Re: qualification of components of an all group

"Henry S. Thompson" wrote:
> 
> James Clark wrote:
> 
> > >  the components of a group whose <nt
> > >  def="nt-compositor">compositor</nt> is (implicitly) <pt>all</pt>
> > >  may not be qualified, > > and therefore call for exactly one
> > >  appearance of the element they identify.
> 
> > What's the justification for this restriction?  * and & don't interact
> > right in SGML, but I don't see why this can't be fixed in XML Schemas.
> 
> There was considerable concern expressed in the WG about the
> implementation cost of unconstrained re-introduction of & into XML
> Schema,

I can certainly empathise with that. Note that one of the things that
was hardest about and-group implementation in SGML is the detection of
ambiguous content models.  

> with the result that we tied it down VERY hard.  I think we
> probably over-reacted to that concern:  we can't even use it for the
> content model for <datatype> because it doesn't allow ?.  That much at
> least I will lobby to have allowed, but I'm less sure about *:  do you
> want e.g.
> 
>  a b a
> 
> to be allowed by (a* & b),
> 
> or only e.g.
> 
>  b a a

The former (ie different from SGML).  This seems both what people want
in practice and easy to implement.

James