[Tagging] Feature Proposal - RFC - Sidewalks as separate ways
wendorff at uni-paderborn.de
Sun Apr 3 11:12:42 BST 2011
There are two things I would like to make additions to.
1) separate or tag-scheme (the long discussion in general):
I'm one of the people who think, both should be used on a
Working on a map- and routing-portal for blind people (therefore most
likely a special case of pedestrian routing in the main focus) we
thought about this issue several hours.
On the one hand we would prefer to have separate tagged sidewalks as
it's necessary to have information about crossings etc.
On the other hand we are active mappers, too. Therefore we know, that
it's more work to map separate sidewalks, if one start to add more
details than already possible with the tags at the street.
We came to the conclusion to try going the following way:
- As application developers we have to generally support both schemes.
The separate ways, because they are better, the tags because they are
more often availlable.
- It's necessary to create virtual, local ways for the sidewalks - but
that should be possible. Problems could arise, where the street has
footway:left=yes and additionally a separate tagged way. I thought about
footway:left=separate to state, that there is a sidewalk on the left
side of the street, but that this footway is mapped separately, too. I
would appreciate comments to this idea.
- If possible and we have the time, we want to support a tool to expand
the tags to separate ways in a semi-automated manner. But that's only
the sketch of an idea yet.
Finally: It's better to tag footway:left etc. than to tag nothing about
sidewalks. It's better to map separate sidewalks - if you are able to
consider the width of the street, the wide of the sidewalks and possibly
strips of green or lanes of parking cars in between.
For crossings it's good to tag them even when footways are tagged not
separately, but it's not possible AFAIK to indicate features of the
sides of the crossings (sloped curbs, tactile paving etc.). If footways
are mapped as separate ways, think about mapping crossings as
5-node-ways like the following ASCII-Art indicates:
| | |
| | |
The vertical lines left and right are the footways, mapped as
highway=footway; footway=sidewalk; width=*; surface=*;
The vertical line in the middle is the street, mapped as highway=*;
The horizontal line, here 5 times + is the crossing way with 5 nodes:
the outer nodes are connected to the sidewalks and not tagged with anything.
The node in the middle is tagged as highway=crossing; crossing=* (e.g.
uncontrolled); crossing_ref=* (e.g. zebra). It's connected to the street
and to the crossing way.
The nodes in between are only part of the crossing way. They indicate
the curbs at both sides of the street. Tagging should consist of
sloped_curb=*; tactile_paving=*; where sloped_curb is a measurement
(necessary for wheelchair users) or a rough value out of 0, low, normal,
high. The latter allows applications for blind people to prevent
dangerous issues (where there is no curb at all - sloped_curb=0), and
wheelchair drivers have better information than nothing. That's not
perfect for wheelchair drivers - I know, but it's possible to survey
nearly on-the-fly, while a concrete measurement would require lot more time.
2) Footway names:
Let a sidewalk be mapped as separate way. Then the question arised in
the discussion, if this sidewalk way should get a name attribute with
the name of the street.
As far as I understood that, most people did not want that. The argument
to not tag the name to sidewalks was mainly, that it's the name of the
street; not name of the sidewalk.
I have a counter argument, that on the other side could be an argument
against separate mapping of sidewalks: A sidewalk is part of the street.
Therefore to be completely correct, the name would belong to the
associatedStreet relation. If not availlable (as I also prefer the
relation to be optional), I think, it's correct to be more verbose and
redundant here and to tag the name to street AND sidewalks.
If we take into account the proposed footway=sidewalk, it's very easy to
prevent sidewalk names from being rendered in graphical maps.
Am 03.04.2011 10:42, schrieb David Paleino:
> Hello Ed,
> On Sat, 2 Apr 2011 22:26:22 -0400, Ed Hillsman wrote:
>> The discussion of the sidewalk issue seems to have stopped. I added
>> some comments in the discussion section of the wiki last week, but
>> there have been no further comments there or here in nearly a week.
> I saw your comments on the talk page, but I took them as generally appreciation
> -- i.e. there was no question? :)
> (yes, the relation is optional)
>> I think each of the proposals (sidewalks as separate ways, and
>> sidewalks as attributes of streets) has merit in different
>> circumstances, and the choice of which to use should be optional. Like
>> Josh Doe, I've been mapping sidewalks as separate ways, but that is
>> because I've done most of my mapping in the suburban areas where I
>> work and live. I've tried it a bit in downtown Tampa, and down there I
>> think it makes more sense to tag them as attributes of roads.
> Could you clarify why in one case you chose the "separate way", and in the
> other case you chose "tagging the main road"? This could be enlightening for
> those thinking that the two proposals are in contrast :)
> (I myself believe they can cohexist -- I just believe that mapping separate
> ways gives the chance of mapping more details, that's all)
>> Except for a few block faces and intersections, where the city hasn't
>> made curb/kerb cuts for wheelchairs yet. This means that some intersections
>> have some sidewalks where wheelchairs can cross in some directions,
>> but not in all. And for these I would code the sidewalks as separate
> This is one of such examples, thanks. However, it seems a bit odd to have
> "mixed styles" in one place/city/... :) -- but maybe it's just a matter of
> taste ;)
>> With regard to routing, sidewalks on college campuses, in parks, and
>> in cemeteries may be interior to a large area bounded by streets, and
>> as a result some may not have an associated street to use for a name.A
>> few sidewalks on a few campuses may have names of their own ("Slant
>> Walk" on the Miami University (Ohio) campus, is one example), but most
>> don't. So routing will need to come up with some other way to refer to
> I guess the named ones would have a name= attached; for the others, I can
> imagine a router saying "turn left at the next footway, 200m" -- just like they
> do with unnamed roads (be them unnamed, or the name hasn't been tagged yet).
>> This is an unresolved issue, but it's one for the routing, not
>> for OSM to deal with (if there is no name, there is no name to tag).
>> Where a street parallels a street, but at a distance, I share the
>> concerns about using relations to associate sidewalks with their
>> streets. Would it work to add a tag "associated_street" and then
>> simply list the name of the street? For example, highway=footway,
>> associated_street="East Fowler Avenue". A value of "none" could be
>> coded if the sidewalk does not parallel a street.
> Well, the relation is optional, but there usually is one already.
> associatedStreet/street, in fact, is used for housenumbers -- so creating one
> for sidewalks wouldn't be "wasted time" -- it's just preceding the housenumbers
> mapping :)
> But yes, being the relation optional, we can just use plain tags (even though
> I'm fond of relations, and believe plain tags can become broken).
>> I hope there will be more discussion of these two proposals.
> I can't sincerely say why they stopped. Probably temperature became too hot to
> continue :) -- I must say I calmed down a bit on my side, I just hope we don't
> start again fighting each other ;)
> Tagging mailing list
> Tagging at openstreetmap.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Tagging