Re: Potential new issue: PSVI considered harmful

Dare Obasanjo wrote:

> > I can imagine doing type
> >annotation in a much more lightweight way than bringing a large complex
> >declarative schema facility to bear.  In fact, why shouldn't I just be
> >able to jam something into the instance or infoset saying "this
> >attribute here is an integer"?
> W3C XML Schema already has this facility through xsi:type[0]. However I
can't even begin to imagine how this system would work without validation.
How would one express simple or complex without looking at some predefined
schema and comparing the instance with that schema? Or do you propose this
only for simple types? Similarly how would one annotate attributes in an
instance documents when attributes themselves are meant to annotate

At a baseline, the type of any attribute may be identified by its QName.

It comes down to how you define "validation" which has become a somewhat
overloaded term, particularly as it related to "schema validation" according
to a specific schema language. But generally, yes, you would compare a set
of constraints that define a type against some part of the instance document
to assign a type.

As Tim says, the benefit of calling this a "type adorned infoset" is that it
doesn't imply any _particular_ process of adorning the infoset with types
even if _some_ process is used.


Received on Thursday, 13 June 2002 15:28:04 UTC