Colloquium Details
Can Lisp Be Saved?
| Author: | Drew McDermott Yale University |
|---|---|
| Date: | November 30, 2009 |
| Time: | 16:00 |
| Location: | 220 Deschutes |
| Host: | Dejing Dou |
Abstract
Lisp is, as its inventor, John McCarthy once observed, "a local maximum in the space of programming languages," and to some fans it seems like it might even be the global maximum. However, it has one glaring weakness: an inadequate type system. The type system, by contrast with the rest of the language, has clumsy syntax and obscure semantics. Plus, the practices of Lisp programmers tend to be based on the assumption that no one's keeping tabs on you when you change or test the value of a variable. As a consequence, the Lisp compiler provides almost no useful information; most debugging occurs at run time.
Unfortunately, languages with good type systems, such as ML and Haskell, have evolved away from Lisp's wonderful XML-like syntax. They rely too heavily on Currying as a representation for multi-parameter functions. There is practically no such thing as an ML function that takes an indefinite number of arguments. When you try to create a type system able to handle this essential Lisp feature, you run into some very interesting and obscure issues.
Fortunately, recent research in type systems has begun to fill in these gaps. I will provide an overview of this research (mostly done by people other than me) and propose that Typed Lisp deserves a place in the world of cutting-edge programming languages.
Biography
Drew McDermott is Professor of Computer Science at Yale University. He was educated at MIT, where he received a Ph.D. in 1976. His research is in planning and knowledge representation, with side excursions into philosophy. He has been one the prime movers of the International Planning Competition, and the language it spawned, the Planning Domain Definition Language (PDDL), that is now the accepted standard for expression of planning domains and problems. His 1996 paper on the heuristic-search approach to planning was just named co-winner of the ICAPS Classic Paper Award. He is a Fellow of the American Association for Artificial Intelligence.