That’s a poor comparison. highway= roads are a functional classification. You can certainly have =service and =footway with access=no . This is not trolling, or not a =service and =footway . A business may physically block them almost fully with their inventory and other junk on their premise, even for emergency exits very illegally and dangerously. That doesn’t mean such thing doesn’t exist.
What is meant by activity anyway? If a sidewalk is always almost fully occupied by street stalls or people sitting with their belongings spread out, is it not an intended use? You can still walk over them. If
(mis)using an existing tagging scheme rather than create new key/values seems like (mis)tagging for the routing software. 
.
With a new tag, local activists can survey roads, and campaign for changes in laws (or enforcement) and eventually get rid of the illegally parked cars.
4 Likes
But it is a criterion for prepending a disused: lifecycle prefix, is it not?
I kind of though it was the whole point to use existing tagging scheme so routers, renderers and other data consumers could use them? 
About misusing, I agree with you. That is why I try hard to avoid misuse and to find reasonable solution which works (just read the whole thread if you believe I didn’t try enough). Only requirement is that the additional tag should not change a meaning or basic tag completely (i.e. it should not be trolltag).
For example, I do not find it acceptable to mark something as sidewalk=yes and then add sidewalk:unusable=yes, because such additional tag completely reverses the meaning of the first one, i.e. is a trolltag. Extra tags should just refine the meaning of the more primary tags, not revert them.
So I’m fine with any other (non-trolltag) options, like:
sidewalk=not_usable, orsidewalk=no+something_that_looks_like_a_sidewalk_but_is_not_usable_as_sidewalk=both, ordisused:sidewalk=both- or some other idea in which most basic router (i.e. without extra support for new tag) will not be trolled into doing completely different thing than it should (i.e. failure mode should default to failing “safe”)
Secondary goals are minimizing damage to the map by less experienced editors editing such ways (see iD-related discussion above) and minimizing amount of work other people will have to invest to implement the idea (ideally, the solution would be just something some volunteer can document on the wiki and others use, and no tools change is needed).
I agree, that would be nice extra benefit. But it cannot be overriding basic requirement that adding an extra tag must not reverse the meaning of previous tags.
That is why (to me) disused:sidewalk=both seems to perfectly describe the situation: it means there is an existing sidewalk on both sides, and it is not usable as a sidewalk. So it allows both for distinguishing existence (or not) of such spaces, and to distinguish space usable as sidewalk from those unusable as sidewalk. And it is already supported by editors, routers etc.
If you feel extra reason describing why it is currently unusable (i.e. disused) is important: sure, extra tag like disused:sidewalk:note=xxx (or better yet more often supported plain note) could added. Or some machine-readable equivalent like disused:sidewalk:reason=* with values like illegal_parking etc could be invented (although I personally am not convinced very many people would bother)
Because, that is exactly the purpose the lifecycle prefix disused: is intended for. It’s wiki (see above) even gives very similar use case as the part of main explanation of lifecycle prefix:
For example, a concrete parking area which is no longer used for parking cars but which still carries a name sign might be tagged as […]
disused:amenity=parking+disused:parking=surface+name=*
While it is certainly possible to invent other tag(s) to accomplish that same purpose (like unusable:sidewalk=both on nonfunctional_sidewalk=both etc.) I fail to see any advantage it might bring to the table (and see a lot of disadvantages, like people needing to add support for it everywhere). Or do you see some advantages which different naming (for about the same purpose) would bring?
1 Like
In ordinary, everday language, there is a sidewalk here, and it is being used, it’s just that it’s being used for parking (obviously not its intended use).
How about sidewalk:use=illegal_parking? A sidewalk is physical infrastructure, like a building.
We don’t tag a church that is being used for something else than it was built for as disused:building=church. We tag it as building=church, building:use=apartments.
not for physical infrastructure, it is for services / activities.
Let me repeat my suggestion from above: obstacle
This is how we deal for example with paths that are overgrown and hardly usable.
2 Likes
If you think so, then it likely heavily differs by country or other. Over here, if you showed people first picture, and ask them wheather they see on the side of the road: (a) footway, (b) cycleway, or (c) (illegal) parking; vast majority of them would answer (c) (illegal) parking.
That is because it is not usable as (a) surface for movement of pedestrians nor as (b) surface for movement of bicycles, but it is being used as (c).
Note that sidewalk/footway is just a footpath (surface desgnated for movement of pedestrians) that happens to be situated on the side of the road, and is often (but not always) somehow physically separated from it (kerb, tree line, bollards…)
I’d prefer sidewalk=illegal_parking (or more general sidewalk=not_usable) as it would lessen the most problematic possibility that someone who has not taken the time to read this discussion might naively add sidewalk=yes in addition to that sidewalk:use=illegal_parking, thus breaking everything for pedestrians)
A sidewalk is physical infrastructure, like a building.
I disagree. What would you call this thing on the left side of the road? sidewalk or cycleway?
I would most certainly call it a cycleway, as it is designated surface for movement of bicycles, and pedestrian traffic is not allowed. Thus, I would tag that highway=residential with additional cycleway:left=track + sidewalk=no . I’m very interested how would you tag it @osmuser63783 and @dieterdreist (and others)?
Yet, physically, it might have been footway/sidewalk instead (for example, if that image was taken from slightly different position, you might likely guess wrongly).
In other words, just because there is paved area separated by kerbs on the side of the road, does NOT automatically make it sidewalk. Sidewalk implies specific purpose (designated pedestrian traffic).
2 Likes
And this is why ultimately the local community has the final say in how various objects are represented, even if that does not mesh perfectly well with what is perceived otherwise to be the general standard. Thinking on your post, it is very true that the cultural idea of what constitutes a sidewalk vary greatly. I myself have experience with two types of ‘sidewalks’: in Brasil and United States. They could not be more different than each other. I wish you all the best in your efforts to find a tag combination that works to represent this situation @Matija_Nalis.
2 Likes
That’s a false dichotomy: it’s (a) and (c). It’s called parking on the pavement where I live, it’s not uncommon. I don’t think anyone would claim that that means there isn’t a pavement.
Same as you, of course. I wasn’t trying to make an argument that any paved area next to a road is a sidewalk/pavement.
I completely agree: this is the key question. Is it designated (intended, set aside in legislation) for pedestrian use? I think this discussion has shown that those who answer “yes” find it hard to support any suggestion that includes “sidewalk=no”.
Amen to that. Maybe it’s all down to regional/cultural differences. No one prevents you from tagging it as you wish, even if many people who have participated here would tag it differently. 
Edit:
Not a bad idea in my view. Neither both nor no. Hopefully deters routers, but also deters retagging as both.
1 Like
But how do you know it is (a) and (c), and not (b) and (c)? At least in my country, only indication if it was originally intended as (a) or (b) is that there will be painted picture of the bicycle every 100 meters or so. Without removing all illegally parked vehicles, you simply couldn’t know. Only thing that all people could at a glance agree is that it is (c) (illegal) parking.
In other words @osmuser63783 and @dieterdreist, why would you decide to mark that as sidewalk=both instead of cycleway=both ?
Or it might even be the combination (even most probably in some countries) sidewalk:left=not_usable + cycleway:right=not_usable (or the other way around).
Thanks, I hope others see it the same way - this is what I want to point out. If you don’t actually know what is on that side of the kerb, is it OK to invent some meaning without any evidence just because you made a random guess what it might have originally being designed as?
So,
-
if one insist on mapping “physical feature” instead of “current usage” (which seems to me to be the division in opinions here), then, for example, one could map that picture at top as a
cycleway=right+sidewalk=left. Or as acycleway=both+sidewalk=no. Or as asidewalk=both+cycleway=no. Or few other combinations, which might (but need not) actually be true.
To me, such ambiguity which is “resolved” only by guessing fails Verifiability principle. We would instead need some tag which means “paved surface on the side of the road without any physically possible usage, and with unknown legally intended usage”. I would argue that neithercyclewaynorsidewalkare that tag, as they have their meaning (dedicated surface for movement of bicycles / pedestrians, respectively). Maybe something along the lines ofundefined_raised_surface=both/left/rightwould be most precise? -
on the other hand, if one insist on mapping “current usage” instead of “physical feature”, situation is much clearer - it is certainly
sidewalk=noandcycleway=no, as they don’t fulfill those purpose. It also (as opposed to above option) actually works for one of the most popular uses of OSM, which is routing. It might also additionally be tagged withundefined_raised_surface=both/left/rightor similar (likedisused:sidewalk=both– note that while it might also be partially incorrect, difference between “unusable sidewalk” and “unusable cycleway” is not really all that important; as opposed to difference between “sidewalk” and “cycleway” where difference is a huge deal)
Yes, sidewalk=not_usable would certainly confuse much less routers than sidewalk=yes + sidewalk:usable=no, but it is still less than ideal, as work would be needed to verify/patch all routers and patch all editors…
e.g. less complex routers will still check only for sidewalk!=no and decide that other values mean OK to route - so all of sidewalk=left, sidewalk=right, sidewalk=both, sidewalk=separate, sidewalk:left=yes, sidewalk:right=separated etc. would mean it is OK to route pedestrians there
A sidewalk is physical infrastructure, like a building.
I disagree. What would you call this thing on the left side of the road?
sidewalkorcycleway?
cycleways are also physical infrastructure
1 Like
Cycling on the pavement is never legal in the UK unless there’s a blue “shared use” sign. Pavements that are only for cyclists (which would of course be mapped with cycleway=* not sidewalk=*) virtually don’t exist. The only one I’ve seen had a dropped kerb connecting it to an on-road cycle lane, in addition to bicycle symbols. That would be easy to spot even if it was blocked.
By the way, sidewalk=bad has 20 uses, in case you don’t want to start completely from scratch.
I’m aware (even if I don’t agree at this time) of your stance that actual use of infrastructure doesn’t (primarily) matter when tagging, but that is not at all what I asked here.
The question was how would you @dieterdreist (in OSM) tag that asphalt surface (with white-on-green pictograms) on which the lady in purple is driving her bicycle on ?
I.e. would you tag that as cycleway:left=track + sidewalk=no (as I would) or as something else?
Because, if you would also seem to tag it as such, than that is common point of reference we all seem to agree on and can base further discussion on. If you however disagree with that tagging (and would e.g. tag it as sidewalk=left + xxxx?) then I’d like to explore that reasoning and lead the discussion in completely different direction.
I think we’re just disagreeing that most people here would want to tag the officially intended use of a sidewalk/cycleway, and you, Matija, want to tag what it’s really being used for. There’s no common ground, really. If it’s custom in Croatia to do it like this, go ahead. But since you’re asking here, it’s probably not.
As for that picture: Since that cycleway has a pictogram of a bicycle, it would be a cycleway and not a sidewalk.
If everything without a pictogram would officially be a sidewalk, and there was no pictogram, it would be a sidewalk, no matter if there’s cyclists riding on it, cars parked there, or a heap of dirt blocking the way.
A sidewalk with a parked car on it, is still a sidewalk. It’s just being misused. But as soon as the city paints parking spaces on the sidewalk, it would (partially) become on_kerb-parking. I’m sure you already understand us, you just don’t want to tag it like this 
2 Likes
While it is certain that there are some disagreements (which I’m working on trying to find common ground from which to proceed, see previous post), I’m not sure where you’re pulling that “most people here” statistics from. If anything, poll above would suggest that majority prefers tagging that as disused:sidewalk=both, with somewhat less people supporting sidewalk=no, and least number supporting sidewalk=yes option.
But since you’re asking here, it’s probably not.
I’m not quite sure how you’ve come to such conclusion, but your guess is completely wrong. I’m asking here, as I’d like to find some solution globally applicable, and document it that so the next person that comes by does not have to invest so much time to find best suitable tagging.
If it were about me and my preferences, I’d just continue clicking “there is no sidewalk” answer in StreetComplete and be done with it, saving all that precious time. But while one of the better solutions, that is still not ideal solution; so if some other option is found which is acceptable worldwide, it maybe might be offered as an extra answer in StreetComplete which adds extra tagging, and it might be documented on the wiki.
If not, I guess I’d wasted some time in hopeless effort, and people will continue mapping such sidewalks depending on their random preference, which will continue to erode usefulness of sidewalk=* tag for both “sides” (as neither could be sure what was actually meant by that tag - weather it is surface on side of the road that was intended as cycleway or as a sidewalk or as something else, or if it is the kind of sidewalk that could be used for routing pedestrians)
You mean “cycleway with a parked cars on it, is still a cycleway”, right? If we are talking about picture on the top of the thread. Or are you not even sure if that is cycleway=yes or sidewalk=yes, yet you still will tag it randomly as one (or both?) of those? 
Also, how exactly would you @Nadjita tag that surface purple lady is driving on ?
1 Like
The majority of the mappers that I know would use neither of your suggestions. I might be wrong, though, but I didn’t vote, because all 3 don’t seem fitting. The suggested
highway=residential
sidewalk:both=yes
sidewalk:both:obstacle=parked_cars
note=sidewalk not usable for pedestrians as it is 24/7 full of illegally parked cars
was ditched, because a) editors might not support it and b) routers would still route over the sidewalk. None of this would be reason enough for me to abandon a perfectly valid and fitting schema.
Then I misunderstood the intention, sorry.
That was meant as a general statement, no matter the situation/picture. Both, cycleways and sidewalks don’t change what they are when there are obstacles on them.
cycleway:<side>=track
cycleway:<side>:oneway=no
sidewalk:<side>=no
Unless this is in a country where you’re generally allowed to walk on cycleways. I’d also add a bicycle=use_sidepath if the cycleway is compulsory. Since there’s bicycle pictograms painted on the way, the intended use is clearly a cycleway.
Thanks. So far, it seems that the common ground is that “purple lady” picture is to be tagged with (more or less detailed variant of) cycleway=yes + sidewalk=no. If someone disagrees, please speak up.
If you’d be so kind, I’d ask @Nadjita @dieterdreist @osmuser63783 (and others of course, especially if they had objections before) how would they tag asphalt part on the left of the kerb in this picture (the right side with car on it is highway=unclassified, in country where it is not legal to drive bicycles on surfaces designated for pedestrians):
(note I’ll address more points raised by you and other users later; I’ll try to take it one step by one step, as otherwise it seems that important points get lost).
That would be a sidewalk:<side>=yes in my country or a separate highway=footway + footway=sidewalk
1 Like
Yep, I agree. In Germany this would definitely be a sidewalk for pedestrians unless there would be some official roadsing declaring it to be a cycleway or a combined sidewalk/cycleway. Nevertheless it is possible of course that in other countries different regulations apply.
1 Like
What’s your point? That you don’t know if it’s a sidewalk or a cycleway if it’s blocked by cars? The one in your first post is clearly a sidewalk, you said as much (“built as one”).
If I genuinely don’t know what it is, I just wouldn’t tag it (neither its presence nor its absence). This suggests another solution: Just don’t add a sidewalk tag. But don’t be surprised when someone else adds one later.
1 Like
Exactly. And the even bigger point raising from it: if something is NOT Verifiable by mapper what it actually is, it should not be mapped by random tags which are not verifiable based on a pure guess.
The one in your first post is clearly a sidewalk,
Is it? “clearly”? How came then that this last (“black car”) picture is not “clear” to you then? It even shows much more of the surface than the first picture in the thread? If you are sure of the first picture, then you must also be doubly sure for the last one too. Yet, you seem to be unsure about the last picture.
you said as much (“built as one”).
Oh, so you too were born in that specific Zagreb neighbourhood several decades ago when there were much less much cars, and walked it often daily? Sorry I didn’t recognize you after all those years! 
joking aside: In other words, I do have local knowledge of that specific neighbourhood spanning several decades, so, against all chances, I am somewhat more familiar with that specific tiny part of the map more then the 99.999% of the OSM mappers (just like I am not so familiar with 99.999% of the rest of the planet).
So, what I can map here with source=local_knowledge there, you (most likely) can’t (not from the picture; you can visit and make friends with older locals and get to know a story over a cup of coffee, of course. The older generation there is generally a friendly bunch and likes to talk about times past!)
So, my source of information is not only that picture. From that picture alone, even I (living in that country) wouldn’t really know what it was initially before it became illegal parking. I might map it as disused:sidewalk=both, or as a disused:cycleway=both but simply because difference between “unusable sidewalk” and “unusable cycleway” is negligible. (as opposed to difference between “sidewalk” and “cycleway” which is huge).
This suggests another solution: Just don’t add a
sidewalktag
I actually agree with you here. This is why I actually did like that idea of adding just disused:sidewalk=both (without adding sidewalk=* tag), and probably why it was the most liked answer in the poll.
To conclude, how would you @osmuser63783 @Nadjita @dieterdreist (and everybody else) feel about:
highway=residential
disused:sidewalk=both
note=sidewalk is not usable for pedestrians as it is 24/7 full of illegally parked cars
one might even invent (and add to that list) suggested tag like sidewalk:both:obstacle=parked_cars or similar, if one thinks that machine-readable reason (in addition to human-readable reason in a note) is important addition for determining the reason why the sidewalk is not usable.
It specifically doesn’t add any sidewalk=* attribute, to keep common ground, and avoid contentious choices. Would that work for you?