Custom Parsing

XML 1.0, 1.1, etc. Ripper is part of the Gorille project, which had as its original motivation the hope of separating character rule evaluation from parsing to simplify versioning issues with new flavors of Unicode. Ripper itself doesn't do such rule evaluation - filters on top of Ripper would do it.
Custom contexts The existence of a separately configurable context object makes it easy for applications to specify default assumptions, like namespace and entity declarations, without having to worry about what's actually in the document. (The document's own values will still override, of course.)
Alternate schema approaches Modularization means that we can start rearranging what happens. If I want to use my own XML format for defining entities and RELAX NG as the schema language, I can build a processor around those assumptions. Yes, it's dangerous, but it's also sometimes necessary to work with high voltage.
Whitespace handling Ripper lets the parser choose how it wants to value or not value whitespace. It would be a good foundation for a whitespace normalization tool or for an application that wants to preserve everything intact.