Kan foot=use_sidewalk samen met sidewalk=* worden gebruikt?

Slightly related on the topic of sidewalks.
I am coming accros a German user who has tagged quite some streets in Amsterdam that have foot=use_sidepath with sidewalk:left=no and sidewalk:right=seperate.

I looked up the last tag on the forum and this seems to be something being discussed more on the german fora.

In my opnion these tags should not be needed when foot=use_sidepath is already there. And possible even cause for the same routing errors as in Google Maps that cyclists and pedestrians are routed following the main road; while the actual sidewalk is part of the more or less parallel cyleway or even parallel residential road. (e.g. the case at the President Kennedylaan)

What do you think?

why marking that sidewalk on the right is separately mapped (so routing along road should be avoided) would encourage router to use road instead of separate geometry

foot=use_sidepath does not contain information about on which side of the street the foot path is, while sidewalk:right=seperate does. This could be used by routers to give a more accurate route for pedestrians (if the street is approached from the left, it could tell to “Cross the street and turn right onto the separate sidewalk”, for instance)

The wiki for foot=use_sidepath says

… tag the main road as foot=use_sidepath if the sidepath is mapped separately.

If the sidewalk is mapped separately, sidewalk:* tags are redundant, I think. The geometry information is in the mapping itself.
Routers should not use the road tagged with use_sidepath. I expect them to find the available connection in the vicinity: the sidewalk way.

If the sidewalk is not mapped as its own way, there should be a tag sidewalk=left|right|both on the main road (or cycleway) and the router can use the main road (or cycleway).


Agreed with @Peter_Elderson, I think *=separate is superfluous tagging.

If you are convinced this is a good idea, I would suggest to make a detailed example, that gives the steps that the router makes and show that there is no other way *=separate to get something done.

Let me give a start, a router works with points (A/B/C/D) and lines (residential/cycleway/footway), is your example something like this where someone wants to walk from Point A to Point D


*=separate can be also used for visualisation, statistics, matching roads with separately mapped sidewalk geometries, QA…

Though I do not see use specifically for routing if access tags were added.

1 Like

Of course some context is helpful. The Pres. Kennedylaan is a one way secundary road for each side, with parallel cycleways (and a parallel road) that have sidewalks next to them.

Pedestrians are not supposed to use the secundary road at all and are supposed to use the cyclewa/parallel residential road.

Ideally I would even give it the foor=no and have the router figure out the alternative road. foot=use_sidepath imho is curtousy to the router.

For a router, foot=no and foot=use_sidepath are roughly the same. It means that the router has to avoid routing over ways that have these tags. The difference is most clear with bikes.

For normal bikes, bicycle=use_sidepath means don’t go there. But if (in The Netherlands) the bike has 3 wheels and is wider the 75 cm or the bike has a trailer that is wider than 75 cm then the bike can use the road. Except of course if the road is specifically closed for all bikes.

So a router for wide bikes would treat use_sidepath as yes.

For pedestrians it is not clear if those exceptions exist. But it is still good to distinguish between road where pedestrians are never allowed and roads where pedestrians are supposed to walk on the bike path of sidewalk.

Is bicycle=no and bicycle=use_sidepath equivalent for regular bicycles in Netherlands?

For example there could weirdly planned layouts (maybe less likely in Netherlands) where say turning left is impossible using cycleway - but possible if going on bicycle=use_sidepath road a bit. In Poland that would be legal. But not on road with cycling ban (bicycle=no).

There are plenty examples in The Netherlands where the road has a T-crossing; but the parrallel cycleway doesn’t. So cyclist would not have a road to make a turn.

They still might do so, crossing a stretch of grass and it might not even be illegal; but the intent of the road design is for them not to cross there an cross elsewhere.

Way too vague to convince me. Most things can also be done with foot=use_sidepath and I am pretty sure the rest can be done with the OSM context for example with PostGIS.

QA based on things that superfluous is the worst QA there is :wink:

I would model that by leaving the bicycle tag out on the appropriate piece of the main road.
The secton between where the cyclist joins the road up to where the left turn actually is.

It largely is, the difference should be that bicycle=no means that it is signed and bicycle=use_sidepath not directly signed but indirectly. I guess that in The Netherlands less than 5% should be mapped as bicycle=no but there is still a large percentage mapped as bicycle=no.

bicycle=use_sidepath is with a good bicycle router largely useless tagging as the router should prefer cycleways over roads. Why it is then still fine to map it is:

  1. There are some constructions where additional “weight” is needed to have a bicycle router chose the correct route.
  2. People otherwise have the hard tendency to tag the road as bicycle=no
  3. You can argue that bicycle=use_sidepath is weaker than bicycle=no and a bicycle router can translate that in it weights.

This last point matches what you added on the Wiki:

Note that in some cases cyclist may be allowed to use bicycle=use_sidepath ways while not allowed to use bicycle=no ways in any case at all.

Any decent bicycle router should also not completely lock out bicycle=no as there are tagging errors and often is it allowed to use it on foot, i.e. with a dismounted bicycle

That is certainly not the case. If you route for a ‘racing bike’ then smoothness=good is fine but smoothness=intermediate should be avoided. There are plenty of roads whether the road has smoothness is good and the bike path has smoothness=intermediate.

There are also cases with the bike path has less optimal intersections, so the router would prefer the road as it makes more sense. Of course it depends on priorities. But you also want to prevent all kinds of detours just there is a bike path somewhere.

Een router kan overgaan op foot, maar dan geldt nog steeds bicycle=no. Dan staat er gewoon foot=yes. Ik kom dat weinig tegen. Bijna altijd als je op een weg niet mag fietsen, dan wil je er ook niet lopen. Ik heb wel eens een weg lopend overgestoken omdat er alleen een verbod on te fietsen stond. Maar dat ging alleen goed omdat het erg rustig was.

Ik denk dat Emvee het heeft over bv. een bruggetje dat alleen gebruikt mag worden door voetgangers, maar wel een heel stuk afsnijdt als je als fietser er lopend overheen gaat.

1 Like

Dismount zou een optie moeten zijn in een router. Vaak is het handiger om maar om te fietsen. Bijv. er zijn voetgangersbruggetjes waar je met een fiets niet overheen wil. Er zijn ook brede fietsen die niet geschikt zijn voor een smal wandelpad.

Dan zit er na enige tijd ineens zo’n wringhekje voor waar je je fiets overheen moet tillen. Jammer voor rolstoelers, dat wel.

Well that’s kinda problematic if there is a foot route relation on that road. JOSM will give an error if you try to upload the changeset:


if there are separately mapped sidewalks, then relation should be marked there (you would walk on sidewalk, not on carriageway, right?)