Validation tests for foot/footway/sidewalk

[naoliv]

First a problem: validating a way with highway=footway + foot=no gives us a footway used with foot= (note that there isn't a value after the =)

• done.

Now the enhancements: highway=residential + sidewalk=both + foot=no doesn't issue a warning (it's as contradictory as using highway=footway + foot=no)

• controversial

Also, highway=residential + footway=both doesn't warn that footway is deprecated (and that sidewalk should be used instead)

• done.

[Klumbumbus]

with highway=foot you mean highway=footway?

[naoliv]

Replying to Klumbumbus:

with highway=foot you mean highway=footway?

Yes, sorry :-)
Also, sidewalk=both (on second paragraph) can be both|left|right|yes and footway=both (third paragraph) can be both|left|right

[Klumbumbus]

typos

[Klumbumbus]

Replying to naoliv:

First a problem: validating a way with highway=footway + foot=no gives us a footway used with foot= (note that there isn't a value after the =)

In this case we could fix this by simply changing !? to =no in trunk/data/validator/highway.mapcss#L67, but I made a ticket for this, see #10859

[Klumbumbus]

Replying to naoliv:

Now the enhancements: highway=residential + sidewalk=both + foot=no doesn't issue a warning

It seems some people tag this way when the sidewalks are mapped as separate ways, however I'm not sure if this is good tagging.
Example: ​way/220407308.
In this case should the sidewalk key set to no or be removed?

Overpass turbo querry: ​http://overpass-turbo.eu/s/6yN

[Klumbumbus]

In 7814/josm:

see #10851 - add validator rules for footway=*

[naoliv]

Replying to Klumbumbus:

It seems some people tag this way when the sidewalks are mapped as separate ways, however I'm not sure if this is good tagging.
Example: ​way/220407308.
In this case should the sidewalk key set to no or be removed?

My understanding is that sidewalk is an attribute of the highway.
It's also what is documented in ​http://wiki.openstreetmap.org/wiki/Sidewalk:

The sidewalk (or pavement) is that part of a highway set aside for the use of pedestrians, often separated from the carriageway (or roadway) by a kerb (also curb). A sidewalk may be separated from the carriageway by only a kerb, by a road verge or alternatively may be at some distance from the road (but still associated with it) or separated from the road by some form of barrier, for example bushes or a line of trees. A road may have a sidewalk on only one side of the carriageway, or both side or have no sidewalks.

If the highway allows pedestrians at the sidewalk then it's wrong to use foot=no (so foot=no should be removed)
Like it's implicit that a car may not drive on the sidewalk, it should be implied that a pedestrian must use the sidewalk (and not the street itself) for walking.

[skyper]

foot=no is definitely wrong. Reminds me about the discussion of bicycle=use_sidepath.

Actually, there are some discussions on the German mailing about the issues with separately tagged side_ways. Tagging it on the highway=* and as separate object without any indication seems to be wrong, especially, as I know some places where you find both. The sidewalk directly at the kerbs and a parallel footway 2-3 metres away.

One solution could be sidewalk(:left/right/both)=separate_way or similar.

[naoliv]

Just complementing, I randomly picked some objects from your overpass query and, from the sample, they all have a wrong foot=no.

[Klumbumbus]

In 7819/josm:

fix #10851 - add validator warning for sidewalk=* together with foot=no, workaround for another validator message

[aceman]

Replying to naoliv:

If the highway allows pedestrians at the sidewalk then it's wrong to use foot=no (so foot=no should be removed)
Like it's implicit that a car may not drive on the sidewalk, it should be implied that a pedestrian must use the sidewalk (and not the street itself) for walking.

This is not to be implied. It depends on the country law, whether pedestrians MUST use the sidewalk. Also, the change here actually goes against ticket #9379. Please see the discussion there about the per-country specifics.

I agree that it is not defined on the wiki whether the foot=* applies to the road surface+any sidewalks OR only to the road surface itself (asphalt). Ticket #9379 idea is for it to apply only to the road surface. And in the ticket here you now take the opposite explanation.

[skyper]

Replying to aceman:

This is not to be implied. It depends on the country law, whether pedestrians MUST use the sidewalk. Also, the change here actually goes against ticket #9379. Please see the discussion there about the per-country specifics.

What happens if the sidewalk is blocked or full of snow ?

I agree that it is not defined on the wiki whether the foot=* applies to the road surface+any sidewalks OR only to the road surface itself (asphalt). Ticket #9379 idea is for it to apply only to the road surface. And in the ticket here you now take the opposite explanation.

#9379 follows exactly the same direction as foot=* counts for the whole highway=*. Since #9379 was fixed, validator warns about foot=yes without sidewalk=*, nothing more.

Regarding this ticket, we could use the same warnings for cycleway=* and bicycle=no.

[naoliv]

Replying to skyper:

Regarding this ticket, we could use the same warnings for cycleway=* and bicycle=no.

Where is this warning that I am not seeing using a test way with highway=residential + cycleway=lane + bicycle=no?

[Klumbumbus]

Replying to skyper:

Since #9379 was fixed, validator warns about foot=yes without sidewalk=*, nothing more.

No, the test was removed completely. see #9686

[Don-vip]

In 7827/josm:

see #10851 - fix unit test

[Klumbumbus]

Replying to Don-vip:

see #10851 - fix unit test

Thanks. My RSS-Reader for Jenkins push me the failed build just now :(

[Don-vip]

No problem, my Jenkins had some problems this week and I wasn't available to fix them until now.

[Klumbumbus]

In 7873/josm:

see #10851 - remove validator warning for sidewalk=* together with foot=no, as there is no clear consensus about this test