Re: What is the advantage of RELAX in comparison to Schemas?

[Jonathan Borden <jborden@mediaone.net>]

Bob Kline wrote:
> On Tue, 30 Jan 2001, Bullard, Claude L (Len) wrote:
>
> > I think yes we agree.  Then would you also agree we are back to the
> > problem of what a namespace signifies?
> >
> > That is, if we must live with the XML Schema as not having
> > context-aware contraints, in many cases dereferencing the namespace
> > URI/URL must return information (eg, a RDDL) that enables the
> > interpreter of the document to locate all of the constraints
> > explicitly defined by the sender as needed for interpretation?
>
> I'll have to thing about that solution.  We've sidestepped Namespace
> complications for this project as much as we could get away with (so
> far).
>

    It sounds like you have some requirements that indeed might be best met
by a combination of schemata. For example a grammar based schema such as
DTD/XML Schema/RELAX/TREX combined with a set of rules such as Schematron.
Or perhaps a set of constraints such as RDF Schema.

    No one type of schema claims to solve all the problems of computer
science. It is most often necessary to apply a set of "business rules" to a
dataset which has already been constrained by a schema. Take for example how
databases have been used for decades regardless of anything to do with XML.

    I think that Schematron fits this role (of applying 'business rules'
quite nicely). The real question then becomes: what constraints can we
declare using XPath vs. what constraints will still need to be tested using
a custom piece of code.


> I'd like to think that the Schema WG might reconsider its decision to
> omit context-aware constraints, though I'm aware that it's unlikely to
> happen any time soon.
>

    While I'm not opposed to subsuming XML Schema's pattern facet which
matches regular-expressions into a more general rule facet -- which would
match against an XPath extended with a regexp() function, it is late in the
XML Schema game and we do have ways to solve such problems outside of XML
Schema.

   A RDDL aware piece of code could implement a multistep validation test

based on natures:

   boolean multistep-validation(document,
http://www.w3.org/2000/10/XMLSchema, http://www.ascc.net/xml/schematron)

or purposes:

    boolean multistep-validation(document,
http://www.rddl.org/purposes#schema-validation,
                            http://www.rddl.org/purposes#rule-validation)


-Jonathan