xml-dev - Re: Schemata are not just constraints [was: "RDF + Topic Maps" = TheFutu

[ricko <ricko@geotempo.com>]

From: "Eric van der Vlist" <vdv@dyomedea.com>

> Even if sometimes I'd wish they are only this, I think that it's quite
> limitative to restrict schemata (either syntaxic or semantic) to a set
> of constraints.

There are two effects: schemas viewed by their proximate effect of producing
some transformed data (which in turn conforms to some other meta-schema),
and schemas viewed by their ultimate effect of constraining the direct data.

In other words,
   a schema language is expressed in terms of ultimate constraints
   a schema implementation actually converts the schema into a
transformation functions which transforms the instance into some proximate
form
   a schema assessor/valdator checks whether this transformation matches the
allowed proximate form
   the report mechanism gives appropriate messages, in terms of the ultimate
constraints.

So, to some extent, schemas as a "set of constraints" hides "schemas as
high-level transformation languages" and "schemas as high-level diagnostics
generating systems."

So how can we test schema implementation conformance?  It seems to me that
we have to have provide a conformance-meta-schema for the proximate form.

Cheers
Rick Jelliffe