[Tagging] Feature Proposal - RFC - Public Transport v3

[John Doe]

Hey, thanks for sharing your views 🙂

07-Mar-2020 03:01:41 Phake Nick :

> [...] for such renderer to work, access restriction for public transit vehicle need to be complete, which is rather difficult not just because of the work it take or the current incompleteness of the amount of keys in OSM that can be used to represent multiple localized types of transportation, but also because of things like mechanical restrictions of bus models that would not be available in any public document and cannot be verified outside of the public transit company.

I feel that this information can largely be approximated from road classifications. PT Assistant already complains if a bus route is using a road lower than a certain classification. As and when turn restriction data is added, routers can adapt the route automatically.

> [...] how can editors know when it is necessary to add a waypoint to show a route correctly?

I think it will always need mapper testing, as it already does. But PTv3 does remove a lot of moving parts, and makes creation and maintenance of routes easier.

> If a new road is being built next to existing road which doesn't affect existing public transit routes, how to preemptively make sure routing engines won't automatically redirected a route to the new road?

In this case, since the platforms being served haven't changed, I think you'll find that
1. The likelihood of the route changing is very low. Even hail and ride routes shouldn't be likely to change much, since the via points used to define them would probably keep the routers on track.
2. Even if it does change, the platforms are the same and the router is still preferring the fastest path between them - users are unlikely to be greatly affected. At worst, accuracy of ETAs could be affected.

What I have in mind is the case of Delhi's NH9, in which a road was changed from two to four carriageways. In such a situation, with the constraint of the existing stops, routers would have to ignore the new inner carriageways and stick to the outer carriageways, which is exactly what happened on the ground 😄 If you have some other examples in mind, it would help us all to discuss their specifics.

> Third, way points are more transient than ways. It's more easy for a node to be deleted or replaced when editing geometry than ways. How to maintain integrity of a route geometry when others edit road geometry?

To verify this, I tried deleting a stop position which was part of a route relation. Both JOSM and Vespucci warn you about the relation, and ask for confirmation. iD does not, but fortunately someone's already made an issue about it and they're considering it - https://github.com/openstreetmap/iD/issues/5846

I've added FAQ entries for all these, thanks for the feedback! I hope I'm able to address your concerns about our proposal.