Notes from the FOSDEM '08 Conference


    So, this year again (February 23-24) I was able to attend the
    FOSDEM - Free Open Software Developers European Meeting.
    This event is held annually at the end of February at the Free University in Brussels.
    Like last time , I will not go into the technical details of the reports, just try to convey the atmosphere of the conference. If you want to look through slides or watch videos of performances, then you can find it on the official Fosdem site .

    This time I focused more on reports that were interesting to me, so I can’t tell you more than what I saw and heard myself.
    I was interested in listening to people who work with FreeBSD and parallel systems (clusters and other HPC), so most of the time I spent in the * BSD user group room.


    After the traditional Friday Beer Party (Belgian beer is just music!), From Saturday morning we were already at the opening of the conference, which is traditionally held in the Janson audience. After the opening speech of the FOSDEM committee, it was the turn of the first report - “Linux in Hollywood”. Everything was funny until the moment when someone did not ask the speaker what kind of free software they use in Hollywood. In addition to Linux itself, he found it difficult to name anything (ffmpeg does not count). The hall grumbled :)
    But what can you do, this is the truth of life - serious video processing at this point in time is provided only by paid packages.
    src = "" />FreeBSD Project Servers
    src = "" />Explanations for various development branches
    src = "" />Freebsd ecosystem
    src = "" />Schedule distribution of project participants by age
    src = "" />Developed World Map

    However, the most interesting thing started after that. I must say, I am very pleased with the increase in BSD-related content every year. This year, one of the first speakers at the conference was Robert Watson himself, a security officer from the FreeBSD core team! And the topic of his report (“How a large scale open source project works”) is very relevant in my opinion - he talked about how the development of the FreeBSD project was organized. Well-coordinated work of thousands of people, multi-level quality control, discussion of new features, resolving disputes between developers - this allows you to create a high-quality operating system without which the current Internet is unthinkable.
    I remember a talk by Alan Cox, a Linux kernel maintainer, that I listened to a couple of years ago. He then praised the organization of work in the FreeBSD team, complained that with each version he had to rake the code, merge patches. Then in his report he talked about isolating commits in different branches so that new pieces of code first pass integration tests and then only fall into the stable version as the guys from FreeBSD do . In conversations on the sidelines, many complained that yes, they say, you have to constantly monitor new versions of the kernel and quickly update.
    Well, okay, I don’t want to argue which is better - everyone chooses for himself. It only seems to me that * BSD systems are not well known to the general public and such reports are very necessary for all of us. By the way, Robert Watson spoke in the largest audience that is available on FOSDEMe and the hall was completely full - people even sat on the steps in the aisles.

    The next report, “Status update of Software Patents”, was a little sad - the situation is getting worse every year. In America, patenting has become the main brake on the widespread dissemination of innovative technologies, patent barriers are everywhere and the IT industry is suffering a lot from this. The situation in Europe, fortunately, is a little better, but large companies carry out powerful lobbying of their interests. The speaker is a lawyer by profession, he defends the interests of the Open Source Community with his colleagues and was very pessimistic. Without the support of governments and without clarifying the dangers of patenting, it will be very difficult to deal with these barriers.

    After lunch, I went to listen to the opening of the FreeBSD Developer Room. Since the guys from Berkeley make not only a unix system, but also PostgreSQL, they made their presentations in the AW corps. It was convenient, in addition, post-Grace workers recently released a seriously revised version of their database.
    The first report was dedicated to this event - the innovations and improvements of the existing functionality were listed. The main thing in the new version is a completely redesigned engine, which has now become much more productive.

    A report on the new version of FreeBSD 7.0 followed, however, I myself was aware of all its innovations, as I carefully follow its development and upgraded the day before. Therefore, I succumbed to the entreaties of friends and went to listen to Andrei Zmievsky with his "Unicoding with PHP6." In general, there is nothing special except the full Unicode support in the new version, the whole report was about how cool it would be to write on a puff now without various crutches. For some reason, it still seemed to me that Andrei was somehow removed, "far from the audience," his report "did not catch." There was a feeling that he did not care about this project, that he just came to report on the work. By the way, I noticed several guys from the RoR-community, they were sitting quietly and only occasionally meaningfully looked at each other smiling :) They did not ask questions ...

    As I was told, the previous report “Perl 6” was also boring - the developers completely changed the syntax of the pearl without adding anything significant in essence. On the sidelines it was in full swing how it will now be difficult to rewrite everything under the new version - as you know, pearl is a hard-to-read language, it will be very difficult to understand the implemented functionality and reproduce the same one working in the new version. In general, there was a lot of criticism of Patrick (the speaker). He made excuses, even specifically emphasized that he "only realized what the community approved." The new syntax is by the way something, take a look if there is time, there are very interesting things.
    Patrick paid special attention to Parrot, an interesting little thing, which makes it easy to make interpreters of new languages ​​according to the bnf scheme.

    In general, after PHP6 I could not stand it and returned to * BSD & PgSQL talks. There they just finished discussing the status of SNMP in FreeBSD (as I understand it, things are going well). This was followed by a report detailing the progress in increasing the speed of the postgres server. Although I already read all this and it has already been discussed in many places on the Internet, it was interesting to know the details, due to which the performance was improved. This is mainly more predictive of the engine and more intelligent control of disk I / O.

    Then there should have been the most interesting - “SWOT Analysis between PostgreSQL and MySQL”. The hall cluttered to the eyeballs. I must say that the conference was attended by many people from MySQL and several people from Sun. Last year, Sun's vice president of open-source technology, Simon Phillips, spoke on Liberating Java. This year I saw him on the sidelines, besides there was a traditional OpenSolaris booth and, as you know, Sun is one of the sponsors of FOSDEM. Now add to this the fact that Sun recently acquired MySQL. Can you imagine what interest was on this report from both sides?
    On the one hand, PostgreSQL is under a BSD license, and on the other, it is still free, but with a powerful corporate MySQL backer.
    In short, an aunt came out before us and said that the report was postponed to Sunday. After a couple of minutes, the hall was almost empty ...
    In general, it was not very interesting further and I decided to take a walk and look at the stands.

    I must say, on FOSDEMe you can pass exams on Linux and FreeBSD administration. This is very convenient, because otherwise it is quite difficult to settle various organizational issues.

    So ended the first day of the conference.


    On Sunday, I first became interested in project build systems and went to listen to a talk about Hudson. The report was short, only half an hour, during which the speaker talked about how this toolkit is configured and "how it works great later." I didn’t listen to the end, since at 10:30 Postgres Future Directions started and I ran there.

    There, Bruce Momzhlan excitingly talked about various new ideas that postgraduates would like to implement in future versions. He admitted that of course everything will not be possible to implement to the next version, but nevertheless in 8.4 the storage device will be improved and the operation of indexes and some built-in functions will be optimized. In the future, they will improve clustering (I think you can even participate) and replication.
    You know, but I really liked how Bruce told all this, he is so passionate about this project! He has the same spark in his eyes, he spoke not on the slide, but from the heart, it seemed that his ideas were born right here. I was very happy for this project, it is in the right hands and these guys know what they are doing. In the end, of course, he broke the applause :)

    Then Magnus spoke, talking about how they ported PostgreSQL to the Win32 platform in native mode. Well, here of course it was about fork (), shared memory, I / O, and so on - a typical set of problems when porting. For a long time they lagged behind the main development branch, but gradually caught up with it and integrated their achievements into it. Surprisingly, at some point they had better performance than the unix version. By the way, I did not know that the main system under which PostgreSQL is developed is FreeBSD. Now I don’t remember who, but one of the speakers said the day before that FreeBSD is their main workhorse. And although they didn’t sharpen the code specifically for it, it was nevertheless claimed that postgres is best rotated on the fra. That's it.
    Of course, the speaker was asked why they started porting to a proprietary system, which he said that this would significantly expand the number of people using this wonderful database. It’s hard not to agree.

    The next 15-minute talk was about Ruby on Rails on FreeBSD with PostgreSQL. Well then, there was nothing extraordinary for me, a year ago I myself had configured such a bundle at home, even with Apache, I checked it for hosting. Everything just works, nothing special.
    Then there was a small talk about “DBI-Link 3.0” - this is Perl-binding for PostgreSQL, also useful to know.
    Next was the “New Enterprise Features in PostgreSQL”, which described various options for backup, management of nested transactions, etc.

    After lunch there was an interesting systematic report about GNU autotools: autoconf, automake, libtool etc. Indeed, such things are very important, since they allow you to standardize the assembly of open projects and get rid of solving many already solved problems. The speaker categorically asked to use only autotools and not to reinvent the wheel every time again. “Don’t even think about creating something similar to autotools, most likely you just won’t be able to do it, but you will waste a lot of energy. Better take our toolkit and focus on the idea that you are going to implement ”. He also told how he once raked someone's project, which was half in autotools, half in anarchy. He brought programming pearls - it turns out that many are trying to use autotools without fully understanding how this works. In general, although I use this in my work, I wanted to refresh my ideas about it. Now I downloaded a manual for myself on these things, re-read it, find out new details, get enlightened :)

    The next report was about a full-text search on the postgres website. It turns out that if the already implemented things are used correctly, then all this works very quickly. And of course, you need to think carefully about what and how to index.
    Of course, I understand that things like DBSight (based on Lucene) are more universal, but as an example of the same Sphinx shows, specialized solutions can achieve the best. There was nothing extraordinary in the report, just an example of the correct approach to solving the problem using the standard PostgreSQL functionality.

    The last thing I managed to listen to this time was a report on the implementation of “Two-Phase Commit with PostgreSQL”, which was implemented in version 8.1. Of course, this has long existed in many commercial databases, but the fact that it is in postgres puts it on a par with them.
    Then my attention was attracted not so much by the report, but by the sitting next to me ... Robert Watson! Of course, I wanted to talk to him better.

    I must say, he seemed to me younger than in the photo, apparently younger than 30 years old. He has a 15 "MacBook Pro with Tiger (I have a big eye, since I myself am sitting on a macbook and recently moved to a leopard, I see the difference right away). True, Robert practically does not use his native" poppy "applications (perhaps except for Safari) - I noticed several terminals with ssh sessions in which he looks at mail through Pine, chat in IRC using something textual, and of course compiles something in eterm in textual Emacs :) Eh, but we are somewhat similar ...
    I do not remember, I said it or not, but FOSDEMe in each building there are several points of access Wi-Fi. Free natures continuously, so that many are sitting with their laptops and something there "hachat" all the time.
    I did not distract him very much, and I also had to listen to the report, he told me that the fry had already been released, it hadn’t replicated to all the mirrors yet, he advised cvsup for safety for Monday or later.
    ( UPD: indeed, the sources were tagged only on the 26th, I have already


    After that I went looking for my comrades because I had to get ready to return home. We bought several books (O'Reily are also sponsors of the conference) and once again walked past the stands. There were already fewer people - many left.
    src = "" />
    There were several dozen people who exchanged PGP keys. This is a traditional Key Signing Party, where people show their passport and exchange their public key with a partner. This year, for the first time, there was also an exchange of SSL certificates, but as I read, Open CA is not yet recognized by serious organizations, they need to pass certification to become a full-fledged certificate authority. Nevertheless, already now it is possible to register a public Open CA certificate on your computer and use cryptographic content with a certain degree of security. By the way, Open CA is supported in almost all open web browsers. Therefore, for example, you can create your own certificate and sign it with Open CA to make crypto content available on your web server.

    A little about the atmosphere of the conference.
    As usual, several strong user groups were present - Debian, CentOS, Jabber, KDE, Gnome, PostgreSQL, * BSD, OpenSolaris, Gentoo (although they were completely blown away something this year). There was a stand with embedded devices, though it seemed to me that there was nothing new compared to the previous year. For the first time, Mozillites were represented in bulk - and they had badges in bulk there, and they handed out posters and shirts sold :) BSD and others became more represented - they even had two large stands this time. Accordingly, there were significantly more people from PostgreSQL, the stand was also visible from afar - and the blue elephant, and badges, and shirts (I bought myself one, could not resist), and they had disc books. KDE and Gnome stands were not far from each other, on the contrary - a traditional bulletin board on the wall with a bunch of pasted posts for and against.

    src = "" />
    According to my observations, approximately 60% of visitors are regular users interested in open source. Of the remaining ones, there are about half of the hackers who often try to stand out with all sorts of wicked T-shirts, laptops glued with all sorts of nonsense, etc. The rest are people who are actually involved in the development of open projects or representatives of companies that seriously use open source projects.

    It seems to me that this year the stands were less popular than talk rooms. It pleases, it means that more and more this event serves to exchange views and to disseminate ideas. Again there was Google, but very imperceptibly, he had no stand at all, but he sponsored the event. If only Sun would take an example from him, otherwise he would overshadow many projects by himself ...

    I didn’t really like the choice of books from O'Reily this year, there was a lack of something solid, eternal (like Straustrup for example). There were many one-day books, which for the most part repeat the manuals and which therefore I do not want to buy.

    I liked the choice of beer in the cafeteria, here the organizers have made progress, well done. To whom it is, but I remained faithful to my beloved Kriek. Maybe you should open a second cafeteria somewhere nearby, or else the people are getting bigger and during the break you have to stand in line to buy your own sandwich. My friends and I eventually ate in a restaurant nearby, but this is already more expensive and longer.

    Oh yes, those who read my previous report remember that girl in a short skirt who comes to FOSDEM every year. So, we saw her again this year, though she was with a guy this time :) Well, okay ...

    These are the impressions.

    28 Feb 2008

    Also popular now: