[Tagging] How to tag recreational route with multiple route types?

[Minh Nguyen]

Vào lúc 10:40 2021-08-12, Brian M. Sperlongano đã viết:
> On Thu, Aug 12, 2021 at 1:31 PM Dave F via Tagging 
> <tagging at openstreetmap.org 
> <mailto:tagging at openstreetmap.org>> wrote:
> 
>     I'm struggling to comprehend the point of this.
>     The *whole* point of route relations was there could be multiple but
>     *separate* routes over the same ways.
> 
> 
> +1
> 
> Route concurrencies (multiple routes sharing a single physical road or 
> path) is most cleanly represented in the data model by multiple route 
> relations, each of which claim the road or path as a member.  This is 
> how OpenMapTiles currently deals with concurrent routes for roads, which 
> will be implemented in their next release.  The more we can get in the 
> habit of creating route relations, the easier it will be for route-aware 
> applications and data consumers to work with route data.
> 
> A "route" is a single feature and deserves a single object (consistent 
> with the one feature, one object concept).  Attempting to pack 
> concurrent route information into individual member ways is simply 
> distributing the route feature across many objects.  This makes work for 
> data consumers and mappers alike and in my opinion is more complex than 
> having a single route relation carry this information.

It sounds like we're talking about cases that aren't concurrencies at 
all, but rather a single undifferentiated multipurpose route, where 
OSM's normal route tags are too rigid. Where I map, multipurpose routes 
are commonly thought of as bike routes even by pedestrians and 
equestrians, so I just tag them as cycling routes. But that could 
conceivably differ from region to region. I suppose if we were to start 
from scratch, it would've been more correct to say route=recreational 
and use [access]=designated and network=[country]:[...] to clarify from 
there. But here we are.

These cases are probably most relevant to recreational routes because it 
isn't impossible to swap a bike for a horse. We don't have this problem 
with road routes, because route=road alone doesn't distinguish between 
national and regional routes or between car and truck routes.

An analogous situation with public transportation routes would also be a 
bit contrived. Maybe somewhere out there there's shared bus rapid 
transit/light rail infrastructure and it's anyone's guess as to which 
kind of rolling stock the transit agency would run on a given day.

Even so, multimodal routes do occur. For example, in the San Francisco 
Bay Area, BART is generally a route=subway system, but its 
Antioch-SFO+Millbrae line involves transferring to light rail on the way 
to Antioch, as a cost-saving measure. Just looking at station signs, 
it's all one line and you might not know about the transfer until you 
get to the transfer station, which is only accessible from a BART vehicle.

So we've modeled it as a route=subway and a route=light_rail, both 
members of a route_master=subway;light_rail, and some of which share 
ways with other distinct routes. [1] If the line again gets extended on 
the cheap, maybe we'll need to make it 
route_master=subway;light_rail;bus and eventually bicycle. :-D But that 
route is multimodal in chronological order, whereas these recreational 
routes are multimodal simultaneously.

[1] https://www.openstreetmap.org/relation/2827687

-- 
minh at nguyen.cincinnati.oh.us