I have mapped in different areas across the United States. In many location, side sidewalks and marked & unmarked crossings had no name. In one location that will remain unnamed did have the sidewalks, crossings, and cycling lanes marked the same as the roadways. Unless a pedestrian street, should those be named except when actual names exist? I do include the cycling lanes in that question.
3 Likes
In my personal opinion, no - separately mapped footways, crossings, paths, cycleways, etc. should not duplicate in their name=* tag the name of the associated roadway.
name=* should only be set if that element itself has a specific name.
There are several different ways that this association has been mapped - refer to, for example, this previous OSM Forum thread: Naming sidewalks in Torronto, and the OSM Wiki pages for Relation:associatedStreet and Proposal:Key:is_sidepath, which links to many other related topics and discussion locations.
8 Likes
I recently updated the wiki page Key:footway - OpenStreetMap Wiki with currently used methods for specifying street name on the sidewalk ways.
Currently name is most used worldwide, but has the risk of other mappers who disapprove removing it later. is_sidepath:of:name and street:name are the next most popular tags, and because they don’t render, they’re unlikely to bother anyone. street and associatedStreet relations are somewhat lower down on the popularity list.
There isn’t an officially approved tag for this, so OSM’s “any tags you like” applies. You can choose one… or make your own if you really want.
7 Likes
Naming sidewalks is ALWAYS wrong
By definition, a sidewalk is used for walking by a side of the road.
The whole road has a name, not just the highway, but it’s the road that we tag to a have a sidewalk and then we don’t even have to map it separatelly…
So that’s the roason why we name the highway and not the sidewalk, tree lines, kerbs, area:highway or anything else.
Seriously, when people do whatever this damages the whole project and makes it less attractive for commercial use
2 Likes
I wholeheartedly agree, but that is not how the mapping community ticks like. Mapping a separate sidewalk indeed turns the “highway” into a “carriageway” instead. From my point of view nothing makes the “sidewalk” less capable of bearing the the name of the street. After all, historically they were the first to receive Pavement 
So mapping a separate sidewalk and not adjusting the “highway” to say, that it maps a carriageway only, indeed impairs data usability.
PS: 1) I consider “foot=no” on the highway/carriageway a bad manoeuvre (seen that, to force routers to use the separately mapped footway.) 2) There may be better keys to record the name of the street to use on a sidewalk than “name”. Could this be applied to “carriageways” just the same?
3 Likes
Those improvements to the Wiki page look really great. Thank you for putting in the work, compiling all of the references, and maintaining a neutral and objective tone despite the “controversial” topic and competing approaches. Well done!
3 Likes
History doesnt matter there, really. But if you want to go that far… all the ways we used to walk and maybe horse travel, we nowdays drive cars there and that’s it.
We name the central highway only because that’s the most basic geometry that is required to be.
Street names apply to an area btw, so:
Should we have one day highway:area mapped everywhere in detail, we could stop naming the central highway way and we map just the area, so that carto can choose where and how it wants to render that name..
But yeah carto…
solely because it’s so integral part of OSM we gonna map the central highway map only for it to render the name and that’s it. Really this is the only reason we tag it like that.
This is another discussion like that i see and again from US btw.
You guys should sort out that mess.
As for foot=no this should be detected automatically, but that’s just what we want and not how routings work. Some countries like mine forbit to walk the street if you got a sidewalk and you’re forbidden to corss a dual carriadgeway, so even with no sidewalk foot=no should be applied there by default. But again, country dependent
Nobody needs to navigate tree lines, kerbs, or areas which are explicitly non-routeable, so that’s hardly a reasonable argument. Routing software which gives useful directions needs to know the name of the street and it’s ludicrous to defer that in the hope that the software will one day soon (maybe even before the Sun becomes a white dwarf) accurately infer it from the nearest parent highway.
7 Likes
This has been talker through as well.
On complicated intersections noone was able to coherently say, which sidewalk as part of which street.
No rules on that are specified, so we dont tag name to a sidewalk.
Instead of trying to do it the only way you can do it yourself, talk to programmers and try to work out post processing of the map data, so it is accessible to routers just like you want. Because yeah, it’s a valid point
But if you keep mapping name to every sidewalk and area:highway (which you do should learn about and logically for your issue it requires a name as well), see that current OSm model does not support it and it’s going to be a mess they way you propose.
History aside, I do not see how consumers will ever sort this out. There are a small number of regions where people spent thousands of hours mapping separate sidewalks and consumers do not know whether a highway=residential is for vehicles only or for pedestrians too due to not recording the fact and there are regions where a highway=residential means, of course you can walk there due to no such mapping of separate sidewalks.
1 Like
I learned to be careful with such claims.
Somewhere there is a sidewalk with an actual name, separate from the road name.
12 Likes
here.
6 Likes
OT: What is the specific meaning of cycleway=sidewalk, insead of using =track/lane/separate (as stated in the wiki)?
=track/lane/separate go on carriageway and are dubious to be placed on cycleway mapped as a separate line
cycleway=sidewalk apparently has multiple meanings, see Tag:cycleway=sidewalk - OpenStreetMap Wiki but here it is clearly in “the cycleway is a separately mapped cycle track”
I was going for the fourth of the meanings that @Mateusz_Konieczny linked to - (“the cycleway facility for the parallel carriageway is supplied by the sidewalk”) but actually all 7 definitions arguably apply.
If separately mapped sidewalks are not to be named for the road then run alongside, how does a router describe them in the returned results?
I’m very much a visual user of OSM data, usually by Carto, but increasingly CoMaps instead of OSMAnd. An unshown line could easily be interpreted as not having a sidewalk, especially in areas where sidewalks are separately mapped. Can any renderer show the sidewalks appropriately from the tags some mappers insist should be applied to the highways only and not separate sidewalks?
4 Likes
Yes, for the same reasons names are applied to all carriageways of the road.
4 Likes
This. Without this information, consumers will reasonably treat it the same as any other unnamed road and not tell you the street name.
1 Like
note that software can also detect nearest matching roads if it wants to display this data
Why should this even be necessary in the first place given this is a thing? It’s bad enough we have this problem with road routes and having engrained this weird idea of checking an unrelated object (like individual ways) for the ref. Do we really want to double-down and repeat this problem with names on sidewalks? I see no compelling reason to do so.
I’m also seeing this as being potentially prone to error in boulevard situations where the frontage street has a different name than the boulevard, with the sidewalk between the two, and any time a cycleway or path of unknown name (thus untagged) runs down the median of a dual carriageway. So there’s a couple compelling reasons not to.