Tony Dean wrote:
Hi, I determined that it's not jre dependent. That is I can get it to
fail on 1.4.x or 5.0. I also determined that I can run the
XQueryXPath sample without any problems which performs xpath query
with predicate. So that got me thinking about what is the difference
between the two scenarios:
(1) my scenario -
I create an XmlObject from scratch as follows:
XmlObject.Factory.newInstance();
This actually creates a XmlAnyTypeImpl object with _textsource =
Xobj$DocumentFragXobj.
When xmlbeans uses the saxon engine in this case since the xmlbeans
engine can't handle predicates, it does so by calling:
Path$SaxonPathImpl$SaxonPathEngine class ->
context_node = Xobj$DocumentFragXobj
resultsList = _saxonXpathImpl.selectPath(context_node)
...
DOMSource rootNode = new DOMSource(context_node)
XpathEvaluator xpe = new XPathEvaluator()
XBeansIndependentContext sc = ...
xpe.setStaticContext(sc)
...
Variable thisVar = sc.declareVariable("this")
thisVar.setValue(rootNode)
this throws a classcast exception :
org.apache.xmlbeans.impl.store.Xobj$DocumentFragXobj
net.sf.saxon.trans.Variable can not properly cast
org.apache.xmlbeans.impl.store.Xobj$DocumentFragXobj object type
(2) case that works
The underlying XmlAnyTypeImpl _textsource object = Xobj$DocumentXobj,
not Xobj$DocumentFragXobj.
This seems to be the issue I describe in this bug:
http://issues.apache.org/jira/browse/XMLBEANS-200
Yana explains the reason for the problem in the bug report, but as my
last comment indicated I think there are some inconsistencies in how
XmlBeans is behaving.
As I describe in the bug, I workaround the problem by always calling:
if(((XmlObject)input).getDomNode() instanceof DocumentFragment) {
input = XmlObject.Factory.parse(((XmlObject)input).getDomNode());
}
before doing any Xquery or XSL transformations on an XmlObject with Saxon.
Cheers,
Raman
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
