[Tagging] Feature Proposal - RFC - value 'basic_network' for keys 'network:type', 'lcn' and 'lwn'

Peter Elderson pelderson at gmail.com
Thu Nov 18 21:11:50 UTC 2021


Volker Schmidt:

> (I still don't understand what this basic network is supposed to be)
>

I think I understand. The governmental bodies call it a network, because
that's what they any road system.  And, if you look at the maps showing
them, it looks like a net.

This net is a system of designated cyclable ways, implemented on the road
with Destination based guideposting, and, where it suffices, just pointers
to the way or lane all cyclists should take if they want a certain level of
cyclability.

This is all visible and verifiable, so if mappers want to map it, fine.
Of course they also want to use this after putting in all the work, they
want to see it on the map and they want their planner/router/navigation to
use it.

You could tag all the ways with a tag that says "designated cycleway with
quality N". That's fine for cycleways; with paths you have to consider
access as well, and with roads you have to consider use_sidepath, separate
cycle lanes (left, middle, right) AND access tags, maybe conditional; etc
etc. ... I can see this can be complicated, for mappers and renderers and
routers.

So the idea became to create a relation, throw in all the ways, paths, etc
having this preferential designation. Then the only thing data users have
to do is check the relation; if the way is in there put in the weight. As
it happens some cycling renderers and routers use route relations for neat
rendering and preferential routing, so add type=route to the relation and
BIYU!

The down side is that it is not a route but a collection. Changing the
relation type loses the rendering and the routing. And maintainability is
reportedly low, and a constant PITA, I can imagine that.
The solution is to have the collection relation contain routes, and the
routes contain the ways. Then you keep the nice rendering and the route
relation preference of the router. Maintainability is served, because
changes only affect the lowest level: short routes. And you can make a
hierarchical network: a network of networks at the top, the routes in the
lower network level, as deep as you want, which neatly fits the cycleway
network hierarchies in Germany. As long as the networks can be found on the
guidepost and shields, there is verifiable ground truth.

There is one thing to do. The route relations have to be separate routes!
They should not overlap (too much), each route needs a starting way and an
ending way, the routes should (for the most) chain neatly together. How do
you choose the routes to achieve this, in a way that less expert mappers
can understand? That's where the guideposts come in. Let the routes go from
guidepost to guidepost. Intermediate signs (bicycle logo with arrow) are
not mapped as features, but they indicate the route between the guideposts.

The down side here is the burden to create and maintain a very fine-grained
system of guideposts and small routes. Another issue is that a large amount
of small routes is created, which do not have a name or other presentable
label, at least not one that's found on the road.

I think that's what the proposal targets by proposing
network:type=basic_network. It allows an application to filter these small
unnamed routes from lists, and also enables them to give them special
rendering. Routers could base specific handling on this attribute.

This is my current understanding. I am still not sure if the result
justifies the means, in particular the huge maintenance task if you want to
cover the whole of Germany. I think it can't be done on this large a scale
without special tooling and error detection. The structure is bound to
deteriorate over time unless it is very tightly maintained by dedicated
mappers, and if it's going downhill, it becomes useless very fast.
Porting it to a comparable foot network is a factor more difficult. In
Nederland, I think maybe for horse riding it could have worked, but Dutch
governmental bodies are now implementing Node Networks for horse riding.

Fr Gr Peter Elderson
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20211118/8e810d33/attachment-0001.htm>


More information about the Tagging mailing list