Which one is the better tag to say that a way is oneway for pedestrians?

osmuser63783 · 2024-08-22T18:20:21.983Z

The tag oneway can be ambiguous on highway=footway and =path: Did the mapper mean oneway for pedestrians or did they mean vehicles (usually bicycles) who may be allowed to use the way?

There’s been a lot of discussion on this and now I am thinking of writing a proposal (following the proposal process) to discourage the use of oneway on footway and path, and promote the use of existing unambiguous alternatives. For the full rationale and an informal poll on whether such a proposal would be a good idea, see the existing thread here.

To say that pedestrians can only use the way in one direction, there are currently two alternatives. For the proposal, we should choose one of them, this will be the one that we can recommend to mappers, and also to editor developers to use in presets.

Which one is better?

oneway:foot=yes
foot:backward=no

0 voters

aighes · 2024-08-22T19:19:27.145Z

Maybe see Key:oneway:bicycle - OpenStreetMap Wiki for reference.

westnordost · 2024-08-22T19:48:11.742Z

The proposal link leads to the general proposals page.

osmuser63783 · 2024-08-22T20:56:28.715Z

Yes, that was intentional, I haven’t written the proposal yet. I am just trying to gauge interest for now.

I’ve edited the post to make this clearer.

Tordanik · 2024-08-23T12:08:02.226Z

Conditional restrictions – and, therefore, foot:backward=no – have already been approved and would not need a new proposal.

What I dislike about oneway:foot is that it appears to also use the syntax of conditional restrictions on the surface, but subtly breaks the inherent logic of that tagging system:

With conditional restrictions, appending a suffix (such as :foot) to a key generally restricts its meaning to a certain group of users or a certain direction. But because pedestrians aren’t included in oneway=* according to the current definition, a hypothetical oneway:foot key would have to expand the meaning of the base key.

Hypothetically, it would be possible to bring oneway:foot in line with conditional restrictions without this special exception. But that would require dropping the built-in assumption that oneway doesn’t apply to pedestrians and always make it explicit who the oneway restriction applies to. That is, use oneway:foot=yes for the use case discussed in this thread and oneway:vehicle=yes for the common traffic sign.

Langlaeufer · 2024-08-23T12:30:23.704Z

oneway:foot pro:

oneway:foot is easy to understand.
oneway:foot is 5 x more used than foot:backward/forward.

foot:backward pro:

If oneway is only effecting vehicle oneway:foot does not make sense. (see post of Tordanik)
foot:backward=no is already documented as the way how to do it.

skyper · 2024-08-23T12:57:23.212Z

How about cycleway? There are countries where a combined sidewalk + cycle track is tagged as cycleway, right? So we need to include it and possibly even bridleways.

SomeoneElse · 2024-08-23T13:01:48.765Z

The “elephant in the room” here is that I suspect that a number of highway=footway;oneway=*;some_other_mode!=no might actually not the best way of tagging a particular feature at all, and only a proper survey will fix that.

The obligatory missing option (!) is just to leave the oneway if no other modes are involved since it’s the least confusing to all concerned.

Langlaeufer · 2024-08-23T13:22:07.260Z

The question here is: What is the better solution for oneway for pedestrians?
This can be used with all kinds of highways (paths and streets).

skyper · 2024-08-23T14:00:35.889Z

You are right, I got confused that despite the title only path and footway are mentioned in the description.

I think it would have been better to only link to the poll in the other topic as now the discussion is split into two topics and I only can quote myself:

Effect of oneway on pedestrians?

Well, I hope after 183 comments that it is clear to everyone that considering oneway=yes for pedestrian is most of the time wrong and that highway=* which are only allowed in one direction for pedestrians is a rare case.

Effect of oneway on pedestrians?

Effect of oneway on pedestrians?

and fix it with less ambiguous tags

Er, no. highway=footway; oneway=yes is not ambiguous.
It only starts becoming ambiguous when you start saying that “other modes of transport can also access that footway” which is very much the exception rather than the rule.

Even if we find some situations where it might not be ambiguous, I have to ask if we want to make exceptions (plus probably exceptions from the exceptions) or if we want to make it clear and consider oneway=* only for vehicles (plus horses?).

Regarding editors, presets could include the information that oneway=* is not for pedestrians and offer another tag.

Regarding oneway:foot vs foot:forward/backward, I would prefer a clean cut and use the latter.

osmuser63783 · 2024-08-23T16:25:10.885Z

Tordanik:

pedestrians aren’t included in oneway=* according to the current definition

Thanks! I can see where you are coming from. If we assume that the oneway tag only ever applies to vehicles and that it is nothing more than a translation into OSM tagging of a traffic sign that, as traffic signs do, only applies to vehicles, then of course a one-way hiking path in the mountains with the oneway tag has simply been tagged wrong and needs to be corrected.

This is precisely the opposite of the argument that @SomeoneElse is making: that oneway on a footway for a one-way hiking path is perfectly fine (unless there is also a bicycle=yes or similar, then it becomes ambiguous).

We discussed this quite a lot in other thread and a lot of examples came up and I came to the conclusion that there is no clear majority for either view, so I am proposing a compromise: Avoid using oneway on footway and path altogether, and use a more specific tag wherever possible. But which one is the more specific tag that we should recommend? That’s what this poll is about.

SomeoneElse:

The “elephant in the room” here is that I suspect that a number of highway=footway;oneway=*;some_other_mode!=no might actually not the best way of tagging a particular feature at all

That’s exactly what I am trying to address. Ideally when someone tags highway=footway bicycle=yes oneway=yes, or when someone looks at a way that has been tagged like this, I’d like to see a validator warning pop up that says “Are you sure? The oneway tag is ambiguous. Use oneway:bicycle and [either oneway:foot or foot:backward] to specify one way restrictions for each mode of transport separately."

But for that we should have some consensus in the community first (hence the proposal idea). Let me know what you think! (For comments on whether such a proposal is a good idea at all, the other thread is probably best.)

rhhs · 2024-08-23T18:54:51.125Z

I made a list of quotes about this issue from the preceding discussion:

foot:backward

foot:backward contradicts the rule to not use forward/backward for oneway things.
tag is extremely ugly
foot:backward=no will never catch on, because it sounds like you’re not allowed to walk backwards
you have to explain what backward means - many mappers would have never had any reason to consider “forward” and “backward” on ways
already should work out of the box because they follow established semantics
abomination that is basically impossible to remember
I find convoluted
Used 262 times
hasn’t proven successful

oneway:foot

the tag oneway:foot=yes should be avoided (because a subtag should not contradict the main tag)
oneway:foot is so close to oneway that it would again weaken the strict ‘vehicle-only’ meaning of oneway
people are more likely to find it when searching for the right tag
may not be very elegant but has a clearly defined meaning
doesn’t make sense because if oneway is applying to vehicles only, a subtag should be about a subset, not invert the meaning
oneway:bicycle/moped/mofa are very frequent (>320k combined)
It just isn’t logical to subtype a tag that doesn’t apply anyway
Used 1188 times
seems to be more intuitive

access:forward=yes & access:backward=no

is more consistent and it has some few hundred uses

alan_gr · 2024-08-23T20:19:25.664Z

rhhs:

doesn’t make sense because if oneway is applying to vehicles only, a subtag should be about a subset, not invert

Can we interpret it like this:

oneway=yes has the default meaning “vehicle:backward=no, foot:backward=yes”. In effect it’s a more user-friendly short cut to represent the latter. (Incidentally I’m still not sure if horse defaults to the same as vehicles).

oneway:foot=yes is similarly a short cut for “foot:backward=no”, for use either when the default oneway=yes meaning does not apply, or when it might be unclear. The equivalents for moped, bicycle, horse etc. have a fully consistent interpretation.

This avoids having to explain to mappers why foot oneway has a completely different format to all other forms of transport: it is not contradicting anything, it is just expressing a difference to the default meaning. Just as adding bicycle=yes to a footway doesn’t contradict the footway tag, it just specifies a non default access.

Langlaeufer · 2024-08-24T06:50:13.342Z

alan_gr:

Can we interpret it like this:

oneway=yes has the default meaning “vehicle:backward=no, foot:backward=yes”.

No! oneway (edit: (on streets)) does not say anything about foot access. A motorway is not foot:backward=yes when you put oneway=yes on it.

Mateusz_Konieczny · 2024-08-24T06:57:13.453Z

Langlaeufer:

oneway does not say anything about foot access

depending on where oneway tag is - in some cases it actually gives info that something is oneway for pedestrians (no matter how much it is disliked by some definition purists)

Langlaeufer · 2024-08-24T06:58:27.524Z

yes but in this cases it is not foot:backward=yes - there it is foot:backward=no!
So the statement that oneway=yes inplies foot:backward=yes ist totally wrong!

oneway never adds a “access=” yes. it only sets one direction “access” to no in the one or the other direction.

On roads:
oneway=yes overwrites all types of “vehicle”:backward with no
oneway=-1 overwrites all types of “vehicle”:forward with no
oneway=no does nothing.

Mateusz_Konieczny · 2024-08-24T07:00:16.488Z

oh right

I guess that it demonstrates how confusing foot:backward=yes foot:backward=no is.

Langlaeufer · 2024-08-24T07:06:19.617Z

It shows that implicit “access”=yes (beyond highway tags) is complicated because you have to consider in which order the tags are interpreted, and this order is not documented well.

1 highway default access
2 motorroad=yes default
3. access tags from general to special
4. oneway=yes/-1 tags set one direction to no

It is dangerous to say oneway=yes is only a shortcut for vehicle:backward=no.

E.g what does this mean?
vehicle:backward=no + bicycle=yes
Is this a oneway for bicycle or not? Does bicycle overwrite vehicle:backward?

dieterdreist · 2024-08-24T09:18:51.000Z

depending on where oneway tag is - in some cases it actually gives info that something is oneway for pedestrians (no matter how much it is disliked by some definition purists)

I would say who put it wanted to express that it is oneway for pedestrians, but as these are only a few fragile exceptions compared to the rest of the usage, “giving info” does not describe the situation.

easbar · 2024-08-30T08:18:34.863Z

Langlaeufer:

vehicle:backward=no + bicycle=yes

Could we not interpret bicyle=yes as bicycle:forward=yes+bicycle:backward=yes ?

Then we would have:

bicycle:forward=yes
bicycle:backward=yes
vehicle:backward=no

The more specific tags overrule the less specific ones. So it is a oneway for all vehicles, but not for bikes.

Is this a oneway for bicycle or not?

not

Does bicycle overwrite vehicle:backward?

yes

But I agree that so far probably oneway=yes+bicycle=yes was meant to be a oneway for bicycles as well, so interpreting oneway=yes as vehicle:backward=no and applying the logic as above wouldn’t be so straightforward.