Most recently, from July 8 to 12, two significant events were held simultaneously - a conference and school . In this post, I want to highlight a few features that we noticed during the conference.
The biggest pride of Hydra and the School are speakers.
- Three laureates : Leslie Lamport, Maurice Herlihy and Michael Scott. Moreover, Maurice received it as much as twice. Leslie Lamport also received - the most prestigious ACM award in computer science;
- The creator of the Java JIT compiler is Cliff Click;
- Korutin developers - Roman Elizarov () and Nikita Koval () for Kotlin, and Dmitry Vyukov for Go;
- Contributors to Cassandra (Alex Petrov), CosmosDB (Denis Rystsov), Yandex Database (Semyon Checherinda and Vladislav Kuznetsov);
- And many other famous people: Martin Kleppmann (CRDT), Heidi Howard (Paxos), Ori Lahav (C++ memory model), Pedro Ramalhete (wait-free data structures), Alexey Zinoviev (ML), Dmitry Bugaichenko (graph analysis).
And this is school:
- Brown University (Maurice Herlihy),
- University of Rochester (Michael Scott),
- University of Waterloo (Trevor Brown),
- University of Nantes (Achour Mostefaoui),
- David Ben-Gurion University of the Negev (Danny Hendler),
- UCLA (Eli Gafni),
- Institut polytechnique de Paris (Petr Kuznetsov),
- Microsoft Research (Leslie Lamport),
- VMware Research (Ittai Abraham).
Theory and practice, science and production
Let me remind you that the SPTDC School is a small event for one and a half hundred people, world-class luminaries gather at it and talk about modern issues in the field of distributed computing. Hydra is a two-day distributed computing conference that runs in parallel. Hydra has an engineering focus, while the School has a scientific one.
One of the goals of the Hydra conference is to bring science and engineering together. On the one hand, this is achieved by the selection of reports in the program: along with Lamport, Herlihy and Scott, there are much more applied reports by Alex Petrov, who contributes to Cassandra, or Roman Elizarov from JetBrains. There is Martin Kleppman, who used to make and sell startups, and now studies CRDT at the University of Cambridge. But the best thing is that Hydra and SPTDC are held side by side - they have different reports, but a common place for communication.
Immersion
Five days of the School in a row is a very big event and a big burden for both participants and organizers. Not all of them made it to the last days. There were those who went to the Hydra and the School at the same time, and for them the last days turned out to be the most eventful. All this fuss is compensated by incredibly deep immersion. This is due not only to the volume, but also to the quality of the material. All reports and lectures at both events were not planned to be introductory, so wherever you go, you immediately dive far and deep, and do not let you go until the very end.
Of course, much depends on the initial preparation of the participant. There was a funny moment when two groups of people in the corridor were independently discussing Heidi Howard's report: one seemed completely ordinary, while the other, on the contrary, thought hard about life. Interestingly, according to (who wished to remain anonymous) members of the program committees, Hydra's reports and lectures by the School at their events could be overqualified. For example, if a PHP junior came to a PHP conference to learn about life, it's a little reckless to assume that he has deep knowledge of the internals of the Zend Engine. Here, the speakers did not spoon-feed the juniors, but immediately implied a certain level of knowledge and understanding. Well, indeed, the level of participants who operate distributed systems and write runtime kernels is very high, this is logical. Judging by the reaction of the participants, it was quite easy to choose a report by level and topic.
If we talk about specific reports, then they were all good in their own way. Judging by what people say and what can be seen from the feedback form, one of the coolest reports at the School was Michael Scott, he just broke everyone, he has an abnormal rating in the region of 4.9.
Metaconference
Long before the start of Hydra and the School, Ruslan assumed that a kind of βmeta-conferenceβ would turn out - a conference of conferences, where all the top participants in other events would automatically, like into a black hole, be drawn. And so it happened! For example, among the students of the School was noticed from DeutscheBank, a renowned expert on multithreading.
And of the Hydra members were seen () and () from Odnoklassniki. (At the same time, Vadim also helped us make two excellent interviews with Martin Kleppman - , and the other for viewers of the online broadcast). There were members , well-known speakers Anatoly Kulakov and Igor Labutin. Of the Javist were ΠΈ . Usually you see these people in completely different places - dot netters on DotNext, javist on Joker, and so on. And here they are, side by side, sitting at Hydra's reports and discussing problems together on bofs. When this slightly artificial division into programming languages ββand technologies disappears, the features of the subject area appear: dynamic runtime specialists communicate with other runtimers, researchers in the theory of distributed computing argue heatedly with other researchers, database engine engineers crowd the whiteboard, and so on.
On the report the OpenJDK developers sat in the forefront (at least I know them by sight, but not the pythonists, maybe the pythonists were there too). In fact, there is something so Shipilev in this report... Ori doesn't say exactly the same thing, but a careful look can reveal parallels. Even after everything that has happened in the latest C++ standards, problems like out of thin air values ββhave not been fixed, and now you could go to such a talk and listen to how people "on the other side of the barricade" are trying to fix these problems, as they argue, one could be impressed by the approaches to the solution found (Ori has one of the repair options).
There were a lot of participants in program committees and community engines. Everyone solved their inter-confessional problems, built bridges, acquired connections. I used this where I could, and, for example, we agreed with Alexander Borgardt from together to write a full-length article about actors and asynchrony in C ++.
In the photo: Leonid Talalaev (, left) and Oleg Anastasiev (, right), leading developers at Odnoklassniki
Fire discussion areas and bofs
At conferences, there are always participants who know the subject on an equal footing with the speakers (and sometimes even better than the speakers - for example, when the core developer of some technology was among the participants). There were a lot of such highly expert participants on Hydra. For example, at some point around Alex Petrov, telling , so many people formed that he could not answer everyone. At some point, Alex was smoothly pushed to the side and began to be torn apart by questions, but the falling flag was picked up by the well-known in Rust developer circles and perfectly balanced the load. When I asked Tyler to help with the online interview, all he asked was "When do we start?"
The debatable spirit sometimes broke even into reports: Nikita Koval arranged a sudden Q&A session, dividing the report into several sections.
And vice versa, at the BOF, in terms of multithreading, they remembered about non-volatile memory, dragged it to this bof as the chief specialist, and he explained everything to everyone (in short, non-volatile memory does not threaten us in the near future). One of the leading this bof, by the way, was , who sits on program committees for an insane number of conferences⦠it seems like five at the same time now. NVM was also discussed at the neighboring bof about "Modern CS in real world" and came to this completely independently.
I can share a super-insider that even the direct participants in the story might not have noticed. Eli Gafni performed on the evening of the first day of the School, and the next day he stayed and began to troll Lamport, and from the outside it seemed that this was a game and Eli was inadequate. That this is some kind of troll who wanted to take out Leslie's brain. In fact, the fact is that they are almost best friends, have been friends for many years, and it's just such friendly jokes. That is, the joke worked - all the people around fell for it, took it at face value.
Separately, I would like to note how much love and effort the speakers put into this. Someone stood in the discussion area to the last, almost for hours. The break has long ended, the report began, ended, the next break began - and continued to answer questions. An interesting story happened to me too - catching Cliff Klick by surprise, I received not only a clear and reasonable explanation of that provocative discussion about the lack of tests but also got a full review of it . I never asked for this: I just asked what you can read about AA (it turned out you can listen ), and instead Cliff spent half an hour talking about the language and making sure he got it right. Marvelous. It is necessary to write a habrapost about AA. Another unusual experience is watching the pull request review process in Kotlin. It's really a magical feeling when you go to different discussion groups, to different speakers and plunge into a whole new world. It's something of a level .
English
Hydra 2019 is our first conference where the main language is English. This brings both its benefits and its challenges. An obvious plus is that people not only from Russia come to the conference, so among the participants you can meet engineers from Europe and scientists from England. Speakers bring their students. In general, important speakers have much more motivation to go to such a conference. Imagine that you are a speaker at a completely Russian-language conference: you have reported your report, defended the discussion area, and then what? Drive around the city and see tourist spots? In fact, really popular speakers have seen enough of everything in the world, they don't want to go to see lions and drawbridges, they're bored. If all reports are in English, they can participate in the conference on a general basis, have fun, be included in discussion areas, and so on. The atmosphere is quite friendly towards the speakers.
The obvious disadvantage is that not everyone is comfortable communicating in English. Many understand well, but speak poorly. In general, ordinary things that were solved in different ways. For example, some discussion areas started in Russian, but instantly switched to English when the first English-speaking participant appeared.
I myself had to make opening and closing inclusions of the online broadcast exclusively in English and participate in a couple of interviews with experts on the record. And it was a real challenge for me, which will not be forgotten soon. At some point, Oleg Anastasiev () just told me to sit with them for the interview, and I was too slow to understand what that meant.
On the other hand, it is very pleasant that the people at the reports asked questions with a bang. Not only native speakers, but everyone in general, it worked well. At other conferences, it is often seen that people are embarrassed to ask questions from the audience in broken English, and can squeeze something out of themselves only in the discussion area. It was completely different here. Relatively speaking, some Cliff Click finished the reports a little earlier, and after that the questions went on in a continuous series, the conversation turned into a discussion zone - without awkward pauses and breaks. The same applies to Leslie Lamport's Q&A session, the host practically did not have to ask his own questions, the participants came up with everything.
There were all sorts of little things that few people notice, but they are. Due to the fact that the conference is in English, the design of such things as flyers and maps is lighter and more concise. There is no need to duplicate languages ββand clutter the design.
Sponsors and Exhibition
Our sponsors helped a lot in creating the conference. Thanks to them, there was always something to do during the breaks.
At the stand it was possible to communicate with engineers of multi-threaded systems, solve their puzzles βfrom the headβ, win memorable prizes and just have a good time.
At the stand it was possible to talk about their own systems, both open and open-source: a distributed in-memory database, a distributed binary log, a microservices orchestration system, a universal transport for telemetry, and so on. And of course, puzzles and contests, stickers with a binary cat and the Suffering Middle Ages, gifts like Martin Kleppman's book and LEGO figures.
Please note that the analysis of Kontur problems is already . The analysis is good, worth a look.
Those who wished could buy all kinds of books, discuss them with colleagues. There was a huge crowd for the autograph session!
Results
The Hydra conference and the SPTDC school are very important events for us as a host company and for the entire community. This is a chance to look into our future, to develop a unified conceptual apparatus for discussing contemporary problems, to look at interesting areas. Multithreading has been around for a very long time, but after the first truly multi-core processor appeared, it took a whole decade for the phenomenon to become mainstream. What we heard this week at the reports is not fleeting news, but the road to a brighter future, which we will follow in the coming years. There will be no spoilers for the next Hydra in this post, but you can hope for the best. If you're interested in these kinds of issues, you might want to take a look at our other events, like hardcore conference talks. or . See you at the next conferences!
Source: habr.com