Do we actually need all that, or is just url= sufficient?
I think it should at least be changed to website=*
since URL means something less specific. I guess for stops which only have one ID it would be fine to just use bare website=*
(probably also fine for railway platforms since there are none with multiple), but I was thinking that using the suffix even if there is only currently one ID is more fool-proof if the stop later gets another ID.
Stations should just have website=*
though, otherwise youād just have a bunch of e.g. website:*=https://jp.translink.com.au/plan-your-journey/stops/morayfield-bus-station
even though the URL is the same for each.
Is that necessarily a problem though?
Itās not, it just seems unnecessarily verbose to have e.g.
amenity=bus_station
public_transport=station
name=Morayfield bus station
website:AU-QLD-Translink-SEQ=https://jp.translink.com.au/plan-your-journey/stops/morayfield-bus-station
website:AU-QLD-Translink-KIL=https://jp.translink.com.au/plan-your-journey/stops/morayfield-bus-station
when just website=https://jp.translink.com.au/plan-your-journey/stops/morayfield-bus-station
would convey the same information in less space.
Ah, gotcha!
I was thinking separate stations all redirecting to the same website.
My understanding of gtfs is that a route just provides extra information to a trip. which would then not make it incorrect to have different directions not share a route. Riders and other systems may consider multiple gtfs routes to be one line irl, even if they have to be two or more routes in gtfs.
An example of this would be line 27 in Stockholm where express departures use a different line number, 27S. In the underlying data these are the same line, but in gtfs they are not which is fine given the purpose of gtfs, that being to inform a rider of what train to take.
Personally I would add gtfs:route_short_name:= to the osm type=route(s) since itās the closest but also keep it on the route master in some way to best support applications which expect route_master to equal route.
In reference to:
is there a difference reference between the Beenleigh > South Brisbane (& return) express vs āstopping-all-stationsā trips?
So far Iāve decided on splitting the route masters into two, because I had a realisation here:
So the questionable use of a semicolon in a gtfs:*
tag isnāt a problem anymore (I think a GTFS ID can actually contain a semicolon, although whether any existing feed actually does that is another question). Some of the routes also donāt have a āfriendā that goes in the opposite direction with the same termini.
You can see the way Iāve tagged things so far in the Trains section on the PTNA page (thereās only three so far because I havenāt updated any of the others yet).
What Iāll probably do is create a route=railway
relation for each line so Wikidata has something to point at (e.g. Caboolture railway line (Q5015753), which IIRC is currently pointing at the CABR (Caboolture to Roma Street) route master), and just keep each train route in a separate route master for simplicity.
Looking at the printable timetables for the Beenleigh and Gold Coast lines and Iām not really sure (not really helped by the fact the timetables donāt use these refs in the first place). Either way they should have different shape IDs.