GTFS input with a missing service_id in calendar.txt and/or calendar_dates.txt used in trips.txt will cause a missing reference

[skinkie]

@ue71603 your opinion please.

GTFS input with a missing service_id in calendar.txt and/or calendar_dates.txt used in trips.txt will cause a missing reference in NeTEx. I think this completely matches what is actually specified as input. But but the meaning is different.

I think that a trip without an availability condition or calendar could be considered always active or an opposing view would be never active. I don't think we have ever defined these semantics. Obviously from a GTFS perspective the calendars are 'sort of inclusive' but GTFS itself also does not support mulitple entries in calendar.txt, it does for calendar_dates.txt.

We could say: if you have invalid (service_id) reference then we could make an empty entry in calendar.txt (already in DuckDB). What do you think?

[derhuerst]

Jumping in here with my opinion, as an outsider watching this project. 🙈

A GTFS feed without service_id is invalid because service_id is required. It is still up to you if badger should try to recover and somehow handle invalid feeds, but the most consistent option would be to just reject them up-front.

[skinkie]

Jumping in here with my opinion, as an outsider watching this project. 🙈

Imagine you have a celebrity meeting you at a concert! Hi there!

A GTFS feed without service_id is invalid because service_id is required. It is still up to you if badger should try to recover and somehow handle invalid feeds, but the most consistent option would be to just reject them up-front.

I have seen feeds in the wild (read: Belgium) for years that are invalid in this perspective. It is a must to handle this.