Re: Types and Context

[Jonathan Borden <jborden@mediaone.net>]

> As I understand it, schema-supporting functions would work on a PSVI.
> If the extension functions were defined in terms of the PSVI, then the
> schema language (or rather the schema syntax) that we used would be
> independent. For example, we could come up with a simplified syntax
> that used the same conceptual schema components as those used in XML
> Schema, which would create a PSVI that holds the same information.

sure. what we still need is a processable incarnation of the "PSVI". no such
thing exists today. you are basically saying that "if we had an API which
provides access to the information in the PSVI then we could easily do
this."

What I am saying is that by defining a _type_ as a membership in a set of
instances (which is what the schema algebra does), then this exact
information of the PSVI is obtained. The PSVI adds type adornments and
validation flags. Turns out that the schema algebra defines these in the
exact same fashion. Lets backtrack to DTDs, so we can all be on familiar
ground. Suppose we define the type of a document as whether the document is
valid with respect to a DTD. In the same fashion we can assign a type to a
node based on the node matching the constraints imposed by a class (type
definition). No magic. This works for any schema language and any definition
of class

So this API we are looking for. I am suggesting that this API is a function
"typeOf". This just another way of viewing the "PSVI" yet using the "1.0"
generation of tools that are widely deployed (we just need to do a bit of
tweaking).


>
> So, given that the XML Schema PSVI is the only well-defined PSVI
> around at the moment, then I think XPath extension functions should
> use that for now.

ok use it. how? hint: how would you define an API for XPath?

-Jonathan