UPA and schema handling

I've been fiddling around with very simple schemas that violate the UPA 
constraint -- and have found that some schema tools flag UPA errors 
(e.g. oXygen), while others (e.g. XML spy) do not. This inconsistency 
is, at best, confusing -- but at worst would seem to lead to 
interoperability problems, since a designer could build a schema with 
one toolset and find it is not acceptable to another.

So am I missing something here?  Is UPA really an inviolable constraint 
[my interpretation], or is it just a guideline, in the manner of 
Appendix E 'Deterministic Content Models (Non-Normative)' in the XML 1.0 
specification?  And if it's just a guideline, would this not lead to 
interoperability problems as I've just outlined?

And, if someone already went down this rat hole, can anyone refer me to 
the corresponding xml-dev (or other) thread ;-)

Best --

Ian
-- 
Ian Graham
H: 416.769.2422 / W: 416.513.5656 / E:<ian . graham AT utoronto . ca>