[Talk-us] Usage of highway=track in the United States

[Mateusz Konieczny]

Feb 25, 2021, 15:30 by kevin.b.kenny at gmail.com:

>>
>> On 25.02.21 13:18, Simon Poole wrote:
>>  > highway=road is a left over from the really old days of OSM before we
>>  > had aerial imagery and the likes and IMHO should in general not be used
>>  > at all any more. It is a far better approach to guess a (low)
>>  > classification than to use "road". Very definitely if you are in the
>>  > position to determine a surface value you are in the position to set an
>>  > appropriate proper highway value.
>>
>
> Exactly. It's unrenderable, unroutable, might as well not be there. It's a similar answer to "don't map it."
>
It depends on router and renderer.

OSM Carto renders it, at least some routers assume it to be usable
As always it depends on a data consumer.

See
https://www.openstreetmap.org/directions?engine=fossgis_osrm_bike&route=36.19116%2C-81.91812%3B36.18927%2C-81.91312
(OSRM car routing is rejected, OSRM bicycle routing is using it)

Rendered also in CyclOSM and Humanitarian map styles.

(and that is without diffing into customization, I can make natural=tree_row in my
install if I would want to do it)

>  
> On Thu, Feb 25, 2021 at 7:53 AM Frederik Ramm <> frederik at remote.org> > wrote:
>
>> It is a discussion worth having. Frankly, I find the idea of erring on
>>  the side of caution has something to it: "I can see there is a road here
>>  but I am 5000 miles away and hence I have zero clue if this road is
>>  available to the general public so I won't claim that it is" - it's not
>>  the silliest idea.
>>
>
> Oh, please. 
>
> I'm talking about roads where I've had my personal boots on the personal ground. I've simply not hiked them all the way to the end, possibly because there's some point beyond which I may not go, but more likely simply because I was using them as approach trails to something else. I don't appreciate the answer that I can't map the portion that I've actually used because I don't know the ultimate destination of the part I haven't used. 
>
You definitely can do it! If you are confident about your guess (or difference is small, like you 
mention) then tag what you guessed with fixme tag.

If someone complains about that - feel free to send them to me to complain.

If you are unsure - use highway=road. And in cases where difference is really tiny:

> Around here, for the 'unimproved' roads, the ultimate destination is the ONLY observable difference among track, service (with or  without service=driveway), residential (motor_vehicle=private) and unclassified (motor_vehicle=private). If a leaseholder builds a cabin or a boat launch, suddenly without anything happening to the road, it changes from track to service, according to the definitions. If there's a leaseholder that builds another cabin, now that's a private residential road.  If it happens to connect to the road network at both ends, but I haven't learnt that because I haven't hiked the whole thing, 
>
All of that can be easily tweaked once such properties are mapped. So it is not
blocking at all mapping road geometry without going N km to the end of a road.

BTW, similarly in Poland road of the same physical characteristics may be residential, tertiary
or maybe even secondary depending on its role in road network. 

It is perfectly fine to map such geometries and tweak classification as two separate stages.

> and it has the two cabins, now it's unclassified?
>
I would not expect two cabins to count as village, so unclassified in such case
seems dubious.

(similarly I am kind of dubious whatever slipway is enough to reclassify road
and whatever single occasionally used cabin deep in woods should actually
reclassify kilometers of highway=track - especially in cases where there
are multiple equally valid ways to reach it)

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20210225/0e2131e0/attachment-0001.htm>