Flock 2017

Two weeks ago I got to travel to Cape Cod (or as I came to call it, Cape Code), Massachusetts, USA for Flock, the annual Fedora contributor conference. I arrived on Monday, August 28 after flying in from Denver, CO where I had been eclipse-viewing (well that happened in Wyoming) and summitting 6[0-5] fourteeners[6].

Tuesday began with a keynote from Matthew Miller, where he presented metrics on the various versions of Fedora in the wild, and talked about where Fedora is heading. After that we had a long session where the presenters each got to give a short pitch for their talks. After that was lunch, and one thing I enjoyed about the schedule this year was the choice to make lunch be two hours each day. That gave us plenty of time for "hallway" type discussions throughout the week. After lunch I went to see Mike Bonet present about Factory 2.0 and the various items that team has been working on in Fedora. Several of their objectives have been related to Bodhi so this was a good session for me to attend. After that I held my "Bodhi hack sesh" session which I thought went pretty well. I think we had about 12 people attend, and I was able to help people get started on a variety of patches for Bodhi. Lastly I attended the dinner and game night, which was a lot of fun.

I started Wednesday morning off with some hallway track discussions, I believe regarding multi-arch support for Fedora's container registry. We will need to get multi-arch support into Skopeo to do this, or we may be able to write a workaround script to accomplish what we need in the short-term. I then attended Pengfei Jia's talk about Greenwave and Bodhi, which is a topic I had spent considerable time on this summer. After this I again enjoyed the long lunch session, as I got to hang out with people and talk about various things. After lunch I attended the beginning of the Fedora Hubs Hackfest which was a lot of fun. Aurilien had made a nice Vagrant image just for the hack session. Unfortunately I was unable to get it quickly due to bandwidth issues, but I did get it running eventually. I ducked out early so I could see Jeremy Cline's talk on the future of fedmsg, where he presented about a number of issues we face with fedmsg and some of his ideas on how to address them. I was particularly interested in the proposal to use a publishing service to reduce the number of sockets in use in Fedora, and to reduce the complexity of configuring fedmsg. I also think this design could work better as we move Fedora infrastructure apps into OpenShift. This talk led into a lengthly hallway discussion about reliable messaging that I stuck around in for quite some time. After a brief rest I ventured out to attend the conference party at Professor Wackenhammer's Arcade, which was a really fun time. I got to spend some time talking to Christian Schaller about his professional journey, and about the workstation development team.

I started out Thursday at the keynote speech by Mike McGrath, What Does Red Hat Want? He described why there are so many large changes happening in Fedora right now, which I thought was helpful. I've found myself feeling concerned at the rate of change in Fedora at times, but it's helpful to think of these changes in the context of what Fedora is - a distribution for innovators. In fulfilling this role we should expect the ride to be bumpy at times. I then attended Till Maas's hack session, the Python Coding Dojo, for a bit. It looked like it was going to be a lot of fun, but then I realized that I was missing Kevin Fenzi's Infrastructure talk. Since I am a member of the Infrastructure team it seemed important that I attend that talk so I had to apologize for ducking out of the coding dojo and I skipped over to the infrastructure talk. When I arrived, the group was in a brainstorming session about short, medium, and long term goals for the infrastructure. We then broke out into small groups to focus on various items in the infrastructure. I found myself in three of the small groups at once! One of the groups was Owen Taylor, Adam Miller, and I and we talked further about how we could enable the Fedora container registry to support multi-arch, as well as how we could add Flatpak support to it. We also discussed how we will enable our user to discover the containers available to them. Another group discussed the DRPM issue that had been discovered in Fedora 26 (due to the inclusion of alternative arches in Fedora 26, Bodhi was unable to produce DRPMs.) Dennis Gilmore was able to fix this during the hack session! Lastly I also had some discussions about OpenShift in the infrastructure and how we would handle Fedmsg with that new architecture. This was a very valuable session for me. After lunch, I attended two modularity presentations back-to-back that were informative. Bodhi still has a lot of work left to do to get modular support in place, so these were good to be able to attend. I then did some hallway chatting, and then went to see Stef Walter talk about the new CI pipeline that is about to be turned on in Bodhi. In the evening I attended the GPG keysigning party (lots of people came!) and then went to dinner with some of my colleagues. Amazon.com ended up picking up our dinner tab which was generous!

On Friday I attended the morning session where presenters were invited to give recaps on how their sessions went, and the conference we done. As it was significantly cheaper airfare for me to stay an extra day, I spent the afternoon enjoying Cape Cod and doing a little hacking. For dinner, I went to eat sushi with a large group and sat with tyll, sayan, and spot and had a great time. I flew home on Saturday afternoon, and slept for 15 hours when I arrived!

One unusual thing about this year's flock was an emphasis on doing rather than talking. The idea was that there would be many hacking sessions and relatively fewer talks. I liked the idea and delivered a hack session that I would say was successful. However, I did later feel like I personally got less out of this year's Flock due to that than I did out of last year's Flock, as there were less sessions that I could attend and learn about things that are outside of my pervue. On the flip side, I did get some pull requests out of Flock and I am sure many others did as well. Perhaps my suggestion for next year would be to find more of a compromise between the two - perhaps fewer hack sessions than this year, but more than we had last year could work well.

I look forward to Flock 2018!

[0] https://en.wikipedia.org/wiki/Mount_Sherman [1] https://en.wikipedia.org/wiki/Mount_Evans [2] https://en.wikipedia.org/wiki/Pikes_Peak [3] https://en.wikipedia.org/wiki/Mount_Lincoln_(Colorado) [4] https://www.14ers.com/photos/peakmain.php?peak=Mt.+Cameron [5] https://en.wikipedia.org/wiki/Mount_Bross [6] https://en.wikipedia.org/wiki/Fourteener