The one with `restriction:bicycle=yes` is a mistake; it should be `restriction:bicycle=give_way`, thanks for catching that.

There are at least 3 tagging schemes that in parallel use for tagging right-on-red for bikes. I've been tagging both with the point-based scheme you mentioned (`red_turn:bicycle=yes`) and with this relation-based scheme that seems popular mostly in France (see osm.wiki/FR:Signalisation_routi%C3%A8re_en_France#Panonceaux_de_type_M, https://taginfo.openstreetmap.org/tags/restriction%3Abicycle=give_way#map), because the CyclOSM map renders the latter. It also should in theory be easier for routers because it actually specifies the from and to ways, whereas the point-based tagging scheme is a bit ambiguous.

How did you decide there were no cycle lanes on Baslerstrasse (way #699662519)? They were there on the ground when I was last there in Autumn, even though they aren't in the aerial imagery.

Pretty sure these are not cycleways (as far as I can tell, there are no blue signs as required per osm.wiki/Switzerland/Map_Features#Cycle_and_Foot_Ways). However I only checked on Zürcherstrasse as that's the only road with Mapillary.

When this is the case, the ways should be tagged as highway=footway footway=sidewalk if they are close (< ~2m) to a road (i.e. legally a "Trottoir") or highway=path otherwise.

Instead of adding new ways that overlap existing ways, you should just add the golf-related tags to the existing ways. Separate ways should only exist if the cart path is physically separated from the track/service road, but in this case it is not.

Ja, ich bin fast einverstanden. Aber finde ich dieses red_turn-Tagging auf Punkte ungenügend, weil es unklar ist, auf welches Way das freie Rechtsabbeigen gemacht werden kann. Die restriction-Relation zeichnet sich hier aus und ist sehr klar - from, via, to.

Ich entferne gleich das traffic_signals:turn:bicycle:forward.

red_turn Tagging ist in #101327539 gekommen

Ich habe es hier gefunden: osm.wiki/Proposed_features/turn_signals. Es gibt auch häufig eine Relation restriction:bicycle=give_way (besonders in Frankreich).

Alle drei Taggings sind kaum gebraucht, also habe ich alle getaggt. Sie stehen nicht in Konflikt miteinander: dieses wird auf Ways getaggt, red_turn:right:bicycle auf Punkte und restriction:bicycle=give_way als Relation.

FYI, some of the edits in this changeset are quite wrong, e.g. surface=asphalt on a gravel road (e.g. way/124916624). This in particular is very hard to see from aerial imagery, you should avoid tagging it without either street-level imagery or on-the-ground experience. Also on this way, you reclassified it from track to unclassified, but track is probably more appropriate here. Unclassified is used more for roads linking villages, not for roads that only tractors and non-motor traffic use.

Also, it isn't generally helpful to add obvious defaults like salt=no on a creek (e.g. way/719455933). You should only tag it if someone might reasonably think it was not the default.

Thanks for your restoration. I would propose these further changes, would you mind taking a look and see if we agree or not? Summary:

* bridge footpaths have bicycle=designated (implied by SSV 2.63.1)
* bridge road has bicycle=use_sidepath (implied by SSV 2.63.1)
* traffic signals for the transition bicycle paths (otherwise it's possible to route through the intersection on bike without hitting any traffic_signal). Alternatively, the transition ways could be moved to meet the road just before the traffic signal nodes?
* split Selnaustr. eastbound to end the bike lane exactly where it ends, i.e. the pedestrian crossing
* split Selnaustr. westbound to start bike lane exactly where it starts, i.e. from Am Schanzengraben.

osmChange file: https://filebin.net/gicggiatr9wn6jcm/changes.osc?t=ldz2y594

Your first request seems a little contradictory: if I should end bike lanes at pedestrian crossings, then I will *always* need to split ways on the traffic_signals node, since in Zürich the tagging scheme is usually combined traffic_signals and crossing. I don't mind ending the bike lanes before intersections, but it does cause many more splits of ways, since each way between 2 intersections becomes 3 ways (no lanes > lanes > no lanes).

The problem with the cycleway=track is twofold: a) shared bicycle/foot without separation shouldn't be represented by it (because it's defined as only covering dedicated separate bike tracks) and b) the footpath already is covering it in data, so keeping it is duplicate. The correct fix is to remove the cyclway=track, and correct the tags on the footpath to what is signed (shared bike/foot without separation, i.e. SSV sign 2.63.1). The tagging standards for Switzerland on the wiki say to use highway=footway bicycle=designated for this sign, which is why I tagged it that way and not with highway=path (which is my personal preference). I haven't been tagging with ID presets btw, but according to this local wiki page.

Could you be more careful when reverting?

FYI, this revert broke the bicycle route relations. Also, it's not clear to me why you reverted the bicycle lanes on Selnaustrasse (they do exist), nor the corrected tagging of the cycleways on Stauffacherbrücke.

I intend to restore these edits unless you have some good reason why they shouldn't be mapped.

Thanks for the history lesson - I had no idea Allmend used to be a military area :)

Honestly, it seems to me all 3 tags are deficient. "track" is defined as roads built for for agricultural/forestry/etc., which this isn't really. "service" is defined as roads proving access to a specific place, which this doesn't (if it were signed "Zubringerdienst gestattet" then service would be much more acceptable). "path" is defined as being for non-motorised vehicles, which applies to the current use of this bridge, but not to the construction standard (I hadn't considered this before, because I didn't realise it was constructed to a high standard).

I think "track" is the most easy to defend: this bridge, and the roads through Allmend, are very much like Waldstrasse or Wirtschaftsweg, despite now only being used for leisure, and it reflects that it was originally constructed with capacity for heavy vehicles. Otherwise I would use "path" with "width=3.5", but this doesn't capture the construction character properly. I think "service" is least good.

Sorry for the long comment, and sorry about the English: reading German is ok, with the help of a dictionary; writing coherent German is still a struggle.

> Ich hatte auch schon Rückmeldungen von Mappern, dass 'use_sidepath" nicht funktioniert oder ignoriert wird. Ich meide dies deswegen.

Do you mean foot=use_sidepath (use_sidepath isn't a valid value for sidewalk per wiki)? I think they mean slightly different things: foot=use_sidepath marks a legal duty to walk on the sidewalk. sidewalk=separate only means there is a sidewalk on this road, mapped as a separate way; on roads where walking on them is allowed *and* there is a sidewalk (Begnugungszonen, perhaps?) use_sidepath would be wrong.

> Ich möchte die OSM-Welt nicht auf den Kopf stellen. Aber du bist forscher. Tipp: wir brauchen nicht 'sidewalk=*" (no|both|left|right|separate), sondern sidewalk:left=* und sidewalk:right=*, mit den Werten (no|yes|mapped[_separate] – oder 'explicit').

The wiki documents it as sidewalk=left/right/both/none, and unlike you say I don't want to boldly establish new tagging schemes. The wiki does provide for sidewalk:left/right, but only for extra attributes, not for existence, as far as I can see. TagInfo backs this up, there are many more instances of, e.g. sidewalk=right than sidewalk:right=yes, even if one thinks the latter is a cleaner scheme.

> Ratonale: even a very, very stupid, or lazy programmed Renderer, can tell, that there's some kind of sidewalk (left or right), even it's not mapped or tagged properly, separately.

I welcome you to try ;) The OSM state of the art unfortunately does not make this so easy, especially in edge cases where there are multiple sidewalks. This is precisely why the hint value of "separate" exists, to stop people assuming that the nearest separately-mapped sidewalk must be a duplicate of the sidewalk tagged on the road.

I came to the conclusion after sending the original comment, that it would be a Sisyphean task to keep the tags consistent, and separate ways only seems to be the way forward, even if it goes against the recommendation to tag directly abutting sidewalks only on the street way.

I'm not sure this is right. See https://www.mapillary.com/app/?lat=47.350975000000005&lng=8.520177777777803&z=17&pKey=F7DwYLusmZiLNoVCfmaWEQ&x=0.24543601761529255&y=0.5714318804435264&zoom=2.6463654223968565&focus=photo. It's a little hard to make out in this photo, but you can see there is an island of grass with an apparently immovable pole right in the middle and a bollard behind it, that the path splits around. Each half of the path on either side is too narrow to be used by multi-track vehicles. I have been there myself by bike many times.

When you add sidewalks as separate ways, you might want to re-tag the existing sidewalk=both to sidewalk=separate on the main street way, if present, or just delete the sidewalk= tag. I am investigating rendering of the sidewalk= tag, but it is problematic when a sidewalk is tagged both on the street way and as a separate way.

See way/28938433 for a similar example of an unsigned pedestrian space near Bürkliplatz, also tagged this way.

Sorry, didn't meant to tag a legal Fussgängerzone, just that in front of the Badi is a large pedestrian plaza. How would you tag that?

re: highway=traffic_signals, see highway=traffic_signals#Tagging_also_cycleway_traffic_signals. This is clearly the case that should be followed when there are explicit footways, too. Note how the traffic signals are marked at the stop line for each way.

Also, marking the crossings where cars have traffic lights as highway=traffic_signal but not the ones where only pedestrians have lights breaks rendering, at least on cyclemap: https://imgur.com/ZqzCNhz

See how only one of each pair of crossings are marked with the crossing circle?

The split of the road is definitely wrong per wiki. The bike track is duplicated and should either be tagged as cycleway=track on the road, or as a separate way, but not both. The right turn for bikes at the traffic light is not allowed anymore: the traffic light covers both directions now. And for the traffic lights: the wiki says not to do it the way it was, if there is direction tags, because it is ambiguous.