[Talk-us] Sidewalks as footpaths

Martin Koppenhoefer dieterdreist at gmail.com
Thu May 8 10:13:43 UTC 2014


2014-05-08 11:58 GMT+02:00 Serge Wroclawski <emacsen at gmail.com>:

> Being an American has nothing to do with a really bad data design.
> I've been an American 35 years and I think this is really not a good
> way to model sidewalks.
>


+1, agree, my main concern is that as pedestrian you can actually cross a
road at any point if it is not too much traffic and there are no other
rules forbidding it. Generally with explicity footways routing gets worse
in my experience because there mostly only few connections from the
sidewalk to the road mapped.
There is no way to distinguish separate footways (e.g. separated by a guard
rail) from those separated only by a curb. If mapping explicit sidewalks
there should also be some entity that ties all road lanes (and sidewalks)
together to one object.

On the pro side you can add surface, width and other information to where
they apply, while with tags on the main centre highway way you would have
to split the whole road for every change on the sidewalk (or any other of
its lanes).




>
> The problem (aside from the issue of data clutter) is that the
> sidewalk data can't be used for pedestrian routing because the
> information about the street is not captured. You can't tell someone
> to follow Main Street, because the path is not labeled as such.
>


well, you should either use a relation to tie them toghether, or add the
common attributes like name etc. to all elements (i.e. also to the
sidewalk).

In the end this is a question of detail, for very detailed mapping there is
a benefit IMHO in mapping sidewalks on dedicated objects, but I'd see this
more like explicit lane mapping, i.e. better use another tag than highway
for this in order to avoid confusion with indipendent footways.

cheers,
Martin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20140508/3d9f13a4/attachment-0001.html>


More information about the Talk-us mailing list