Open Source Summit Europe 2017

In September, I attended Open Source Summit Europe. I realized this week that I haven’t written about it yet. The conference is massive. This edition had 2187 attendees from 65 countries according to the post-event email. The great part of the big event is the chance to meet and socialize with people I don’t otherwise get to meet. The side effect is that you often have more than one session you want to attend. The other end of that is you’re about wiped out at the end of every day. Even before the social events for the day start.

I did not get to attend all the talks I wanted to. I did skip a few sessions and spend some time recovering. Here are the talks that I liked.

The keynote by Reuben Paul about security was quite fun. It’s rare you see someone show a live demo on stage and have it work well. His talk is a reminder to developers that obfuscation is not security. When you build something, think about it from the point of view of an attacker. In the world of IoT, the question is not “if” you’re compromised, it’s “when” you’re compromised.

My colleague Robert Kratky talked about modular documentation. This is one of my favorite talks from the event. I’ve taken copious notes about it. The summary of the talk is to build modular use-case oriented documentation. “How to make an omlette” rather than document that talks about knives, chopping, onions, and eggs. While references need to exist, documentation needs to solve users problems.

I don't remember what I'm passionately arguing about

The first evening, there was a CentOS-Fedora-EPEL BoF hosted by Jim, Peter and Brian. This session teased out problems in the ecosystem and how it some of our solutions aren not perfect. For instance, a package in EPEL cannot override a system package, because it’s meant for RHEL as well. A package in CentOS SIG can override system packages. This is the recommended route for non-EPEL packages into CentOS.

On the second day, I attended a half day session about CHAOSS project. It was a good introduction to metrics and what other communities do. Amar and I attended the session together. We’ve come back with a long-list of things we want to do to track the health of our community. The session was quite long, technical and educational. If I had spare cycles to contribute, that’s where I’d be spending my time. This session also gave us a chance to talk to our friends from Bitergia.

My talk was on the last day of the conference. I like my talk being on the first day because it lets me be stress free for the rest of the event. I talked about testing products where there’s a wide range of configuration options. The premise of my talk is that “Unless you can prove it with a test, a feature is assumed to not work.” My approach is that the product needs focus on solving use-cases rather than features. We need to view every feature with the lens of what problem it helps solve. This will let us narrow down configurations which work best for use-cases. This reduces the permutations of configurations which need to be tested.

Image credit: Linux Foundation OSS EU (license)






Leave a Reply