I figured out why I was getting different results with Xalan and dom4j, it had to do with some parsers defaulting to namespace aware and others not.
> Or you can bind your namespace URI to some prefix of your choice and have a > much nicer expression. > > my:one/my:two/@three > > So in other words, when dealing with namespaces its good to use prefixes in Unfortunately, I don't have any control over the xml or the DTD, nor even certainty of exactly what URI will be in use when it runs in production. I know I could transform it first, but the whole reason I got started on this was to avoid having 15 levels of DOM code iterating through nodes and casting back to Elements. I'll just end up with *[name()= on every level. I have a spec that specifies all the elements, and says there will be a particular default namespace and a prefix, and I have sample and test documents from several sources which do *not* match the spec, but do have a default namespace, but no prefix, and I have suggestions from other groups that they may or may not drop the default namespace or change it, and I have another default namespace under the XML-DSIG stuff, which is probably not going to change. And I have to create classes and deliver iterations _now_. So, I really do have to write code that ignores the namespaces, and it really is the most stable and trustworthy thing to do, for these particular documents, until the industry groups and other departments get their act together. -Tom _______________________________________________ dom4j-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dom4j-user
