No, it’s about cycleway=crossing-ways and these can be on combines foot/cycleways. But let’s stop nitpicking, it doesn’t change this proposal.
if the cycleway is a sidepath of a road, then the same rules apply as if you were driving on that road
for sidepaths this depends on the layout (position of traffic lights / road markings / sidewalk and cycleway)
Good point. The cars in the road can’t turn onto the cycleway though so I still see it as a different situation than a two cycleway intersection where bicycles are free to move from one way to the other. I guess I should rephrase my crossing description from this:
where two different types of traffic cross and turning from one way to the other is generally not allowed or not possible.
To something like this:
where two different types of traffic cross and either one or both of the traffic types may not turn onto the intersecting way.
So a street where HGV are allowed turning into a street where they are forbidden would be a highway=crossing 
? Or any street turning into one where cyclists are forbidden?
Seriously, if they are allowed to turn into each other or not, can be derived from the access-tags and the highway=*-value. The only reason highway=crossing existed in the first place was, because we didn’t have separately mapped sidewalks and people wanted to add crossings. Which is why its original meaning was “place where pedestrians can cross”, and then people added tags to additionally allow bicycles and horses to cross. The essential definition was “Pedestrians can cross a street here”
The reason for footway=crossing was that because you were crossing the street, you were no longer on the sidewalk, so footway=sidewalk would be kinda misleading. From this logic, a street crossing a sidewalk would also mean that you are no longer on the street, but on the sidewalk. A tag like residential=crossing might be useful or not - who knows. I’m just saying that it might make sense to add it, mirroring the meaning of footway=crossing and cycleway=crossing 
And I’m getting that you’re saying that for you, it doesn’t make sense 
Friends?
1 Like
What I meant was: if you are on a crossing with a street that has an accompanying cycleway, then crossing this cycleway is treated the same as crossing the road itself, because they present a unit. From a “who has the right of way” perspective that is.
yes, I understood, this is why I wrote it depends on the local situation, as the “accompanying cycleway” is independent from the street, it can have different signage / right of way, although this is rare in Germany, it could still happen also there (e.g. by putting yield signs on the cycleway, or letting the cycleway pass right of a traffic light).
No these are not crossings. Clearly my attempt at defining crossing based on traffic types is not something that can be strictly applied 
. I wasn’t suggesting that any access restrictions would be implied by a crossing node, just trying to generalize which types of way intersections get a crossing node and which do not. A railway=level_crossing is another type of crossing we map. This is a clear case where the two crossing ways are dedicated to significantly different traffic types that do not mix. Meanwhile at the intersection of two streets or two cycleways, both ways are dedicated to the same general traffic type. And yes of course crossing nodes can also exist without a way intersection where sidewalks are not mapped as separate ways.
So can we maybe generalize that there are 3 types of traffic: pedestrian, road-based and rail-based? And whenever any of these 3 meet in a node, we have a crossing? Not even sure if this applies and what the gain could be, but I guess that’s how it could be defined…
Let’s take a step back here. We don’t actually need to agree on a precise definition of a crossing, not if the main goal is to decide how to tag a continuous sidewalk. Big picture: in OSM, when two roads meet, we call it a junction, and when a road meets a cycle path or a footpath, or when a cycle path meets a footpath and someone wants to get from one side to the other, we call it a crossing.
We need to decide whether we are happy with (1) a tag for continuous crossings, and leave the question of how to tag visual continuity at junctions for a later day; or whether we want to come up with (2) a tag or tagging scheme that is able to express visual continuity any time two highways meet, that is, at both crossings and junctions.
I am clearly in favour of (1) because I think it’s much easier to define and we only need a single new value (such as crossing=continuous or sidepath) or maybe a new key and two values (such as crossing:continuous=yes). By comparison, (2) is a can of worms. I don’t doubt that it would be useful for some data consumers to have this information, but it seems harder to define and there are a lot more questions to consider. And we already tag junctions differently from crossings, so would it need to be the same tag?
1 Like
Yeah something like that. It’s more complicated though.
Bicycles use both roads and dedicated cycle paths for example. Anyway, no need for us to further digress in search of a perfect definition. I was just trying to explain my way of thinking.
Which tag should we go with? Please feel free to vote in the poll even if you have not participated in the discussion so far.
- crossing=sidepath
- crossing=continuous
- crossing:continuous=yes
0
voters
Topic example:
There is a changing pattern on the footway, not on the cycleway.
Indicating a sort of line.
Just to make sure: are we refraining to make a decision on how to map driveways that go over (uninterrupted) sidewalks?
Or, if the scheme being voted accommodates that, can someone give an example of how it would look like?
1 Like
To me it still looks like a continuous sidewalk. Does the change in pavement have any meaning for the pedestrians or for the car drivers? Maybe it is only to mark the zone where the cars are allowed? It would be very unclear as marking for pedestrians to wait.
I think the proposal here is for continuous footways across streets, not driveways. In my opinion the intersection of a driveway and a sidewalk is adequately mapped as an untagged, connected node.
3 Likes
In Greater London, we have the TfLCID datasets, which include in the traffic calming data two broad types of crossings of side roads as “side road entry treatment” (TRF_ENTRY) [1].
-
Continuous crossings, locally referred to as “Copenhagen crossings” (the term used by London Borough of Waltham Forest), which have a continuous and unmarked sidewalk across the side road and usually no tactile paving. They also have a give way marking on the side road before it crosses the sidewalk, giving priority to pedestrians crossing the side road. In these cases, the continuous=yes tag might be useful when documented in the wiki with an accepted definition. Blended ‘Copenhagen’ Crossings - Enjoy Waltham Forest
-
normal pedestrian crossings with a speed table or flush kerbs. The pavement is not continuous, the give way markings on the side road are between the crossing and the main road. Tactile paving is often present, but not always.
These were all imported by the paid “mappers” under contract to TfL, who performed the import as sidewalk=yes + continuous=yes, without checking the imagery which is part of the dataset to determine the type of crossing. Without a highway=crossing tag, these would have provided little to no value to data consumers, even if they had connected the imported nodes to highway ways 
(I’ve re-tagged these as highway=crossing with a fixme, so that other users can finish the job.)
There’s a pull request [2] which was submitted after the TFLCID import fiasco was unceremoniously abandoned by the contractors, which is presumably why the maintainers have not seen the point in commenting.
[1] https://cycling.data.tfl.gov.uk/CyclingInfrastructure/data/points/traffic_calming.json
[2] Replace continuous=yes + sidewalk=yes with highway=crossing by rskedgell · Pull Request #52 · cyclestreets/tflcid-conversion · GitHub
1 Like
We have an overwhelming majority in favour of crossing:continuous=yes. I’ll write up a proposal to introduce the new key and two values yes and no.
This will include some notes on suggested usage, for example, there seems to be consensus that driveways and other minor service roads such as parking lot entrances that cross (uninterrupted) sidewalks are so unremarkable that the shared node should typically be left bare, without any tags.
There is one question left that we should maybe address. When you create a crossing in iD, JOSM and others editors, the first thing they typically ask is what the type of the crossing is, this sets crossing=* to some value. When crossing:continuous is yes, what should go in crossing=? Should we say that crossing:continuous=yes should be used with an existing value of crossing=* such as crossing=unmarked or uncontrolled? Should the proposal recommend that when crossing:continuous=yes is set, crossing=* should typically not be set at all? Or should we leave the question out of the proposal and just focus on the new key?
2 Likes
These continuous sidewalks could appear with any type of crossing, but the ones on all the pictures are unmarked so far. If you add examples to the new wiki-page, I would expect complete examples, including highway=crossing + crossing=unmarked + crossing:continuous=yes. Hope that makes sense.
1 Like
Since a continuous crossing may be marked with paint or not, and also may be signalized or not, automatically choosing a default crossing= value when the user specifies crossing:continuous=yes doesn’t really make sense. Probably better to add a continuous attribute to the existing presets.
2 Likes
Thanks all for your input!
I’m now creating a proposal at Proposal:Continuous crossings - OpenStreetMap Wiki
Let me know what you think. If the people here are happy with it I’ll move it to “Proposed” stage soon and send out announcements.
3 Likes