Common Lisp as a Vehicle for Implementing Statistical Environments
Ross Ihaka, University of Auckland

R was originally created by implementing a rudimentary Lisp interpreter
and then providing it with S language compatibility and statistical
capabililties.  Additional extensions such as exception handling,
namespaces, compilation and a CLOS-like object system have since been
added.

An alternative to this retro-fitting strategy is to build systems,
on top of a more complete Lisp implementation.  Using existing language
capabilities makes it easier to create a fully featured statistical
system and should lead to improved systems.

This paper will look at the advantages of moving to Common Lisp as a
base for building statistical systems and examine preliminary work on
this task.