[Tagging] cyclist profiles - was:Feature Proposal - RFC - value 'basic_network' - cycle_network?

[stevea]

On Nov 27, 2021, at 3:53 PM, stevea <steveaOSM at softworkers.com> wrote:
> 
> On Nov 27, 2021, at 3:25 PM, JochenB <JochenB at wolke7.net> wrote:
>> 
>> Am 26.11.2021 um 02:44 schrieb stevea:
>>> So, do so ..., but please don't do so with network:type ... .  Because it really, really is confusing to say this is a "type of network."  
>> On the one hand, I would like to counter: correct, but too late. 100% of our node networks have this key with the unhelpful name. Even with 'network=lcn/rcn/ncn/icn' and 'cycle_network=<country>:<state>:<destrict>:<...>' it is too late, here too it should actually be added which type of Classification takes place in the key
> 
> I'm certain that "poor legacy tagging" has badly stung OSM before so that "too bad, we're stuck with it."
> 
> This is the first I have heard exactly this expressed about the problem being solved.  Or maybe understood that.

To fortify this, I'm OK (and don't think of it as "broken or unhelpful (legacy) tagging" for us to have invented network:type=node_network.  The reason for this (as I've described and I'll briefly repeat) is that "networks" of routes, routes made up of ways (as were the original specifications for such routes in OSM going way, way back to our beginnings) didn't need "an extension to the network key" as they WERE the entirety of expressible objects in the network key.  When node networks (primarily in Western Europe, especially BeNeLux countries  — but certainly elsewhere in the world), a new "mathematical" distinction needed to be invented to describe the difference between vertex-oriented routes (the original kind) and node-oriented routes (the new kind).  Maybe "mathematical" isn't quite right, more like "geometry" or "geometric-geographic" distinction between these two fundamental objects in the data structure known as our relational database that is describing graph-oriented data as relations:  with vertices and nodes as members and little else except the tags on the relations.

Yet, extending network:type beyond this, to express the PURPOSE of the route seems all, completely wrong.  It is (as I contend, even as JochenB has defended this as untrue) "lazy syntactic sugar."  The PURPOSE of the COLLECTION of routes in a NETWORK which are BASIC already has a syntax component in OSM.  It is the value portion of the cycle_network=* key (I contend, and nobody has presented a counter-argument).  True, these values are MOSTLY geographic in nature (and should be) and only occasionally or even rarely used to describe the PURPOSE of a network, but THEY CAN, and in my vigorous opinion, SHOULD.  No need to re-invent a wheel, simply extend an existing syntax element to capture the purpose of the network in any given geographic area.  Yes, this will have to be crafted / designed to wide agreement.  But that is the "work undone" here.

I don't consider "poor legacy tagging" what we do for node networks, even if my previous implied that, so my apologies for any confusion, which I hope to address here.