, Martin. ---Steve Deach sdeach@adobe.com
Received on Monday, 15 August 2005 14:47:00 UTC