Thank Frank for suggesting to make note to get addressed in the spec. Here's a candidate (it's rather XML generic, *not* ChangeSummary specific during which work I happened to discover the issue):
XSD defines how XML formats. For XSD <element xmlns:global="" ref="global:element1"/> the XML should be <global:element1 xmlns:global="" ... and for XSD <element name="element1" ... the XML should be <element1 ... Similarly, for XSD <attribute xmlns:global="" ref="global:attribute1"/> the XML should be xmlns:global="" global:attribute1="..." and for XSD <attribute name="attribute1" ... the XML should be attribute1="..." Before SDO provides API to distinguish local element (<element name="element1") from global element without NameSpace (<element xmlns:global="" ref="global:element1"/>), and local attribute (<attribute name="attribute1") from global attribute without NameSpace (<attribute xmlns:global="" ref="global:attribute1"/>), EMF's ExtendedMetaData (thank Frank) could be a temporary work-around for XML writers to output compliant formats. Besides above note candidate, do we want to fix DataObjectXMLStreamReader.java addSingleValue method and the related ones to use EMF's ExtendedMetaData to report null and "" respectively for NameSpace please? Currently we only report null because XSDHelper#getNamespaceURI returns so. Of course, if we change XSDHelper#getNamespaceURI implementation to return null for local Property/attribute/element and "" for no-NameSpace global Property/attribute/element, no change needed for DataObjectXMLStreamReader.java wrt this issue. As for my current ChangeSummary work, I'll follow what exactly Frank suggests not to worry too much about the corner cases yet. Thank Frank again. Thanks for your attention. -- Yang ZHONG