5th European Lisp Workshop

I just attended the 5th European Lisp Workshop at ECOOP 2008, organized by Didier Verna. As Didier pointed out in his introduction at the workshop, it’s the third Lisp meeting in Europe in a short period of time demonstrating that Lisp is alive and well.

The workshop started with a keynote by Mark Tarver, “Lisp for the 21st century”. His basic thesis is that Lisp’s problems are mostly social and outlined ten qualities work future work in Lisp needs in order to be relevant and noticable. He then described his Qi language (pronounced “Q-I”, something I didn’t know), which brings strong type checking to Lisp. The end of his talk focussed on what he calls “computational adequacy” (one of his ten qualities) and how Qi fails in this regard. Essentially, computational adequacy means a language having enough stuff to easily get things done that people want to do. The main argument was that Lisp lags in this respect. It wasn’t as inflammatory as you might think since Mark seems genuinely concerned with Lisp.

Michael Wessel then gave a talk about descriptional logics and a way to use them in Lisp that is simpler than current approaches. I found the talk a little confusing and dense, but managed to get some clarification from Michael afterwards. He works on semantic web tools and we seem to be in agreement that multiple, specialized ontologies will probably be the norm and there need to be tools to create the ontologies, since no one wants to do it manually.

Next, Pascal Costanza and Charlotte Herzeel talked about work done by their student Leonardo Uribe with QLisp, a Lisp for the simulation of quantum computation. They went into the history of parallelism in Lisp dialects, namely CmLisp, *Lisp and Paralation Lisp, and a little on how QLisp was implemented using them. Charlotte talked a bit about adding parallelizing constructs to Lisp and why it is important. I rather liked the talk, since I considered doing work on this a couple years ago, but got onto other things instead.

After Pascal and Charlotte, I gave my talk, “Adaptive Libraries and Interactive Code Generation for Common Lisp”. I showed one way to realize a library for objects representing many types and how to evaluate code using it. Since there are ambiguities, I use interaction to resolve them and remember the context of the operation. Then I generate code specializing the use of the library using the contexts so that the ambiguities are removed and just for fun, I generate the code interactively. Didier asked for a demo, which I couldn’t give for a variety of reasons, the main one being that I didn’t have access to my code(*). Once I submit my thesis, I’ll make a demo and put the code out there. I like to think my talk went well, but I’m not the best judge of such things.

In the afternoon session, Rich Hickey presented his work on Clojure. I like what he’s done with it, especially his use of interfaces. I think I’ll play around with it for some of my work instead of trying to use Java directly. My short synopsis: Check out Clojure.

Lastly, Pascal gave a great talk entitled “make-method-lambda Considered Harmful”. It boiled down to this: make-method-lambda is a macro-like facility the depends on runtime values, so you get unintuitive results when compiling versus interpreting — sometimes. The proposed solution was to deprecate make-method-lambda and use keyword arguments to call-method in method combination. The bonus is that you can now use closures for the lambda forms of methods programmatically.

Overall, I thoroughly enjoyed the workshop and commend Didier for his efforts. The intimacy of the workshop allowed for better conversation than I have found at larger meetings. Discussions at the breaks and in the evening were enjoyable as well. My only regret is that I couldn’t stick around to attend the Dynamic Languages Symposium today; I have to head back home so I can start a new job.

As for the venue, I didn't care for Cyprus all that much. The conference website claimed it was 3 km from my hotel to the conference venue, but it's more like 5 or 6. I walked there on Monday morning and learned of the error the hard way. There was lots of garbage on the side of the road and the cab ride back from the conference to my hotel seemed excessively expensive. Drivers there seem to totally ignore the concept of a lane.  I also wasn't impressed by Larnaca airport. It wasn't awful, but it isn't making me want to come back.

I'm looking forward to ILC 2009.

(*) It’s a long story, but basically, I took my wife’s laptop instead of mine and didn’t have time to get my environment up and running on it before leaving.