Re: Types and Context

[Jonathan Borden <jborden@mediaone.net>]

Simon St.Laurent wrote:

> >
> > 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."
>
> Better still, a usable XML representation of the PSVI.  Then we could
> use existing XML tools on the PSVI without needing to add all kinds of
> PSVI-only tools to the system.
>
> I'd really like to have a viable route that isn't an API...
>

Let's not get too afraid of this so-called 'API', I am suggesting that this
amounts to a _single_ extension function named "typeOf" ... that is testing
whether a node is 'valid' with respect to a schema _and_ ascertaining the
type of a node are the same thing (under the schema algebra formulation).

The problem with using _only_ an XML representation of the PSVI is that it
doesn't alone replace the "typeOf" function ... assume an XSLT processor
that attaches an "xsi:type" attribute to each and every element in an XML
document. For example

 <foo>123</foo>

becomes ('psvi')

<foo xsi:type="simonsVeryOwn:integer">123</foo>

whipee!!!

the reason that the "typeOf" function is more interesting is that it
"understands" the type hierarchy so that the test:

typeOf(node, "simonsVeryOwn:integer") and
typeOf(node, "xsd:integer") both might be true, yet simply testing the value
of a type adorning attribute doesn't tell us this.

Note that "typeOf" is not unique to XPath, and a similar function would be
introduced for DOM 8 or whatever the version is intended to support the
'PSVI'

-Jonathan