footways* that can be used by bicycles in only one direction, but in both directions by pedestrians should be tagged as highway=footway,oneway=yes
footways* that can be used in only one direction by both bicycles and pedestrians should be tagged as highway=footway,oneway=yes,foot:backward=no or highway=footway,foot:backward=no,bicycle:backward=no, even when bicycles aren’t allowed at all (like for via_ferrata).
*) and the same for cycleway, bridleway, pedestrian, path, steps, corridor, via_ferrata, platform, trailhead
I can somewhat follow the reasoning behind this, but isn’t this, say, a bit ‘academic’? The first category seems more like a rare exception for these kind of highways while the second category appears much more frequently, yet it requires much more complicated tagging this way. Out of all the ways tagged as highway=path,oneway=yes currently probably only a small fraction actually falls into the first category. So is it even realistic to establish this strict ‘vehicle-only’ meaning for the oneway tag? What is currently done to establish these rules among mappers? Does anyone believe this ‘correct’ meaning of oneway will be known by the majority of mappers anytime soon?As it stands now the data is very hard to use, because the tagging cannot really be trusted (can a way tagged as highway=footway,oneway=yes actually be used in both directions by pedestrians)?
Btw I did not mention horse here, but this might be another thing to consider.
Update: After the following discussion I became aware that ways that are oneways for bicycles but not for pedestrians are actually rather common and not as rare as I thought.
I don’t follow this example. You say “can be used in only direction by both bicycles and pedestrians”, but also “even when bicycles can’t be used at all” which seems to contradict the first part.
My experience and the reality in my part of the world (western Europe) is exactly the opposite to yours: I know many ways that can be used in both directions by pedestrians but only in one direction by bicycles. On the other hand, I actually don’t know regular ways that are a oneway for pedestrians (apart from things like access ways to oneway doors or waiting queues and obviously the via_ferrata you mentioned).
The tagging you describe is unnecessarily complex: highway=footway, foot:backward=no is sufficient to mark a foot path as oneway.
highway=footway, oneway:foot=yes is another valid option.
On a highway=path you need to add a bicycle tag to exclude bicycles, but a simple bicycle=no without a direction suffix is sufficient.
Ok, interesting. Can you share some examples? And if you had to guess, how many of the footways (or paths) currently tagged as oneway=yes can actually be used in both directions by pedestrians?
The tagging you describe is unnecessarily complex: highway=footway, foot:backward=no is sufficient to mark a foot path as oneway.
If it is a oneway for pedestrians it is most likely also a oneway for bicycles (if they are allowed), so we either need bicycle=yes,oneway=yes, or bicycle:forward=yes, don’t we? Also if we were talking about cycleway instead of footway we would need the tags I mentioned
On a highway=path you need to add a bicycle tag to exclude bicycles, but a simple bicycle=no without a direction suffix is sufficient.
Yes, if bicycles aren’t allowed there. If they are (but only in one direction) we need the tags I described, no?
Btw out of the current 3656 ways tagged as highway=via_ferrata there are exactly zero tagged as foot:forward/backward=no. There are six tagged as oneway:foot=yes, and 1787 tagged as oneway=yes. Actually overall there are only 86 (265) highways tagged with foot:forward(backward)=no, but there are 56706 paths tagged with oneway=yes.
I think highway=path is a different case to footway/via ferrata/steps. It is so generic that it’s hard to interpret without looking at specific access tags. That question would probably require breaking it down between various combinations (no access tags / bicycle yes or designated / foot yes or designated).
For footway, steps, via ferrata on the other hand I imagine oneway=yes usually applies to pedestrians.
I just had a quick look around Frankfurt: overpass turbo
Out of the >300 footway (and >700 path) the search shows it seems >95% are legitimate oneways for bicycles only.
I only found one exception: the not correctly tagged entrance Way: 592490633 | OpenStreetMap . There might be more errors, but they are really rare.
Right.
Correct. According to the general definition of oneway to exclude pedestrians this is wrong. For some reason oneway=yes is the recommended tag according to the Wiki for via_ferrata. I tend to say that this is some kind of edge case where oneway=yes can be used without ambiguity. There’s only one type of traffic possible on such a “highway” after all (although seeing a horse there might be funny).
For footway, how does this arise - is this something to do with that German signage I have seen mentioned elsewhere, that indicates that a way is primarily a footway but bicycles can also use it? Otherwise the idea of a “footway” having signs for bicycles seems counter-intuitive.
Thanks, so you’re right and this is not nearly as unusual as I thought. A common case seem to be ‘sidewalks’ to be used by bicycles and pedestrians where bicycles need to stay on the right side of the road while pedestrians can use both sides.
Yes, it is an edge case, but since this matter is already quite intrigue it would help to not have an extra rule for via_ferrata. One could easily think the same about footway as it very much sounds like it is only used by pedestrians. Looking at highway=footway,oneway=yes I doubt many people recognize the oneway tag is only valid for bicycles (if they are allowed). Also for some paths (in the mountains for example) it might seem so obvious that they are only accessible for pedestrians that oneway=yes sounds like the right choice.
Yes, exactly. You can either have a shared foot/bicycle way with equal rights to both parties. Or there’s a footway that allows bicycles, where pedestrians have priority and bicycles need to go slowly.
I also tried to search in some more rural areas, but there is hardly anything to be found. In the Alps I found a few footways with oneway tag which actually seem to be oneway for pedestrians.
As a rule of thumb:
As a mapper, I would always use oneway=yes for vehicles only and add explicit tags for pedestrians.
As a data consumer, I would treat oneway tags on ways that are really pedestrian only (footway without bicycle tags, via_ferrata and similar) to restrict the direction for pedestrians.
I think that this keeps the ambiguity in the database as low as possible, and interpretes the data we have as correctly as possible in most cases.
Yes, it certainly depends on the highway type. But footway is not so clear as it is often used for ways that allow bikes as well.
In the Alps I found a few footways with oneway tag which actually seem to be oneway for pedestrians.
These cases certainly exist. Since there are so few ways with foot:forward/backward=no and foot:oneway=yes there must be many that are ‘falsely’ tagged as oneway=yes.
This sounds right. I’m just afraid that the ‘vehicle-only’ meaning of the oneway tag is not sufficiently known among mappers. Recommending it explicitly for via_ferrata certainly does not help with this. Maybe the wikis for via_ferrata, steps and footway should explicitly state that foot:backward=no should be used to restrict the walking direction (while oneway=yes is discouraged)?
It would certainly be difficult for anyone editing in ID to become aware of this. For path, footway, and steps One Way is listed in Add Fields and brings up the oneway= tag.
I don’t know whether the vehicle-only meaning was ever a true reflection of actual practice or only an aspiration. It’s possible that on footway, oneway=yes (at least in the absence of specific access tagging) was understood to apply to pedestrians from the start, much as it still is for via ferrata. But I haven’t looked into the history of the relevant wiki pages.
I don’t know whether the vehicle-only meaning was ever a true reflection of actual practice or only an aspiration.
Yes, that’s the problem. Part of it is also that the ‘vehicle-only’ meaning of the oneway tag is not explained convincingly in the wiki(s).
The oneway wiki currently says:
In the past highway=path+oneway=yes was used by some mappers to mark that pedestrians may move only in one direction, these situations should be corrected, as oneway in OSM only applies to vehicles
“In the past” does not seem like it reflects reality. These tags are still used like this and weren’t corrected. Otherwise there would have to be more foot:backward=no or foot:oneway=yes tags. I found >9000 footways tagged as oneway=yes without a bicycle tag. Also “oneway in OSM only applies to vehicles” isn’t all that convincing. Just looking at steps, aerialway and via_ferrata we see that this rule is not strictly applied in other cases, so why should one follow it elsewhere?).
The counterexample is highway=pedestrian oneway=yes. These are by definition wide enough for vehicles. It’s also common for them to allow buses, taxis, and/or loading at certain times of day.
What I imagine happens quite a lot is: A mapper sees a pedestrian street so they tag highway=pedestrian. They see a oneway traffic sign (that is clearly only meant for vehicles) so they tag oneway=yes. Looking around European city centres with the help of Overpass Turbo, that seems to be a fairly common scenario - far, far more common than pedestrian streets that you are only allowed to walk down in one direction. If we were to invent a rule that oneway applies to all traffic, you could equally argue that it’ll never catch on, citing highway=pedestrian as an example. If a router (such as Valhalla) refuses to route pedestrians against the oneway direction, that’s a misinterpretation of the tag.
Practically, if someone has tagged a highway=via_ferrata with oneway=yes, chances are they meant that climbers are only allowed to use it in one direction - they didn’t mean vehicles. If someone has tagged highway=pedestrian with oneway=yes, chances are they meant vehicles, not pedestrians. Not sure about footway - that requires looking at examples of how people have actually used the tag.
Looking a bit more at the wiki history, I think that from roughly 2013 to 2019 the page implied that oneway=yes could apply to pedestrians for some highway types, including the words “Although unusual, oneway on pedestrian highways (path, footway, track) is possible in some countries, for hiking trails for instance (during the high season crowding or for security reasons)”. For much of this period, there was even a photo example of a hiking trail tagged highway=path, oneway=yes. No other form of tagging was suggested for these cases.
Only in 2019 was an addition made suggesting different tags. And the suggestion then was oneway:foot, not foot:backward - although the latter was introduced a few months later I think.