The first thing that surprised me was that when I arrived at 9:00 (the keynote started at 9:30), the courtyard was already full of people. And I had some interesting conversations in the first half hour.
The opening keynote was boring.
Then I planned to go to the talk “Creating more meaningful Fedora release notes”, but the talk was cancelled at the last minute. So I moved on to the “Upstream Meetup”. Tomas and Lenka prepared interesting questions to warm up. And the answers and discussion was so interesting that we didn’t get beyond this opening slide. But that didn’t matter at all. Together in the group we shared our experiences on how to manage an upstream project and how to work with contributors. Tomas took notes and I’m looking forward to posting them somewhere.
The discussion was so interesting that I skipped “How to build Collaboration and influence Open Source Project” where I originally wanted to go.
On my lunch break, I walked around to four food trucks and chose a Japanese (or Vietnamese?) one where I had some great chicken pieces with rice Kaarage Don.
Then I watched the talk “Learning from Nix: how other package managers can do better”. Nix has a nice handling of hermetic builds, and Evan showed a mockup of how a similar thing could be done in DNF. Plus there are folks at Conflux working on similar things, so maybe we can look forward to a similar thing in the RPM world soon. Otherwise, the NIX packaging language looked very complicated. SPEC file is a trivial thing compared to that. In follow up discussion Daniel Riek mentioned that he has a quick'n'dirty tool that rebuild packages for specified AWS flavour - this gives him 10-20% performance boost. He promised me that he will clean up this script and next year he will present it at DevConf. :)
Then I looked at “Do you like your changelogs”. In the talk, Laura and Frantisek gave a live demonstration of how developers think about and work with changelogs, and combined it with data Laura had collected in her master’s thesis. The talk was entertaining and informative. We will definitely use the data from it in the Packit team, as automating Changelogs is something we want to pursue in the near future.
In the followup discussion about Towncrier I talked with Sviatoslav Sydorenko who pointed me to Chronographer GitHub CI application that checks if commit has towncrier entry. And to sphinxcontrib-towncrier that makes nice documentation [example]. We use towncrier in Mock, this will be useful.
I only hopped on the RPM developers meetup for a bit because I had to move and get ready for my own talk. It was called “Indiana Jones and obsolete projects”. The room was fully packed. In addition to the actual content, I had prepared a theatrical insert: I was dressed as Indiana, and I had a bullwhip. Tomas helped me with the dramatic entrance and a moment later he purposely gave me an excuse to try out the bullwhip. At the end we paraphrased the famous scene where Indiana pulls a gun against the sword. Content-wise, I went through several innovations: cell phones, smartphones, SSDs, CDs… and showed how one technology replaced another despite people initially rejecting the new technology. Similarly, I then thought about the life cycle of software projects. The biggest compliment for me was Martin’s comment, “This was the first lecture where I didn’t fall asleep today.” You can check the recording and the slides.
I stayed in the same room to listen to “Let the user speak”, where representatives of different projects shared their experiences with Packit, Testing Farm and tmt. Which for me as a member of the Packit team was pleasantly refreshing.
But after this session was over, I found myself terribly tired. So I staggered home and hit the bed.