[PATCH] Stop whitelisting footway=*, and add footway=separate to blacklist candidates for sidewalk=*

[JeroenHoek]

To resolve #10851, #10976, #15439, and #19389 the validator checks if ways have footway=* set with one of the values typically used for sidewalk=*. This seems to be a fairly common mistake, so it's good to have a warning here.

Issues

There are two minor issues with the current validation approach I would like to resolve.

footway=separate is missing

The validator checks for left, right, both, none, and no. separate is missing from this list.

Whitelist blocks all other possible footway values except for three

While the above values are clear examples of misstagging, footway is not an invalid key, and is used to provide a refining of highway=footway. The validator acknowledges this for access_aisle, crossing, and sidewalk, but excludes everything else. This seems too strict compared to the other rules — we don't whitelist values for highway=* for example. Mappers can (and do) create new values, and some are gaining broader acceptance. This check seems superfluous with the checks for values that probably mean sidewalk=* in place.

Patch

In the attached patch:

• A check for footway=separate is added.
• The whitelist for footway=* is removed.

Are there any issues I am overlooking or reintroducing with this patch?

[skyper]

According to #21242, sidewalk=none should be replaced by sidewalk=no, so JOSM is correct with the special treatment.

[anonymous]

Really? This isn't reflected in the current documentation of sidewalk=*. I'm not seeing any recent mention on the Taggin-ML either.

I'm not opposed to deprecating sidewalk=none, but was this discussed somewhere else than on the Taggin-ML and the wiki-page for sidewalk=*?

[JeroenHoek]

(Comment by me.)

[anonymous]

Now OSM Wiki discussion appeared: ​https://wiki.openstreetmap.org/wiki/Talk:Key:sidewalk#Is_sidewalk.3Dnone_valid.3F

[mkoniecz]

that was me, sorry

[JeroenHoek]

I've removed the removal of the special treatment of footway=none from the patch.

[JeroenHoek]

Just to clarify: this patch is no longer in conflict with any current developments. sidewalk=none is now considered deprecated in favour of sidewalk=no; this patch is limited to the items listed in the first post.

[skyper]

Ticket #21912 has been marked as a duplicate of this ticket.

[stoecker]

Milestone renamed

[taylor.smock]

@skyper: Did you have any additional problems with the patch?

My one problem is way[footway=none][/footway:/] { -> way[footway=none][/^footway:/] {. If this is correct, I'd like to see an assertMatch/assertNoMatch test.

[skyper]

Replying to taylor.smock:

@skyper: Did you have any additional problems with the patch?

No, though, I did not test it with real data.

[JeroenHoek]

Patch

[JeroenHoek]

@taylor.smock I just followed the example above to remain consistent. This seems to work.

[taylor.smock]

I was thinking of tags like surface:footway (although footway:surface is way more popular). Although maybe it was for footway:left or footway:right. I'd have to dig through the history to figure out why there was a regex.

[taylor.smock]

In 18523/josm:

Fix #21235: Stop whitelisting footway=*, and add footway=separate to blacklist candidates for sidewalk=* (patch by JeroenHoek)

Patch notes by JeroenHoek (modified):

• Add a check for footway=separate.
• Remove special treatment for footway=none, none is added to the general footway=* check.
• The whitelist for footway=* is removed.