ECLM 2011 Notes

This last weekend I was in Amsterdam to attend the European Common Lisp Meeting. This was my third participation in a organized Lisp meeting (after the first ZSLUG in Zurich and the ELS 2011 in Hamburg) and I am happy I’ve decided to go. I was only present at the meeting itself since going to the dinners and city tour would have been way out of my budget. Anyway, the ECLM was a nice venue. I enjoyed most of the talks and still had an opportunity to talk with fellow lispers. I enjoyed talking with Luís Oliveira and meeting Zach Beane.

The first talk was given by Nick Levine and it can be viewed in two parts. In the first one, he talked about his experiences of trying to write a CL book for O’Reilly. It was quite interesting to see how hard can it be to prepare a book, especially for a publisher who was (is?) not very lisp-friendly. The second part was mostly about the community, although presented with a rant on libraries. This is a topic that has been debated several times. Thanks to Quicklisp, the problem now is not installing libraries but finding them and knowing which ones are good. I am not sure if creating another site as suggested would be a good thing since resources are already scarce. Perhaps more thought must be made in how to improve the current ones. CLiki still seems to me the best starting point. Still, Nick Levine talk was good and entertaining. One of the best in the meeting.

The following talks were mostly about companies that use CL as their main programming language. Jack Harper talked about the company he recently started, Secure Outcomes, that produces a unique portable fingerprint scanner. I must say his talk was quite inspiring! He talked about how to get a startup running and the decisions that took him to choose Lisp as the main development language. In addition, he also explained why prefers Lispworks to any other implementation.

Next, it was the talk given by Dave Cooper. I must confess his talk was the weaker of the day mainly because he talked about two different subjects without any connection. He started talking about GDL, the main product from his company, Genworks. I’m sure GDL can be a great thing but I didn’t get much from his talk. About halftime, the talk suddenly changed to the Common Lisp Foundation. This was the interesting part of the talk since he explained the aims of CLF, the people behind it, etc. However, it was not clear how it will distinguish itself from ALU in terms of operation (in terms of purpose, CLF just focus on Common Lisp while ALU in all Lisp dialects) and this was the main concern that was expressed during the questions time. After presenting CLF, and since there was still some time left for the next presenter, he went back to GDL.

Afterwards, it was the turn of Luke Gorrie to present his lisp-hacker startup Teclo Networks. His talk was an expanded/updated version of the one given in Zurich. Still, it was also quite interesting. He started by telling how a group of hackers with a Lisp and/or Erlang background got together to improve the mobile TCP/IP communications. Then, he showed us how TCP badly misbehaves in a mobile network and how their product, Sambal, can give 10% to 27% improvements. Another interesting point of the talk was that CL is used as their main development language. In short, it is used to develop and study all their algorithms. They have a TCP/IP stack fully implemented in CL! Moreover, all their analysis and maintenance tools are also all in CL. However, in the actual product boxes they have reimplemented the algorithms in C. The reason: extreme pragmatism. Luke concluded by hinting that the sales of their product is going very well!

In the afternoon the talks started with Paul Miller from Xanalys. The talk was dedicated to Link Explorer, a windows desktop tool to analyze data. The application is quite impressive and was developed using just CL. Paul also gave us a demonstration of the tool as well as some notes on future development.

The best and most awaited talk, Quicklisp, technically and socially, was given by Zach Beane. The talk focused on several aspects of Quicklisp. Zach started by giving an overview of the famous library problem of CL, the solutions that existed before QL, explaining their advantages and disadvantages. Also, and very important, what people were actually using and what difficulties they were facing. In a survey he did, most CL programmers were installing libraries by hand, including Zach! Then he proceed to how Quicklisp was developed, some technical issues, what is the role of Quicklisp and what is the reception after one year. The talk focused then on the social impact of Quicklisp in the community. One of the things that makes Zach happy it’s the number of emails he gets saying that people are back to using CL and contributing more to the community (i.e., making libraries available) because of QL. Finally, some indications of what is to come. My perception is that the possibility to enable hacking as it was possible with clbuild is one of the most exciting future features for Quicklisp. Zach’s talk was excellent from all points of view!

The last talk of the day was by Hans Hübner. This was my second favorite presentation. Although the topic, code style and conventions, can start some heated discussions, I must say that I agree with almost everything Hans Hübner mentioned. However, like everything, some common sense is always necessary. One of the main points was that lispers should not use constructions which are not part of the standard language when the standard provide options, just because you want to save some typing. It is more important for another programmer to understand faster what is written than forcing him to look for the definition of the unusual constructs. The if*, bind were examples given. Hans also talked abut the 80-column rule, style guides, etc. In the end, it always depends on the project, the people, etc, but code style is important and should not be ignored.

The meeting ended with several lightning talks. The most interesting bits were: Marco Antoniotti announced ELS 2012, to be held in Zadar, Croatia, around April-May; Christophe Rhodes talked again about swankr, a swank and slime for R; the announcement of ABCL 1.0.0 by Erik Huelsmann.

Some words on the organization. Organizing a meeting of this kind is not easy and Edi Weitz and Arthur Lemmens must be congratulated for making a great event. Not all was perfect but everything went smoothly. I wish that it continues to happen in the coming years!