[Tagging] "Feature Proposal - RFC - Sidewalk
wendorff at uni-paderborn.de
Tue Mar 22 16:17:45 GMT 2011
Am 22.03.2011 16:04, schrieb M∡rtin Koppenhoefer:
> 2011/3/22 Josh Doe<josh at joshdoe.com>:
>> On Tue, Mar 22, 2011 at 6:32 AM, M∡rtin Koppenhoefer
>> <dieterdreist at gmail.com> wrote:
>>> 2011/3/21 David Paleino<dapal at debian.org>:
>>>> On Mon, 21 Mar 2011 21:12:55 +0100, M∡rtin Koppenhoefer wrote:
>>>>> 2011/3/21 David Paleino<dapal at debian.org>:
>>>>> To not be misunderstood: I prefer explicit sidewalks (=separate ways)
>>>>> as I wrote above. But you should not map them as if they were
>>>>> independent ways.
>>>> They aren't: have you checked my proposal? They're part of the same relation
>>>> (street, or associatedStreet).
>>> If you need a relation for every sidewalk, it is clear that you are
>>> redefining footway, because not interpreting this relation will lead
>>> to misunderstanding for all footways (they would be understood as
>>> independent ways and routing would work worse then with no sidewalk
>>> mapped at all).
>> Adding footway=sidewalk is not redefining highway=footway, just like
>> saying service=parking_aisle is not redefining highway=service, rather
>> it is a refinement.
> I disagree here. In the case of service it is a refinement, but in the
> case of footway it is not, because highway=service is the tag to use
> for smaller service ways, but highway=footway is not the tag you use
> for lanes (a kind of which sidewalks are), ...
Slightly disagree here:
A sidewalks is a lane for a pedestrian in a low-traffic residential road.
Towards a young child I would define a sidewalk as a way - the child
should not cross the street otherwhere than at marked and much as
possible secured crossings.
A wheelchair user will interpret the sidewalk as a separate way, not a
lane, as long as there is a curb higher than a specific treshold - and
that's the case nearly everywhere, where no marked crossing or at least
a driveway to a house is located.
I know: A healthy, perhaps a little bit tired of life adult will cross
the street wherever he wants; but a local car driver will also use the
street where we objectively have to tag access=destination to have a
I consider a sidewalk with the following assumptions:
1) It is intended for a sidewalk user (pedestrian, child on bike,
skater) to use the street itself for driving. Therefore there is no
constant change intended.
2) It is not allowed for a car to drive at the sidewalk (crossings
excepted). Therefore there is no constant change intended either.
But is that really a lane? Doesn't that indicate the sidewalk being a
You can ask: But I can cross the street directly everywhere I want,
that's not modelled in the data by modelling explicit ways.
My answer would be: Yes, that's true. But what's the drawback of
applications here? The worst case would be miss a cut where no crossing
is present in OSM. I would like to have crossings everywhere, where
- crossings are marked (islands, zebra crossings, pedestrian traffic
signals and so on)
- it is common to cross (left and right of street intersections; where a
footway joins the sidewalk etc.) AND there is no better marked
alternative around AND it's not forbidden to cross.
> ...it is a tag that you use
> only on independent ways. The sidewalk is already comprised in the
> main road according to our data model,
Where is this data model? I would say: nobody thought about sidewalks at
creating the data model - it's not defined.
If you claim that highway=residential includes sidewalks, that's new to
me - and neither proofed nor (well) documented
> and adding a separate highway=footway indicates that there is a barrier between the footway
> and the road.
There is! Ask the next wheelchair user or old man/woman with a walking
frame about the barrier a curb of normal height is for him.
> To overcome this, you would have to use highway=footway
> on lanes / sidewalks, what is not in accordance with the current
As I wrote above: there ARE arguments to define a sidewalks as a
dedicated way, and the current conventions are not common ones in this
More information about the Tagging