Open barbeau opened 6 years ago
I think one of the biggest questions here that I didn't explicitly say above is "what route_short_name
and route_long_name
should be shown to users for this trip (and any other required fields in GTFS routes.txt
)?"
Right now those values don't exist in GTFS-realtime spec, so if the route isn't in the GTFS data the consumer doesn't know this information.
So in GTFS-realtime TripDescriptor, here's the current definition of route_id
:
The route_id from the GTFS that this selector refers to. If trip_id is omitted, route_id must be provided.
I have always interpreted this as saying that the route_id must match a route_id in the GTFS data.
Quick informal poll of people at Transit Data Workshop says no, right now you shouldn't add routes via GTFS-realtime.
There does seem to be interest in fleshing this out, though, especially in context of changing trip stop sequence order/geometry/adding stops on the fly.
We have a draft spec for GTFS-ServiceChanges (v3.1) which is a good way to add new routes in real-time: http://bit.ly/gtfs-service-changes-v3_1
Summary:
Right now we have rule E004, which flags route_ids that appear in the GTFS-realtime feed but not the GTFS data.
I saw the following
ADDED
trip in STIF's feeds (caveat - it's actually a 3rd party, not their official feed - see https://groups.google.com/d/msg/onebusaway-developers/3lDMoMeF2zQ/u5UXvx5TAgAJ), which has a route_id that doesn't appear in RATP's GTFS data:From a quick scan, it looks like the only reference in GTFS-realtime for
ADDED
is for trips: https://developers.google.com/transit/gtfs-realtime/reference/#enum_schedulerelationship_1...says:
Is adding a new route via GTFS-realtime allowed?
I think we need community clarification on this. If it is allowed, we need to modify E004 to check if the
ScheduleRelationship
isADDED
, and if so no error should be logged.Steps to reproduce:
Run the normal server validator on RATP's data:
Expected behavior:
All route_ids that appear in the GTFS-realtime feed should appear in the GTFS data
Observed behavior:
There is an ADDED trip that includes a route_id that doesn't appear in the GTFS data - is this ok?