W3C home > Mailing lists > Public > www-xml-schema-comments@w3.org > April to June 2000

Re: Note on DTD-related questions

From: Dan Connolly <connolly@w3.org>
Date: Fri, 16 Jun 2000 13:21:33 -0500
Message-ID: <394A702D.CCDF1031@w3.org>
To: Murray Altheim <altheim@eng.sun.com>
CC: w3c-xml-schema-ig@w3.org, www-xml-schema-comments@w3.org

Murray Altheim wrote:
> I'm sorry that I haven't been able to provide a detailed summary of
> comments on the hard work that Michael, Eve and Norm provided in
> their response to Ann Navarro and my comments on the XML Schema
> specification [1].
> While my present schedule has prevented this summary from being made
> public, I think that given the current deadline for comments that
> I'd at least like it to be known that I don't believe the solutions
> provided by [1] to be sufficient to enable definition of XHTML or
> particularly XHTML's modularization using the XML Schema grammar;

That seems like a hasty conclusion. I have been working
on the details of various modularization issues
(in http://www.w3.org/XML/2000/04schema-hacking/), and I have
not found any of them to be insurmountable. I am
pretty close to being able to convert the XHTML DTD modules
to schema syntax by machine.

> For example, the ability to easily redefine a content model is not
> a feature available in XML Schemas, except for the less-than-
> interesting case of appending element types at its end. Ie., it's
> impossible to add a <preface> near the beginning of a content model
> if that is desired without modifying the base Schema itself. In a
> DTD this is accomplished rather easily by redeclaring the content
> model in almost any number of places.

Putting declarations in a DTD doesn't automatically teach all
the consumers what to do with the <preface> element.

> This is a common requirement
> that is not satisfied (as far as I can understand from the
> Structures spec and the "Note on DTD-related questions".

Is it really a common requirement to be able to arbitrarily
override a content model this way? i.e. are DTDs really
used to represent contracts that are that weak?

If a langauge designer intends that sort of weak contract,
they can express it using <any/>, no?

In general, "the ability to easily redefine ..." just about
anything doesn't seem like a good thing. A big part
of the value of schemas is being able to keep definitions
straight, and to relate one definition to another. If I define that
all cars have 4 wheels, I don't want somebody else to be able
to say "no, I'm changing his definition: some cars
have 6 or 10 or 18 wheels". It's fine if they say
"vehicles are a genearlization of cars, and vehicles may have
1, 2, 4, 6, 10, or 18 wheels." But to just erase one
definition and put something else in its place seems
to muddy things.

> And any
> possible solution shouldn't require any rocket-science: the solution
> in modular DTDs is fairly straightforward if the DTD is regular
> and well-documented.

The problem is: even though a DTD may be regular and well-documented,
we can't exploit the regularity by machine. We can't check constraints
like "the the value of the expires attribute should be in YYYY-MM-DD
form" when they're expressed ala
	<!ATTLIST reservation
		expires %Date.fmt; #IMPLIED>

We could develop a specification for stuff like %Date.fmt;
and %foo.cm; and %bar.attrGroup; or something, but we've chosen
to use <simpleType/> and <attributeGroup/> and such in stead,
for reasons that I think are clear.

> Without going into great detail I think the consequences of the
> rather profound differences between the flexibilities inherent in
> highly parameterized DTDs (given both the pros and cons therein)
> and the well-structured and inheritance-based constraints in XML
> Schemas mean that certain features of DTDs are either impossible
> or impractical in Schemas, and vice-versa. I do not find XML Schemas
> to be amenable to XHTML modularization *if* that is a priority.

I think it is.

> I
> might note that the idea of striving toward the creation of a data-
> centric schema for a document-centric vocabulary might in the end
> be less than necessary or important, ie., the project itself might
> be a lesson in futility, less than valuable for any particular
> community, a marketing exercise perhaps.
> Again, I apologize that I have not been able to be more explicit
> in my response, and would endeavour to clarify any comments that
> I have made if there seems to be any misunderstanding. I hope that
> this note and any subsequent comments fulfills my responsibility
> in commenting on these specifications.
> Thank you all for your patience,
> Murray
> [1] "Note on DTD-related questions", Eve Maler, C. M. Sperberg-McQueen,
> Norm Walsh, 20 April 2000
> http://lists.w3.org/Archives/Member/w3c-xml-schema-wg/2000Apr/att-0019/01-dtdissues-reply.html
> ...........................................................................
> Murray Altheim                            <mailto:altheim&#x40;eng.sun.com>
> XML Technology Center
> Sun Microsystems, Inc., MS MPK17-102, 1601 Willow Rd., Menlo Park, CA 94025
>       In the evening
>       The rice leaves in the garden
>       Rustle in the autumn wind
>       That blows through my reed hut.  -- Minamoto no Tsunenobu

Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Friday, 16 June 2000 14:24:45 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:49:52 UTC