This leads to another topic where there is just as much need for action.
You can find the is_sidepath-scheme here:

It looks like a stub (note also the talk page), because the idea is very
simple but still solves some big problems with separate paths:

  * Separate paths are thus tagged with "is_sidepath=yes" (or simple
    "sidepath=yes", need to be discussed). "footway=sidewalk" is already
    in use to solve this issue on sidewalks, but there is nothing for
    e.g. separate bike paths.
  * The associated highway is still tagged with
    "cycleway/sidewalk=separate" (as well as the access tag
    "use_sidepath" if using the sidepath is compulsory).
  * The main categories of the highway, "name" and its classification
    like "primary/secondary", can be assigned to the separate way with
    Keys like "sidepath:of" or "sidepath:of:name". Other values like
    "lit" should anyway be tagged directly on the separate way.

What is achieved with this? Routers recognize very easily whether a
separate path is associated with a road. For example, if you want to
avoid main roads in routing, you can do so with this scheme. Routers
can, without misusing "name" on the separate way, provide a street name
("Turn right on XY street"). This variant also has advantages for
rendering. For example, at smaller zoom levels, renderers could display
"cycleway=separate" as "cycleway=track" while skipping separate paths
for better clarity.

The problem remains that physically non-existent road crossings ("wildly
crossing the street"), which in reality represent a crossing possibility
for many users, are still not available for routing. In my opinion, this
problem is not very relevant if separate ways are well mapped (which
they often are unfortunately not!) and all essential routable
connections are in the database. At the beginning and at the end of the
route, people can use their brains ("destination across the street") if
their routers do not solve this task for them. If there are major
detours on the way, routable connections or suitable access tags (e.g.
"foot=yes, wheelchair=no") are missing. Or it is a crossing possibility
which I do not want to get from a router because of its dangers. In this
context it might be more important to find a way to distinguish between
physical/formal and "wild"/informal connections and to raise awareness
for the use of access tags for different user groups (like wheelchair,
bicycle with trailer).

- Alex -

Am 19.09.20 um 00:54 schrieb Alan Mackie:
> On Fri, 18 Sep 2020 at 21:35, Tobias Knerr <> wrote:
>> On 17.09.20 02:35, Taskar Center wrote:
>>> This is yet another example why "sticking" the sidewalks onto the
>>> highway (as a tag) rather than mapping them as separate ways is
>>> appearing to be less and less practical. Please see our sidewalk schema
>>> proposal
>>> <>
>>> from several years ago.
>> Your sidewalk proposal unfortunately doesn't really address the crucial
>> shortcoming of separately mapped sidewalks: The lack of a reliable
>> mechanism for figuring out which section of road a given sidewalk way
>> belongs to.
>> I agree that we should be able to give sidewalks their own geometry, but
>> we _also_ need the relationship between sidewalk and road. So far, all
>> the proposals attempting to support the former end up sacrificing the
>> latter.
> Was this meant to be one of the purposes of associated street relations?
>> There have been some promising discussions recently around the
>> sidepath_of idea, but that's still just brainstorming. Until a practical
>> solution is found and actually used in the database, sidewalk mapping
>> will remain a choice between two options that are broken in different ways.
> I hadn't heard this one. Do you have a link to the discussion? I would
> personally prefer sidewalk_of or walkway_of if we were to go this route
> though. Sidepath sounds like something that's branching to me.
> Both associated street and sidepath_of still have the issue of when you're
> allowed to jump from one to the other, kerbs can be stepped over by most,
> railing less so (they're often to keep pedestrians out of blindspots). It
> must be difficult to tell if a sidewalk is separated specifically because
> the transition from one to the other is more thoroughly blocked and not
> simply as an added level of detail  with no more than the normal impediment
> to foot traffic.  The only thing I've seen discussed that might work for
> this was in a talk about way and street areas.
>> As for the main issue of the thread: I would welcome a clear definition
>> for the meaning of width. In my own mapping and when writing the
>> relevant code in OSM2World, I have counted sidewalks etc. as part of the
>> road's width if they are mapped as tags on the main way. But I would of
>> course change that if there finally was a documented and widely
>> agreed-upon recommendation. I don't care so much which one it is - but
>> we need one.
>> _______________________________________________
>> Tagging mailing list
> _______________________________________________
> Tagging mailing list
Tagging mailing list

Reply via email to