KR> We should be returning the instance that not only matches the keys
KR> specified, but the properties as well. In this case, the query
KR> will return an instance even though the virtinvalid property
KR> doesn't exist.
KR> This is something we'll want to fix in the future.
Eh? IMHO, this is not really the job of the providers. We're
definitely not equipped to be enforcing schema in the same way that
the CIMOM is.
I'd say that if the CIMOM isn't going to prevent you from inserting
completely invalid keys into a reference, that it doesn't make sense
for us to do anything other than ignore them as well.
--
Dan Smith
IBM Linux Technology Center
Open Hypervisor Team
email: danms(a)us.ibm.com