Routing at highway=crossing with foot/bicycle/horse=no

General talk Tagging general discussion
, , , ,
42

In my opinion any access tag on a crossing node is not needed, not wanted and superfluous as all “access” should be tagged on the crossing ways.

and if the crossing is only mapped as a node and sidewalks aren’t mapped explicitly?

43

Either map them in that area or come up with specific tags like sidewalk:crossing:foot=no. But access is the wrong tagging for such a situation.

But I would be curious about such a real life situation. Since usually as a pedestrian you can cross a road everywhere as long as there is no physical separation and as soon as there is a physical separation, you should map the sidewalk explicitly.

44

This is not an example of an editor “enforcing its own standards”

It is the established convention for a path-road junction to be connected, and for that connection vertex to have a highway=crossing tag + crossing=* refinement. I don’t know why @Minh_Nguyen decided to muddy the waters by bringing iD’s validator into the conversation. When we built this feature we considered all of the existing documentation, best practices, and what kind of prompting would lead to users make helpful edits.

1 Like
45

I probably should have been more explicit in what I meant.

So because 0.27% of the crossing nodes is tagged with bicycle=no/dismount, that is a good idea? I do think that is a good idea but that has been reverted.

I think I am also on the “descriptive” camp but I use statistics to see what makes sense.

If you look at highway=crossing#combinations you do not see bicycle/foot on the first page while the page does not list tactile_paving=*, kerb=*, button_operated=*, traffic_signals:sound.

That makes to me the adding of the bicycle key to the page not something that is descriptive but prescriptive.

Good idea to come up alternative to tag if a crossing if for pedestrians or cyclist or both. I do not see added value in it but if it keeps mappers happy that want to map this it is better than using the “access tags” it is a plus.

46

The thing is that the crossing-proposal explicitly mentions bicycle=* and horse=*, and since crossing=* is a refinement of highway=crossing, it does make sense to put it there as well.

But if we all agree that using access-tags to denote who’s allowed to cross isn’t the best idea, I think we can now switch to the point where we decide where to go from here. I’m sorry if my posts were a bit emotional and angry. I do value @Minh_Nguyen’s contributions to the wiki, and I can understand how this specific one came to be. Please accept my apology, and let’s move on to finding a solution.

3 Likes
47

Yes, it is a very small number, although we probably all agree that these are likely mistakes, good candidates for retagging to crossing=no.

Purely to say that I didn’t dream up the idea of highway=crossing for cycleways one day and put it on the wiki in a fever dream. I’m quite sure it was already common practice before iD introduced the crossing ways validator rule, but it’s hard to prove that with an Overpass query. Maybe there’s an old mailing list thread about it or something.

Since the 2008 proposal predates footway/cycleway=crossing ways by several years, it clearly intended to apply access tags to highway=crossing nodes. The approval means we probably shouldn’t quietly delete that recommendation, but we could add guidance about how the practice has become somewhat rare (compared to the use of highway=crossing without access tags) and is problematic for routers.

As I mentioned early in the thread, I think these access tags are largely unnecessary in practice. If a bicycle-exclusive cycleway crosses a road, that cycleway will be mapped as a way even if sidewalks are only modeled as tags in the area. A router isn’t going to tell a pedestrian to cross there, and it probably wants to warn the driver on the road to beware of the crossing regardless.

The 2008 proposal’s inclusion of these tags was really just a way to avoid losing information from crossing=toucan and crossing=pegasus in regions where crossing_ref=toucan/pegasus are inappropriate, but eventually it got conflated with access tagging in general. The vote happened at the same time as the one for highway=path that commandeered access tags as a replacement for highway=*way tagging, so it isn’t so surprising that we got that wrong too. All we need to do is downplay the significance of that recommendation and move the access tags to the ways where appropriate.

2 Likes
48

From this numbers we don’t know whether the cycle path, bridleway, or road there allow pedestian usage. E.g. It’s still possible that there are cycleways with foot=yes, but the intersection is for cyclists only. If highway=crossing is for pedestrians or cyclists or …, crossing=no should be interpreted to mean that no one is allowed to cross here.

overpass turbo highway=crossing with foot|bicycle|horse=no → 22233 nodes

49

That’s true, and I view that query of mine as less informative than:

50

It would IMHO also be quite nice if JOSM (and iD?) validators would warn about the problematic tags (like bicycle=no/dismount on highway=crossing nodes).

2 Likes
51

That would be doable, probably after revising the documentation.

2 Likes
52

I did remove the one bicycle=no from the page.

Any objections on forehand if I further edit the page?

53

I removed the suggestion to tag “access”=no
Tag:highway=crossing: Difference between revisions - OpenStreetMap Wiki

5 Likes
54

By the way, in the future, when questioning a wiki edit on a wiki talk page, please tag the user in your comment so they can see it. Few users have every page they edit in their watchlist and monitor their watchlist religiously. This is a common reason why talk page discussions stall or become one-sided.

You can tag a user by adding {{ping|User name}} or a link to their user page. It’ll show up in the notification list as long as you add it at the same time you sign a comment with ~~~~. Of course, pinging the user here on the forum works too. :blush:

3 Likes
55

I did update the page further to indicate access-tags are better done via the crossing way.

4 Likes
56

Hi @Langlaeufer ,

Germany :germany: should be good now.

Some of the fixed foot-navigation errors:

I did not fix bicycle and horse though.

57

Sind noch 1157 Kreuzung mit bicycle=no oder horse=no

58

the same as previously, though I got more convinced that it is a really annoying tagging scheme