[www-tag] <none>

Message-ID: <3DDE8086.6010807@textuality.com>
Date: Fri, 22 Nov 2002 11:07:50 -0800
From: Tim Bray <tbray@textuality.com>
To: noah_mendelsohn@us.ibm.com
Cc: WWW-Tag <www-tag@w3.org>
Subject: Re: Let's get some principles nailed down

> I'm unconvinced that using the Infoset is the right way to accomplish this 
> goal.

Are you talking about the W3C Infoset spec, which was clearly not
intended to be the basis for this kind of thing?  Or are you talking
about the while idea of an formal definition of XML as an tree
data structure that a parser, or DOM program, or SOAP processor,
or whatever could produce and manipulate?  As much as it might seem 
attractive to simply define XML as the syntax, I think the can of worms was 
opened long ago (I *know* we
wrestled with it in the DOM WG in 1997, and I think that was one
inspiration for the InfoSet spec).

It seems to me to be more of a "wave-particle duality" thing rather than a 
"one true definition" thing. XML is syntax ... XML is a data model ... The 
data model is produced
by parsing the syntax, the syntax is produced by serializing the data
model ... some specs are more efficiently and interoperable written
to the syntax, others to the data model, other serializations of the
data model are possible for specialized purposes,  but all are equally "XML

If there are any architectural issues in the XML corpus that need
the collective wisdom of the TAG, it's this rat's nest!
a) the syntax-data model duality itself and how to live with it, b) 
reconciling the multiple definitions of the data model (InfoSet, DOM, 
XPath, XQuery, ...)
into a solid basis for moving forward, acknowledging that some existing 
code will break in the process.
c) best practices for maximizing interoperability in a (furture) world
where XML 1.x syntax is not the only serialization of the
XML data model in widespread use.


[speaking only for myself, not any WG I'm affiliated with or my
employer]

Received on Friday, 22 November 2002 15:05:54 UTC