RE: multiple types

[Gavin Thomas Nicol <gtn@ebt.com>]

> > The operative word is *conforms*... type is painted on
> > data by a set of assertions.
>
> I think this is a really important point. Example of this are "Document
> Types" aka DOCTYPE which defines a document type as validity
> w.r.t a DTD (a set of constraints). Another example is XML Schema regular
> expressions which define a type as a pattern conformed to. Another
> example is what Murata Makoto recently posted about "Hedge
> Regular Expressions" which related to TREX, RELAX and perhaps
> what Examplotron is growing into. Yet another example is
> Schematron which directly defines conformance in terms of
> assertions. So yes.

Just to follow up with one last though before I disappear for a
couple of days.

I think people are getting confused between instances and type
definitions. In the context of a type definition, it makes great
sense to use "is-a" and "has-a" and "extends", etc. This is
again, defining the *type*, not an instance.

When a chunk of data comes along, one then asks "does it conform
to the definition of this type?" (does it have a...?).

These are very different things from actually generating the
XML.