Re: It's time for practical XML!

Ron Bourret (rbourret@dvs1.informatik.tu-darmstadt.de)
Tue, 6 Oct 1998 14:02:38 +0200


> I must admit, though, that there is an aspect of the XML effort that
> bothers me, which is the overlap of some of the efforts. I am thinking of
> DCD/XSchema/XData/RDF for instance. I would like to see convergence or at
> least common strategies, nice and clear with pictures and examples. Maybe
> it's too early for that?

Overlapping proposals are an unfortunate reality of working out "standards". It
is easy to think of XML, XLink, namespaces, etc. as having been handed down from
on high, but I have no doubts that the process was just as messy as what you see
now in the schema arena -- witness the overhauls of namespaces and XSL, for
example. The only difference is whose efforts are overlapping and how public
the process is.

It has always been my hope that XSchema would help speed up, rather than hinder,
the standards process by gathering and formalizing a large amount of input early
in the process. (Of course, my motivation for helping in the project was not to
forward standards, but simply to develop a simple schema language that I could
use right now.)

I do agree that the time is ripe for a convergence -- XSchema, DCD, and
Schema-oriented XML all appear to be workable schema languages and all have, I
believe, been at least partially implemented. In spite of their differences,
they actually have more in common than not and cover roughly the same ground.

This latter leads me to believe (rather naively, no doubt) that the W3C
committee working on a schema language (assuming it exists) could simply choose,
in one-from-column-A-two-from-column-B fashion, from the following areas and
mold it into a common whole:

Elements:
* element names
* element nesting (e.g. does an attribute definition need to be inside an
element definition?)
* element value type (attributes vs. PCDATA elements)

Interaction with DTDs:
* degree of DTD duplication, including degree of entity support
* mappings to and from DTDs
* interaction with DTDs

Usability features (as desired):
* container elements
* authoring features (documentation, root, etc.)y
* extensibility elements/strategy
* parameterized element types
* global attribute definition
* modules
* reuse/inheritance strategy (XLinks, ID/IDREF, extends, etc.)
* etc.

Other:
* data type support
* namespace support

Use:
* validation against a schema file
* how to associate a schema file with an instance file

Of course, it was naivete that led me to spend a good part of the summer on
XSchema, so who am I to say? Still, it would be nice...

-- Ron Bourret