Live from EuroBSDCon 2013, day 3

Live from Malta today attending the EuroBSDCon 2013 conference. Today is the first day of the conference itself. Many more people have shown up as expected and there have been tons of very interesting talks all the time. It is both good and bad that there are several tracks: you can select the topic you are most interested in, but sometimes great talks overlap!

Keynote

Today's opening session was led by Theo de Raadt, the founder of OpenBSD. His keynote focused on explaining how there is no real research happening on operating systems any more and how new, risky technological changes can be tested in a real-world system like OpenBSD.

Theo later focused on the change of their time_t definition to be long long on all platforms to transparently cope with the Y2038 problem, which is actually a real issue that will hit us all even before 2038.

Google stand

Google is a platinum sponsor of the conference, so we had a stand in it. The stand is staffed full-time by a recruiter and part-time by the 5 Googlers attending the conference. I stood by the stand a few times during the day and got to talk to some people about their current work and about how being an SRE at Google is.

Hallway talks

Again, hallway talks are a very powerful aspect of these conferences. Here are two of the highlights from me:

The first one is my chat with Colin Percival, the author of Tarsnap. Tarsnap is a secure online backup system for the truly paranoid. In talking to him, I learned a bit about how his standalone work is, a little bit about Amazon EC2 and S3 and the current status of FreeBSD on EC2. In his presentation later on about this topic, I liked his clear explanation of how libarchive implements all the logic regarding archive management and how bsdtar is just a thin layer over it, which is a great example for my my CLI Design series.

The second one is my talk with Valeriy 'uwe' Ushakov — a long-time NetBSD contributor that I had not met for years. From him, I learned that VirtualBox was actually not discontinued by Oracle, which is something that was misreported in the news (and thus confused me until now) at the time Oracle announced they would stop supporting some virtualization technologies... but not VirtualBox!

NetBSD

Today has been a strong day for NetBSD. There have been several talks throughout the day about technologies in NetBSD and they were popular enough based on the number of attendees. Of special interest are an update to the toolchain changes required to move towards a fully BSD-licensed world, a very detailed description of how NPF works and how it scales, and my presentation about Kyua. More talks are coming up tomorrow.

I also got to meet some NetBSD developers that I either had never met before or that I had not seen for years. It is always good to put faces to people and to catch up where due. Also, as one of my bosses said: "People that work together but don't meet in person at least once every 6 months become email enemies" and it's very easy to see how this is true in many open source projects — especially in NetBSD. I'd like to encourage NetBSD developers to attend these conferences more frequently and interact with people outside of NetBSD. There is a lot to learn from other projects (and viceversa)!

Kyua

My talk on Kyua and automated software testing happened today. I focused half of the talk in explaining why writing automated tests is very important and the other half in detailing the architecture behind Kyua.

Quite a few people attended although there were not as many questions as I would have liked at the very end. However, I would like to think that the presentation was successful and that the topic was interesting enough: I did not spot many laptops open.

Please send me any feedback you may have! You can also check out the slides although they don't have much content on their own.

Ports, KDE 4, Firefox

There were quite a few presentations on ports-related stuff also spread out throughout the day. One of the sessions included a description of major changes happening in the FreeBSD ports tree, although I could not go to that one. Other presentations included all the work that OpenBSD had to do to get KDE 4 running — not a small amount at all — and a very interesting presentation on how to keep up with Firefox's packaging in OpenBSD — not a small amount either.

It is clear from all these presentations that maintaining packages for large third-party projects is painful and that such pain is shared by the various BSD systems. That said, it is good to hear that such packages are being tackled and that the maintainers are working more and more with upstream to minimize the pain down the road.

More tomorrow.