Lists Home |
Date Index |
6/5/2002 9:32:51 AM, "Thomas B. Passin" <email@example.com> wrote:
>Now this is really frightening! How could anyone possibly outguess the
>system to know what the result would be???
Right. I think this is the most compelling response to the eminently
sensible point that Dare made -- "standardization not
balkanization should be the order of the day." The problem is that
W3C XSD is "self-balkanizing" -- it's clear from this thread that the
spec is so complex and obscure that in practice only its most basic
features (roughly those it shares with DTDs an RELAX NG plus the
basic datatypes) can truly be counted on to be interoperable across
implementations and authoring tools, or
understandable by any but the most specialized experts.
It would seem to me that The Way Forward lies in
a) encouraging people to use the RELAX NG non-XML syntax to
author and exchange schema definitions, translating that
to DTD, RELAX NG XML, or (a subset of) W3C XSD syntax as
appropriate. (I THINK tools exist to do this ...)
b) encouraging the W3C Schema folks to define conformance level(s)
that correspond more closely to the features of DTDs and
RELAX (+ basic data types) so as to return us to a common
set of basic schema functionality layered on XML per se.
Those that need the advanced features of W3C XSD or RELAX,
of course, should use the best tool for the job.
c) voting with your feet for whichever REALLY provides the right
mix of features, formalization, usability, and interoperability
on a project-by-project basis.
d) cooperating under the aegis of the ISO DSDL effort to create
real standarization -- a menu of choices that reflect best
practices learned by the actual application
of these various schema languages.
[my humble opinion, not that of my employer, of course]