[Tagging] PTv2 public_transport=stop_position for stop positions that vary based on train length

Jarek PiĆ³rkowski jarek at piorkowski.ca
Sat Aug 8 01:16:53 UTC 2020

On Fri, 7 Aug 2020 at 20:54, Andy Townsend <ajt1047 at gmail.com> wrote:
> https://www.openstreetmap.org/node/12004813 is a
> "public_transport=stop_position" for a local station and is part of
> https://www.openstreetmap.org/relation/6396491 among other relations.
> The problem is that train lengths vary, and there are a number of stop
> positions, each of which are actually signed on the platform for the
> benefit of the drivers.  From memory I think that there's at least a
> 2-car stop, a 4 car stop and 6/8 and 10/12 car stops.  The problem is
> that the current node doesn't correspond to any of them.
> As I asked on the changeset that added the one above
> https://www.openstreetmap.org/changeset/40603523 , how should these be
> mapped and how should the PTv2 relations be set up for the different
> services that use them, given that different train services will use
> different stop locations here depending on train length?  Should each
> stop position be mapped and should there therefore be different copies
> of each relation for all the possible train lengths?  Should a "pretend"
> average stop position be added which is actually never correct but will
> at least look nice to data consumers that use PTv2 data?  Given that we
> don't know the actual stop position perhaps the railway=station object
> (in this case https://www.openstreetmap.org/node/7154300845 ) should be
> used instead?

Neither of these seem like great options. Personally: if the stopping
positions for trains of different lengths have actually been surveyed,
I don't have a problem with adding several stop_position nodes with a
note=* or description=* or even an invented
stop_position:train:cars=10;12. Don't put any other tags on them and
they won't render, and it's hardly the most cluttersome of our various

There is however a problem if the Midland Main Line trains come in
different lengths. I suppose you could create a separate route
relation for each set service (from, to, calling at, and train length)
and add the stop_positions to the appropriate relation, but that's
pretty extreme relationing.

(As a bit of background, I believe this is a German influence in the
original PTv2 proposal: a given train ref (e.g. "ICE 13") refers to a
single train that will have the same amount of carriages
(extraordinary circumstances excepted). That's more specific than a
"Midland Main Line: Sheffield => London St Pancras" relation.)

> Maybe the public_transport=stop position should be omitted entirely?
> This last option seems extreme, but one reading of
> https://wiki.openstreetmap.org/wiki/Tag:public_transport%3Dstop_position
> where it says "However, marking the stop position adds no information
> whatsoever when it is placed on the road at the point closest to
> highway=bus_stop or on the tram tracks closest to railway=tram_stop. In
> that case it can be abandoned. " might actually support that (and that
> seems to be the view of one side of the argument in the USA).

Note that this sentence was added in various edits in 2020, well after
the initial proposal was approved. "adds no information" might well be
true (given enough computing power), but "it can be abandoned" is
someone's opinion.

I'm personally fine with not having stop_position unless the actual
stopping position is ambiguous due to some very unusual geometry, but
to be clear, this is against the initially approved proposal.

> Maybe the "correct" answer is none of the above?  With a "local mapper"
> hat on I've managed to avoid PTv2 since it basically isn't relevant
> anywhere I normally map things, largely because I don't tend to do that
> near any actual public transport infrastructure, but with a DWG hat on I
> haven't been able to avoid the question, hence me asking here.

I hope the DWG hat is fireproof :) ptv2 is a topic that arouses passions.


More information about the Tagging mailing list