[Tagging] public_transport=platform rendering on osm-carto

Paul Allen pla16021 at gmail.com
Fri Jun 22 09:01:27 UTC 2018

On Thu, Jun 21, 2018 at 11:34 PM, Martin Koppenhoefer <
dieterdreist at gmail.com> wrote:

> I am with Jo here: the stop_positions at bus stops do not add any
> information, at most they would add convenience for data consumers (if
> every bus stop had them), but OSM traditionally values mapper convenience
> higher than simplicity of data evaluation.

I don't fully comprehend all the implications of PTv2.  I don't fully
comprehend the capabilities of routeing engines as
currently implemented or how they may be implemented in future,.  I'm
considering only the aspects of this that
affect buses, not other PTV2 categories.  So there may be extreme
Dunning-Kruger ahead...

I'm starting to think (probably incorrectly) that the stop position on the
highway is more valuable in a relation than the
platform but the platform is more valuable than the stop position in

As far as routeing goes, the stop position is important.  When I switch
between bus/foot I do NOT (in most
circumstances) walk through the platform itself and there's only a 50%
chance I will walk past the length of
the platform.  Platforms are often offset from stop positions, such as
If I get off the bus there and turn left I don't walk past the platform.
If I get off and turn right I'd only walk through
the shelter if there were people blocking the rest of the sidewalk.  I know
it's only a matter of a metre or so, but
from a strictly technical perspective the routeing is from the stop
position, not the platform.

Even if the underlying mechanics is taken care of by the database engine,
it takes more CPU and memory to
find the segment of a highway closest to a platform than to find the
highway that a node (stop position) is on.  Of
course, I could be completely wrong about that.

If there is any merit to the idea I put forward that stop position and/or
platform should be interspersed in the
route relation rather than gathered at the beginning (I'm not convinced
there is merit in the idea, but I would
find it easier to figure out what is going on in a bus route relation if
they were interspersed) it feels more
natural to me to have the stop position than the platform.

Taking all that into account, it feels more natural (to me, when thinking
of bus routes) that platforms are
physical objects unrelated to bus route relations but which should be
rendered and that stop positions are
a necessary part of bus route relations but which should not be rendered.
I don't even see a necessity to
"link" the two via an auxiliary relation as some seem to have suggested.

As far as I understand it, I actually have the choice to map it this way
(although whether I choose to include
platform or stop position in the relation they currently have to be
gathered at the start of the relation) without
breaking any rules.  If I'm right, the only downsides are that I have to
map two distinct objects (platform and
stop position) and I have to repeat information (the name of the stop, at
least) for both.

Or maybe I'm completely missing the point of just about everything.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20180622/bc6f4bab/attachment.html>

More information about the Tagging mailing list