Tuesday, April 19, 2005
Linux.conf.au Day 2
I was part way through writing up my notes from today and discovered they were already twice as long as yesterday. So I stopped and cut things back a lot. It's still pretty long though...
Today was the second and final day of the mini confs. After looking at the program yesterday and picking the talks I wanted to hear, I realised my day was going to be pretty similar to yesterday – OpenOffice.org in the morning and Debian in the afternoon.
The OpenOffice people decided to give away some OpenOffice books to people who had filled in their evaluation forms (I hadn’t dropped mine in because I wasn’t there for the whole thing). The first one drawn went to a guy in a yellow shirt (one of the conference helper/volunteers) which I thought was pretty cool. He tried to convince them to redraw it, but the audience wouldn’t let them. Which again is pretty cool.
The first session was entitled “To OOo or not to OOo? That is the question”. It was given by a guy named Kevin Russel from a group called
He talked first about some of the myths about why people move to OpenOffice, and made some really interesting points. Like the fact that one of the advantages of Open Source software is that the money you spend to implement it stays in the local community (in the form of consultants etc) rather than going overseas (in licence fees). Someone in the audience also made a good point that in a government organisation it actually costs a lot to implement a $0 piece of software – apparently this guy has actually had senior managers sign a cheque for $0.00, because that was part of the required software procurement process.
The second talk (and the one I was most looking forward to from OpenOffice) was a case study De Bortoli Wines. The speaker, Jonathan Coombes, had been called in after the upper management had tried to force the Finance department to use a buggy release of OpenOffice without proper training.
Coombes talked about some of the interesting problems he faced, like how they dealt with De Bortoli’s problem of having several fairly remote regional offices. The solution ended up being have very little in the way of hardware outside of the main office. Basically most of the PCs don’t have any hard drives. They just boot straight from a DVD, and the individual user data is stored on a USB drive. That means upgrades can be rolled out (and rolled back) really smoothly and quickly.
Coombes also talked about how they implemented a blind trial – they didn’t tell the users that they’d changed operating systems to Linux. And most of the users didn’t even notice they weren’t running Windows. A couple apparently noticed the lack of a START button, but even that wasn’t a big deal.
Next up I snuck into the Debian theatre to hear Angus Lees talk about the Non-Free branch of Debian. The basic outline of the problem is that the Debian philosophy is that the Debian distribution should consist of only software they the source code is open. This means that a lot of things like firmware and fonts shouldn’t be included in the Main branch of the code. But, if you don’t get the drivers for your network card when you install Debian, how do you go on the web to download them?
The talk was really interesting, and gave me a great insight into the Debian mindset (something I hadn’t really realised existed). Half way through the talk it kind of degenerated into a freeform discussion, but that was just as cool (and to his credit, Lees seemed quite happy to let it go). One of the most interesting people in the discussion was a big bearded American who said he was the HP Open Source & Linux CTO (or something like that – I’m sure someone will tell me who he was - turns out he is Bdale Garbee. Thanks again Stilly). He was pretty convinced that the market will move towards more open source drivers in the near future, so Debian should stick to its guns and maintain a hardline reading of the FSG.
After lunch we had a talk from Steve Kowalik about his pet project, Linda. I’m still not quite sure what Linda is supposed to do other than check Debian packages. However, the talk was a really interesting insight into how a random person develops for Open Source. Kowalik had come up with an idea for something he thought would be useful, and so written it. He talked about how he wished he would get more bug reports so he could make it better (and I guess so he knew people were using it), he talked about his development cycle and talked about what he wanted to do in the future.
If I said “fortunately Kowalik’s talk was only half as long as it was meant to be”, people might get the wrong idea. But it was lucky, because it gave another Debian developer James Cameron (no, not that James Cameron) a chance to talk about an app he wrote. His problem was that if he was updating his friend’s Linux builds, it would take forever to download the required packaged (the bandwidth in the outback being pretty crappy), and he’d find himself downloading the same packages at each friend’s place.
His solution was to write an app (called apt-walkabout) that would basically bundle all the package requests together so that he’d only have to download them all once and he could do it somewhere with better reception. Very cool.
The Herding the N00bs talk by Matthew Palmer turned out to not be quite what I was expecting. It was more about turning random Debian users into hardened Debian developers. But it was still pretty interesting. He talked about the various things he’d like to implement in the Debian mentor program (stealing fairly heavily from the Debian-Women mentor program). He also gave a pretty good run down of the things someone needs to do to get a package into Debain, and how to get one of the cool @debian email addresses.
The final talk for the day was on why Debian needs certification. The basic conclusion was that Debian doesn’t need certification, but it would make the rest of the world much happier if there was a bit of paper they could check for. You can have the best understanding of something in the world, but often if you can say “yes, I have that qualification!”, you probably won’t get the job.