Re: [Moderator Action] Re: Non-deterministic content model

Murata Makoto <mura034@attglobal.net> writes:

> > Several people have pointed that out, but I don't see anyone
> > explaining why. There is a reason...
> 
> I do not think your argument holds water.
> 
> > Suppose your scheam had some appinfo that guided processing:
> > 
> >  <element name='A'>
> >   <complexType content='elementOnly'>
> >    <element ref='test:B' minOccurs='0' maxOccurs='unbounded'>
> > 	<annotation><appinfo>turn left</appinfo></annotation>
> >    </element>
> >    <element ref='test:C' minOccurs='0'/>
> >    <element ref='test:B' minOccurs='0' maxOccurs='unbounded'>
> > 	<annotation><appinfo>turn right</appinfo></annotation>
> >    </element>
> >   </complexType>
> >  </element>
> > 
> > Given <A><B/></A>, it's not clear whether one should
> > turn left or turn right.
> > 
> > The "no non-deterministic content models" restriction in
> > the XML schema spec makes sure that you can always correlate
> > the elements in the input with bits of appinfo (or
> > other type information) in your schema.
> 
> You do not need deterministic content models for this purpose.  
> You only need strongly-unambiguous content models, which can represent 
> any regular language.  Remember that deterministic content models 
> cannot capture all regular languages.  (More about this issue, see 
> Anne's paper.)
> 
> Strongly-unambiguity: there is only one possible parsing.

I agree this would be a good compromise.  Note that Koike's content
model is not strongly-unambiguous, however, and Dan's answer is still
correct for why _that_ is a problem.

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/

Received on Saturday, 26 August 2000 06:53:53 UTC