Debian @ International Linux Kongreß 2001

Trip Report

On November 26th, International Linux Kongreß 2001 started with a Clustering Workshop which was held by many experts within this sector. This workshop was thankfully sponsored by the German Ministry of education and research. This was the first time that an expert workhop was organized right before the regular Linux Kongreß and it was also a success, according to what both attendees and organizers told me. Also for the first time, Linux Kongreß was organized outside of Germany, hence, becoming more international. This time it took place a couple of kilometers inside of the Netherlands, at the University of Twente in Enschede.

The Debian Project was invited to attend the conference with both a special developers workshop and a booth in the foyer run by Debian people. This was a great opportunity which was made possible through sponsorship by the German Ministry of education and research and GUUG. Because of this, Debian People did not have to register for the conference through regular channels since their attendance was fully sponsored. Many thanks for this.

Since the offer reached us somewhat late we had to improvise a little bit. Hence things didn't go as smooth as for other conferences where plans were made a couple of months in advance. On Wednesday, November 28th, we planned to hold a workshop for advanced Debian people (like developers and advanced and interested users). We had to decide who was going to attend the workshop and the conference within two weeks or so as well as decide what we will actually do with the workhop. It was meant to last for the entire afternoon on the day before the conference was officially opened.

So for the workshop we were planning something new. An introduction talk should give a short overview about the Debian Project and the techniques used to maintain the Debian Project and our GNU distributions. Following should be time for discussions with the attendants. The introduction should raise some questions that had to be discussed later. Since about a dozen of Debian developers planned to be present during the workshop there should be experts for all kind of questions, so the attendants should be able to get sufficient answers to all questions.

Wednesday

This day was a little bit stressing since I used part of the night getting my own stuff in shape. Finally, when trying to print the slides, because we didn't get a confirmation that there will be a beamer in the workshop room (and indeed, there wasn't), my printer tightly folded the slide inside of itself, so I had to disassemble the printer in order to get the slide out again. Don't you love these kind of things? On the next day I didn't get up as early as planned so I was admittedly only just in time when I finally left Oldenburg, just after I found out that there will be a traffic jam 10km in advance. Oh well, good start...

I arrived late in Enschede, since I got into another 1.5 hour lasting traffic jam near Osnabrück. Well, it wasn't exactly a good start for the day. After I managed to navigate through the streets of Enschede -- since the macro map ended at some point, and the micro map started at another point, one had to connect these points without a map, but interpolation and thinking helped -- I finally found myself lucky parking right next to the Debian booth, by accident admittedly.

At the inside of the conference area the Debian booth was already in preparation. Joost and Thomas were around, reassembling their boxes for demonstration. Thomas brought with him a nice poster of FAI and also borrowed a demo PC from the GUUG for demonstrating how FAI works. This was a permanent attraction during the show and a lot of people watched it.

The workshop was quite a success. A couple of people were around who were not yet known as Debian developers (I guess, some of them could become prospective developers..). We discussed various topics related to Debian. Among them was a journey to the inner working of archive maintenance including buildd's and how the ftp site is structured.

Since Russell Coker was around he also gave an introduction to the portslave package. It provides a program which uses Hayes' AT commands to talk a modem and does fancy stuff with it, used for turning your favourite Linux box into a dialin-server. It will display a login: prompt at which the user can enter a user-name and password. If the user sends PPP data then portslave will run it's own pppd instead and authenticate the user via PAP. When the user-name and password are received they will be verified via a RADIUS server.

Martin Schulte from the Linux Kongreß Team invited us to join the Speakers Dinner at the evening. We were quite lucky, since this wasn't planned. Hence we went to the Debian woody house that was reserved for us to dump all secondary crap so we don't have to carry everything to the dinner. The woody houses were... err... wood houses with simple beds and only electricity and a heating. No water, no kitchen, no toilette... This trip was going to become interesting... However, we found out that there were indeed some toilettes and showers in another building outside, so things suddenly became less critical.

We moved to the speakers dinner with most of the Debian crowd and ended up in the bar just before the dinner was opened -- we thought. Just after finding a table and actually sitting down to start talking again, next to a table with other speakers, we were told that this was not the area where the Kongreß Dinner was going to take place. Bummer.

Again we grabed our bags, laptops and jackets and moved through a maze of corridors after which we finally entered another large room in which the Speakers Dinner was to be celebrated. Strangely, we were the first people to enter the room (for the moment ignoring the fact that there were waitresses and waiters around). The largest table became the main Debian table on which some more chairs were added so all of us got a seat there.

We enjoyed the evening very much, since we actually met a lot of people and were able to talk to them and discuss various topics. Also, interesting discussions about Free Software versus proprietery Software came up ending in the question "Does Free Software actually use its power to come up with impressingly new ideas and use the freedom to implement and try them?"(*) An amazing (or depressing, for what it's worth) number of Free Software Projects target at reimplementing software that is already known in the commercial and proprietary market.

Since Free Software isn't bound to marketing droids and company bosses dictating the goals and features of a particular software, it should be perfectly suited to implement new ideas and come up with drastical changes. However, looking at many Free Software projects this doesn't seem to be the case. New questions came ub as: Why are companies required to come up with new ideas so often? Why are special design centers needed for a new GUI to appear? Maybe the Free Software Community lacks a number of cool artists and philosophers?

Thursday

People with a good sleeping bag had a good morning, people without were cold during the night. Fortunately nobody turned the heating on. I was one of five people who went upstairs for bed, others stayed downstairs where groups of two beds were combined to a loft bed. There were 15 beds in every house, three of six wood houses were rented by the Kongreß and most Debian people stayed in wood house no. 4. I guess we were quite lucky that the wood houses were only about one kilometer away from the conference building, they looked like they were somewhere outside in the woods...

After rebuilding the booth (since we secured our machines by moving them into a locked room instead of leaving them at the booth) some of us went to the opening by Jos Vos and the Keynote held by Ted Ts'o. We really enjoyed the keynote, even though it contained a slap in the face of Debian. However, Ted described one problem that free software has (which is not bound to Debian, fortunately).

User interfaces and documentation are designed, implemented and written by developers. This is a known fact. However, another fact is that developers think different than an average user. This results in a large gap between understanding user interfaces by developers and users. This again results in average users not being able to understand a program, its user interface or some of its structures. Additionally, some documentation partially can only be understood if one has a deep view inside of the software, hence don't require its documentation anymore. Usability testing, dragging virgin people to the computer and forcing them to fulfil a certain task could be very helpful and an interesting practice for the developer to watch.

Most of us frequently went to the lecture rooms to listen to the talks or got cought by discussions with visitors. This was very interesting, because during the talks only very few people were around at the foyer and at the booths, however, after the talks were over and a break was in place, the foyer was nearly overcrowded. This day covered talks about networking stuff (netfilter, wireless, iptables etc.) and ended with user-mode Linux and a keysigning party.

The evening was fun again, the buffet at the social event even revealed some vegetarian food besides vegetables. Again one had to pass through a maze of corridors to enter the proper room of the Bastille where the social event took place. After the dinner we went back downstairs to the "pub" but were thrown out at half past one in the morning. Doesn't that suck? Some people used this as an excuse and escaped to bed, others walked around the building into the Vestingbar where life music was enjoyed until five o'clock in the morning.

Friday

Unfortunately somebody turned on the heating after he came back to the woody house early in the morning, hence it became quite worm upstairs which wasn't a good base for sane sleep. Also, after getting up and actually standing with the head meeting the roof, one got a heat shock. A good reason to escape the house fast if ones own cycle shouldn't collapse.

Only very few people got up early enough to listen to the first talk by Ted again. However, the people who made it to the lecture building got a free shower provided by the dutch nature, wonderful, there were ten minutes of walking required to walk from the woody house to the lecture building, enough time to get wet. It kept on raining for the entire day. Thank god this was Linux Kongreß and not LinuxCamp.

The first talk on this day covered the mandatory ext2/ext3 roadmap which seems to be given every year at Linux Kongreß. Because of the social event and the opportunity to hang around at bars late, the lecture room wasn't as filled as it was on the day before. But anyway, Ted gave an interesting talk revealing some details about current problems of and further plans for the main Linux filesystem ext2/ext3.(**) Backwards compatibility is an important goal even though more features are to be added.

Further talks during the day covered several issues of high availiability computing (Linux-HA), such as Distributed Replication of Block Devices, LVM and Lustre, a cluster file system. The LWN editor was also tracking 2.5 and revealed what the kernel hackers are up to. Another talk covered Rule Set Based Access Control whose speaker was interested in including it into the Debian distribution. However, since a kernel patch is required before its userland will be of any use and it is far from getting included into the main kernel tree, it's not likely to be added as default by Debian.

During the day a girl from Cobalt NL came along with a new RaQ system on which FAI should install Debian. Well... a good idea, probably. However, Cobalt uses special bioses and network cards that weren't initially supported. Also problematic, the RaQ wasn't able to boot a bzImage, only a zImage. After working on it for a couple of hours it looked as it should be possible in general to get Debian installed on the box, just not during the congress.

The final keynote was given by Jon "maddog" Hall who talked about the past and the future of GNU/Linux. A lot has happened in the past years. New Linux companies were founded or went to the stock marked, have disappeared or gotten significantly smaller these days. Nevertheless the number of Linux servers, supercomputers and embedded systems are growing at a large rate. Maddog tried to explain why this phenomenon is happening, and what the next steps should be to move Linux into the marketplace.

Unlike other Linux Kongresses this one provided a small breakfast after the first two talks and even lunch after the next two, and also a tea break before the last two talks. This was very nice and kind from the organizers, since people didn't have to care about eating and could fully concentrate on talks and discussions. This was more than just a couple of juices but a small meal. During these breaks we took the opportunity to exchange key fingerprints, of course.

Concluding my report now, I'd like to express big thanks to the GUUG, the BMBF, the NLUUG for making the Linux Kongreß possible as well as special thanks to Martin Schulte of GUUG who was the head organizer of Linux Kongreß and who had invited the Debian Project to attend and have a workshop of our own. I'm looking forward to attending the next Linux Kongreß in Karlsruhe, a couple of days in advance of LinuxTag 2002.

Post Scriptum

(*) Some new ideas that were invented through Free Software include BIND (internet nameserver, without it, the internet wouldn't be able to exist), c-news and INN (Usenet news servers, electronic bulletin boards etc.), themes (themable widget libraries, think of Gnome and KDE), Enlightenment (even though some people may miss some functionality, but it's look is definitively new), X11 (the ability to export displays over the network), xiafs (who of you does remember the filesystem Frank Xia designed?), HTML (of course, crediting Tim Berners-Lee), Emacs (ever saw a lisp interpreter that can actually edit files? Lacks a decent editor, but hey...), Languages like Perl, Python and Ruby.

(**) Like there are programs out in the wild that want to create more than 2^15 directories, which is not possible with the current ext2. Or new and fast machines that compile a whole bunch of source files into object files while the filesystem can only store second-significance for timestamps which is able to screw up make. Hence, at least microsecond-significance would be fine. Or small 1 gb files (small as in movies tend to be larger, think of video production) that are to be removed and the filesystem has to follow all inodes in order to free them. Instead, if subsequent blocks would be used this could be faster. There is more...

Report from Pro-Linux.