Stefan Lindbohm

”Next stop is Röjan. We’re ahead of schedule so the train driver will show you where to pick wild strawberries.”

@jon@gruene.social ”You just see this train”, he said not knowing exactly who he was speaking to 😅

@jon@gruene.social I am honored (well, and offended) by their description of us as big tech and the risk that Railfinder might soon take over all European ticket sales any second now :)

@jon Longest rail route in EU - the game CHOOSE YOUR CHALLENGE [ Fewest route km no matter the travel time ] [ Fastest including disruptions ] [ Vacation mode - prioritise good sleep & nice stops along the way ] Only real life experiences count. When do we organise the event? Any other challenge variations to add? I’m personally getting more committed as this discussion goes on. @Jonas_Bostrom

@jon@gruene.social @HaTetsu@mastodon.com.pl I’d guess that, yes, Norwegian operators are the ones providing realtime data to Entur. Static schedules are probably coming from BaneNOR but not realtime. So Entur is just showing the static data? Meanwhile SJ simply has no realtime system of their own. Their app shows data from Trafikverket straight up. Hence no realtime data gets sent outside of Sweden and no data is shown for outside of Sweden in their app.

@larstransportworld@mastodon.social I had fun with these timetables before 😅. No country has so many coach groups/connecting trains as Denmark :)

@bovine3dom@masto.ai Have you heard about our new business model: make -2% margins net on sales, pay hundreds of Euros in compensation when things go wrong. Forbes should have me on their cover. @jon@gruene.social

@jon@gruene.social One reason could be that they have training to validate certain ID’s and those are the ones they accept, or? Like it’s not about exactly about the country, but just they need a list of approved ID types for practical reasons. Can’t exactly be the first time a German driver uses these ferries of course, so even then it seems a bit stupid. But maybe not nefarious?

@jon@gruene.social @stadsplanering.bsky.social Night train 91 and even X2000 Stockholm - Malmö are among trains that typically don’t get big delays though. On this itinerary and currently, there’s a lot higher risk of delays or even cancellations for the Talgo’d ECE connections through Denmark. (Traveling daytime around Sundsvall or night train Stockholm - Hamburg would be a whole different story.)

@cycling_on_rails@mastodon.social This is how it works, but I’m not sure I’d call it a loophole. Well, maybe the fact that many operators seem very eager to get you to click that cancel button. Rule 1 when I help people in these situations is to keep the original contract of carriage, as that’s the key to all reimbursements etc for actually making it to the destination. Only cancel if you decide not to travel at all. @partim@social.tchncs.de

@cycling_on_rails@mastodon.social In this case I’d say passenger rights are already strong enough. You have a right of alternative transport (any bus/train) and hotel + food where necessary. You have to give the operator a chance to organise it for you, but if they fail to do so within 100 minutes you can expense it on your own and claim the money back. Yes you need to expense it, and yes maybe operators should have to communicate it more clearly, but otherwise you have pretty good protections.

@patrick@mendeddrum.org You don’t even need to have a reduction card for any singular ticket solution to hit pricing problems. The base offer will by definition have to be the least price competitive of all operators involved, i.e. prices will be the equivalent of a full flex ticket, never comparable to the various non-refundable/low-flex options you can get from each operator. See e.g. classic international tickets (”CIV”/”tourist” tickets) or the national tariff system in Czechia OneTicket @jharbers@mastodon.online @erpu@eupolicy.social

@jon If there’s one thing I’ve learned working with this it’s that you never run out of reasons to manually override some supposedly standardised data :D

@partim@social.tchncs.de Oh, nice! I think efficiency is much needed for the 22 GB datasets, as you say. Imagine importing 30-40 countries once we get open data everywhere. So probably worth it to accept some complicated stuff for the sake of performance. Did you see my writeup on NeTEx the other day btw? Sounds like we both did the same thing of just jumping in and realising how bad things are 😅 https://gist.github.com/stefanlindbohm/58d11afbb9af2d5ec88263f62322caaa

@partim@social.tchncs.de Would you like to join a soon to launch open source project that aims to fix this? We’re a few orgs who are coming together to write tooling to unify this mess so we can use NeTEx data in our respective apps.

@jon DB does a lot of manual entry/overrides on the data from other countries that are not published to third-parties by the other carriers. Check-in times for Eurostar missing is another example (and part of the reason we don’t yet sell Eurostar despite having ticketing access). We’re basically redoing the work DB has done in their database on a case by case basis.

@jon Absolutely. I do want to give Norrtåg credit for generally doing a lot more than most other Swedish operators for connections though, despite this case where arguably they should stop 10 min to allow connections both to and from the night train (they come/go different lines). Their train crew has hunted down SJ operations center and cancelled intermediate stops to save time just to make connections for me in the past. Stark contrast to my experience with other operators. @Jonas_Bostrom

@bovine3dom@masto.ai You got us! Just wait until we can make -2% margins on a MUCH larger scale. If this regulation passes, it could even happen overnight, some believe. @jon@gruene.social

@patrick@mendeddrum.org (Reason being that the lower prices are motivated by various restrictions, but restrictions are different for different operators, so can’t be mixed in a single ticket which needs to have one set of conditions.) @jharbers@mastodon.online @erpu@eupolicy.social

@bovine3dom@masto.ai This is extra fun as a reseller that connects to online booking API’s, because when we send a request for a connection like these, the API’s respond something like: ¯\_(ツ)_/¯ (Of course their own systems don’t issue tickets that break their internal connecting rules.) @jon@gruene.social @sintzoff@mast.eu.org @R1Rail@pouet.chapril.org

@jsl@hachyderm.io The data will be enforcably open in 2029 when the latest timetable data regulation enters in full force. Until then it’s already possible by purchasing the industry data source MERITS. We do this at Railfinder and have the same data as Rail Planner (and more or less same as DB Navigator minus local transit). @jon@gruene.social

@partim@social.tchncs.de @cycling_on_rails@mastodon.social I think this is just due to how the legal text is structured. According to the Swedish Consumer Agency’s interpretation you simply have a right to food and hotels, and ”if you had to expense it you have a right to reimbursment”. There’s an English translation but the Swedish text is clearer: https://www.konsumentverket.se/varor-och-tjanster-process/forsenat-eller-installt-tag/

@jon The two Atocha stations don’t even have an official transfer connection between them 😵‍💫. I should probably ask what is intended there because we don’t currently support a transfer between them at all.

@partim@social.tchncs.de I think NeTEx is a massive net of leaky abstractions. The different frames have no meaning that I could identify for users of the data, but simply just originate from having been created in different working groups as different XSDs. The only things relevant for data users are the actual objects within each frame (each object type can anyway only exist within a certain frame type). Validity cascades down from the frame, so if that is just flattened to the object you’re fine.

@partim@social.tchncs.de If we’re a few people who can spend some time, we should be able to create something that works well. My choice would also to build this in Rust (although it’s new to me). I was thinking something like a hybrid SAX parser that creates a stream of individually self-contained objects. Maybe there are better ideas. I’ll check what process we should have for new people to join at this time, I’ll get back with a DM!

@jon@gruene.social I think the main thing we need to figure out a solid solution for is: how do we define the planned basis to use in lieu of a collection of tickets in regards to passenger rights? For example, if you have a 100% reduction card of some sort, what arrival time is used as basis for passenger rights? Is it always the scheduled timetable as it was when the train departed from its depot that day - or would it somehow be possible to prove you planned based on an earlier timetable.

@jon@gruene.social One way would maybe be to allow using a printed/saved journey itinerary from a journey planner that respects timetable rules/MCT’s? Then there would be a dependence on journey planners existing for your journey, but that is a lot lower of a barrier. Are there other solid starting points to consider?

@partim@social.tchncs.de It’s still about half the number of accommodations per meter of train, comparing only the night coaches. Seat coaches are probably fine, although I think it sounds very wrong to have to bring a full set of 7 coaches through the night just to have that option for the few who want the lowest price possible. I’m admittedly not at all an expert of train operations, but this whole thing just sounds totally counter to all discussions I’ve heard of actual issues with our current night trains.

@jon@gruene.social If anything, it is very much in line with OSDM implementations. DB and others are already offering the ”single ticket” exactly as proposed.

@jon@gruene.social I guess there’s some logic there - passenger rights do have exemptions for factors outside of the operator’s control. Same with law enforcement (you can make a guess how I experienced this one, it involves Germany). @quixoticgeek@social.v.st

@HaTetsu It was Sundsvall västra (west station) that was cancelled to make it to Sundsvall C (final stop). Not many people would get off there but even if it wouldn’t have been a big deal to get back :) @jon @Jonas_Bostrom

@jon Looking at historic times for these trains, it looks like the connection being intended in the other direction is a possibility. 94 has had 05:42 before too, but 7182 had/has most days 05:35. Luleå > Kiruna sounds like a connection someone might want to make possible. Except of course, 94 is now cut in Boden so that’s gone for the current timetable year. (Edited: messed up the direction of the Norrtåg first 😅) @Jonas_Bostrom

@partim@social.tchncs.de I think you overestimate Gävle/Uppsala and underestimate the alternatives on the route. There are frequent options with regional Mälartåg all day. The bigger reason for seats on these trains is between Boden and Kiruna/Narvik, where services are few and a day train adds a lot, relatively. Because of the long distance it’s also more than just a short stretch in the evening/morning. But these currently often run with two classic seat cars. @sebwilken@mastodon.social @cycling_on_rails@mastodon.social

@partim@social.tchncs.de Nitpick, but regular services with high speed trains north of Stockholm are X55 not X2. For some reason all SE high speed trains are incorrectly marked as X2 in DB/Hafas. @sebwilken@mastodon.social @cycling_on_rails@mastodon.social

@partim@social.tchncs.de There are 24 recliners in the corridors of the solo compartment coaches, that’s it. Coach map here: https://www.trafikverket.se/contentassets/6947aea2c65b4cfba5e6f4cf8f680588/print_om-taget_bakbelyst_for-print_2026.04.14_lowres.pdf

@pettter@social.accum.se I counted all seats as seats only. Including the recliners that are positioned in the corridor of the single person compartment coaches, hence the available seats in the night coach sets.

@aslakr@mastodon.social Indeed :)

@moof I don’t have anything on the phone, not even a case.

@jon@gruene.social The somehow is what I’m thinking needs a concrete proposal. That would make it a lot more likely that an MEP will get it through, no?

@quixoticgeek@social.v.st @jon@gruene.social Timetables include a connection path between all Paris stations with a given minimum time. So a late train to one station making you miss the departure from another would apply for sure. Question is what happens if you follow the connecting time across Paris but the RER itself gets you delayed - not sure what happens today if you do this on a through-ticket?

@partim@social.tchncs.de Also yes, translating the schema to human is kinda what I attempted to get going with my doc. Would love to add more stuff that we see need explaning, if you have more examples. The validity is maybe first off in that list?

@jon@gruene.social The regulation from February/March whenever it was is already there to mandate all this data to be enforcably open in 2029 at the latest. So that part should be possible to rely on already!

@grlodi@mastodon.social Press goes on headlines, a fancy demo unit in Stockholm C and ribbons being cut by ministers as usual :). But I’d definitely expect SOME opinion piece or something by now pointing this out. It seems totally crazy to me from what I gather listening to people who have experience in night train operations.

@partim@social.tchncs.de Looks nice! We would need something that manages nesting of elements, but I was thinking of just creating a CSS file based on some semantic markup for this and just text editing a simple HTML file for the relevant parts of the schema 😅. Then we could take inspiration for styling from anything we like.

@partim@social.tchncs.de You mean for documenting the various elements/attributes/fields in a nested but humanly readable way? Haven’t seen rustdoc HTML exports, do you have an example? I was thinking of imitating some good API docs out there, think something like Stripe’s API docs or even Swagger. Needs to be adapted (and not as messy as Swagger), but as a point of inspiration.

@partim@social.tchncs.de I think a generic NeTEx parser will be needed regardless too, for use with any of the profiles out there. Something that simply provides a stream of objects within any file you point it at. (And maybe with a feature to index the byte offsets of any tags with an ID to support seeking/random access.) Then profiles can later be handled through classes that consume this stream in whatever way necessary.