Last week I had a great time at LinuxCon Europe in Dusseldorf and I want to share my thoughts not only about the presentations I went but also about the general atmosphere and the city ’cause it was a very whole experience for me – perhaps I’ll do the city talk in another post.
It was my first time ever participating in a conference so I can’t really compare this experience to anything else, but it was a very big event and the organizers were really well prepared. The overall content was very varied and most of the time I was interested in more than one presentation during the same time interval, which meant I had to carefully choose my schedule.
Of course I was excited about the Linux: Where are we going keynote that features Dirk Hohndel and Linus Torvalds. It was a very fun and light talk. Compared to the rest of the presentations I went to, it wasn’t all that technical. Somebody raised the issue about the way the community can be rude or mean – I never experienced this, but there are certainly times when it is so – and I guess there are arguments on both sides. This was discussed in the Kernel Developer Panel as well.
Jono Bacon’s talk, Building Exponential Communities, was one the best out of the all the talks at LinuxCon! He very nicely captures the essence of building a (open-source) community: a level of empowerment, a sense of purpose and a sense of belonging. Couldn’t have it any better, really. I am not really into community management but these ideas are still relevant to me: even as a small piece of the puzzle it is very important to understand the foundations of communities in open-source.
On my first day, I got only 4 hours of sleep ’cause I was very excited. I had rehearsed the night before my presentation in my hotel room. The walls were kinda thin and I tried not to talk really loud, but hey, who could blame the other hotel guests for mistaking my murmurs with a broken record player or some magic mantric incantations – sorry, guys.
Enhancing Real-Time Capabilities with the PRU - so PRU stands for Programmable Realtime Unit and it’s a very interesting architecture TI proposes. Basically, you want to get real-time operations over fast (that means very little latency) and deterministic so that’s why you have this specialized processing unit but you also get to connect that unit to another processing unit, an ARM processor in order to take care of the more complicated or more resource consuming operations. The very interesting part I guess is how the two parts communicate and there’s one from PRU to ARM which is sending interrupts and then from ARM to PRU who does polling.
Technical Feasability Study: Is Linux Ready for Medical Devices - the short answer is yes, but no. That means that you get comparable performance with Linux, but there may be other factors to take into account when choosing Linux for your products – for example, the technical support a commercial OS offers. Check the presentation out as it has more figures and goes through the steps in more detail.
Karen and Sarah (btw, they were really cool!) proposed we do a mockup presentation and that was really helpful. Last time I spoke in front of people was during my graduation project’s presentation and that didn’t go very well… I also met a part of the OPW interns and saw a preview of their presentations – really nice work! The rehearsal lasted about 45 minutes so I still had time to go to another presentation. I went to Christoph Lameter’s talk about slab allocators but as the time for my talk drew nearer I just couldn’t focus anymore. The Kernel Internship Report was really well done. I liked the fact that Sarah had some interesting statistics about the interns overall impact on making the Linux kernel cleaner. I was very nervous up until my time to talk. And then I saw we had a microphone. Don’t know why but microphones intimidate me thinking my voice would sound funny. Just when I started talking I managed to look in the crowd and find some familiar faces and calmed down a bit. At one point, I did utter a Romanian word (but I think it was in a very silent voice and nobody noticed). Overall, I think I managed to get my point across. You can also find my talk there. :)
The first talk of the second day was also from ELCE and it opened a new perspective for me. Do you ever think of how the user interacts with the API or the tool you build, how user efficient is it? Well, the talk Building tools from the outside in: bringing user-centered design to embedded Linux tackles this aspect. Belen is an interactions engineer and it was the first time hearing about such a job (guess I’m still clueless). It was a very dynamic presentation and it made very clear why and how the Yocto guys build the Toaster project.
The Chromium OS audio system talk was moderately intelligible to me because I wasn’t familiar with how any audio system works but managed to get an idea which was the whole point. I went to another talk (couldn’t find the slides, but I found an article) who admittedly sounded rather promising but unfortunately the project was still in its initial stage – hope to hear more about this initiative and its results.
I didn’t know much about the big.LITTLE architecture and the Cut power consumption by 5x without losing performance: a big.LITTLE software strategy presentation was a very nice way to find out about it – like how it can be improved. The measurements were made for rendering scenes from browser, a very specific scenario.
Josh explained really well the whole stack of the OS in the Chrome OS internals talk. Although there was a lot of information for me to process, the thing that I found very distinguishable is the security approach Chrome OS takes. It is a very well structured presentation and worth checking out!
High-speed data acquisition with the Linux IIO framework - I am starting to work with IIO and this was a great way of understanding the capabilities of the IIO framework. I mainly looked at how sensors can use IIO (usually having very little data to transmit when a hardware interrupt occurs) but never at how large streams of data could be sent using IIO and DMA.
How to design a Linux kernel API - this was very instructive and Michael Kerrisk really made a great point: it is really important to expose coherent and documented interfaces.
Open Source: A Job and an Adventure - there’s a lot of jobs in open source, not just developers. That’s a thing I haven’t really thought about before. Dawn talked about different ways to enter the open source community and get to actually work in open source. At the end of the talk though I felt a little bit discouraged by the fact that all these people had been a part of open source when it wasn’t that widely used and they were really avant-garde. One of those times when I wish I was born in another period – like when I want so bad to be at a Rush (yes, the Canadian band) concert in the 80s.
The best part of the closing event was the Spaceteam game. Basically you and your friends are part of a spaceship crew trying to go to hyperspace. Each one has only a part of the control system and has to yell out the instructions they see on the screen and cannot be activated using their own control panel ’cause another teammate has that function. The instructions were not always very scientific or serious – things like ‘dry clothes’ or unutterable things – and sometimes the scene looked like utter chaos but these things made it look more fun.