I played around with several Python and Perl XML parsers, and settled on XML::LibXML with perl. I may be too quick to judge, but I tend to use the setup where the test example works with minimal adjustments.

It's for a work project to pull jpegs out of a system that has a RESTful API and returns XML objects. So far, it's been pretty easy to fake it using text files with the header and query info, and openssl s_client (all the requests to this system are required to be over HTTPS, which is good). The overall goal is to automate it, so I'm over 1/2 way there. I'm looking forward to getting back to other projects.