- From: Martin J. Duerst <mduerst@ifi.unizh.ch>
- Date: Sun, 15 Jun 1997 17:39:42 +0200 (MET DST)
- To: David Rashty +972-2-6584848 <RASHTY@WWW4.HUJI.AC.IL>
- cc: Dani Ilan <standard@NetVision.net.il>, Stefan Fuchs <sf@bezeq.co.il>, Israel Ervin Gidali <gidali@vnet.ibm.com>, Nati Guedalia <natig@ncc.co.il>, Gil Mor <gilmor@microsoft.com>, Mati ALLOUCHE <matia@vnet.ibm.com>, Moshe Shalom <moshe_shalom@easx.co.il>, Yevgenia Palanker <pal@actcom.co.il>, Doron Shikmoni <P85025@VM.BIU.AC.IL>, Francois Yergeau <yergeau@alis.com>, Gavin Thomas Nicol <gtn@ebt.com>, Glenn Adams <glenn@spyglass.com>, www-html@w3.org, Khaled Sherif <sherifk@caivm1.vnet.ibm.com>, Jeff Rosenschein <jeffr@accentsoft.com>, Chris Wendt <christw@microsoft.com>, John McConnell <johnmcco@microsoft.com>, Yaniv Feinberg <yanivf@microsoft.com>, Erik van der Poel <erik@netscape.com>, Uri Postavsky <urip@rtlsoft.com>, Uzzi Ornan <ORNAN@cs.technion.ac.il>, Chris Lilley <Chris.Lilley@sophia.inria.fr>, "Mark H. David" <mhd@WORLD.STD.COM>, Edward Resnick <Edward.Resnick@Israel.Sun.COM>, ILAN Hebrew List <ilan-h@taunivm.tau.ac.il>, Jonathan Rosenne <rosenne@NetVision.net.il>
On Sun, 15 Jun 1997, David Rashty +972-2-6584848 wrote: > tHERE ARE TWO METHODS > 1) tO USE THE <PRE> TAG, THIS IS THE LESS USED METHOD > 2) TO uSE THE TAG WHICH ALIGN THE TEXT TO THE RIGHT AND ADD LINE BREAKS. > THE BEST SOLUTION USED IS TABLE, WE USE THE FOLLOWING > <TABLE WIDTH="X%"><TR><TD ALIGN="RIGHT" NOWRAP> > > THE NOWRAP PREVENTS THE LINE FROM BEING WRAPED IN THE WRONG PLACE SINCE hEBREW > IS WRITTEN FROM RIGHT TO LEFT AND THE LINE BREAKING IS DONE LEFT TO RIGHT > > tHERE ARE ALSO OTHER SOLUTIONS LIKE <P ALIGN="RIGHT"> I see. So there is indeed considerable effort going into making some visually coded text appear correctly on old browsers. And texts marked up this way won't display nicely, or even readably, on browsers conforming to RFC 2070 or Cougar (which included RFC 2070). So there always have to be two document versions, one visual for old browsers, and one with real BIDI for new browsers. The distinction between iso-8859-8 and iso-8859-8-X would then only have the effect to help negotiating these versions; i.e. old browsers would send Accept-Charset: iso-8859-8, and new browsers would send Accept-Charset: iso-8859-8-X. Is this a workable model? It implies that servers store two versions. Servers that don't want to store two versions can use visual order and also use <BDO> in order to tell newer browsers to treat things the way it is intended. If everything is tagged with <BDO>, the difference between iso-8859-8 and iso-8859-8-X disappears. > >>Having the "charset" parameter value affect generic line parsing > >>mechanisms seems not very desirable. As far as I understand, even > >>now special markup is neccessary to have acceptable results. > >>Visual order can be kept by adding <BDO> (bidirectionality > >>override) at the right places (although that was not the original > >>puropose of BDO). BDO has the advantage that it won't affect old > >>browsers that don't do BIDI at all. > > When you specify in the charset for Visual order, you command the browser > to parse the text as English Text. this is an exception. This is not what RFC 2070 nor Cougar is doing. Overloading BIDI semantics with "charset" parameter may have been necessary for email, but we want to move away from it. And having to tell developpers: Implement the full BIDI algorithm, but then please also have a method for switching it off, depending on the "charset" is not what will get a lot of work done in this area. Regards, Martin.
Received on Sunday, 15 June 1997 11:44:52 UTC