[Talk-transit] Ideas for a simplified public transportation scheme

Jarek PiĆ³rkowski jarek at piorkowski.ca
Tue May 7 19:14:20 UTC 2019

Hi all,

I wrote some point-form notes of the discussion so far for people to
refer or respond to. I asked questions in 7a, 7c, 8c, 10c, 11c, 13b,

1. Majority of world's public transit is buses
2. Majority of world's bus stops are simple signs (or sometimes no
signs at all) and will never [1] have a designated "platform"
3. Many editors feel that highway=bus_stop is sufficient tagging for these
4. We should also support other transit modes which may more often
have more details
5. Proposals that suggest to change existing widely-used tags such as
highway=bus_stop for semantic reasons are unlikely to succeed
6. I am unaware of any successful proposal which successfully rolled
back a previously accepted proposal. To those disliking PTv2 I suggest
either ignoring it, or adapting it into a PTv3 keeping in mind the
issues it tried to address, some of which are below.

7. For public transit routing, it appears that having highway_bus_stop
nodes ("locations where people wait for buses") arranged in order in a
relation is sufficient, per the comments about OsmAnd in Stockholm
7a. Correct me if I'm wrong but I haven't read a lot of opposition to
this "Stockholm-like" lightweight route/route_master relation scheme
7b. This bus stop node does not have to be placed on the way that
buses travel on
7c. From what I'm understand, this bus stop node does not have to be
connected to a pedestrian highway either, with routers presumably
jumping from the nearest highway?

8. A stop_location (to use ptv2 terminology) on the way that vehicles
travel on could help with things like calculating and showing the
likely route the bus will take, but this can also be calculated
without the stop_location node by projection of other stop objects
onto the way
8a. stop_location is specified as optional in current wiki description
of PTv2 [2]
8b. Stephen initially said "we need stop positions so routers can get
people from stop to stop on the buses/trains"; a subsequent message at
06 May 2019 13:53:10 -0500 (if I understood correctly) said that
stop_position can be optional in cases of simple geometry (which is
presumably vast majority of them)
8c. What changes would people like to make? Clearer guidance as to
when stop_position and stop_area might be needed for buses - in cases
like ambiguous geometry due to multiple parallel highways? Or would
people prefer to ignore routing of buses between stops and advocate
for removal of stop_positions in all cases?

9. There are some cases that do not cleanly fit into hw=bus_stop
"PTv1" tagging, for example a sign-only stop served by both buses and
trams, or a waiting platform served by both buses and trams
9a. Because we must retain hw=bus_stop per #3 and #5, any
accommodation of these cases must either be initially of tags, or
guidance on how to place highway=bus_stop tags

10. Meaning of public_transit=platform tag is dependent on context, it
unifies/duplicates some existing tags, arguably it sometimes describes
imaginary things, and it is disliked by many editors
10a. Some of the dislike is due to disagreements as to what "platform" means
10b. Please remember that OSM's British English naming conventions are
in a foreign language or dialect to vast majority of editors
10c. public_transit=platform does not appear to be needed for walk+PT
routing. It could be replaced with bus_stop, tram_stop,
railway=platform, highway=platform as appropriate. Please correct if
I'm wrong.

11. stop_area is disliked by many editors as being pointless in
majority of cases involving surface transit
11a. stop_area relation was recommended by the OsmAnd blog post
introducing their public transit support, but evidently OsmAnd works
without it in Stockholm
11b. stop_area is currently mentioned but not recommended in the wiki [2]
11c. Would anyone like to see changes here? Should we explicitly
*discourage* stop_area except where needed, and what would be the
criteria for needing it? [3]

12. Many of the currently mapped tram systems have a railway=tram_stop
+ public_transport=stop_position node on the rail, so we should
probably not change this scheme either without good reason

13. There is currently no clear way for tagging stops that also have
physical platforms, except for PTv2
13a. This exists as physical feature in real world and should be
supported, in a manner compatible with platform-less stops
13b. Should we add bus_stop/tram_stop on one of the nodes of the
platform way [4]? Next to the platform? As pointed out by Markus, we
can't do what might be the most intuitive method of the platform
way/area sharing bus_stop tag because the platform is also a highway=

14. Discussion hasn't really touched much on railways or subways.
Should we take this to mean that people are generally happy with PTv2
tagging on those? If we change the guidance for surface transport,
PTv2 for subways doesn't necessarily have to change.

[1] Or at least not in the timeframe we should concern ourselves with.
As mentioned by RobinD/DC Viennablog, stops with designated platforms
are more accessible (easier to level-board or deploy a bus ramp) and
might feel safer to riders, but that doesn't mean most stops will be
upgraded anytime soon.
[2] I mean https://wiki.openstreetmap.org/wiki/Public_transport
[3] e.g. in Toronto there are some stops across the street from each
other with different names. They are logically the same stop (stop
group? stop area?), for vehicles travelling in different directions.
Of course that could also be computed from spatial positioning and
route relations.
[4] as in https://osm.org/way/395511322 (and yes that really is a bus
bulb platform)

More information about the Talk-transit mailing list