Follow up question: what's the best way of incrementally consuming XML in Python that's character encoding aware? I have a very large file to consume but I'd rather not have to fall back to the raw SAX API.