RE: RE: Resctriction problem on complexType

Mike Kay writes:

> I suspect you've hit one of the cases where the algorithm in 
> the spec fails to allow restrictions that in fact make perfect sense.

Note that there is an excellent chance that Schema 1.1 will be revised to 
allow such seamingly sensible restrictions.  Basically, what's changed 
since Schema 1.0 and Schema 1.1 is that enough smart people have worked 
through the algorithms to the point where we think we can safely say: "if 
the the restricted content model allows no instances (specifically no 
lists of elements to be matched by the content model) disallowed by the 
base, then it's legal".  In the 1.0 timeframe, we knew that's what we 
wanted, but weren't sure it was a practical check for processor builders 
to implement, given the specific features of our content models.  So, look 
for that as a likely change in 1.1.

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------