GTFS Map to use duckdb

[davidr1234]

As per @ue71603 request an issue to discuss - @skinkie feedback for https://github.com/skinkie/reference/pull/108

Here's what was discussed so far: me: "I'm not sure about the benefits still. What's the ROI of this? Do I even want to do those checks in this particular case? @ue71603 you probably already have considered this and decided against it, why was that?

As for the adaptations - should we decide accordingly - I am not yet clear on how we generally handle this.

• Does a call to "gtfs_import_to_db.py" suffice?
• What's the data structure of the db? I can find out myself, but that's not the point.
• What would be a good example on how to best read the routes, trips, stops, and stop_times from db? Do we maybe have a script for this purpose too? that maybe fills in dataframes with the data?"

Answers by @skinkie

1. does a call... Yes, this creates the GTFS schema, and does some fixes along the way.
2. what's the... For the added files:
3. what would be... gtfs_convert_to_db.py DuckDB can directly give you the dataframes.

[skinkie]

We can, but the first step is basically the question if GTFS would be the right way to start. I understood you created this code for visualisation purposes of the output (which is good). But if the medium long term goal is to visualise the data, why don't we start with NeTEx in the first place?

From a ETL or MVC perspective I wonder if our view, that being the map itself is something which should accept concepts, which could be stops and shapes (GTFS) or scheduled stop points and routes (NeTEx). My focus would be after the current list of issues is resolved, to get back to the original GUI for editing and visualisation (as input).

[davidr1234]

@ue71603 what are your thoughts?

[ue71603]

We have student work on the list for that.