Is the problem of backward compatibility so important? I don't see any
reasons why a W3 browser or any other tool could not support several different
DTDs. There are extremely powerful SGML parsers in the public domain. Once
again I say that they can be easily integrated in W3 browsers while keeping the
current parser parts. Then according to the DOCTYPE declaration of the document
received from the server you could decide what parser you'll use.
Nobody said that every HTML document with hyperlinks or inlined images has to be
rewritten. Most current documents conform to HTML 1.0 or HTML 2.0. If people
want to keep them as they are nobody prevents them from doing so. This is a matter
of browsing tools. If a given browsing tool is able to show only 1.0 and 2.0
documents that's all. If it receives an HTML 3.0 document it could try to parse
it anyway and warn the user that the document might not be well interpreted.
Regarding the size of the documents, I can tell you that if you want quality in
your information you'll have to pay the price for disk space. BUT when I see
HTML documents full of images (and everyone knows how big an image can be) I
must say I doubt they will triple in size.
What HTML 3.0 proposes is to add more powerful constructs to do tables, maths and
so forth. Who said it would take twice as much work to create a hyperlink ? It
all depends on the tool you use to produce your document. Of course if you use
emacs or vi you'll have to type in more tagged information. But with editors like
HoTMetaL or other WYSIWYG editors all thos mechanisms could be hidden from the
user. Just make a link with a point and click interface. Of course the new
NMLOC and NMLIST elements would be invisible both in browsers and WYSIWYG editors.
Well the DeRose/Durant solution uses clink and notloc. It's another solution.
Eliot Kimber made the first proposal of a HyTime HTML DTD and I don't know how
close the proposal I made is close to his.
I see the BASE attribute as something useful to help the resolution of external
identifiers in entity declarations. It may be seen as a way to build location
ladders (with only one rung) but this is not the way I see things.
Declaring the entities in the document type declaration could also be an optimization
advantage for browsers. Once the DOCTYPE declaration is parsed they know all
the entities that are going to be used in the document instance and therefore
they can get them in advance.
Thank you for your remarks, Joe.
Best regards,
Christophe
-- Christophe Espert - E-mail: espert@cln46fw.der.edf.fr ======================================================================== EDF - DER | High Text 1, Avenue du General de Gaulle | 5, rue d'Alsace 92141 Clamart CEDEX - FRANCE | 75010 Paris - FRANCE Tel: 33.1.47.65.43.21 ext. 6635 | Tel: 33.1.42.05.93.15 Fax: 33.1.47.65.50.07 | Fax: 33.1.42.05.92.48 ISO 8879:1986 - SGML | ISO/IEC 10744:1992 - HyTime | ISO/DIS 10179 DSSSL