On Jul 12, 2018, at 1:46 PM, Jarek Piórkowski <ja...@piorkowski.ca> wrote:
> Damien's question appears to be about nodes like
> https://www.openstreetmap.org/node/438843513, which has
> name=Berri-UQAM, operator=Société de transport de Montréal.
> short_name=STM seems inappropriate here, we could do
> operator:short_name=STM or something but it seems a bit much.

Thank you for your analysis and reporting to the list, Jarek!  Yes, I agree 
that operator:short_name=STM is a bit of "overkill" (getting over-specific on 
the key side).

> The nearby station https://www.openstreetmap.org/node/26233453 has
> name=Jean-Drapeau, network=STM, operator=Société de transport de
> Montréal which seems like an attempt as good as we might get. Commuter
> rail station https://www.openstreetmap.org/node/548900549 has
> network=RTM, operator=Réseau de transport métropolitain which fits
> that scheme as well. Similar with a random bus line on North Shore
> https://www.openstreetmap.org/relation/3472432

I find that operator=* is a key which certainly applies to "underlying rail 
infrastructure" objects (railway=rail), especially when the rail is 
freight-oriented, though I have also seen operator=* set to the value of the 
passenger operator when the underlying infrastructure is one of 
[railway=light_rail, railway=subway, railway=tram] on more passenger-oriented 
rail.  Though, I seem to recall more frequently (I'd have to do some Overpass 
Turbo queries to confirm this) network=* is applied to the passenger (not 
freight) elements instead of operator=*, both are used, both seem correct.

Without getting "lost in the weeds," there are/were three "levels" of railway 
route relations:  #1 is/used to be route=tracks (largely if not completely 
deprecated in North America, but maybe still used in Germany), #2 is 
route=railway (a grouping of what we in N.A. call "Subdivisions" or "Branches" 
or "Industrial Lines") and #3 is route=train relations for passenger rail.  We 
can (and do) have passenger rail as route=train relations all over N.A. withOUT 
the "underlying infrastructure" of route=railway relations, but I, others and 
indeed OSM consider this incomplete and rather sloppy.  The Germans use all 
three (or did).  The Bottom Line for what we in N.A. should do is to use BOTH 
of the "middle-" (#2, route=railway) and #3 "higher-" level (route=train) 
relations to describe "track infrastructure" and "passenger rail routes."  OK, 
thanks for reading all that, it makes a better OSM.

> Looking through map very casually I didn't see any operator=STM on the
> subway. I did see it on a bus line
> https://www.openstreetmap.org/relation/270258 but changing it to
> network=STM and operator=Société de transport de Montréal seems like
> it'd be fine there IMO.

Yes, again, I agree.

> To me "operator" looks a bit more little technical than the other
> tags, so to me it would be alright to use the longer more formal name.
> But I wouldn't edit-war anyone about it. I'd say run a query, see
> which is more common currently, ask people here (as you've done), then
> after a week change the minority tags to match.

You saying "more technical" might be agreeing with me that operator=* is at a 
"lower/middle level" (infrastructure on track, not "higher level" as applied to 
the different relation of route=train for passenger rail).  So I think we are 
largely in agreement:  if you (and Canada) want to move into the direction of 
putting operator=* on freight rail (and maybe sometimes passenger rail), yes, 
that seems correct.  If you additionally want to use the network=* key for, in 
this example, STM, yes, that makes perfect sense to me as well.  So does your 
suggestion/approach of "run a (OT) query...change minority tags to match."

Thank you for good discussion,
Talk-ca mailing list

Reply via email to