why invent the wheel again and again if there are platform independent open source libraries out there?
I'm not against code reuse at all, but the Pd data types DTD is so simple, that relying on a big library like Xerxes would be overkill IMO, and it will make compiling more difficult. Instead the xml handling can be embedded using something like tinyxml, which is exactly that, tiny.
Yes, XML is also handled with a custom parser in pool. The general XML format is really not likely to change and if this is handled properly there should not be a problem. In pool it's only a few dozens of lines of code.
best greetings, Thomas