fundamental facets - inquiry from the XML Schema Working Group

["C. M. Sperberg-McQueen" <cmsmcq@acm.org>]

In the course of our work on XML Schema 1.1, the XML
Schema WG has encountered a question which I hope readers
of this list can help us answer.  The short form is
simple:

  Do you care about (do you USE) the fundamental facets
  of the XML Schema simple datatypes as exposed in the
  post-schema-validation infoset (PSVI)?

The longer form is easy to explain:  the XML Query and
XSL Working Groups have asked us to define two totally
ordered subtypes for durations, analogous to the two
duration types known in SQL and in the XQuery/XPath
Functions and Operators specification.

Defining the types is easy, but there is one catch: if we
don't take special steps to make something different happen,
these totally ordered types will inherit the values for
all of their fundamental facets from their parent type,
so that any software which looks at the fundamental facets
will be told that they are partially ordered.  This is
not actually false, but it's not really as helpful as it
might be.

We face a choice: we can either try to write special rules
that explain how a processor knows to change the value of
the ordered facet from 'partial' to 'total', or we can
ignore the problem.

It has been suggested that we ignore the problem because
no software in existence actually uses the fundamental
facets as exposed in the PSVI for anything -- they serve
(it is said) purely a documentary function, and solely with
regard to the built-in types.

In order to avoid having the fundamental facets provide
misleading or (in some cases) inaccurate data, some have
suggested that we remove the fundamental facets entirely from
the PSVI, leaving them only as a form of documentation for
the built-ins.

If users currently rely on having the fundamental facets in
the PSVI, that would be a bad idea.  If no one uses them, it
might be a good idea.

Your views, please?

Thanks.

-Michael Sperberg-McQueen
 Co-chair, W3C XML Schema WG