[OSM-talk] sidewalks

[Tom Chance]

On Friday 07 Aug 2009 23:15:39 OJ W wrote:
> On Fri, Aug 7, 2009 at 10:15 PM, Martin
>
> Koppenhoefer<dieterdreist at gmail.com> wrote:
> >> How is routing going to know that you can cross the road if you're on
> >> a sidewalk footpath and there's another one 8m away across a
> >> residential road?
> >
> > AFAIK that's an open question. IMHO this will have to be indicated by
> > relations.
>
> would it be useful to have a 'sidewalk' tag on such footpaths, which
> can be interpreted as "you may travel from this footpath to the
> nearest highway without restriction"?

I'm not clear about the benefit of getting this complicated! That's a lot of 
extra work! Also, how will a routing engine know if a way without parallel 
footways is one without pavements or one where the local mapper hasn't entered 
all of them in? At least details like POIs and house numbers can be optional, 
in that an area without them just has less scope for namefinder, routing, etc.

Can we not assume that every highway except rural trunk roads and motorway has 
a sidewalk, unless it has an additional tag like sidewalk=no? Or even foot=no!

If a pavement/sidewalk deviates significantly, just add a footway / cycleway / 
other way branching off from the main highway as appropriate. If the 
pavement/sidewalk is really quite separate, as in your Milton Keynes example 
(http://osm.org/go/eu4qDpI_3--) then by all means add extra parallel ways.

This approach is standard for cycle lanes on the edge of roads, and for cycle 
lanes that are quite distinct from roads they run parallel to.

Regards,
Tom