This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
XSLT and XQuery both allow an element to be deep-copied without changing the type annotations on the nodes being copied, but with the copied nodes having a different base URI from the original. For this and other reasons, it is desirable that the validity and type of an element should not depend in any way on its base URI. Currently we have no rules preventing assertions (and perhaps CTA) producing results that are dependent on the base URI of the node being validated. The best way of preventing this problem occurring would be to lose the base URI of the nodes while constructing the XDM instance that is supplied as input to the assertion. I believe that the resolution of bug #5207 makes it clear that XDM allows the base URI of a node to be "absent" in such a way that operations dependent on the base URI would fail; I think we should probably construct an XDM instance with this property. It would be a good idea to wave this past other QT people to check that they share the same understanding.
I think I disagree. Their gun, their foot, their bullet. If people write base-URI-sensitive patterns, it's because they care, and if the base-URI is gone, they get a different result, that's good, not bad. I'm not at all convinced that QT's handling of the base URI property is correct, I certainly don't want to change XSD based on it.
During its 2009-04-17 telecon, the schema WG decided to resolve this issue without any change. The likelihood for a user to stumble over this on purpose is very small. Michael, if you would indicate your concurrence with or dissent from the WG's disposition of the comment by closing or reopening the issue, we'll be grateful. If we don't hear from you in the next two weeks, we'll assume that silence implies consent.