I apologize for tossing that term around cavalierly. Perhaps a better way to put it is that, if we’re already able to identify highway=crossing tags that occur because of iD’s former validator suggestion, then we can just as well remove those tags.
I also want to point out that I’m not really going out on a limb here. The one-click addition of this tag was one of the most frequent complaints about iD within the mapping community for four years. The original topic of this (already concluded) RfC was all about clarifying the situations that had to be mapped as crossings for other reasons.
Meanwhile, anyone who does want to identify every point of conflict between a footpath and a roadway, no matter how minor, can already do so. I don’t really know what problem we’re trying to solve that the existing solution can’t solve.
(Since tone is hard to communicate in text, I’ll clarify that I appreciate the back-and-forth as we figure out the best way to contribute )
I think it might be helpful for my understanding for there to be a clarification on what exactly defines a crossing, and why you think these don’t fit that definition.
At least in the way I use the words, I definitely wouldn’t think of sidewalk-driveway intersections as “crosswalks” but definitely would think of them as “crossings” and, regardless, I’d appreciate input on how to add information to those locations and also make it clear that they are not what anyone would consider typical crosswalks.
A router or other data consumer can see from the fact that a driveway and a sidewalk connect that there is some potential for conflict there.
The driveway-sidewalk node fails the “duck test”: most of them wouldn’t be called pedestrian crossings by someone unfamiliar with OSM.
In OSM many of them are tagged highway=crossing only because of a validator rule that has since been removed because people complained.
What a bare node doesn’t tell you is what this intersection looks like: for example, whether a blind user should be warned about a kerb.
We could just tag the above using two barrier=kerb nodes on either side of the shared node without tagging the shared node… @Minh_Nguyen, is that what you’re suggesting? Then how would you suggest tagging the absence of any obstruction? That’s the problem you are trying to solve @Lumikeiju right?
OK, that might be a source of the misunderstanding: crossing is an example of OSM’s preference for British English. The Vienna Convention heavily influences discussions about navigation mapping; it uses the term “crossing” where Americans would refer to a “crosswalk”.
Actually, I’ve been under the assumption that any characteristic of a pedestrian crossing – markings, a sign, curb cuts, tactile paving – would warrant a highway=crossing node and a highway=footwayfootway=crossing way. This includes the green bike path markings in Seattle, but not the vast majority of driveway–sidewalk connections.
Just to make sure we’re all on the same page, here are some pictures of driveway-sidewalk intersections that I have in mind when I say I wouldn’t really call them a pedestrian crossing*:
*I am using pedestrian crossing here as the British term for what I imagine an American might call a crosswalk (although I’m not familiar enough with the term crosswalk to confidently say they’re equivalent). Here is what the Highway Code has to say about crossings.
(Some confusion arose in the original RfC because some people were using crossing in a more mathematical sense, to mean intersection, and then of course they are “crossings” between two ways.)
Oh, to clarify, the markings I’m referring to are cycleway markings, not footway markings. They’re added by SDOT where cycleways cross certain service roads. The adjacent continuous crossing of the sidewalk across the driveway isn’t marked. I brought up these markings to highlight that we recognize that crossings (places where the path of a pedestrian/cyclist intersects that of a car) are areas of interest for safety reasons even though they are not crosswalks.
Interestingly, the qualifications for these markings are apparently the same as the mapping guidelines developed for these types of crossings in Quebec (refer to this post in the GitHub issue for iD adding crossing tags at these spots)
[SDOT] Driveway approaches: at driveway crossings with 5 or more housing units and commercial buildings, a green panel shall be painted for a PBL crossing a driveway shown above in Figure AO.
Though I guess they don’t follow their own instructions because I’ve seen these markings on non-protected bike lanes, but as I understand it, “paint is/isn’t infrastructure” is a whole 'nother can of worms
Guidance: Engineering judgment should be used to establish intermittent breaks or interruptions in the buffer space, such as for driveways, transit stops, or on-street parallel parking lanes, in order to convey access points or an otherwise general legal movement to cross the buffer space (see Figure 9E-6).
Figure 9E-6. Examples of Markings for Buffer-Separated Bicycle Lanes
…
I’ve been trying to consistently map these interruptions as a highway=crossing node and highway=cyclewaycycleway=crossing way. Obviously, it isn’t a crosswalk, but the markings do distinguish it as a bike crossing. Regarding the topic of this RfC, I don’t think crossing:continuous=yes would be appropriate in this situation, since the bike infrastructure is literally interrupted by the driveway.
This is essentially the issue I’m facing. I appreciate that crossing = crosswalk != intersection (and thanks all for describing this conflation explicitly), but I was hoping this tagging scheme would allow me to tag sidewalks whose intersections with driveways are not just more accessible because they lack curbs, but actually safer, because they signal to everyone involved (through e.g. surface continuity and raising the driveway to the sidewalk’s level rather than lowering the sidewalk to the driveway level) that that space is primarily meant for pedestrians, and only secondarily for vehicles.
At least the first two (the third image is a bit hard to understand) are precisely the kinds of intersections I would like to explicitly map as being continuous for the pedestrian.
Am I misinterpreting the intent of the crossing:continuous=yes scheme? And if so, what should I use instead?
It seems to me that crossing:continuous=yes would definitely apply to the situation where traffic_calming=table would also apply from the driver’s perspective. But if there isn’t a table and both the driveway and sidewalk continue uninterrupted at the same level, would that really be a continuous pedestrian crossing only?
If we agree that these examples aren’t pedestrian crossings per se, then you could distinguish them by their lack of a highway=crossing tag, or maybe by a not:highway=crossing tag. But not crossing=no, which awkwardly means crossing is prohibited.
That was the intent of crossing:continuous=yes: There is no interruption of the sidewalk. Not the pedestrians are crossing the carriageway, but everyone else is crossing the sidewalk.
The documentation gives this very driveway and sidewalk as an example of a connection that should not be tagged as a crossing, let alone crossing:continuous=yes. This is the guidance that is now being challenged. Did the community overlook or misunderstand this guidance when it was proposed in this thread back in November and subsequently approved?
The reason for this was, that by default, driveway/footway-crossings have continuous sidewalks, and we wanted to discourage a mass tagging of all these non-crossings with crossing:continuous=yes, which is why it’s discouraged to do so.
On top topic of marked service road / driveway crossings: I’d only use an empty node (or at least without any crossing information) on a sidepath-driveway crossing if only if it would be crossing:continuous on any other roads. For anything else, I’d go ahead and mark the sidepath with markings / changed surface as a crossing way.
Part of it because a driveway isn’t a regular street (e.g. in the context of Germany, street restrictions on the highway still apply on despite the lack of a nearby signs which would exist at a street intersection) and also because visible driveway crossings are the exceptions, not the rule unlike for any other street where the sidewalk is to be expected to discontinue.
Oh, interesting. Is this assumption documented somewhere? It never occurred to me that people (and routers) would have this expectation, perhaps because in my vicinity it’s very often not the case. I’m also not sure how common this is around the world.
When the road that crosses a continuous sidewalk is a minor service road, such as a driveway or a car park entrance, there is community consensus that this is not worth tagging explicitly
Which means: if a minor service road, such as a driveway, crosses a sidewalk without any crossing-tags, then it’s a continuous sidewalk.
If it’s the other way around (the sidewalk crossing the road), then add highway=crossing, and so on.
also around here it is usually not the case. The sidewalk gets interrupted for almost any driveway, and you can feel lucky if the kerbs are lowered where you cross, which is now usually done, but only since a few years in my area, so that most places have 2 raised kerbs for every driveway-crossing.
I’m sorry, but I really don’t see how “this is not worth tagging explicitly” implies “it’s a continuous sidewalk”. I mean, I do understand the reasoning, but don’t think it’s nearly explicit enough in the current phrasing. If that assumption is to be documented as an established community consensus (which, to be honest, I’m not even sure it should), then IMHO it would have to be stated directly.
Please let’s not expand the meaning of highway=crossing, footway=crossing, and cycleway=crossing to include every intersection between a residential driveway and a sidewalk or bike path. Many such driveways serve just one house with just one or two vehicles and thus the potential for a vehicle/pedestrian/cyclist conflict is very low. The experience is quite different from crossing an actual road. If some people are unhappy leaving these intersection nodes untagged, please come up with a new tag. Maybe something like highway=driveway_crossing (just don’t choose highway=crossing2). Please don’t dilute the meaning of existing tags that are for actual road crossings.
I totally see your point for residential driveways, but there are also a lot of driveways that go into apartment buildings, as well as commercial and industrial buildings — essentially, places where it’s expected that there will be non-trivial amounts of vehicle traffic. These intersections definitely should be taggable to convey whether they veer towards prioritizing cars or pedestrians in their physical design and presence/absence of barriers for each type of user. (I’m not saying that highway+crossing+crossing:continuous is definitely the way to do it, but from @Nadjita’s description it seems to be.)
Nothing prevents you from tagging high-traffic driveways (maybe they aren’t driveways then, but that’s a different thing) with the appropriate tags. In this context, crossing:continuous=yes simply means: the road goes over the sidewalk. The fact, that it’s discouraged to do so, should not stop you from using it for places that you’d rather want to call a crossing.
)
