8th European Lisp Symposium - Confession 53

2015.04.23 10:48:49

I'm currently sitting in a lecture hall at the ETH in Zürich. It's 8 AM9 in the morning, and despite having had a long night's sleep, I'm still quite tired. It's amazing what two days of just listening and talking can do to you. It also feels so much longer than that, just because of all the things that happened. I suppose I should start from the beginning though.

The European Lisp Symposium is an annual conference organised by fellow Lispers. This year it took place at the Goldsmiths university of London from the 20th to the 21st of April and featured two keynotes1, twelve talks, and thirteen lightning talks. Aside from the symposium's talks, there was also a lot of discussion and chatter during the coffee breaks and even at the end of the days, long into the nights. With a new record of 89 registrants, there were a lot of people to meet and talk to.

When I decided to take part, it was mostly in order to meet the people I knew from Freenode's #lisp channel, but I was very pleasantly surprised to find that the various talks too turned out to be very interesting to me, even opening new ways to solve existing problems in my projects2. Of the people I've met, everyone was nice, interesting, and fun to talk to. I spent the most time with Robert Strandh (beach), Christian Schafmeister (drmeister), and Masatoshi Sano (@snmsts), but I would have loved to speak more with others as well, if only there had been more time to do so. Hopefully I'll be able to join again next year and have a chance to do so then.

My journey began on Sunday afternoon, flying to London. The airport was quite crowded and the flight fully booked out, crying children and babies included. Though the amount of people for the flight was not altogether surprising given that it departed at acceptable hours and was just at the beginning of the school holidays in Zürich. Regardless, the flight went well thanks to being able to drown out the noise with headphones. Arriving in Luton, the next challenge was to find my way through London to Greenwich, where my hotel was located. It turns out that Google Maps does not plan the best routes through the city, as it took me through districts I had no idea about, and I had to wander about aimlessly in the night more than once before reaching my destination. As I would learn on my trip back home, it's much better to ask the clerks at the railway stations for directions.

I arrived at the hotel around 9 'o clock local time, which was unfortunately too late to join the pre-ELS dinner that had been arranged. Though with over forty people, it would probably not have been very enjoyable to me to begin with. I had other things on my plate anyway. Apparently my booking (through booking.com) did not go into Mercure's systems, so they had to arrange something for me on the fly. Fortunately that was solved without a hitch, so after a quarter of an hour of standing around at the reception I was finally allowed into my freezing cold air-conditioned hotel room. Eager to meet people, I left my room again soon after and tried to lure whoever I could get into the hotel's bar by offering chocolate on the mailing list.

After waiting for a very brief while, I met a German guy3 and we talked about all sorts of interesting things. Turns out he's sneaking Common Lisp into high-performance computing. Later we were joined by Christian Schafmeister, who had just arrived after a long plane ride, and the discussion turned to Clasp, LLVM, and I forgot what all else we talked about until after midnight.

The next morning I woke up way too early, even escaping my phone's alarm. I made my way through the streets over to Goldsmiths, grabbing a croissant on the way. The weather was very nice; blue skies all over, and a nicely cool temperature to boot. I couldn't have asked for anything better. Finding my way through the campus I was surprised to find that a bunch of people had already arrived before me. I picked up my badge4 and sat down to wait for everyone to show up.

The first keynote was lead by Zach Beane and showed a really nice overview of what happened in the five years of Quicklisp's existence, and gave a taste of all the things to come in the future. I was very surprised and humbled by the brief shout-out he gave to me for my library contributions. I wanted to speak with Zach about a few plans5 regarding dists and versioning in Quicklisp later on, but I never got the chance to do so, unfortunately.

After a brief coffee break followed the first session of talks of which I most remember the Racket-powered computer vision and the lisp-backed visual data-flow system. There were a surprising number of Racket-related talks overall, suggesting to me that I really should invest some time some day soon to take a good look at it. I'm not quite sure what to think of the visual data-flow system, it seems really slow and inconvenient to work with, even if it offers a nice way to write automatically parallelised programs. I don't remember much about the last talk about Processing in Racket, which I'm really sorry about.

For the lunch break I remained in the building and spent the time talking to Robert Strandh, his wife6, and Christian Schafmeister. I'm not sure why, but for the entire time that I was in London I never felt hungry, so I barely ate anything and still always had the same amount of energy as I always do. It'd be neat if that could continue on for a while longer, to be honest.

After lunch we had a change of plans, due to the second keynote being rescheduled for the next day. What followed in its place was the second session of talks, which included a very interesting one about executable pseudo-code. Using this system (composed of a couple of CL macros), translating pseudo-code from textbooks to similar-looking, but actually executable code, became trivial and even rivalled the speed of optimised C++ implementations thereof for a specific example. The group is now testing the same for more algorithms, to see if this holds true. Exciting! We also heard about a Racket based system of language-generating specifications in order to allow for a better system to constrain an application's permissions in situations like that of Android, and we heard about a new algorithm to implement fast processing of lists in reverse order, using the stack for an implicit reversing operation.

A second coffee break with cookies ensured, followed by another Keynote talk, this time from a Googler at the ITA-Software group, speaking about unwanted memory retention despite garbage collection. He spoke about a long journey through the madness of debugging GC, with the end-result being that –if I understood correctly– the problem didn't lie in the GC at all, but instead in the way mmaped resources weren't being properly cleaned up by hand. Oh dear!

Closing off the first day we had a round of lightning talks, where I too took part and raced through my presentation about Qtools. We also heard about asynchronous hash-table operations, a mobile game engine using MOCL, OpenCV interaction from CL, an Erlang-like system for CL, and using Common Lisp in high-performance computing.

I had invited Christopher Hurley (Mithent) over for the conference banquet, and he luckily managed to make it fine, so I spent a large part of the remaining evening talking with him. It was really nice to finally meet in person again, if only we didn't live so far apart we could have that opportunity more often. Later, Masatoshi Sano joined us at the table and for a long while we spoke about the various cultural and lingual challenges involved in learning a new language. I was too embarrassed to try to say anything in Japanese to him, my mind completely locked up when I tried to form a sentence. I clearly still have a long, long way to go! By comparison, his English was immaculate, especially once he started to feel more comfortable talking to us.

After the banquet I once again hung around the hotel bar with a bunch of lispers, talking about this and that, but mostly compilers. I had to excuse myself around midnight, since I didn't want to be exhausted for the next day.

We started into the morning with an introduction to Gendl, a system that engineers use to build models for complex machinery. Apparently it has a long history and is mostly used in the aircraft industry. Unfortunately we didn't have time to get into any more involved stuff with the system, but what we did get was nonetheless a very interesting way to build and compose models using Common Lisp.

Following the subsequent coffee break we had a talk about Lambdatalk, a lisp-like extensible markup system, that reminded me a lot of my FuncEM project from back when I was still clueless about lisp. It was unfortunately also rather clearly apparent to me that the author didn't have much experience as a web-designer. The previews and examples he showed were stuffed with the kind of design I did when I first started out. Stuffed with shadows, gradients, and all sorts of other effects that don't do anything but distract from the important things. Anyway, before I go off on another rant about bad design, I'll move on to the second talk, which was about a symbolic pattern-matching system in Clojure. I was immediately reminded of the chapters about that in PAIP, which I should really pick up again soon. The main motivation behind this kind of system was the hope of better engaging students for the author's CS course. I myself would definitely be interested! We then had a talk about various data structures optimised for fast persistent access, which are currently used for Franz' AllegroGraph database.

My lunch break was once again spent talking to Robert Strandh about various things, this time less Lisp related, and more about general politics, ethics, and all the various hard problems that are fun to complain about, but almost insurmountably difficult to do anything about.

The afternoon was the highlight of the entire symposium. We had absolutely amazing talks, most notably Christian Schafmeister's about Clasp and his struggles of writing a C++-interoperating Common Lisp implementation in order to build gigantic molecules, Robert Strandh's first-class global environments that allow for a clean bootstrapping environment in CL, Eitaro Fukamachi's Woo beating node.js in HTTP server benchmarks and now shooting for the stars by aiming to be the fastest HTTP server ever, Miroslav Urbanek using CL in high-performance computing to crunch numbers for quantum physics simulations, and finally Chris Bagley introducing a modern approach to writing and experimenting with OpenGL comfortably in CL. Now I'm noticing that I put all the talks in the ‘most notably’ listing, though I think that is well-deserved.7

But the excitement wasn't over yet for me, as I had another lightning talk to do. Since I knew I had a lot to say on Radiance, even if I did cut it down to a minimum already, I had to ramp up my talking speed to eleven and raced through it as quickly as I could. I hope people managed to follow along regardless! The rest of the lightning talks were about lisp job hiring at RavenPack, information about the Common Lisp Foundation, an update on the status and plans for the common-lisp.net project, a tool to quickly install and set up CL implementations, and closing off with physics simulations for chocolate.

While the symposium had ended, the journey wasn't over quite yet. I spent the rest of the evening going for a drink with a couple of folks (beach, drmeister, and splittist were there too), where they gave me valuable advice on how to schedule my work. I left them around eight, as I was getting a headache and wanted to head back before it got dark. After relaxing in my hotel room for a brief while and swallowing a pain-killer, I felt well enough again to head back down to the bar. A couple of people once again gathered, and probably talked late into the night. I had to leave them early, even though there wasn't anything that I wanted to do more than stay and talk to them. I was too anxious about oversleeping and messing things up with my journey back home in the morning, so I had to go to bed on time.

Fortunately my alarm clock did its job proper; still tired I beat myself out of bed, did my packing, and checked out. I then set out for Greenwich Station, which confused me a lot at first, as there was a front entrance that didn't have any clerks, only ticket machines that didn't take my 50£. After wandering about worried for a good while I found the main entrance, and a very nice clerk printed out a good travel-route for me. Thanks to that, my journey to Luton was very comfortable and easy, with the only hiccup being a delay in the Tube line due to some clog up in the system somewhere. So I arrived at Luton about two hours too early. I spent my time there drinking a hot chocolate and paying five pounds for their WiFi. Getting onto the plane went by very quickly, and I'm quite sure the plane wasn't even half-full by the end. There was quite a bit of turbulence during the flight though, so my sketch didn't go as well as the ones from the flight over, though even those weren't great to begin with. Oh well!

I arrived at Zürich airport in the afternoon, euphoric about having successfully undergone such a wild journey, but also still dead tired from travelling, listening, and talking so much. I originally wanted to do post-recordings of my lightning talks to put on Youtube still8, but I just didn't have the energy to. I couldn't even get myself to stay up until midnight as I usually do, so I just called it an early night. And here we are already!

In conclusion, it was an absolute blast. I really hope that I can do it again next year, but depending on university and similar circumstances, it might not be as easily possible as it was in this case. A tremendous “Thank you!” goes to everyone who attended, and especially those who took the time to write a talk and organise the event. I'm also extremely thankful for all the people I had the chance to speak to personally; I enjoyed every minute of it.

Now that this is all over, I have to focus all the more on university, so that I may pass the coming exams in summer. Due to this I'll also most likely have to heavily tune down the amount of work I can spend on my lisp projects, maybe drop it altogether for a while. Still, if you have anything you'd be interested in knowing or speaking to me about, I'll be available. Send me a mail, hit me up on twitter, or preferably find me on Freenode's #lisp IRC channel.

Edit: Zach Beane also wrote a nice article about his ELS experience: http://lispblog.xach.com/post/117542924493/the-8th-european-lisp-symposium-2015

[1] Originally three, though one was cancelled due to unfortunate illness. Get well soon, Bodil Stokke!

[2] Mostly Chris Bagley's CEPL, which I hope to use to implement Parasol's shaders with, but Eitaro Fukamachi's talk about Woo also reminded me to finish off the Radiance driver for it, and probably switch to it in production soon.

[3] Whose name escapes me, I'm so sorry! I'm awful at remembering names!

[4] Which were very stylish. The only mishap was that they didn't include a field for the IRC nickname or twitter username, and printed the name a tad too small, so you had to squint to read it from afar. Hopefully that'll be corrected for the next one.

[5] Mostly I'm interested in adding GIT capabilities to Quicklisp and allowing some kind of ‘version fluidity’, analogous to allowing certain systems to be upgraded, while keeping others in place.

[6] I don't think I ever caught her name, but if I did, I must have forgotten again. As I said, I am terrible with names and am very sorry about that. Edit: I've been helpfully informed that her first name is Kathleen.

[7] The videos are going to be put up in a couple of weeks, so that you may experience the excitement as well if you couldn't make it to the conference.

[8] I'll do that later today and link it here

[9] Now that I'm done writing, which is almost all I did for the time, it's 12 'o clock. If only writing didn't take that much time to do, I'd do it oh so more often!


Written by shinmera