Open nvkelso opened 8 years ago
This is comparing two different things: The roads
query is looking at the railway
key, but the transit
query (via mz_calculate_transit_level
) is looking at the route
key.
The difference is that railway=
is generally tagged on the way which represents the geometry of the physical track, whereas route=
is generally tagged on a relation containing not only the collection of track geometries but also the stations and/or stops which make up a regular named service or line.
We should include route=monorail
(49 instances) and route=narrow_gauge
(2 instances), but we need to be clear that these are not the same tags - and we shouldn't expect a route=light_rail
to necessarily contain all/only railway=light_rail
.
Also, it looks like there are railway=subway
ways (32,543 instances) - should we look at including those too?
The transit layer does not include railway=narrow_gauge and railway=monorail, while the roads layer does. Seems like they should be consistent (and pack them at zoom 10 like subway and light_rail).
roads.jinja2
functions.sql