DebConf24 – Busan, South Korea

I’m finishing typing up this blog entry hours before my last 13 hour leg back home, after I spent 2 weeks in Busan, South Korea for DebCamp24 and DebCamp24. I had a rough year and decided to take it easy this DebConf. So this is the first DebConf in a long time where I didn’t give any talks. I mostly caught up on a bit of packaging, worked on DebConf video stuff, attended a few BoFs and talked to people. Overall it was a very good DebConf, which also turned out to be more productive than I expeced it would.

In the welcome session on the first day of DebConf, Nicolas Dandrimont mentioned that a benefit of DebConf is that it provides a sort of caffeine for your Debian motivation. I could certainly feel that affect swell as the days went past, and it’s nice to be excited about some ideas again that would otherwise be fading.

Recovering DPL

It’s a bit of a gear shift being DPL for 4 years, and DebConf Committee for nearly 5 years before that, and then being at DebConf while some issue arise (as it always does during a conference). At first I jump into high alert mode, but then I have to remind myself “it’s not your problem anymore” and let others deal with it.

It was nice spending a little in-person time with Andreas Tille, our new DPL, we did some more handover and discussed some current issues. I still have a few dozen emails in my DPL inbox that I need to collate and forward to Andreas, I hope to finish all that up by the end of August.

During the Bits from the DPL talk, the usual question came up whether Andreas will consider running for DPL again, to which he just responded in a slide “Maybe”. I think it’s a good idea for a DPL to do at least two terms if it all works out for everyone, since it takes a while to get up to speed on everything.

Also, having been DPL for four years, I have a lot to say about it, and I think there’s a lot we can fix in the role, or at least discuss it. If I had the bandwidth for it I would have scheduled a BoF for it, but I’ll very likely do that for the next DebConf instead!

Video team

I set up the standby loop for the video streaming setup. We call it loopy, it’s a bunch of OBS scenes that provide announcements, shows sponsors, the schedule and some social content. I wrote about it back in 2020, but it’s evolved quite a bit since then, so I’m probably due to write another blog post with a bunch of updates on it. I hope to organise a video team sprint in Cape Town in the first half of next year, so I’ll summarize everything before then.

It would’ve been great if we could have some displays in social areas that could show talks, the loop and other content, but we were just too pressed for time for that. This year’s DebConf had a very compressed timeline, and there was just too much that had to be done and that had to be figured out on the last minute. This put quite a lot of strain on the organisers, but I was glad to see how, for the most part, most attendees were very sympathetic to some rough edges (but I digress…).

I added more of the OBS machine setup to the videoteam’s ansible repository, so as of now it just needs an ansible setup and the OBS data and it’s good to go. The loopy data is already in the videoteam git repository, so I could probably just add a git pull and create some symlinks in ansible and then that machine can be installed from 0% to 100% by just installing via debian-installer with our ansible hooks.

This DebConf I volunteered quite a bit for actual video roles during the conference, something I didn’t have much time for in recent DebConfs, and it’s been fun, especially in a session or two where nearly none of the other volunteers showed up. Sometimes chaos is just fun :-)

Baekyongee is the university mascot, who’s visible throughout the university. So of course we included this four legged whale creature on the loop too!

Packaging

I was hoping to do more packaging during DebCamp, but at least it was a non-zero amount:

• Uploaded gdisk 1.0.10-2 to unstable (previously tested effects of adding dh-sequence-movetousr) (Closes: #1073679).
• Worked a bit on bcachefs-tools (updating git to 1.9.4), but has a build failure that I need to look into (we might need a newer bindgen) – update: I’m probably going to ROM this package soon, it doesn’t seem suitable for packaging in Debian.
• Calamares: Tested a fix for encrypted installs, and uploaded it.
• Calamares: Uploaded (3.3.8-1) to backports (at the time of writing it’s still in backports-NEW).
• Backport obs-gradient-source for bookworm.
• Did some initial packaging on Cambalache, I’ll upload to unstable once wlroots (0.18) hits unstable.
• Pixelorama 1.0 – I did some initial packaging for Pixelorama back when we did the MiniDebConf Gaming Edition, but it had a few stoppers back then. Version 1.0 seems to fix all of that, but it depends on Godot 4.2 and we’re still on the 3 series in Debian, so I’ll upload this once Godot 4.2 hits at least experimental. Godot software/games is otherwise quite easy to run, it’s basically just source code / data that is installed and then run via godot-runner (godot3-runner package in Debian).

BoFs

Python Team BoF

Link to the etherpad / pad archive link and video can be found on the talk page: https://debconf24.debconf.org/talks/31-python-bof/

The session ended up being extended to a second part, since all the issues didn’t fit into the first session.

I was distracted by too many thing during the Python 3.12 transition (to the point where I thought that 3.11 was still new in Debian), so it was very useful listening to the retrospective of that transition.

There was a discussion whether Python 3.13 could still make it to testing in time for freeze, and it seems that there is consensus that it can, although, likely with new experimental features like disabling the global interpreter lock and the just in time compiler disabled.

I learned for the first time about the “dead batteries” project, PEP-0594, which removes ancient modules that have mostly been superseded, from the Python standard library.

There was some talk about the process for changing team policy, and a policy discussion on whether we should require autopkgtests as a SHOULD or a MUST for migration to testing. As with many things, the devil is in the details and in my opinion you could go either way and achieve a similar result (the original MUST proposal allowed exceptions which imho made it the same as the SHOULD proposal).

There’s an idea to do some ongoing remote sprints, like having co-ordinated days for bug squashing / working on stuff together. This is a nice idea and probably a good way to energise the team and also to gain some interest from potential newcomers.

Louis-Philipe Véronneau was added as a new team admin and there was some discussion on various Sphinx issues and which Lintian tags might be needed for Python 3.13. If you want to know more, you probably have to watch the videos / read the notes :)

Debian.net BoF

Link to the etherpad / pad archive link can be found on the talk page: https://debconf24.debconf.org/talks/37-debiannet-team-bof

Debian Developers can set up services on subdomains on debian.net, but a big problem we’ve had before was that developers were on their own for hosting those services. This meant that they either hosted it on their DSL/fiber connection at home, paid for the hosting themselves, or hosted it at different services which became an accounting nightmare to claim back the used funds. So, a few of us started the debian.net hosting project (sometimes we just call it debian.net, this is probably a bit of a bug) so that Debian has accounts with cloud providers, and as admins we can create instances there that gets billed directly to Debian.

We had an initial rush of services, but requests have slowed down since (not really a bad thing, we don’t want lots of spurious requests). Last year we did a census, to check which of the instances were still used, whether they received system updates and to ask whether they are performing backups. It went well and some issues were found along the way, so we’ll be doing that again.

We also gained two potential volunteers to help run things, which is great.

Debian Social BoF

Link to the etherpad / pad archive link can be found on the talk page: https://debconf24.debconf.org/talks/34-debiansocial-bof

We discussed the services we run, you can view the current state of things at: https://wiki.debian.org/Teams/DebianSocial

Pleroma has shown some cracks over the last year or so, and there are some forks that seem promising. At the same time, it might be worth while considering Mastodon too. So we’ll do some comparison of features and maintenance and find a way forward. At the time when Pleroma was installed, it was way ahead in terms of moderation features.

Pixelfed is doing well and chugging along nicely, we should probably promote it more.

Peertube is working well, although we learned that we still don’t have all the recent DebConf videos on there. A bunch of other issues should be fixed once we move it to a new machine that we plan to set up.

We’re removing writefreely and plume. Nice concepts, but it didn’t get much traction yet, and no one who signed up for these actually used it, which is fine, some experimentation with services is good and sometimes they prove to be very popular and other times not.

The WordPress multisite instance has some mild use, otherwise haven’t had any issues.

Matrix ended up to be much, much bigger than we thought, both in usage and in its requirements. It’s very stateful and remembers discussions for as long as you let it, so it’s Postgres database is continuously expanding, this will also be a lot easier to manage once we have this on the new host.

Jitsi is also quite popular, but it could probably be on jitsi.debian.net instead (we created this on debian.social during the initial height of COVID-19 where we didn’t have the debian.net hosting yet), although in practice it doesn’t really matter where it lives.

Most of our current challenges will be solved by moving everything to a new big machine that has a few public IPs available for some VMs, so we’ll be doing that shortly.

Debian Foundation Discussion BoF

This was some brainstorming about the future structure of Debian, and what steps might be needed to get there. It’s way too big a problem to take on in a BoF, but we made some progress in figuring out some smaller pieces of the larger puzzle. The DPL is going to get in touch with some legal advisors and our trusted organisations so that we can aim to formalise our relationships a bit more by the time it’s DebConf again.

I also introduced my intention to join the Debian Partners delegation. When I was DPL, I enjoyed talking with external organisations who wanted to help Debian, but helping external organisations help Debian turned out to be too much additional load on the usual DPL roles, so I’m pursuing this with the Debian Partners team, more on that some other time.

This session wasn’t recorded, but if you feel like you missed something, don’t worry, all intentions will be communicated and discussed with project members before anything moves forward. There was a strong agreement in the room though that we should push forward on this, and not reach another DebConf where we didn’t make progress on formalising Debian’s structure more.

Social

Conference Dinner

The conference dinner took place in the university gymnasium. I hope not many people do sports there in the summer, because it got HOT. There was also some interesting observations on the thermodynamics of the attempted cooling solutions, which was amusing. On the plus side, the food was great, the company was good, and the speeches were kept to a minimum, so it was a great conference dinner, even though it was probably cut a bit short due to the heat.

Cheese and Wine

Cheese and Wine happened on 1 August, which happens to be the date I became a DD at DebConf17 in Montréal seven years before, so this was a nice accidental celebration of my Debiversary :)

Since I’m running out of time, I’ll add some more photos to this post some time after publishing it :P

Group Photo

As per DebConf tradition, Aigars took the group photo. You can find the high resolution version on Debian’s GitLab instance.

Talking

Ah yes, talking to people is a big part of DebConf, but I didn’t keep track of it very well.

• I mostly listened to Alper a bit about his ideas for his talk about debian installer.
• I talked to Rhonda a bit about ActivityPub and MQTT and whether they could be useful for publicising Debian activity.
• Listened to Gunnar and Julian have a discussion about GPG and APT which was interesting.
• I learned that you can learn Hangul, the Korean alphabet, in about an hour or so (I wish I knew that in all my years of playing StarCraft II).
• We had the usual continuous keysigning party. Besides it’s intended function, this is always a good ice breaker and a way to for shy people to meet other shy people.
• … and many other fly-by discussions.

Stuff that didn’t happen this DebConf

• loo.py – A simple Python script that could eventually replace the obs-advanced-scene-switcher sequencer in OBS. It would also be extremely useful if we’d ever replace OBS for loopy. I was hoping to have some time to hack on this, and try to recreate the current loopy in loo.py, but didn’t have the time.
• toetally – This year videoteam had to scramble to get a bunch of resistors to assemble some tally light. Even when assembled, they were a bit troublesome. It would’ve been nice to hack on toetally and get something ready for testing, but it mostly relies on having something like a rasbperry pi zero with an attached screen in order to work on further. I’ll try to have something ready for the next mini conf though.
• extrepo on debian live – I think we should have extrepo installed by default on desktop systems, I meant to start a discussion on this, but perhaps it’s just time I go ahead and do it and announce it.
• Live stream to peertube server – It would’ve been nice to live stream DebConf to PeerTube, but the dependency tree to get this going got a bit too huge. Following our plans discussed in the Debian Social BoF, we should have this safely ready before the next MiniDebConf and should be able to test it there.
• Desktop Egg – there was this idea to get a stand-in theme for Debian testing/unstable until the artwork for the next release is finalized (Debian bug: #1038660), I have an idea that I meant to implement months ago, but too many things got in the way. It’s based on Juliette Taka’s Homeworld theme, and basically transforms the homeworld into an egg. Get it? Something that hasn’t hatched yet? I also only recently noticed that we never used the actual homeworld graphics (featuring the world image) in the final bullseye release. lol.

So, another DebConf and another new plush animal. Last but not least, thanks to PKNU for being such a generous and fantastic host to us! See you again at DebConf25 in Brest, France next year!