has been uploaded. Adds caching... parsed/compiled src files, autogenerated .class files, etc., are all writen to cache.jar to be reused in future invokations of the interpreter. This should help with the major slowness of the interpreter, all the auto-generated classes that are created, etc.
Also, the read/parse/eval has been seperated out. Input is handled by an AbstractFileSystem implementation (ie. local filesystem, .jar file, etc.), parsing is done by an implementation of Parser, and compilation/eval is done by an implementation of NodeEvaluatorFactory/NodeEvaluator.
While it may seem overkill to make the parser a plug-in component, there is a good reason, which hopefully I will demonstrate in a future release ;-)