Re: Enlightenment via avoiding the T-word

[ht@cogsci.ed.ac.uk (Henry S. Thompson)]

Ronald Bourret <rpbourret@rpbourret.com> writes:

> Nicolas LEHUEN wrote:
> > Yes, but as validation and typing are tightly bound in terms of algorithm,
> > why not perform them both in the same path, for the sheer sake of efficiency
> > ? When validating, you collect a lot of useful data (you know what XSDL
> > types you are matching against elements), so why no keep this information in
> > a PSVI to give it to the next layer ?
> 
> >From my reading of the schema spec, this is what a validator does. My
> point is that it would be nice if the PSVI was clearly factored so I
> could write a processor to just add type information without performing
> validation. This would be much cheaper and would be useful in
> applications where I trusted the source of my documents.

What factoring of the PSVI would you like that isn't already there
implicitly?  The W3C XML Schema REC defines a number of properties for
use by conformant schema-validators.  It is already open to you to
make provide a subset of those properties of your own choosing in the
output of some non-validating processor that you write.  If a
downstream app needs only your properties, then it can run either on
top of a conformant schema-validator, or on top of your processor.

That's the whole point of the Infoset abstraction -- it provides a
vocabulary for specifying application profiles, that is, what
processors supply or require in the way of information about XML
documents.

ht
-- 
  Henry S. Thompson, HCRC Language Technology Group, University of Edinburgh
          W3C Fellow 1999--2001, part-time member of W3C Team
     2 Buccleuch Place, Edinburgh EH8 9LW, SCOTLAND -- (44) 131 650-4440
	    Fax: (44) 131 650-4587, e-mail: ht@cogsci.ed.ac.uk
		     URL: http://www.ltg.ed.ac.uk/~ht/