The issues with encoding should be clarified in docs, I think.
This module generates us-ascii XML documents. This is mostly ok if your feeds are for HTML only because HTML has its own way of encoding non-ASCII characters into ASCII (HTML entities). You only have to jump through a hoop to get everything encoded right.
But as Atom feeds are more and more used outside of simple case of "subscription to a stream of HTML pieces" this issue should be dealt with.
Should be easy, actually, as current Perls shine at Unicode support.