Names and schemas

[Robin Cover]

When Paul Prescod says:

> Schemas do not define markup. If they did, then the idea of an element
> type being constrained by multiple schemas at the same time would be
> impossible. But it is not. Schemas constrain markup. Schemas define
> classes of markup. Schemas do not define markup (or elements, or element
> types). Even SGML never claimed that DTDs defined element types. I
> consider this a fundamental error. SRCDEF is just the technical
> manifestation of the error.

I'm made to feel more in doubt than ever about the level of abstraction
that's envisioned for "schema" (and XSchema) in this discussion.  It's
not clear to me that the principal goal of a new schema language would
be to define classes of markup, with (syntactic?) constraints, though
certainly that could be one role.

>From what realm(s) of discourse are these definitions being drawn?  I
have taken for granted that a "schema" is intended to offer something
like a class definition for an object, or like a formal notation for
an attribute definition, but at some level higher than necessarily
commits to a particular markup model, and at a level which would
ideally abstract over markup languages which proclaim disinterest
in object semantics.  If this is the case, I see no reason why a
schema could not define one or more markup models: "I am a BLORT
(object) and I know how to generate a view of myself that conforms
to the SGML FARBLE-DTD, as well as a view corresponding to the
the XML FARBLEX-DTD", etc.  Also, if schemas are for defining
object (element) and attribute models, and not particular sets of
such definitions, then the notion of a document root should fade
into the background, and provide room for namespace semantics that
do not privilege the document itself as dictating a namespace scope.

If the goal of XSchema is to support the expression of semantic
constraints, with respect to which XML is said to have express
disinterest, then I see some potential for interesting work here.
Not to discount mechanisms recently designed in the corresponding
WG4 effort, of course, which indirectly or directly support a
whole range of notations not allowed under 8879:1986.

When I say "semantic constraints," I'm talking about something like
Tim Bray's "Strong Data Typing" (everyone has read it?)
and not "processing semantics" such as might be defined in methods
in particular programming languages.  However: once a schema does
address itself to primitive relational semantics and data-type
semantics, and admits of virtual attributes that exploit a
notion of "self" and simple inheritance, the lines blur.  I don't
think that's a reason to be afraid of the territory, but a reason
for XSchema to embrace something more interesting and useful than 
simply the design of an instance syntax for XML (for which any number
of proposals already designed might work suitably well -- see the
logs from the past two years).

That said (and perhaps too much), I welcome clarifications from
Paul and others on what is meant by "schema" and what authorities
(domains) are being used for the definition. and (especially)
the extent to which it's agreed that a schema should (not) be
constrained by the notions of particular markup models.

-rcc



  
-------------------------------------------------------------------------
Robin Cover                    Email: robin at acadcomp.sil.org
6634 Sarah Drive           
Dallas, TX  75236  USA          >>> The SGML/XML Web Page <<<
Tel: +1 (972) 296-1783 (h)     http://www.sil.org/sgml/sgml.html
Tel: +1 (972) 708-7346 (w)
FAX: +1 (972) 708-7380
=========================================================================

xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To (un)subscribe, mailto:majordomo at ic.ac.uk the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at ic.ac.uk)