Re: Indexing requirements for HTML and other document types

Dave Raggett (dsr@hplb.hpl.hp.com)
Fri, 3 Jun 94 11:41:18 BST


From: Dave Raggett <dsr@hplb.hpl.hp.com>
Message-Id: <9406031041.AA06476@dragget.hpl.hp.com>
Subject: Re: Indexing requirements for HTML and other document types
To: fielding@simplon.ICS.UCI.EDU
Date: Fri, 3 Jun 94 11:41:18 BST
Cc: www-html@www0.cern.ch

Roy,

Thanks for responding. My message was perhaps too open ended. :-)

I do believe that we should decouple indexing info from the evolution
of the HTML DTD. The question is what data model is needed and how to
represent this as cleanly as possible in such a way that minimizes
problems for current/future browsers.

The META element allows one to include a set of attribute/value pairs.
Perhaps we need to support a data model which caters for multiple
inheritance. We also need to say something about the role of the
info, e.g. for indexing, for HTTP, ...

A further issue is dealing with indexing info in the document body
in an open ended way.

One route is to formally extend the HTML DTD as explained by Bert Bos
One can then use FIXED attributes to model inheritance as suggested
by Dan Connolly. This requires some sophistocation to parse ...

Another route would be to add a ROLE attribute to META, e.g. ROLE="INDEX"
and another tag for declaring isa relations, e.g.

        <ISA OBJECT="fido" CLASS="dog">
        <ISA OBJECT="bob" CLASS="democrat">

If we need to elaborate this much further then Dan's suggestion of a
separate language (Prolog or Lisp) would make more sense.
--
Best wishes,

Dave Raggett

-----------------------------------------------------------------------------
Hewlett Packard Laboratories              email: dsr@hplb.hpl.hp.com
Filton Road                               tel:   +44 272 228046
Stoke Gifford                             fax:   +44 272 228003
Bristol BS12 6QZ
United Kingdom