What about the way along the sidewalk centerline that functions as a crossing footway? I guess you could similarly step up from the intersection node and read the width=* of the driveway, but tagging that is very rare. In contrast, it’s quite easy to draw a new way from edge to edge and that also smoothly allows for curb nodes.
This was shared in that older thread: File:Tactile paving on sidewalk at parking lot entrance along Taylor Street, San Jose, California.jpg - Wikimedia Commons
This might be a crossing with table and not a continious sidewalk?
The wiki page gives the impression that crossing:continuous=* is only intended for ‘actual’ crossings over streets and not for driveway intersections (i.e. in combination with highway=crossing).
Driveway intersections like that should absolutely not be tagged the same as regular street crossings. It pollutes the data set and makes it impossible to accurately find the ‘real’ crossings. The primary purpose of the tag is not just simply signalling that this is a point where pedestrians and drivers might come into conflict, it has important additional implications.
Like @ezekielf said, finding intersections between pedestrian paths and other highway=*s can be done automatically just from the existing highway tags. If there is no other meaning in the tag besides that such an intersection exists, it doesn’t really provide any useful information.
I’ve split this topic of sidewalk-driveway crossings off into its own thread, since this one was supposed to be about crossing:markings=surface.
what is the difference between a “real” crossing and a driveway crossing? amount of traffic? Are we also going to use different tags where a road has very few traffic?
What about driveway crossings like this:
This driveway isn’t a service=driveway. There was a whole thread about it.
Anyways, for a minor service road, a better criterion for whether there’s anything worth tagging as a crossing is whether there’s any affordance for crossing – whether the bikeway or sidewalk gets interrupted in any manner:
Forum mods can afaik split threads properly and move posts to new threads.
@mods-general is it possible to move the posts starting from this one to a new thread?
https://community.openstreetmap.org/t/crossing-markings-surface/125774/27
(No idea if it’s somehow possible to integrate that with the manually created thread: Intersections of Sidewalks and Driveways)
These posts in between should probably stay here:
What do you suggest we name the new thread, and where do you think it should be placed (in which category)?
Is it possible to merge threads? This is the one I created to continue the discussion. Apologies for making a bit of a mess to clean up.
I can move a thread into a second thread, and it will then merge. Tell me which threads or parts of threads to merge, and starting at which messages.
That’s why the @mods-general are paid the big money. 
Thanks!
All posts in this thread starting from https://community.openstreetmap.org/t/crossing-markings-surface/125774/27 except
https://community.openstreetmap.org/t/crossing-markings-surface/125774/31 and https://community.openstreetmap.org/t/crossing-markings-surface/125774/33 should be split off and merged into https://community.openstreetmap.org/t/intersections-of-sidewalks-and-driveways/125873 if possible. I think the category is fine as it is.
The current wiki text is still exactly like this and telling you to use surface:colour instead. I didn’t invent anything new when adding the example.
I’m in favour of adding solid back as a proper value, why does it matter if it has any legal meaning? It makes the crossing look completely different and therefore deserves to be tagged imo
Done. 
Thanks, @apm-wa. To clear up any confusion following this merge, note that this thread is continuing below now from this post above.
I think that if the footway (usually, but not always a different surface to the roadway) continues through the connection between the ways, then there is no crossing.
Otherwise, locations like this: Node: 11229703947 | OpenStreetMap end up with vast numbers of crossings (18 crossings in 102 metres). Some locations similar to this, where each house has more than one side-by-side parking space at the front (essentially a continuous car park, with garden fences every second space) would result in the footway becoming a continuous crossing, potentially hundreds of metres long.
As a pedestrian, I would want to know where there are so many driveways, and would prefer to be routed on a sidewalk that doesn’t have those, if possible.
E.g. I would prefer to be routed on east side of Way: Sundale Park (46874652) | OpenStreetMap rather than on west side which per aerial imagery has driveways.
How should this be tagged then?
you don’t need an explizit tagging for that, you can simply use the information that sidewalk and driveways are crossing.
However, the linked example clearly shows how more detailed but incomplete information can destroy the routing. There are missing crossing options over the streets at the junctions and the access roads outside the properties can certainly be used to cross the street (so they are not private there).
No, a simple count of sidewalk-driveway intersections cannot communicate enough information to satisfy the outcomes I listed in this post above, while separately mapping them can.
No, this is an example showing that incorrect (access=private) or missing (no footway=crossing) information can break routing - which applies to all mapping and is not specific to how “not-crossings” impact the map.
By missing crossway information, do you mean that crossways that physically exist in some way have been omitted, or that the mapper should have mapped “places where people are likely to cross the road” as crossings?
E.g. at this location the router adds a pointless detour down a cul de sac, whereas a real person would likely cross where I have circled in red. But it looks to me that there may be nothing special there that indicates a crossing.
