Am I the only one that cringes at the thought of a interpreted language built ontop of an interpreted language? Yes I know the java vm is very well built and runs almost as quick as a true compiled language but still you have to admit it's silly building a language ontop of a perfectly good language...
Hopefully someone else hasn't answered this while I've been going on a murderous rampage here at the college (long story).
It probably isn't actually an interpreter built on top of the JVM. Here is my impression w/o having read the developers docs....
a) The Groovy source code can be compiled into Java bytecode and ran in the JVM just like regular Java, which means that this code has the potential to become machine code if the JVM decides to compile it.
b) You can run it using the Groovy interpreter ('groovy sourcefile.groovy'), which afiak, takes the JVM completely out of the loop, yet gives developers the same Java API to use. It is possible that the interpreter is written in Java, but I would bet it is written in C probably and was probably constructed with the usual suspects (lex, yacc, etc).
c) You can start the interpreter and use it like a shell similiar to Python and Lisp, which is a nice way to diddle around trying to figure out a short algorithm or do some sysadmin stuff.
Just my initial impressions from what I've read so far..... I'll have some more time this weekend and we'll see what is what then.