How to tag *unusable* sidewalks?

Just had a thought.

Would just using the normal sidewalk=both + barrier=cars (or maybe parked_cars?) work?

TI says that it has been (just!) used before: barrier=car x 4 & =Cars x 3, & also 1 each for barrier=parking, Parking & parking_pathway.

Would just the barrier= tag be read by routers to say you can’t get through there?

not really: it would block whole highway=residential, not just its sidewalk.

For barrier=* to (potentially) work, one would need to map sidewalk as a separate way highway=footway parallel to that highway=residential, which is quite counterproductive given that such sidewalk cannot be used. So solution is wanted which just adds extra tags on existing highway=residential, and yet which only affects sidewalk part of it.

1 Like

Personally, I would probably not map the cars unless they were truly permanent: in most places I know of, how many cars there are depends on the time of day e.g. during the day on a weekday there will be far fewer. Also, we don’t just map for routing purposes, but also for renderers, data analysis, and other reasons, so I dislike sidewalk=no just to discourage routing when there actually is a sidewalk there.

If you are keen to map it, another option could be sidewalk:smoothness=impassable (with sidewalk:both=yes)?

1 Like

smoothness=impassable does not mean what you seem to think it means. It only means no wheeled vehicle can use it. It does not apply to pedestrians. (in fact, many mountain paths for pedestrians are smoothness=impassable).

Good point, though in this case “impassable for wheeled vehicles” would, technically, not even be wrong: it is hard to see how any wheeled vehicle (bicycle, kick scooter) could use the sidewalk.

But yes, if the goal is to discourage pedestrian routing along this road, it probably wouldn’t achieve that…

What about simply sidewalk:both=parking. It implies that sidewalks exist but are effectively used for parking. If for some reason all the cars go away it reverts to a regular sidewalks.

It’s not a barrier= with definite semi-permanent presence. obstacle= and sidewalk:*:obstacle= should be used.

2 Likes

This “irregular” illegal behavior can be quite “regular” persisting. Should use informal:parking:*= , as parking:*:informal=yes is sorta troll tag. This can further be used for other non-compliant attributes.
disused:*= is used on activities to disable POI and functionalities, not structures and physical compositions that continue to exist. (similarly sidewalk:*:width= should not be used as width= is for the physical width which is unchanged; compared to other possibilities, eg sidewalk:*:width:practical= from maxspeed:practical= ) At least it should be sidewalk:*:disused= to show there is still a sidewalk. And then it may still be argued it is not “disused”, but in use for other purposes.

1 Like

this is not a big problem, extra complexity for other tools, editors and data consumers is a bigger issue

note that at least iD mappers would be likely confused by this new tagging and probably not see it - the same goes for all other editors going with “show typical tags in nice formatted views”

or note=sidewalk not usable for pedestrians as it is 24/7 full of illegally parked cars

D: sidewalk=both_but_blocked_by_illegally_parked_cars_all_the_time ? (similar to what @IanH mentions but mentions illegality)

if sidewalk is never passable then it de facto does not exist

if sidewalk is never passable then it de facto does not exist

I think the sidewalk-property describes the presence of physical infrastructure, not whether it is usable, dangerous, etc.
In this particular instance it is confirmed that the sidewalk is usable sometimes (OP wrote that illegally parked cars are occasionally towed)

4 Likes

I think this is what we cannot agree on. The existence of a sidewalk and its illegal use by cars as a parking space are 2 completely separate things. There have been some good suggestions how to tag this, but it seems they are not satisfying the OP, because routers wouldn’t use them. So this is basically a problem of tagging for the router. I can totally relate to that, and maybe something like class:foot could be pushed (similar to class:bicycle) to discourage usage of certain ways for pedestrians, but denying the existence of a sidewalk is not the way to go for me, sorry.

4 Likes

Any tagging scheme that could be perceived as implying the sidewalk does not exist would eventually be removed by a new mapper or other community member. Whether they understood the original intention or not.

2 Likes

I agree it would be better to distinguish between physical infrastructure (highly verifiable, and useful also for rendering and other purposes) and usability (important for routing but more subjective, and potentially subject to more frequent changes). What’s more, with sidewalk=no,

  • for the casual mapper it is impossible to see if the sidewalk is blocked only right now or always
  • the tag would need to change when the city starts enforcing the rules, even though the physical infrastructure hasn’t changed

In my view the best options that have come up so far are the various ideas around obstacles, irregular parking, or practical width. They’re not in the poll so I haven’t voted.

Why not use one of them and then try and get routers to use them? (I guess a real world example of a road with a lot of traffic would help here)

1 Like

If anything, if marking parking at all, I’d make sure it is marked as illegal, e.g. sidewalk:both=illegal_parking. However regardless of it needs adding support in all (old, current and future!) routers and other tools. It’s never-ending effort, so I want to avoid that.

I’d imagine that by default routers will simply route pedestrians on all roads which have sidewalk* !~ no|none, which would fail to work correctly with such tags and will happily route pedestrians where they should not be routed. So IMHO, routing should be correct even with such dumbest routers (as all new ones will start as, and many will remain).

That is the main reason why I’ve given up on my own idea of sidewalk:both=not_usable too (and haven’t even put it in the poll)

I agree that sidewalk:*:obastacle is less bad than sidewalk:*:barrier; however suffers of same “never-ending effort” problem as above.

There is a big leap between informal and illegal, though. For other suggestions, also see “never-ending effort” problem as above.

Ah, I should really check iD in more detail. I’ve always assumed that if it cannot interpret the tag that it show it in row form, and not hide it? If it hides them, that sounds like a horrible design decision that is likely to break all things that are not in “most popular” list.

or note=sidewalk not usable for pedestrians as it is 24/7 full of illegally parked cars

Hopefully iD does show that note tag at least? I’ll try to do some checking later this weekend.

That seems to be major difference in opinions. Sidewalk is in my definition some surface that pedestrians use to walk on the side of (otherwise dangerous to them) road. If they cannot walk on it, then it is not sidewalk, but a parking, shoulder, verge, or something else.
At best, if not usable for intended purpose, it is disused:sidewalk=* (which is currently my favorite idea). Same as shop=greengrocer which has completely stopped selling things and nobody works there anymore for years is no longer a shop=greengrocer+produce=none but instead disused:shop=greengrocer (even if all physical shop infrastructure is still there!)

Because of “never-ending effort” problem defined above. It is not just current routers etc. It is all the new ones that will ever be written, and it is all the old ones that people use on old devices and cannot upgrade as newer versions require better hardware/OS. I’d like to avoid leading them to their deaths by default just because they didn’t upgrade on time :wink:

So, after reading all suggestions, my current best match is:

highway=residential
sidewalk=no
disused:sidewalk=both
note=sidewalk not usable for pedestrians as it is 24/7 full of illegally parked cars
  • highway=residential is self-explanatory I hope
  • sidewalk=no - there is no actual sidewalk that pedestrians may use. It solves the main issue of routing (why we tag sidewalk=* at all) and It should also help with iD editor and others that hide less popular tags from mappers not to make catastrophic changes.
  • disused:sidewalk=both (or =left or =right) - makes it known in no uncertian terms that there is however existing physical infrastructure made for sidewalk, but it is not currently usable. It may easily become usable in the future (as it is disused, and not abandoned or worse). This should clearly define the physical state and appease users who are against just pure sidewalk=no as partially technically incorrect.
  • note=sidewalk not usable for pedestrians as it is 24/7 full of illegally parked cars (or other text - it might be blocked with waste disposal containers / recycle bins or other things) - this is human readable explanation which would hopefully be shown to mappers in various editors which should hopefully make them more careful not to damage existing mapped state (I’ve checked and iD at least shows it, as do JOSM and Vespucci). Also existence of note key might be checked programmatically by some editors like StreetComplete and potentially damaging quests might be skipped if it exists (as most roads will not have note text).
4 Likes

iD by default only shows tags it can understand and hides tags from user. The idea is to hide tags from user and be a general purpose editor which is far more ambitious than what SC tried. Often it works, but when it fails it leaves mappers confused and unaware that they caused data loss.

In fact right now even sidewalk tag is not shown at all, see Add `sidewalk` field by tyrasd · Pull Request #454 · openstreetmap/id-tagging-schema · GitHub

1 Like

I’ve never seen the tag list closed :slight_smile: Maybe I’ve opened the list one day and did never close the tag list since then - I can’t remember. But I’m a regular iD user and iD has never hidden any tags from me:

2 Likes

I created a new test account in private window and this section is collapsed by default, except cases where iD failed to match any feature to the line/point.

If opened and not closed - it will be remembered.

1 Like

I’ve checked it now: iD remembers the state of the tag list. If the list is closed at closing iD then it will be closed at the next use of iD. If the list is opened at closing then it will be opened at the next use.

1 Like

There is a big leap between informal and illegal, though

not necessarily, an informal feature could be legal, but often these are illegal. Informal settlements, to give a well known example, are usually illegal.