[Tagging] Feature Proposal - RFC - value 'basic_network' - cycle_network?

[Sebastian Gürtler]

Am 28.11.21 um 01:48 schrieb Brian M. Sperlongano:
> On Sat, Nov 27, 2021 at 5:02 PM JochenB <JochenB at wolke7.net
> <mailto:JochenB at wolke7.net>> wrote:
>
>     Am 24.11.2021 um 05:10 schrieb Adam Franco:
>>     As I've been reading this long discussion the thought that keeps
>>     coming up in my mind is that "relation are not categories":
>>     https://wiki.openstreetmap.org/wiki/Relations_are_not_categories
>>     <https://wiki.openstreetmap.org/wiki/Relations_are_not_categories>
>     These giant relations are bad, but they are not categories. They
>     would be categories if one could also recognize without these
>     relations that the ways are part of the network. But this
>     information is only mapped in OSM when the ways are included in
>     the relation.
>
>
> I would beg to differ, as these giant relations seem quite
> specifically to be categories.  I see them as no different from taking
> every way that is part of the US Interstate Highway System and placing
> them in a single relation called "Interstate Highway System".  Surely
> the US Interstate Highway System is a "network" as the roads are all
> interconnected.  However, there is absolutely positively no need to
> shove them all into a relation.  Instead they are all members of
> individual route relations, tagged type=route, route=road, network=US:I.
>
> It seems that the challenge here is that you have all of these cycling
> ways which are certainly part of an interconnected network, though
> they are not part of any named and numbered route.
>
> I see this as very simple - we have a tag for this, network=lcn[1]. 
> These ways are all part of a local cycling network, so tag them that way.
>
> The wiki page for network=lcn indicates that this should be tagged on
> route relations, however, I note 2,500 usages of that tag on ways. 
> While this is small compared to the 38,000 usages on relations, it's
> clearly in use on ways directly.  I would propose to solve this
> problem quite simply by updating the documentation to reflect this
> real-world usage -- network=lcn is is being used on ways, and could be
> tagged directly on highway=cycleway features.
>
> If you want to query for the "basic network", now you just query for
> highway=cycleway + network=lcn.
>
> [1] https://wiki.openstreetmap.org/wiki/Tag:network=lcn
> <https://wiki.openstreetmap.org/wiki/Tag:network=lcn>
>
I try to add not too much information in this discussion, we shouldn't
discuss on the tagging only ways versus relations too much here, for the
proposal aimed at another thing.

But: I think it is really unfeasible to avoid the relations.

* Quite special situation but really a thing where I have no other
solution: We have situations where the "basic network" crosses other
routes that are signposted as well but there are strictly separated so
that if you come from one route you won't see the signs of the other.
You only give a representation of that if you create route relations. If
you only tag the ways then you can't discriminate the possibility that
there are signposted branches on the crossing (which make up a node).
(Examples
<https://wiki.openstreetmap.org/wiki/User:Segubi/Elemente_NRW-Radwegenetz#Crossing_of_different_networks>)


* I join the strong opinion that the routes in mention are also routes
in osm-sense, they differ from streets as you have no characteristics of
the ways as in streets, the bicycle routes use all kinds of ways "as a
guest" the ways are not continuously cycleways. For cars: A motorway is
a motorway, a national street is a national street, they may not use
temporarily a foot path for a few 100 meters and continue again as a
motorway. As Peter Elderson stated - this would be close to tagging for
the renderer.

("If you want to query for the "basic network", now you just query for
highway=cycleway + network=lcn." doesn't work! And you would have to tag
the nation wide network with lcn, where there have been lots of
discussions on it without an agreement for over a decade.)

* A agree that the huge network relations are not useful, and that these
risk to be as useless as a mere collection (which the superrelations
like "Radverkehrsnetz NRW" in fact are not. These are distinct entities
which you can define on the basis on ground). A reduction to at least
linear route relations makes it easier to maintain the network as you
can check and correct the small units that you have seen on ground
instead of searching for the single ways and handling the branches which
may be missing in the data.

Sebastian


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20211128/a5ce2f4b/attachment-0001.htm>