`highway=service` areas neither provided nor accepted

[ManDay]

Carto renders area=yes; highway=service, but iD suggests area:highway=service for the "Road Area" preset and complains about the former not being a line.

[ManDay]

as from Matrix

ManDay: area:highway=* and highway=* + area=yes have different semantics. Use area:highway=* in combination with a linear highway=* if there are clearly defined movement directions/lines, highway=* + area=yes if vehicles are expected to move around the area at will in any direction.

So I suppose it's only about support of the latter.

[tyrasd]

there is a preset for highway=pedestrian + area=yes: https://github.com/openstreetmap/id-tagging-schema/blob/v3.2.2/data/presets/highway/pedestrian_area.json

Other types of highways are very rarely used on areas, as far as I can see, and the wiki also doesn't allow it for many of them (e.g. https://wiki.openstreetmap.org/wiki/Tag%3Ahighway%3Dresidential). So this would only be affecting highway=service + area=yes (which seems to be used very rarely on situations like industrial loading/unloading areas). Or are there other highway types I overlooked here?

[ManDay]

It may not be exactly canon, but I subscribe to the perspective that OSM effectively encodes a vectorized (as opposed to voxel/pixelized) volume/surface description, to the effect that both lines and areas are nothing but some arbitrary building blocks for that description.

When a surface area is too complex to be described by a line (pretty much everything which is not "road shaped", i.e. translational symmetry along a line), an area has to be used.

So, bluntly put, I think everything which exists as a area=no should, in theory, also be supported as an area=yes for cases where the shape of the entity demands it. So something like highway=path with area=yes does make sense, although it is perhaps uncommon. In the end, your call which combinations you deem relevant enough to support them as a distinct rendering. In my personal opinion, highway=service with area=yes is fairly common, though.

[matkoniecz]

So, bluntly put, I think everything which exists as a area=no should, in theory, also be supported as an area=yes for cases where the shape of the entity demands it.

This is blatantly wrong.

Mapping for example area of motorway carriageway is done with area:highway=motorway not highway=motorway + area=yes

Mapping area of tree row is done with landuse=forest not natural=tree_row + area=yes

Mapping area of waterway=river is not done with waterway=river + area=yes

highway=service + area=yes is widely misused where area:highway=* should be used.

[ManDay]

I think these are separate tags, though I agree I wasn't precise enough when I said "surface description".

area:highway=* is to designate the surface material associated with a highway=* path (of an area=no highway), whereas highway=*; area=yes is to designate the surface usage (of an area=yes highway).

You are correct w.r.t. the forest, but I already said that my view is not strictly canon (and tagging schemes are a matter of frequent dispute anyway).

I think the description I gave is simple, consistent, and compatible with the existing area:highway=* and highway=* semantics, therefore it makes sense to adopt it. The fact that other tags (like the forest example) are not necessarily compatible and sometimes messy notwithstanding.

[matkoniecz]

area:highway=* is to designate the surface material associated with a highway=* path (of an area=no highway), whereas highway=*; area=yes is to designate the surface usage (of an area=yes highway).

That is completely wrong and is your personal tagging idea.

Mapping for example area of motorway carriageway is done with area:highway=motorway not highway=motorway + area=yes

highway=motorway + area=yes is a clear tagging mistake.

[ManDay]

No, what you quoted is actually established practice (and described in the wiki). What you say about a motorway is correct but is in no contradiction to what I said. The fact that you will rarely to never find the specific combination highway=motorway; area=yes is because motorway usage is inherently linear. A motorway therefore does not require the semantics I described (where usage is of the kind area=yes), but you shouldn't dismiss the principle just because you found an example to which it does not apply.

[matkoniecz]

You claimed general rule covering everything:

So, bluntly put, I think everything which exists as a area=no should, in theory, also be supported as an area=yes

bolding original from https://github.com/openstreetmap/id-tagging-schema/issues/431#issuecomment-1126297896

because motorway usage is inherently linear

not on toll plazas ( https://commons.wikimedia.org/wiki/File:%C3%9Ajharty%C3%A1n_l%C3%A9gifot%C3%B3.jpg ), motorway air strip sections and numerous other places

(I will unsubscribe as it is getting quite offtopic here, sorry)

[ManDay]

Yes, that's correct. That's what I claimed should be supported. And if you agree that there are features where usage is not a network of linear paths (which, by the way, I suppose might still the case for many toll stations, though), then indeed, that would be a case where highway=motorway; area=yes is appropriate. Again, that is in accordance with what's written on the wiki.

Perhaps it would help if you could clearly formulate the objection that you have in the sense of "if that is done like X, then Y breaks", or something of the sorts. I'm afraid I have not exactly understood why you started this discussion.

[tyrasd]

@ManDay I appreciate your enthusiasm, but this issue is not a good place to discuss potential changes to the way the OSM tagging system. Please consider moving the discussion to a place like the tagging mailing list for example.

---

Back on topic:

I found that there are two more highway values which are documented to be allowed on areas, although they both seem to be somewhat unclear or controversial situations:

• highway=footway – quote: There is no clear consensus on this topic
• highway=unclassified – the wiki is slightly inconsistent on this one: the highway=* overview page lists the unclassified as a line-only tag

[matkoniecz]

highway=unclassified – the wiki is slightly inconsistent on this one: the highway=* overview page lists the unclassified as a line-only tag

this seems clearly wrong, I edited wiki: https://wiki.openstreetmap.org/w/index.php?title=Tag%3Ahighway%3Dunclassified&type=revision&diff=2326550&oldid=2310410

highway=footway – quote: There is no clear consensus on this topic

in some areas people widely use area=yes + highway=footwat/pedestrian to map areas of sidewalks.

[ManDay]

It was not my idea to start this discussion. I just tried to respond to @matkoniecz when they started criticizing my perspective. I think the issue which I filed about highway=service; area=yes is not at all proposing any changes to the tagging system. It is well confirmed by practice, wiki, and carto.

The only thing I tried to convey with my remark was that I'd think it useful if iD would not only "support" that in the scheme, but more generally area=yes as I explained it above. But it was just a remark. If I had meant it to be central to the issue I'd have said so in the opening post.

[tyrasd]

Ok.

the issue which I filed about highway=service; area=yes is not at all proposing any changes to the tagging system.

I think there might have been a slight misunderstanding, because the issue title did originally mention all highway=* presets. At least I got slightly confused there as well.

For iD we will continue to support the documented/approved usage of tags.

[richlv]

Thank you for fixing this. With iD complaining about highway=service + area=yes, people went around "fixing" such occurrences, in most cases breaking things - that is, instances where an area was mapped, not outline of a linear way.