[Tagging] Feature Proposal - RFC - "is_sidepath" as a sidepath concept

Jens Glad Balchen balchen at saint-etienne.no
Fri Dec 2 21:34:20 UTC 2022


On 02.12.2022 13:31, Alex wrote:
> Paths and ways along a road can be mapped separately in OSM, but those 
> separate geometries cannot be identified as part of the road, or only 
> with the significant effort of using geometric processing (which most 
> applications can't perform). 

I strongly agree with the concept of tagging information that is hard or 
impossible to compute accurately or reliably. I want to emphasize that 
from the start since the objection has been raised in multiple 
discussions that something is theoretically computable. I would rather 
that we not obstruct valuable tagging because the same information could 
theoretically be computed by some brilliant algorithm not yet 
constructed and/or extremely detailed tagging, that in combination would 
require significantly more effort than just tagging the desired 
information in the first place.

> Therefore, a sidepath concept using the tag "is_sidepath" as an 
> additional tag on ways (in particular cycle ways or foot paths) is 
> proposed to indicate whether a way is related/attendant to a road 
> (i.e. adjacent and parallel to a road) or whether it runs 
> independently/isolated without any relationship to a road. 
> Furthermore, an extended set of sub-tags is proposed to allow to 
> explicitly tag important road attributes on the sidepath itself.
>

_Mainly_, I have concerns about the concept of a cycle path or foot path 
being attendant to or a sidepath of another road.

In Norway, we no longer have cycle paths and foot paths. We have 
cycleways, footways and carriageways. It may seem like a small 
difference in terminology, but it makes a large difference as part of an 
overall mindset. Roads are roads, and different types of roads are 
simply meant for different types of travel.

To me, this proposal sounds similar to a hypothetical proposal that we 
start tagging roads that runs adjacent to railroads as 
"is_sideroad=yes". It makes very little sense to me to establish the 
railroad as the "primary" line of travel and the road as an "attendant" 
line of travel. Just because they are parallel does not mean there is a 
meaningful relation or hierarchy between them.

I'm concerned that this type of tagging establishes and codifies a 
hierarchy, a relative importance, and an implied dependence, that I 
don't see in many of the posted examples, and that I'm not aware exist 
in formal road structures.

An exception of course is if the road was specifically built to 
facilitate access to the railroad by non-rail means, and the road was 
tagged as such, it would make sense to relate it to the railroad it serves.

I know the term "sidepath" exists and is used, but I can't see that it 
is semantically different from cycle path, cycle road or cycleway, 
except for this rather vague property of somehow being an add-on to the 
"proper" road.

Can you provide any thoughts on how you see the "is_sidepath" 
relationship in principle?

_Additionally_, I'm curious about what you will do with the tag. I've 
read the use cases, but I don't fully understand them. Perhaps you can 
elaborate?

Rendering: How and why would a renderer treat a cycleway differently 
when is_sidepath=yes?

Routing: How does it help routers to know that a cycleway runs parallel 
to a carriageway? If the preference is to ride on cycleways, I assume 
the router will pick the cycleway regardless. If there is no preference, 
I assume the router will pick whichever line is shortest.
I see a value in being able to capture a name.

Data analysis: How is capturing the quality of a cycleway obstructed by 
the lack of an is_sidepath tag?

_Finally_, I agree that any use of is_sidepath or similar conceptual 
tagging, like in the railroad access road scenario, seems most valuable 
when it actually includes a direct reference to the other line. Simply 
knowing that there is a line out there that is parallel and adjacent to 
this line seems like very vague information. Duplicating some 
information from the related line seems a bit odd, because it still 
leaves no way to automatically tell which line we're trying to 
reference. But I guess it can be hard to implement a direct reference in 
practice, if the two ways don't have exactly matching line segments?

Cheers,

Jens
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20221202/1985da33/attachment-0001.htm>


More information about the Tagging mailing list