Actually, the picture shows is a change in the paving there, something that might be construed as “structural”. Still, in this case as well as when there is only ground markings, I’d find myself obliged to add sidewalk:right:kerb=no, as such a miss from what usually is taken for granted is not accommodating to the visually impaired.
I do not see: The sidewalk is the part of the street set aside for pedestrian use. That is the definition. My update of the wiki does nothing to change that, not in the least.
The definition according to whom? The closest term, if any, is still “footway”. Sidewalk is a specific part with a specific design, which is most commonly found and recognized worldwide to be physical.
UK’s CIHT and IHE has some recommendations.
Cycleway “A part of the highway allocated for use by cyclists, including both cycle lanes and
cycle tracks.”
Footway “A way within the boundaries of a highway which also includes a carriageway, over
which the public has a right of way on foot only.”
Pedestrian lane “A part of a carriageway allocated for use by pedestrians. Includes both areas delineated by road markings, or by a reserve of other physical barrier.”
This mostly follows the Highway Act. Except “pedestrian lane” is not legislated. https://www.legislation.gov.uk/ukpga/1980/66/section/329
Now indeed, “footway” in reality refers to the physical sidewalks in UK technical terms; and it can well be argued if a “pedestrian lane” is contained in a “carriageway”, it can’t be a “footway” under this definition. But considering OSM history, and how “footpath” as well as “cycleway” is laid out here and used in OSM, it is more logical and fitting to treat “footway” as the equivalent to “cycleway”, including both pedestrian lanes and the physical sidewalks.
For further comparison, DMRB has explicitly noted the “definition of carriageway can differ from those used in the Highways Act 1980 c66 [Ref 14.N] and the SI 2016 No 362 (TSRGD) 2016 [Ref 23.N]”. https://standardsforhighways.co.uk/tses/attachments/10442706-b592-42c8-85f8-2a0c779a8e37
Both DMRB and TSRGD defines “carriageway” as for vehicular travel only. The Traffic Signs Regulations and General Directions 2016
The Highways Act “carriageway” is closer to DMRB’s “paved width”, which includes hard shoulder, thus pedestrian lanes could instead be closely associated with the “paved width”. In US documents, such difference exists between “traveled way” and “roadway”, with pedestrian lane defined on the roadway. https://www.fhwa.dot.gov/environment/bicycle_pedestrian/publications/small_towns/fhwahep17024_lg.pdf#page=102
I agree (or I wouldn’t have opened this topic at all, obviously)
Even if this is accepted, you have to ensure everyone adds either sidewalk:*:separation= or kerbing for it to be useful
It is sidewalk:right:kerb=*, and you do not always need to specify it, only when it differs from default. For example, in Croatia, vast majority of sidewalks are assumed to have kerb separating it from street, so if there is sidewalk:right=yes (or sidewalk=right) adding sidewalk:right:kerb=yes is unneeded (and unwanted needless stating of the obvious), so you would only need to sidewalk:right:kerb=no in rare cases where there is no kerb.
Same as in Croatia, for example, highway=motorway is assumed to be lit=yes, foot=no, bicycle=no, motorcar=yes, minspeed=60 and shoulder=right (among other things). You would not go specifying all of them, they are implied. You would only tag some of them if they unexpectedly differed from that defaults.
How do you suggest to check the usage of sidewalk= while we are at it?
Then what about when there is somehow a bike lane? Despite having the same physical treatment, one is claimed to be a sidewalk=left , the other only a cycleway:left=lane . This is confusing and inconsistent.
But the fact that TMTOWTDI was there before this issue, why bring it up now?
If that is confusing and inconsistent, it is unfortunate, but is an accepted fact of OSM that there are multiple ways to map same real-life elements, and has been for a long period of years.
The only difference pertaining to this thread is that specific lane value (as opposed to yes value, which is already accepted everywhere) in cycleway:left=lane is de-facto, while sidewalk:left=lane is only (and that slightly) in-use. Whereas sidewalk:left:kerb=no is noticably more popular tagging for saying the same thing.
The documents you quote are self-evidently not in accordance with OSM tagging. Just because UKRLG calls something a “footway” does not at all mean that it is nearly the same thing that OSM calls a “footway”. Thus, UKRLG definitions are irrelevant in OSM tagging.
Case in point: take a look at osm.org for many ways tagged as highway=footway. Notice that very may of them are not“within the boundaries of a highway which also includes a carriageway”, so they would not match UKRLG definition of “footway”. They however match OSM definition of footway just fine (hint: UKRLG would call lone highway=footway as “footpath” instead)
This is my problem. We already have enough issues with using terms that have poor definition. Sidewalks and shoulders are two physically different thing. I’m fine with different terms to other variations.
For context, the word “sidewalk” comes from American English. To speakers of this dialect, the general notion of a sidewalk is one of the most basic transportation concepts, right up there with “intersection”. We learn the meaning of a sidewalk as kindergartners, more or less by intuition, based on the predominant construction practices here, which are obviously influenced by regulations. A sidewalk is a physical structure; the idea of identifying one by a road marking, sign, or legal code is completely foreign to us.
Yesterday, I encountered this sign at the approach to a bridge. The structure behind it was built as a sidewalk and remains one, even though today it’s closed to pedestrian traffic:
If OSM has repurposed the word “sidewalk” to refer to a related but distinct British English or non-English concept, particularly one that depends on a third country’s laws, then miscommunication is inevitable. It’d be as if Indianans commandeered the word “motorway” to refer to an elliptical auto racing track (true story). Maybe that’s justifiable, but if we want to settle these debates, we need to take a step back and explain where we’re coming from. What is the word, in its original language, that each of us is promoting? How is it defined, and why is this definition the optimal definition of sidewalk in OSM?
I don’t recall the video – version 2.0 was over five years ago. But a lakeside promenade, as you’re describing it, would only be a sidewalk if it runs beside a road that also rings the lake. It was probably just an oversight.
See the talk page discussion. You had unilaterally rewritten this gallery to suit your perspective, which apparently was not shared by whoever originally created the gallery. Both versions have caused a lot of befuddlement among North American mappers. To a less experienced mapper, it isn’t merely incorrect; it’s incoherent. But maybe that’s just because the gallery needs better examples and more explanation?
I think you’ve proved my point from the wiki talk page discussion. In the U.S., there have always been shoulders that pedestrians use informally. Even the states that categorically ban pedestrians from roadways still allow pedestrians to hug the edge of the roadway if there aren’t any other suitable pedestrian facilities. (Some states explicitly allow cyclists to use motorway shoulders for the same reason.)
As a fairly recent safety measure, transportation departments have begun endorsing the conversion of some shoulders to “pedestrian lanes” as a “quick build” precursor to an actual physical sidewalk. The only physical difference is the painted pictogram, flexible bollards, and some signs. The only legal difference is that pedestrians are no longer in a legal gray area for walking on one. If you haven’t encountered one of these pedestrian lanes before, you’re more likely to refer to their European counterparts as shoulders, even though they only visually resemble shoulders.
I’m not sure if this was irony, but for what it’s worth, various passages on that page indicate that shoulders aren’t exclusive to motorways. If they are, we’ll need your help coming up with another word for the majority of shoulders that aren’t along motorways and also aren’t primarily intended for pedestrians.
By the way, “traveled way” is obscure jargon. Most American English speakers have no word for this concept, but some might refer collectively to “the travel lanes”. To generalize:
through lanes + turn lanes = travel lanes
travel lanes + buffer + bike lane + street parking + hard shoulder = roadway
roadway + curb + soft shoulder + verge/berm/tree lawn + sidewalk + greenspace + noise barrier = right of way
footway:right=lane seems like a pretty elegant solution to me. We don’t worry about renderers or routers mistaking a street for a cycleway just because it has a cycleway:right=lane tag; why should it be different with footway?
And where do you see sidewalk= being used for pedestrian lanes in Taginfo? How can you be sure what sidewalk:*:kerb= is for, which can be sidewalks without kerbs, but still with other barriers.
You are changing the central idea of “sidewalk” from a physical walkway to one encompassing marked or signposted areas only. The level of infrastructure and comfort is then in doubt, not confidently assumed. Alternatively as I said, lack of kerb can still mean there are railings or other protection, but sidewalk:*:separation= has not been used yet.
The sidewalk= vs cycleway= comparison is not about the syntax. It is about the same physical condition being a cycleway:*=lane for bikes, while you force pedestrian lanes to use sidewalk:*= with the only reason being it exists, ignoring its semantics. sidewalk:*=separate would not be applicable for pedestrian lanes. The entire set of concept is not transferable from sidewalks to pedestrian lanes.
For me, there’s a difference between a footway and a sidewalk. A sidewalk is a structural element of a street, next to it. How it’s separated from the carriageway is probably down to local legislation. In Germany, 99% are separated by a kerb and raised above the level of the carriageway by ~12cm. A sidewalk is usually only used by pedestrians, but it can be a shared space with cycleways, in which case it’s still a Sidewalk. Further structural elements include the carriageway, shoulder and verge. Then there’s also the embankment, the median strip and maybe others I’ve forgotten, or simply don’t know the English word for.
A footway, on the other hand, would be more like a space on the street “where can I walk”. So if there’s a sidewalk, then the footway would usually be on the sidewalk. Since there’s usually no sidewalks outside cities in germany, most of the time you either have to walk on the shoulder or the road verge.
This would mean, for me, that sidewalk:<side>=yes implies something like footway:<side>=sidewalk, much like we have cycleway:<side>=lane/track and track would mean, at least over here, on the sidewalk. I completely agree that in case of a footway painted on the street, footway:right=lanesounds like a sane choice ifsidewalk:left=yes always implies footway:left=sidewalk. And the problem is: it doesn’t, because footway=* is a key to precise a separately mapped highway=footway / highway=path and not to specify where pedestrians are supposed to walk
We have similar issues with the cycleway-key. We use cycleway:<side>=track, but over here, in the majority of cases, this “track” is actually the sidewalk, so cycleway:<side>=sidewalk would make more sense. I’m sure it’s not the same all over the world, but it’s like this in Germany. But then we also have cycleway=crossing, which is kinda misusing the “where are the cyclists are supposed to drive”-key as a “is this cycleway track crossing the road”-kinda key.
So taking this all into consideration, I think the only way to tag this right now is to use lanes:
As much as I’d like sidewalk:right=lane to be valid, it seems like a contradicting tag if we consider a sidewalk a structural part of the street and not just “the area where pedestrians walk on” footway:right=lane would be great, but footway is currently used in a weird way. Maybe this works, maybe not. I’m not sure yet.
Disclaimer: this is, if you consider sidewalk a structural element of a street and not the intended use of a part of the street.
I’m just going to leave these two links here, because they’re worth reading in conjunction with this one:
What’s clear to me from all these discussions is that we are lacking a system for tagging poor pedestrian infrastructure, like the one we have for cycling infrastructure, from shared_lane and lane to track, etc.
Rest assured, if this was not shared by whoever, it would not have remained, original author already invited here. I did not like the gallery back then. Now browsing the history, I think I know your reasons. This is unnecessarily combative. I rework that.
PS: Rewording complete, dubious flag left in place though, as I do not know, it the update heals that. I am sorry for any inconveniences that caused.
Also added shoulders to the introductional text of the attributes section, which here exist almost exclusively as breakdown-lanes. Obviously, there are places where shoulders are much more common and serve quite a host of uses.
I’ve tried to document various methods mentioned to tag pedestrian lanes at Sidewalks - OpenStreetMap Wiki.
Feel free to improve and add other possible methods if I missed some.
That is the same here. I would not place too much emphasis on sidewalk being a term that comes from American English. All the sidewalk documentation in the OSM Wiki has been started by Germans, so it is not guaranteed, that they used the term in the US meaning.
Here picture, where the separation between footway and carriageway changes from kerb to road marking.
For me it is just sidewalks, certainly no shoulders there. Mind you, local law has a definition of sidewalk, and markings are literally allowed as separators.
I agree, there is some confusion. But it is relatively small (and insignificant for data consumers) compared to for example far more problematic OSM use of highway=footway in a way that is contrary to what Footway actually means in English (i.e. in OSM, footway not only does not need kerb/etc separation, but in fact does not need to have nearby carriageway at all). With sidewalk the confusion is at least only partial - it seems to me that if OSM is to be used around the world, and not just in one country, that there will be necessarily always be confusion for English speakers (given tags are in mostly in English). There is even often serious confusion between just different dialects of English; now imagine how it must be in much more culturally diverse parts of the world!
I sometimes wonder if it wouldn’t have been better if OSM decided to use wikidata-alike numeric IDs instead of UK English names for tags / values - it often seems to create way more confusion and flamewars than it helps with “intuitively” understanding the meaning
That being said, I do not wish to enforce any particular use. I would just plead that whatever people decide to map those “pedestrian lanes” (or whatever one want to call them) that they make sure that:
open appropriate tickets so routers (and other data consumers) actually make use of such way of mapping
Choose key from that list in link I’ve sent you, click on Values, and type lane and press Enter.
Ummm, I think you might’ve confused me with someone? I’m the one who opened this thread in order to find out how people map it (and replied to you only to clarify something you asked about). I do not intend to promote any way specific of mapping those. I only intend to invest time to document what people say how they are mapping those things, in order so next poor soul that comes along does can see that in short form and decide for themselves quickly, instead of instigating yet another long-winded discussion without consensus. IOW, I only intend to describe how it is being tagged (so interested data consumers can handle it), and not prescribe how it “must” be used.
That is true that it would bring extra details. As you said, it has not been used yet. If you care about it, I suggest to link into other thread by clicking on port reply arrow and choosing Reply as linked topic, and try to describe how it would be used (or even jump straight into https://wiki.openstreetmap.org/wiki/Proposal_process).
Thanks, I’ve not seen that suggested so far, I’ve documented it in the wiki too!
I have no problem with using the :lanes-tagging schema but please be careful with the value of lanes=*. Bicycle anf foot lanes do not count!
Two more points to consider:
How about horses? Are they allowed on the right lane? I guess not.
Thanks! I’ve added a warning to the wiki that while :lanes“extension covers all kinds of lanes for all kind of vehicles and is not restricted to motorized traffic”, lanes=* does not count neither foot nor cycleway lanes.
(also, you links seem to link to wiki.openstreetmap.de instead of wiki.openstreetmap.org, which doesn’t work for me)
They are not defined (not only for the right, but also for the left lane!), so it is unknown. In places where they are an realistic possibility, it should be defined (as might other things like kick_scooter, snowmobile, hand_cart, electric_bicycle which might be region-specific)
It is unclear. See wiki part about “some mappers consider oneway=* valid for all kinds of movement, including foot” and below, as well as Talk:Key:oneway#Pedestrian_oneways
is the first element empty/undefined on purpose? I see it documented for lane maxspeed to be omitted when non-existent, but cycleway lane example (which should be similar to pedestrian lane example for :lanes extension) at Lanes - OpenStreetMap Wiki uses regular no as example for forbidden access
Thanks for pointing out that lanes=* only refers to the lanes with motorized traffic, that was new to me. Your usage of access:lanes=yes|no + foot:lanes=|designated would imply that pedestrians are allowed to walk on both lanes of the road. I’m not Croatian, so I don’t know if that’s correct or not.
I do not know how deeply we want to dive into :lanes-tagging here or if a new topic would be better.
Additionally, I only know the German law well enough to talk about details like horse and which lanes are completely prohibit.
Yes, I left the values empty as I do not know the rules in Croatia.
Even in Germany, no for bicycles is disputed as there are too many exceptions like turning left on the next intersection, overtaking, or if the lane is blocked. We have use_sidepath as access value but we miss a use_sidelane or similar for access tags per lane, see:
In the U.S., there have always been shoulders that pedestrians use informally. Even the states that categorically ban pedestrians from roadways still allow pedestrians to hug the edge of the roadway if there aren’t any other suitable pedestrian facilities. (Some states explicitly allow cyclists to use motorway shoulders for the same reason.)
- it seems to me that if OSM is to be used around the world, and not just in one country, that there will be necessarily always be confusion for English speakers (given tags are in mostly in English). There is even often serious confusion between just different dialects of English; now imagine how it must be in much more culturally diverse parts of the world!