FOSDEM 2026

The purpose of this trip was (ostensibly) to attend the FOSDEM 2026 conference. Since I’m no longer working at Beeper, there is not really any work-related reason for me to attend the conference. But I wanted to do it to give myself an excuse to go to Europe, and as an opportunity to catch up with former Beeper coworkers, and meet up with a few other friends as well.

I departed Prague Friday morning on the earliest flight. The flight was an uneventful and I slept through most of it. I’ve flown into Brussels Airport Zaventem many times at this point. It’s probably the European airport that I have the most experience with at this point. I’m at the point where I have navigated it enough that it feels familiar. (It’s crazy when you start to know where things are at foreign airports.) Overall, I think the airport is pretty good, but the best part is how well connected it is to the train system.

Once I arrived in Brussels, I headed straight for HSBXL (an open source hackerspace in Brussels) where there was a Matrix hackathon. Things were just starting to get going. Projects were being proposed, and people were discussing what they were going to work on during the hackathon. I ended up working with Henry on the Nexus client which is a Matrix client that uses Flutter with the Gomuks FFI SDK. I have a bit of experience with Gomuks, so I was able to help out a bit. Honestly, though, I was not very useful. I helped work through like one issue, but Henry did all of the actual work.

Instead, I spent most of the time configuring Niri and catching up with Sam and Addison who had also made the trip to FOSDEM from the states. Towards the end of the morning, Tulir arrived and was actually helpful to Henry, since he actually built the FFI SDK. It was pretty cool to see how well the SDK worked for developing the Nexus client.

After the hackathon was over and people who actually did things shared what they had worked on, Addison and I headed to our hotel to get checked in and then we went to dinner. We went to a burger place that was kinda mid. I was quite tired considering I’d been travelling since early morning, so I turned in for the evening.

On Saturday, I met Sam for a late breakfast at a cafe located halfway between my hotel and Université Libre de Bruxelles (ULB) where FOSDEM is hosted. It was nice to catch up and hear how he is doing. He is moving to London soon to work at Adobe, so I think I’ll have excuses to make a trip to the UK every once in a while.

Once we got to ULB, the most interesting talks looked like the ones going on in the Browser and web platform track.

The first talk I attended was about MathML Core which is a cool way to encode mathematical typesetting in HTML. I have used MathJax in the past for this, but MathML appears to be something I could use for many cases going forward. I’ll have to try it out sometime.

The next talk was about WebViews (the embedded browsers that pop up in-app on mobile apps). I didn’t realise that the WebViews are entirely separate from the browsers you have installed on your device. This naturally causes fragmentation and confusion because the features are also often different between WebViews and the mobile browser.

The last talk I listened to in that devroom was about WebTransport. It’s a more modern protocol than WebSocket, and provides things like the ability to have unreliable delivery of packets which is useful for things like streaming where dropping packets instead of attempting to resend is actually the preferable behaviour. It seems like it’s basically a lower-level version of WebSockets, and is built to take advantage of HTTP/3. I don’t really have any use for it at this point, but it’s nice to learn about the new technology.

At that point, I went over to the big auditorium for a talk about a Git workflow the speaker refers to as “Atomic Flow”. The core of the flow is that each commit should “atomic”. You should be able to checkout any commit, and everything should build. All commits should be easily revertable, and you should have a linear history. I’m a big fan of AtomicFlow, and I have used a lot of the principles of that flow in my projects. I wonder if there is anything that I can learn and apply to the Git processes that we use at Can/Am.

After that talk, it was stack time so I went to get a Belgian waffle from the food truck and then went over to the Matrix stand to catch up with people. I had a nice conversation with Neil and caught up with Kim. I also caught up with Tulir and Tobias about how things were going at Beeper.

Then I went back the browser devroom for another couple talks. The one that I found the most interesting was one about the WASM garbage collector, but by that time my brain was fried so I decided that acquiring nutrition would be the next order of business. Addison and I found an Indian restaurant that was pretty decent, and then I headed to bed.

On Sunday, I had a slightly late start, but I managed to get to the Go dev room for the talks I really cared about. The first talk was about the new Green Tea garbage collector in Go 1.26. The new GC is optimized for modern processor caches and takes advantage of vector operations. I didn’t quite understand what was going on, but I’m always interested in how people are taking advantage of new processor features.

The next talk was about Go’s reflection. It talked about how type information works, and it honestly felt pretty obvious. It was kinda nice to know that Go’s reflection does more or less what you would expect it to.

By this point, it was lunchtime so I went and got some noodles from one of the food trucks and then went to the Matrix devroom for Hubert Chathi’s talk about DMLS. Messaging Layer Security (MLS) is an IETF standard for end-to-end encryption in messaging systems, but relies on a central server. Hubert talked about two approaches for making it work in a distributed manner. Both of them are (not very) conveniently named DMLS (Distributed MLS and Decentralised MLS). One of them uses this thing called a Puncturable Pseudo-random Functions which sound cool and scary at the same time. It’s seems like a very active field of cryptography research. It definitely sound like there is not going to be any movement towards using MLS in Matrix any time soon, since there are so many foundational problems to resolve before then.

That was pretty much the trip. I enjoyed catching up with friends at FOSDEM, and it was nice to see Sam (despite living in the same state, we seem to only see each other when we travel overseas to the same conference). I am not sure if I will attend FOSDEM next year (it’s not the best time of the year to travel to Europe) but I have enjoyed making the trip the last few years. It has been especially nice getting to know members the Matrix community through events like this.