Because #22 pre-generates a restricted index of all GTFS stops prior to loading them. Current code loads full matrices and then reduces size straight away, but would be a perfect use case for arrow. The loading for each batch of vertices is also quite slow, and arrow should considerably improve the efficiency of that.
Because #22 pre-generates a restricted index of all GTFS stops prior to loading them. Current code loads full matrices and then reduces size straight away, but would be a perfect use case for arrow. The loading for each batch of vertices is also quite slow, and arrow should considerably improve the efficiency of that.