Version 0.10 of XML::Records is now available on CPAN. It allows you to pull "records" out of XML documents and turn them into Perl data structures (similar to Grant McLean's XML::Simple, but without having to read the entire document into memory), create simple tree structures (same as XML::Parser::EasyTree) from them, or generate PerlSAX events from them, allowing you to build DOM, XPath or similar structures from selected parts of a document. Since it's now a subclass of XML::TokeParser, it allows you to read token-by-token as well as record-by-record.