W3C home > Mailing lists > Public > xmlschema-dev@w3.org > March 2002

Re: Choice

From: <noah_mendelsohn@us.ibm.com>
Date: Mon, 11 Mar 2002 09:09:36 -0500
To: Jeni Tennison <jeni@jenitennison.com>
Cc: xmlschema-dev@w3.org
Message-ID: <OFAE04E015.45A6902B-ON85256B79.004CD5A9@lotus.com>
I think we generally agree.  I still have some suspicions regarding XSL 
performance, and the degree to which tools can "grok" what a stylesheet is 
doing.  If I want to say:  "either this attribute or those elements" or 
"the integer value of this attribute must match the number of elements 
that occur as children", I'm not sure it should require a theorem prover 
for a tool to figure out what's going on.  I also worry about performance 
(and lack of bounds on performance) and streaming characteristics of full 
XSL, when considered as a basis for extended schemas.  Then again, XSL is 
a recommendation (and a good one for many purposes), and Schematron has 
shown how nicely it can be applied to certain problems.  Relax-like 
technology has some interesting pros and cons too.  I think the schema WG 
should compare a variety of approaches to co-occurrence constraints before 
settling on one.

Noah Mendelsohn                              Voice: 1-617-693-4036
IBM Corporation                                Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142

Jeni Tennison <jeni@jenitennison.com>
03/09/02 07:56 AM
Please respond to Jeni Tennison

        To:     noah_mendelsohn@us.ibm.com
        cc:     xmlschema-dev@w3.org
        Subject:        Re: Choice

Hi Noah,

> Also: Schematron is relatively declarative; my concern was primarily
> with the possible jump to non-declarative languages. It's much
> easier for tools to reason about constraints captured declaratively
> (e.g. in schema facets or derivations, or even Schematron
> assertions) than imperatively (in a Java loop).

Yep. Naturally I was thinking of XSLT as the Turing-complete,
W3C-approved, declarative language of choice.

> The ability to build tools that manipulate and derive information
> from schemas is of key importance. Writing programs that validate
> documents or types may not get us that.

Absolutely. In my mind, there are three primary purposes to schemas -
validation, documentation and tool support. A rule-based approach is
great for validation, and as Schematron shows can be combined with
documentation, but really suffers in terms of providing support for
tools (for helping people author XML, or for analysis prior to
query/transformation for example). I think that the object-oriented
approach of XML Schema provides a big win in this area, and I'm
certainly not advocating that this is lost.

However, I think that at the moment XML Schema's validation power and
flexibility suffers because of its focus on tool support. I am simply
arguing that incorporating a rule-based approach seems a neat way of
correcting that balance.

Anyway, I'm sure that the XML Schema WG are considering all kinds of
changes; just thought a little bit of user input couldn't hurt once in
a while.



Jeni Tennison
Received on Monday, 11 March 2002 09:24:44 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:55:55 UTC