Bill-
I have written a similar solution for a customer before. The problem is that processing the XML documents from the repo at high volume tends to cause it's own problems. I haven't reviewed your code but one example of this is if you read the document from the repo for every call. The engine doesn't reliably clear session data, and their documents, fast enough and you can run out of Java heap. It's very obvious when this happens: the engine drops all calls while it resets the heap! The workaround I had to put in place is to store the documents into a common session and only load from the repo if the session didn't exist. Another one is that the complexity and size of the scripts required to parse the XML is so intense that you end up lowering system capacity, similar to if you had the engine reading a VoiceXML document.
It's a fun idea though; good for refining those scripting skills.
Please remember to rate helpful responses and identify helpful or correct answers.
