I’ve been using the semicolon-delimited approach: street:name=ABC Street;XYZ Avenue.
2 Likes
I’ve been leaving them with no is_sidepath:of:name tag
2 Likes
Me too, but with is_sidepath:of:name.
¿Porqué no?
Where the two crosswalks meet closely I try to be mindful of keeping the sidewalks going to the curb (and tag the curbs), with an intersecting node between the sidewalks at roughly the centrelines of the pavement. I tag each sidewalk segment with its respective is_sidepath:of:name, so that a router would reasonably pick up on the intersection of the two. For example at the northeast corner of this intersection:
I would tag the E-W sidewalk (“1”) with Eglinton, tag the other (“2”) with Laird, but would extend the Eglinton sidewalk west (in brown) to the curb (the black dot).
Where you’ve got a bigger distance between the crosswalk curbs I would tag the intermediate segment with both street names, because… well, they’re “sidepaths” to both at that point. 
2 Likes
Hm this is actually quite convincing to me. And then a smart router could do something like not announce a change in street if you are coming from and return to one of the two streets listed. I was mostly concerned that routers would just announce “turn slightly right on Laird Drive Eglinton Avenue East”, but if you have both this is a very tractable problem!
I’m interested to hear from more community members on this. If there’s a clear consensus in Toronto I’ll start tagging like this.
In Quebec, I usually keep these with no name.
We’re now at 28000 sidewalks, crossings, and bike tracks tagged with names, and 2350 unique names. This is largely through the astounding effort of @bphennessy, thank you!
5 Likes
Almost all of the old City and York are covered now! I also found a whole bunch of areas that never had their sidewalks mapped while doing this, so this has been a great excuse to fill in those gaps in the data.
5 Likes
A quick reminder to update the sidewalk tags on the roadways as you do so. Where sidewalks are drawn separately, the roadway should have sidewalk:both=separate (or :right=separate + :left=no or the other way around).
2 Likes
Thank you for all your work!
But, I simply have to ask (actually a question for @bphennessy and not for Jarek): why exactly did decide to use is_sidepath:of:name=* instead of street:name?
After skimming through the thread again, Jarek asked in the first post and in the following discussions they said they will use both and afterwards it looked to me like most would support street:name=* over is_sidepath:of:name=*. Also with my post where i tried to explain my reasoning, with some upvotes.
I did not read someone favouring is_sidepath:of:name=* (maybe i missed something, quite possible :D )
And then you went ahead and used the is_sidepath:of:name=*.
I’m curious as to why. Again, i really don’t want to diminish you work or anything, i want to learn why.
Other people had already started mapping when I jumped in and I just used the same tag they were using.
2 Likes
Hi @Negreheb , I had noticed your thumbs-down on my celebratory posts earlier and was curious… Glad to finally find out why 
I used it because it was more widely used at the time, and my impression at the time was that people favouring street:name had an opinion but hadn’t done much work. And also because Any Tags You Like and I liked it more.
To this day is_sidepath:of:name is more widely used once you discount the uses of street:name on street parking areas.
If someone does the work and brings a proposal for street:name for sidewalks and MUPs and it is accepted, it will be an easy mass retag from is_sidepath:of:name and I won’t have a problem with it.
2 Likes
I agree footways are part street and inherit its name, same as road. I just wish we had better technology for it, like associatedStreet Relation where all the meta-data tags go and in addition to all the member 
1 Like
Would it be possible to summarize the choice for Canada in the Canadian osm wiki, and link to this discussion? Before doing this for the whole of Quebec (using is_sidepath:of:name or street:name), I would prefer to have a page in the Canadian wiki for reference in case someone complains or ask why we chose the sidewalk street name tag. I will translate it to french after its creation. Thanks!
2 Likes
Looks like the sidewalk ways in Quebec have name=* tags, per footway=sidewalk | Tags | OpenStreetMap Taginfo Quebec ? If no one’s complained about that, I wouldn’t retag to or double-tag with is_sidepath:of:name - semantically it means the same thing, and software support for name is bound to be wider.
“Support” isn’t binary. Obviously, if Québec’s sidewalks go from name=* to is_sidepath:of:name=* or street:name=*, something will change: most renderers will drop the labels and most geocoders will stop indexing the ways by name. That’s presumably by design, not necessarily an indication of poor support among renderers and geocoders. If you think renderers really ought to label is_sidepath:of:name=*, see how stylesheet authors would react to a feature request along those lines.
1 Like
I don’t really want to rehash the entire first half of the thread. I’ll just say that I’m now using is_sidepath:of:name not because I think name is incorrect on sidewalks, but because I care more about getting data not deleted from OSM and getting routers to use it than I care about being precisely right.
But I’ll just say, I think name is exactly as valid on Way: Rue Berri (974396950) | OpenStreetMap as it is on Way: Rue Berri (27208094) | OpenStreetMap.
Nor do I. Just wanted to be clear that this is one of many attributes that can be “supported” by doing nothing, depending on the use case. Basically, @ChaireMobiliteKaligrafy, don’t let the lack of apparent support in renderers dissuade you from applying the alternative keys. I also invite you to participate in some of the recent discussions elsewhere about how to address diagonal connections and other nuances.
I was also referring to the lack of support in any pedestrian routers that haven’t been told about is_sidepath:of:name or had a chance to implement it yet. OsmAnd is just one example I know of. Or any other kind of data consumers.
I really don’t care about the renderers very much in this particular case, because on a graphical map the connection between a street and a sidewalk is more obvious than in most other applications of OSM data.
1 Like
So I suggest keeping name for now in Quebec and change them later when routing engines will have broader support for is_sidepath:of:name or street:name. There is no urgency at all here 
. Right now we prefer removing names on small segments at intersections (diagonals).
i know there isn’t much i can say to change anyone mind, but try to see what @Minh_Nguyen is trying to point out.
Don’t map things that has little sense in the grand sheme of things
For instance, take a look at our rival rendering:
Not even all street names render, coz that would clutter things. If there is a carridgeway and its named, that completelly suffices. No reason to double that to multiple strips of sidewalks, cycleways, tram tracks and whatelse.
If you need it for routing, make sure it actually works and make it bulletproof
because if any commercial routing is going to use your mapping, they’d like to solve any bug reports of their consumers and if the logic behind your mapping prohibit that coz of faulty logic design.
They won’t implement that and rather cast names of nearby carriadgeways for those sidewalk routing.
Just saying.
think it through