39C3

A few days ago I attended the 39th Chaos Communication Congress (39C3) in Hamburg, Germany, as part of the KDE presence there.

KDE

Like at 38C3 in the previous year we had a small KDE assembly as part of the Bits & Bäume Habitat.

There were two talks with KDE contributors in the main program:

• Bettina, Caro and Joseph from KDE Eco presented their work on the “End of 10” campaign.
• Shinjo spoke about security breaches at South Korean telco operators.

The End of 10 campaign also had organized a meetup and a install party.

As Jonah has already reported we got very positive feedback from attendees about KDE’s work.

Itinerary

Itinerary was a frequent discussion topic for me, both with the KDE team and attendees in general.

• The MapLibre-based vector tile map integration got pushed over the finishing line (see also Jonah’s report).
• Reaching the end of the year and people being interested in their travel wrap-up, a few issues regarding handling of cancelled reservations in the statistics got fixed. And as people’s trip lists are growing longer and longer that also got a better grouping.
• We talked about possible ways to improve the stop picker, in particular the country selector which is becoming increasingly annoying to use as it gets longer and longer as Transitous coverage keeps expanding.
• Ideas for a better first-start onboarding page.
• We need to investigate whether we handle Zügli’s ticket revocations correcty that were mentioned in the Deutschlandticket fraud investigation talk.

Android

I also had the opportunity to talk to developers of F-Droid and other FOSS Android applications, who share a lot of the pain we are also dealing with in bringing KDE applications to Android.

A big concern especially for people not associated with a bigger umbrella organization is the upcoming requirements by Google for developer verification. While KDE might be less affected by this directly, any negative effect on the larger FOSS ecosystem is of course also bad for us. The continuous close-down of AOSP development is also not helping, making it significantly harder for Google-free Android variants.

None of that is entirely surprising, and it increases the pressure on fully open Linux systems becoming a viable alternative on more mobile devices. Both that as well as collaboration on adjacent infrastructure such as fully open push notification infrastructure has been making good progress in 2025 fortunately.

Transitous

We had a Transitous meetup on the last day at Wikimedia’s assembly, hosted by Jonah, Julius and myself. We should organize and announce this earlier next time, but the space was nevertheless full. Meeting notes are in the wiki.

We had quite a few conversations around Transitous beyond this as well:

• It looks like we should be able to get SIRI-FM data for a number of operators in Germany in Q1 2026. That’s facility monitoring information such as current or planned elevator outages, something highly relevant for wheelchair routing.
• We talked about finding better ways to maintain data augmentation than the currently used MOTIS Lua scripts. While there’s options on where to store such information (e.g. Wikidata or OSM), the main challenge is matching GTFS data lacking stable identifiers to such an external data set.
• We met with the team behind esel.ac, a community-run bike sharing system in Aachen. It’s using OpenBike and produces a GBFS feed that we now also have in Transitous. Community-run mobility services and a community-run routing service are of course the perfect match, and I’d like us to use those cases to showcase what the whole stack of open standards, open hardware and open software can do together. We discussed possible extensions to the GBFS feed such as booking links, geo fencing zones and return constraints for this.
• There was some prototyping towards using Wikidata as a source of “temporary POIs” for Transitous. That’s events that are tied to a specific location (e.g. a conference) which then can be used as a destination for routing. We managed to produce an abomination of a SPARQL query that resulted in a technically valid set of events. However we haven’t found a proper solution yet for reliably excluding events we would not want in there, such as those of fascist parties.
• There’s people working on implementing deck layouts in NeTEx. While motivated by accessibility and seat booking, this could provide us with train coach layouts as a byproduct as well. There’s public transport operators involved, so this is hopefully not just theory but also going into production in the not too distant future.

It also looks like we might have another iteration of the Transitous Hack Weekend in Berlin, next weekend already (January 9-11). That’s very short notice and not entirely finalized yet. If you are interested in joining please get in touch in the Transitous Matrix channel.

OSM

The OSM assembly was conveniently directly next to the KDE one, so I could easily drop into conversations about indoor mapping, indoor routing or indoor positioning there. Interest in all parts of this seems to be increasing, we probably should improve the introduction material for this a bit.

There’s also a plan to have an (offline) meeting in the next months to get some of the pending tagging proposals and open questions e.g. around “thick” walls, stairs and fractional levels sorted out and over the finishing line.

We also had the opportunity to discuss the FOSSGIS e.V. becoming a possible umbrella organization for Transitous and/or the Open Transport Community Conference. Especially the latter is becoming slightly more pressing as we got a few sponsorship offers while looking for a venue, and that’s something we can only make use of with an organization behind us that can actually handle money.

Weather and Emergency Alerts

I also met with FOSSWarn to discuss the next steps on the public alert distribution server:

• Better monitoring to notice system failures, task queue backlogs, source feed outages, increasing parser or push notification submission error rates, or suspiciously large subscription areas. As a byproduct this might also provide interesting insights in the alert data.
• Better ways to deal with rate limits on push servers. This needs to be ultimately resolved at those servers, as public alerts are very prone to produce notification bursts. This is also a blocker for scaling this up further and e.g. deploying this in a default KDE Plasma installation.
• Performance improvements for the alert area vector tile generation. That wasn’t a focus initially as this was meant purely as a diagnostic tool, but it has become popular to the point that FOSSWarn wants to integrate this directly into the app. This will probably require geometry simplification on ingested “inline” CAP geometry, which would benefit everyone by smaller and cheaper to parse/render CAP data.

You can help!

Events like Chaos Communication Congress are enormously useful for bringing together, connecting and enabling collaboration between people from different areas or initiatives. The sheer size and diverse set of attendees help a lot with that.

Attending events however incurs cost for travel, accommodation and entrance. Your donation to organizations like KDE e.V. or FOSSGIS e.V. support such activities.