This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
The definition of ·ordered· in 4.2.1 requires that all values in the value space of the datatype inquestion must be from the same primitive datatype. The algorithm for calculating the {value} of the ordered schema component give the value partial for a union some of whose basic members are ordered but whose value space is not a subset of that of a single primitive datatype. E.g., if A is the union of decimal and string, and B is the union of dateTime and decimal, neither A nor B are ·ordered·, but for both the {value} of the ordered facet is partial. The definitions should be harmonized one way or the other.
A portion of the proposal for bug 5062 (http://www.w3.org/XML/Group/2004/06/xmlschema-2/datatypes.b5062.html) provides a solution for this bug; if that proposal (or the portion that changes the term definition of ordered) is adopted, this bug can be marked CLOSED (not just FIXED, since I'm the reporter of this bug).
The proposal for bug 5062 mentioned in comment #1 was adopted with amendments by the XML Schema WG on 14 March 2008; after some confusion about the details of the decision the adoption was clarified and reaffirmed on 21 March 2008. http://www.w3.org/XML/Group/2004/06/xmlschema-2/datatypes.b5062.html (member-only link) That proposal addresses the issue raised here, so I'm marking it resolved.
As suggested in comment #1, I'm going ahead and marking this issue CLOSED.