Cliff and Peter:
I like to check with you guys if Xerces-J and XML Spy still require explicit
declaration of the xml namespace, i.e., whether the declaration
xmlns:xml="http://www.w3.org/XML/1998/namespace"
is still necessary.
I recall that Cliff has indicated that Xerces-J 1.4.4 works much better than
earlier versions. I also have the strong suspicion that the remaining error
Error: constraint 3.x.3: Attribute 'form' cannot appear in element
'attribute'
may have been solved by replacing our version of xml_lang.xsd with either
http:/www.w3.org/2001/03/xml.xsd or http:/www.w3.org/2001/xml.xsd
because Anli Shundi has warned me that there is a problem with the use of
the 'form' construct in xml_lang.xsd.
Similarly, Peter does not seem to have any problem sharing with me a working
version of the CPP/A schema even though it makes use of the xml:lang
attribute in a way similar to msg-header-2_0.xsd.
David:
If neither Xerces-J nor XML Spy requires the explicit xml namespace
delcaration, I suggest we don't even say in the spec that some parsers may
require the explicit xml namespace declaration (a comment in the XSD may be
sufficient). Therefore, only one change needs to go into both Appendix A
and the separately available msg-header-2_0.xsd. We should simply fix the
import element for the xml namespace to use a schema location of
http:/www.w3.org/2001/03/xml.xsd.
I don't think it is a problem even if two MSH's have to use different
versions of msg-header-2_0.xsd, one with the explicit xml namespace
declaration and the other without the declaration. The schema instances
(i.e., the ebXML messages) may or may not carry the optional
xsi:schemaLocation indication for the
http://www.oasis-open.org/committees/ebxml-msg/schema/msg-header-2_0.xsd
namespace. If not, the parser must still be able to resolve the above
namespace. Even if they do specify the default schema location, i.e.,
http://www.oasis-open.org/committees/ebxml-msg/schema/msg-header-2_0.xsd
the receiving MSH should still be able to configure its own parser to use a
locally cached version of the schema, one that is understandable to that
parser, instead of the one hinted at by the xsi:schemaLocation declaration.
Regards,
-Arvola
