PyRXP_ claims to be the fastest Python XML parser around... but this has a price. Once XML is parsed, you get a tree structured as tuples, lists and dictionnaries. While this is very Pythonic, it's not so trivial to write/read code handling such a tree. 

People at Reportlab argue about that for speed. Dealing with classes in Python is mem hungry, so they use only basic data structures (tuples,list,...). I can understand that but i believe it's an `extremist` point. Finding a balance between the DOM API (which is `extremist` too :) and such structures could turn to be more popular, maybe a Python `xml_node` builtin type ...


.. _PyRXP: http://www.reportlab.org/pyrxp.html