Matt Flaherty wrote: > Okay, fair enough. Thanks. I'm zeroing in on a possible solution anyway. > I figure I can implement a Traversal.Handler that will return null from > all its methods and traverse from the root element. The enterElement > method would call Node.getXPath() on the input argument and use the > result to query the other tree
This should work OK and if it is not, you can always replace Node.getXPath() and the XPath evaluation by something faster, more low-level, that you'll write yourself. > and do a comparison node for node. This part is potentially expensive depending on how you compare the nodes. > Would this be too costly to work on a realistic document tree?
