For me the fact that highway=bus_stop is only allowed on nodes is a plus,
not a minus.

The reason why I would like us all to use nodes for representing the stops
is that a route relation with  a single node for each stop and then a
continuous linked string of ways is conceptually about as simple as it can

I would prefer no to use stop_area relations in the route relations as it
adds a layer of indirection, which I feel is unneeded.

Oddly I wouldn't mind that for the ways though. For those it would indeed
be beneficial to group them in sub route relations. We need support in
editors though. I created some tickets to improve JOSM in that direction
(show continuity line if first way in sub relation connects to last way in
route / same for last way/next way in route). That would have to be part of
another proposal though.

I'd like to simplify first and then propose these sub relations for the
ways. Drawing the continuity line is one thing, visualising the whole
itinerary is another that become harder to do.

Anyway, you are saying that you think it's alright to duplicate details
between stop_position node/platform node, platform way and stop_area
relation and that is exactly what I think is the main problem with the
current way of mapping. The details should go onto 1 single object that
represents the stop and that object should be the (only) one added to the
route relations. stop_area relations can be useful to show which objects
belong together, but only if there is one for each stop. If they just group
everything for several stops that are near to one another, they can just as
well be left out.


Op do 26 jul. 2018 om 09:09 schreef DC Viennablog <>:

> Hello,
> I have been following the discussion about p_t:v2 and would like to also
> throw in my point of view.
> In my opinion, a few things need to change, but some should stay as they
> are now.
>    1. I would love to have the highway=bus_stop tag depricated. The whole
>    confusion I think comes from the highway=* and railway=* tags. Those should
>    mostly be dropped in regards to stop positions.
>    2. The public_transport tags should become the new main attraction. I
>    would leave the stop_position tag. That would go as follows:
> *On a node on the road / tracks*:
> public_transport=stop_position
> bus=yes / trolleybus=yes / tram=yes / train=yes
> name=*
> optional tags: network, local_ref, alt_name, uic_name, note, description...
> *On a node, way or polygon/multipoligon on the side of the road/track *(*as
> a platform*):
> public_transport=platform
> maybe optional: bus=yes / trolleybus=yes / tram=yes / train=yes
> optional tags: name, network, operator (as in the company that actually
> maintaines the station, which can differ from the line operators), note,
> description ...
> Those two elememts could be put in a* stop_area *relation, and multiple
> stop_area relations that belong together in a * stop_area_group *relation,
> that can be tagged with:
> name=*
> optional tags: network, operator (as explained above), alt_name, uic_name,
> note, description...
> Then, the tags like railway=halt and railway=station, which in my opinion
> are to specific for osm could be removed, making mapping the stops of
> raillines much easier. (In german, that is the endless discussion whether a
> trainstop is a station (Bahnhof) or halt (Haltestelle). That is quite
> unimportant in most cases. If need be, we could map the infrastructure
> (landuse or building) as a public_transport=station or
> public_transport=halt instead)).
> In terms of the route relation, we then could add the stop_area relations
> to the route relation with a *stop *role.
> 3. One problem adressed some time earlier was the great amount of
> redundance that the route relations add to the database. There was a, not
> even that bad, though somewhat inpractible sugesstion, to sum up some roads
> in route part relations, and always when those routes would need to be
> added to a relation, we add the relation of routes instead. I had a look
> around vienna and found a few instances, where that would work great, and
> some, where that would potentially yield the same amount or more entries
> than in the first place. But that sugestion could also be a possibility at
> also allowing that, in addition to single roads in a route relation.
> I hope my description of my idea is understandable.
> Kind Regards
> Robin D (emergency99)
> PS: I mapped all bus-lines in Vienna the way that I think the p_t:v2
> scheme is supposed to work (keeping to the wiki as much as it doesn't
> contradict itself). The problem with the highway=bus_stop tag is, that it
> is only allowed on nodes. So as soon as the platform becomes
> "micro-mapped", the bus_stop has to move from the platform to the
> stop_position, which to me, as a new mapper in 2014, screamed: "That node
> is the most important part of the whole thing". Which obvously it isn't but
> tell that to an over-motivated new mapper... So if we could stream-line
> that scheme more, and finally sunset the bus_stop tag and some of the
> railway-tags, then the wiki page could be written much simpler, allowing
> for the newest mappers to understand it. And then, we could give them
> Vienna, Austria as an example 😉.
talk mailing list

Reply via email to