[xsd-users] GML schemas, referenced instance not typed

[James Clark]

Hi Boris,

I'm working with James Boyden trying to use the GML schemas. Following 
our recent discussion of derivation by restriction problems, we have 
discovered some new things and run into some exciting new errors.

What you might find interesting, is that the restriction errors we ran 
into earlier in the GML schemas are -technically- valid (although 
evidently a very unusual way of constructing things); there is a bug in 
Xerces-C++ http://issues.apache.org/jira/browse/XERCESC-1281 that cannot 
deal with the obfuscated way the restriction is set up; while elements 
like gml:methodName may appear on the surface to violate restriction 
rules, on closer inspection gml:methodName is substitutable for 
gml:name, which makes everything okay.

When we eventually got a response from the GML developers, it was 
revealed that there was a secret version of GML 3.1.1 with some 
depreciated elements removed; I tried it out with xsd and it seemed to 
be just what was needed to get around the restriction issue. However it 
revealed some more errors as the schemas use unions (which I know xsd 
doesn't support yet):

jclark at polygon: ~/projects/xsd
$ xsd cxx-parser ../gml/3.1.1s/base/gml.xsd
../gml/3.1.1s/base/basicTypes.xsd:29:12: error: expected 'list' or 'restriction' instead of 'union'
../gml/3.1.1s/base/basicTypes.xsd:52:54: error: expected 'list' or 'restriction' instead of 'union'
../gml/3.1.1s/base/basicTypes.xsd:73:62: error: expected 'list' or 'restriction' instead of 'union'
../gml/3.1.1s/base/basicTypes.xsd:94:61: error: expected 'list' or 'restriction' instead of 'union'
../gml/3.1.1s/base/basicTypes.xsd:101:59: error: expected 'list' or 'restriction' instead of 'union'
../gml/3.1.1s/base/basicTypes.xsd:122:61: error: expected 'list' or 'restriction' instead of 'union'
../gml/3.1.1s/base/basicTypes.xsd:143:62: error: expected 'list' or 'restriction' instead of 'union'
../gml/3.1.1s/base/gmlBase.xsd:36:42: warning: element '_Object' is implicitly  of anyType
../gml/3.1.1s/base/gmlBase.xsd:36:42: info: did you forget to specify 'type' attribute?
../gml/3.1.1s/base/temporal.xsd:238:10: error: expected 'list' or 'restriction' instead of 'union'
../gml/3.1.1s/base/temporal.xsd:298:66: error: expected 'list' or 'restriction' instead of 'union'
../gml/3.1.1s/base/temporal.xsd:310:47: error: expected 'list' or 'restriction' instead of 'union'


I hacked around these unions by replacing them with basic things like 
string, integer etc, which seems a perfectly reasonable kludge for our 
purposes. But I've now found a new error that I don't know how to 
handle. Perhaps you can shed some light on this one?

jclark at polygon: ~/projects/xsd
$ xsd cxx-parser ../gml/3.1.1s-hacked/base/gml.xsd
../gml/3.1.1s-hacked/base/gmlBase.xsd:36:42: warning: element '_Object' is implicitly  of anyType
../gml/3.1.1s-hacked/base/gmlBase.xsd:36:42: info: did you forget to specify 'type' attribute?
ice: referenced instance 'http://www.opengis.net/gml#metaDataProperty' is not typed
Aborted


Looking at the definition of metaDataProperty (gmlBase.xsd:224), it does 
appear to have a type= definition.. I'm not sure what the problem is, 
exactly. I'm using xsd 1.5.0.

<element name="metaDataProperty" type="gml:MetaDataPropertyType">
	<annotation>
		<documentation>Contains or refers to a metadata package that contains metadata properties.</documentation>
	</annotation>
</element>
<!-- =========================================================== -->
<complexType name="MetaDataPropertyType">
	<annotation>
		<documentation>Base type for complex metadata property types.</documentation>
	</annotation>
	<sequence minOccurs="0">
		<!-- <element ref="gml:_MetaData"/> -->
		<any processContents="lax"/>
	</sequence>
	<attributeGroup ref="gml:AssociationAttributeGroup"/>
	<attribute name="about" type="anyURI" use="optional"/>
</complexType>


I've placed the developer's cut and my hacked version of it up on my 
webspace so you can view it if you feel the need:

http://www.geosci.usyd.edu.au/users/jclark/gml/3.1.1s/
http://www.geosci.usyd.edu.au/users/jclark/gml/3.1.1s-hacked/
or a tarball if you prefer:
http://www.geosci.usyd.edu.au/users/jclark/gml/3.1.1s-hacked.tar.gz

Many thanks,
James Clark