[RFC] Refining `service=alley` with 2 new service tags - Proposal

I like the approach. However, key:residential is currently only a classifier of landuse=residential, and we don’t have subtupes of highway=residential. Not that this is unseen before though – see the mishmash at key:service.

1 Like

That’s true but hopefully a data consumer interested in buildings will first check if the way has building=residential and someone interested in streets will look at highway= first, so the clash shouldn’t be too much of an issue :slight_smile:

Just so I’m clear, what you’re proposing is

highway=service
service=alley

for “back lanes of Toronto” (e.g.), and

highway=residential
residential=alley

for the narrow Italian street you had the Streeview link for?

Not a bad idea, but under such a scheme what is this lane in Toronto? It’s still just a ‘service’ space, really, but it serves only residences…

Yes exactly.

The one you linked to (this lane on OSM) would still be highway=service service=alley. It might be in a residential area, but it serves the back of residences - garages, sheds, parking spaces - and as far as I can see, no one actually lives on it, so it wouldn’t be highway=residential.

In contrast, the narrow Italian street provides access to the front of properties.

I think you can probably see where I’m going with that though, right?

  1. residential=alley could easily be misinterpreted to mean “any ‘back lane’ in a residential area”, and,
  2. “laneway homes” are slowly becoming more prevalent in North American cities: do we change a highway=service, service=alley to highway=residential, residential=alley when some percentage of the frontage of the lane is occupied by these “backyard suites”?

True. The distiction between an alley that is a residential street and an alley that is a service road (in the OSM sense) gets muddy when you look at a case like this mews lane where the former stable buildings have been converted into houses. But with highway tags there will always be edge cases. Isn’t it better to have two tags to choose from than to use the same tag (highway=service service=alley) for both the Toronto example and the Rome example, only because they are both known as alley in English?

1 Like

Both highway=service service=alley and highway=residential residential=alley say to a human that the way is an alley and an alley, respectively. The only apparent difference between the two kinds of alleys is that the latter serves residences. It turns out we already have a tag for that, one of the oldest tags in OSM: abutters=residential. Or is residential=alley saying something beyond the fact that it’s an alley?

while this is undisputed, I do not think it meets the expectations of a residential road, where in the extreme case even tiny motorcars might not pass physically. I understand that the idea is to express this fact with the subtag “residential=alley”, but to me it seems as if it would lead to many problems for actual data users, at least in a transitioning period, where motorists will be led into such streets just to discover there isn’t actually a street. In contrast, a new highway class will lead to missing streets in the transitioning period, which seems much saver and likely less problematic, at least for motorists.

1 Like

Using highway=service in medieval cities is tagging for the renderer and should be robot-replaced by highway=residential.