[Tagging] Routing through the end of a barrier

[Kevin Broderick]

On Fri, Aug 1, 2025, 10:01 Greg Troxel <gdt at lexort.com> wrote:

> Michael Tsang <miklcct at gmail.com> writes:
>
> > Hi all,
> >
> > If there is way which intersects the end of a linear barrier (for
> example, a
> > chain), are you expected to be able go to through it?
> >
> > For example, if a chain has 3 nodes, A-B-C, and a highway goes X-B-Y, I
> don't
> > expect to drive a car through that highway. However, if a highway goes
> X-A-Y,
> > should you be able to drive a car through it?
>
> I would say this is a tagging error.  When you  have a highway, it's a
> line feature, but it's understood to have width on either side of the
> line.  A barrier that ends in the middle of the highway will block half
> the highway.   It is extremely unlikely that this exists in raelity.
>
> Thus, the question is whether a router should  guess:
>
>   - A) tagging error: barrier stops just short of highway
>     + ignore barrier
>
>   - B) tagging error: barrier actually continues
>     + way is not routable
>
>   - C) tagging correct: barrier literally stops at centerline of highway,
>     blocking one direction
>     + way is passable in one direction, based on where barrier is and
>       drive-on-left or not
>
>
> I vote for
>
>   - routers should pick A
>
>   - validators should flag this
>

I would second this conclusion, and I'll add that my first guess would be
that someone tagged "barrier=" on a point on the way, then either they or
someone else made the barrier a line to satisfy a validation warning.

My second guess would be that they added the barrier as a line but
misplaced the first click and left it off-center.

Guessing that they intended to indicate the way only has a barrier halfway
across and allows travel in one direction would be a distant third. If
that's actually the case, I'd be thinking hard about whether there is
actually enough physical separation to warrant mapping the road as a way
for each direction, with the obstructed side ending on either side of the
barrier. I realize that's a mapping theory rather than a routing one, but I
think it's how I'd be inclined to map that scenario.

I guess possibly four is that you have some sort of obstruction halfway
across the road, effectively narrowing the roadway to one lane but allowing
traffic in each direction (just not simultaneously). While that's not
terribly uncommon on forest tracks and similarly minimally maintained
roadways, I wouldn't expect it to be likely anyone would actually map
things that way.

>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20250801/6c672fe6/attachment.htm>