[Tagging] Clarification on the role link in route relations

[Brian M. Sperlongano]

On Thu, Jan 13, 2022 at 4:21 PM Dave F <davefoxfac63 at btinternet.com> wrote:

> On 13/01/2022 18:30, Brian M. Sperlongano wrote:
>
> On Thu, Jan 13, 2022 at 1:16 PM Dave F via Tagging <
> tagging at openstreetmap.org> wrote:
>
>>
>> I'm intrigued by your Point 3.
>> Please provide a link to it in OSM & details of your rendering (Github,
>> website?).
>>
>> DaveF
>>
>
> The example is in the github link that Minh provided:
> https://github.com/ZeLonewolf/openstreetmap-americana/issues/79
>
>
> Thanks for that, but it provides no info about this new rendering. You
> mentioned Opentilemaps. Is that what they're using?
>

This is getting deeper than the tagging list probably cares about, but yes,
this renderer is using an OpenMapTiles / mapLibre vector tile stack, and
the new version of OpenMapTiles that was released this month adds support
for concurrent road route relations.  The behavior of OpenMapTiles with
concurrent road route relations is that each way contains a list of routes
that the way is a member of (regardless of role).  The front end renderer
then takes that data and uses it to draw one or more highway shields.  The
issue that prompted this discussion was that in a few places, shields were
appearing on ramps TO a highway rather than just ON a highway.  This is
because those ramps TO the highway were added to the road route relation.

I've just added a couple comments describing the possible technical
> solutions to this issue as well at the link one of the objects in question
> on the map, which I'll also paste here:
> https://www.openstreetmap.org/way/587338079
>
>
> Cheers. That way is an 'exit from', not a 'link to' so it shouldn't be a
> member of the VT 15 relation. The roads heading to the V 15 should be
> included with link roles.
>

You've contradicted yourself here.  First you say that way shouldn't be a
member of the relation, and then you say it "should be included with link
roles".  So it's not clear whether you favor including the ramps (with link
roles) or whether you favor not making them members of the route relation.


> However, this is a developer, not OSM database problem. They should have
> filter rules to remove any data they're not interested in.
>
> Your Point 3 - When you say 'Remove relation link memberships from the
> map' do you mean from their  rendering or OSM database? If the latter, I
> disagree.
>

As a community renderer author, I feel that we have a special
responsibility to:
1. Render correctly when data is entered in correctly
2. Render incorrectly when data is entered in incorrectly

Intentionally rendering bad data badly provides appropriate feedback to the
community resulting in better data for all users in the end.  I would
necessarily take a different approach on a commercial product, where I
would attempt to suppress mis-tagging rather than expose it.

So the decision space here is quite simple.  If there is a community
consensus that roads which connect to a route, but aren't on a route,
should be included in a route relation, then OpenMapTiles has a
responsibility to exclude that data when marking which roads are part of a
route (and I would have a PR posted quite quickly to accomplish this).

However, if we agree that such roads should NOT be part of a road route,
then my intent would be to continue rendering them in that wrong location,
in order to give feedback to mappers that there is a problem in the data.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20220113/64725185/attachment.htm>