Scheme job at AMARC

The AMARC (the World Association of Community Radio Broadcasters) is seeking a Scheme/Web developer. The job description is in French.

A great MSLUG meeting

Yesterday's MSLUG meeting was a true success. About 25 people attended the meeting, and 15 of us went for a beer after the talks. And again, there were a few new faces. Maybe that's because it was our first meeting of the season, or because of the new format (that I called a SchemeCamp). Or both. But I know that many people liked the idea of having short presentations. A variety of subjects helps gather more people, I think.



Let me briefly summarize the 4 presentations:

1. Christopher Diggins talked about Cat, a language he is currently developing. Cat is a stack-based language, like Factor, but it is functional and statically-typed. The goal of the talk was to show how Scheme can be compiled to Cat, but Christopher went short of time. I'm sure he will give another presentation in a near future.
   
   The slides from his talk can be found here.
   
   
   
2. Étienne Bergeron presented a number of results from his Ph.D. thesis work. Étienne is working on a JIT synthesizer, i.e. a JIT that generates FPGA configurations from Scheme code. He mentioned that, in the course of his work, he had to reverse engineer the Xilinx protocol using innovative techniques, but which seems to cause a number of IP problems. The JIT synthesizer is currently limited to small expressions, but the results are interesting and promising.
   
   
   
3. Adrien Piérard presented part of his Master's thesis work, namely a portable interpreter of Scheme, with an emphasis on serialization/deserialization issues for mobile code. The ultimate goal of his work is to provide a portable and interoperable implementation of Termite.
   
   
   
4. Finally, David St-Hilaire and Étienne Laurin described the implementation of the Tetris game in Scheme. Their implementation targets Tcl/Tk through PS/Tk, but they are now working on a SDL port. It can be obtained from Snow. Note that David and Étienne are undergraduate students and they worked on this project in their spare time.
   

In short, we had a great time! Thanks to all who contributed to this success.

Another Scheme job in Montreal

There is another opening for a Scheme programmer in Montreal:

This person would work with Gambit to help in the development of the LORNET project. In particular, the initial project would be continue the development of a workflow/dataflow engine (based on Termite). This work is part of the TELOS system which is a Semantic Web based OS. Taking place between research and applications, the work is very interesting and provides great challenges.

Next meeting of the Montreal Scheme/Lisp User Group

Our next meeting will be this Wednesday, August 29th. The details of the meeting are not fully settled, but at least Christopher Diggins will talk about its translator from Scheme to a statically typed functional stacked-based language. The details will soon be posted to the MSLUG web site.

By the way, we will be experimenting with another meeting format this fall, inspired by the BarCamp and DemoCamp type of events. At each meeting, three or four people will give short talks (about 15 minutes) about one of their favorite projects (academic, commercial, open-source, or hobby). Many people have already showed an interest in giving such a talk. (I will certainly give one myself, about the grammar IDE we are currently developing at Nü Echo.)

Gambit-C 4.0 released

Gambit-C 4.0.0 has finally been released. Its website has also been revamped, and is powered by MediaWiki.

This release is a very good step forward, in terms of community support. At last, Marc got it right this time. Welcome to the open-source world, Marc! What we really need now is a whole set of useful libraries to help the Gambit-C community grow and not reinvent the wheel for each new application.

As users of open-source software, we should all give back to the community.

Musical composition program in Scheme

Kjetil Svalastog Matheussen just announced on the SISC mailing list that he has almost completed the implementation of a nice musical composition program in Scheme (it's an applet that requires Java 1.5):

The applet/program is a special kind of composition program which is going
to be used in music education in Norway, age 9 and up or something
like that. It might look a little bit like a toy, but its not, and its
also possible to make high quality music with it.

Almost all of the graphics and gui is made directly in SISC, while the
signal processing is programmed using Java. The source is GPL, placed
here.

I did not play a lot with it, but it seems really well done and demonstrates that even interpreted code can be fast enough for most GUI applications.

Disruptive change

In my previous post, I claimed that R6RS will break with the Scheme tradition. But don't get me wrong. I'm not against change. It's rather the opposite.

But I see R6RS as a disruptive change w.r.t. the long-standing tradition. As I wrote, I think it will kill innovation in the Scheme community. Where will there be room for innovation? Development tools, maybe, libraries, of course, but there is not a lot of room left for new language features.

And if an implementations doesn't implement it and is only R5RS/R4RS-compatible, I fear that it will eventually be regarded as an old implementation. Is this an intended side-effect of the Steering Committee? I doubt it. And I hope not.

To me, the only reasonable thing to do with this standard is to call it something else. It is a successor of Scheme, clearly. Probably a good one. But it's not Scheme anymore. Let's call it CommonScheme. Or Scheme 2.0.

R6RS preliminary voting results

They're out. The preliminary results of the ratification vote. Seems that the draft will be ratified. Congratulations to the R6RS committee! They did a really good PR job.

When I look at the people who voted NO, I see a lot of big names: Jonathan Reese, Will Clinger, Marc Feeley, Felix Winkelmann, Jeffrey Mark Siskind, Aubrey Jaffer, Chris Hanson, Shiro Kawai. People who I have a lot of respect for. (I'm surprised Manuel Serrano did not vote at all.)

What does this mean? Well, I think R6RS will break with tradition. The question is to know whether it's a good thing or not. I think it's not. R6RS will kill innovation, like Jeff Siskind pointed out. Every Scheme implementation, to be of practical interest, is itself a new language. But you could (almost always) run R4RS/R5RS compatible code on them. The previous standards allow really different (and divergent) systems to be developed. Simply consider how Gambit-C, Bigloo, Stalin, Chicken, and PLT-Scheme are different in their implementations strategies. I'm not sure R6RS will still allow this. This should not call this new language "Scheme" anymore.

And the committee was so irrespective of the SRFI process, that's just disgusting. Many brilliant people contributed to this process.

Finally, I don't understand why Matthias Felleisen qualifies the draft as "perfect". He probably means that it's good enough for ratification, but it's certainly not perfect.

The only thing I'm sure is that R6RS will create a schism...

Bye bye vacations

I'm back from vacations. I was in Gaspésie, a wonderful region of the Québec province in Canada, with my family. We had a really great time. But it's over. Now back to work!

I noticed that Erlang is getting more and more attention. Even Sam Ruby is betting on Erlang on the long term (5-10 years) for the implementation of web architecture. That's good.

Also, I was unable to vote against the R6RS ratification. I didn't have access to the internet during my vacations, and arrived too late Sunday to send it.