[
Lists Home |
Date Index |
Thread Index
]
Mike Champion scripsit:
> "Infoset" people can also be minimalists, or they
> may have stopped worrying and learned to love the PSVI, the XQuery
> type system, etc.
Oddly, even being an editor of the Infoset I'm still more of a syntax
person.
> Another way forward, which I doubt if many people will agree with,
> might be refactor things along SGML "markup for authors" and
> XML "infoset for programmers" lines. Agree on a basic syntax
> for XML 2.0 that removes most the stuff that the infoset throws away
> and causes the DOM (which basically tries to live in both the
> syntax and InfoSet worlds) fits, such as DTDs, entities, entity references,
> CDATA sections. That's not to say that people should stop using
> entities and CDATA sections, just to say that they "properly" belong in
> the SGML world where "syntax sugar" is respected and supported.
What I'd like to see for XML 2.0 is a clean separation into two layers:
a "top layer" that understands elements, attributes, namespaces, ids, and PIs;
and a "bottom layer" that understands character references (by name or number),
comments, and literal sections. Each layer would be separately processable;
perhaps it would be good for all bottom-layer constructs to be introduced
by "&", e.g. "&-- ... &--" for comments and "&< ... &>" for literal sections.
This would mean a break with the reference concrete syntax of SGML, but
perhaps would still be declarable as a concrete syntax -- I don't know enough
to say if that's true.
--
One art / There is John Cowan <jcowan@reutershealth.com>
No less / No more http://www.reutershealth.com
All things / To do http://www.ccil.org/~cowan
With sparks / Galore -- Douglas Hofstadter
|