Using the CLI with default parameters, matching sometimes generates stubs instead of streets. Here's an example from SF:
Those stubs are ~<2m in length. The matcher usually fails silently on the rest of the street segment. This is true for most of downtown SF:
I don't see a reason in the data for why this fails. The (current) OSM geometries are tagged highway=residential or highway=tertiary. The OSM geometries are sometimes 2-10 blocks long, but that's true in other areas that do match properly.
Using the CLI with default parameters, matching sometimes generates stubs instead of streets. Here's an example from SF:
Those stubs are ~<2m in length. The matcher usually fails silently on the rest of the street segment. This is true for most of downtown SF:
I don't see a reason in the data for why this fails. The (current) OSM geometries are tagged
highway=residential
orhighway=tertiary
. The OSM geometries are sometimes 2-10 blocks long, but that's true in other areas that do match properly.Here's a geojson for that first image:
stubby_matches_sf.geojson.txt