Thanks for the explanation from a UK viewpoint. Do you maybe have an example (image) of a detached house that is not single family? I still find it a bit hard to understand the difference.
I am aware of this. I normally map the landuse=residential around blocks of houses so I have a single building type within a residential area. For larger residential areas you can just use ; separated values.
I am however starting to doubt the use of residential=* a bit after some comments here. Maybe some more general residential values are enough like urban, rural and one in between. The building types can then be defined on the buildings them self.
Just my opinion, but this whole thing seems pretty unnecessary. And by that I mean the gist of the original comment as well as the whole residential=* thing in the first place. Since itās just not a level of abstraction that is at all useful or even verifiable. I can guarantee if you got two people in a room and showed them a satellite image of any semi-dense area that they wouldnāt be able to say where itās urban, rural, or even where the line is. Let alone how ambiguous the idea of a single family residential area is. Not to mention OSM isnāt a zoning map anyway, and for good reason.
Not that there arenāt single-family residential areas, but at least in America they are already being tagged place=subdivision in a lot of cases. Probably thereās a lot of overlap between that and residential=suburb. Which thereās already place=suburb for BTW. Same goes for residential=urban, which is already largely mapped as place=quarter or place=borough. I donāt see the point in creating a duplicate tagging schemes for any of those. Especially in the case of place=suburb/residential=suburb. Just better define the already exiting tags.
In the meantime thereās benefits to mapping things as landuse=residential + place=suburb. Landuse=residential isnāt really a tag that needs further levels of abstraction. Most landuse tags donāt. Like look through the mess created by the various industrial=whatever tags out there, most of which overlap with either man_made=works or craft.
No - although thatās the point. It isnāt verifiable from imagery alone. You cannot know the circumstances of those in a dwelling without knowing those people.
But examples I can think of:
house shares - unrelated people renting the house together, e.g. students or young professionals
room lets - individuals rent out the bedrooms but shared areas, e.g., kitchen, living room, are accessible to all
lodgers - family home with a spare bedroom rented out by a stranger
I think the confusion arises because the term is not used in this way in the UK and Ireland. You seem to be interpreting the words literally as describing the people who live in the building. I would have done that too before living in Spain where I became aware of the analogous concept of a āvivienda unifamiliarā. My understanding is that in this context it means something more like āresidential structure of a kind that would typically be occupied by a single family who do not share any common areas with other residentsā.If a particular house happens to be shared by unrelated people, that would not change its nature as a āsingle family houseā.
Good question. There was a guy awhile back on the east coast of the United States who deleted a rather large residential area just because there was a convenience store near the middle of it. He wasnāt at all receptive to me telling him it didnāt really matter either I could see same thing happening with these tags. Like someone deleting a residential=single_family area just because thereās a few apartment complexes mixed in with the single family houses or whatever.
Youāre certainly right. Which is why I recommended not using this as a value. It isnāt what it says it is - especially in British English. You have to know the additional context.
We use building=detached to describe this sort of residence, not building=single_family.
place=subdivision is not all that commonly used. I was one of the first mappers to use this tag back in 2008 or so, but it never caught on. tagging and talk-us mailing list discussions from around that time seemed to indicate acceptance of named landuse areas as an alternative, not only for residential subdivisions but also for office parks, strip malls, and other planned developments.
These planned developments are completely orthogonal to the OSM English term suburb as well as the American English term āsuburbā. Indeed, all over the Midwest and Plains states, there are plenty of these named residential landuse areas by reservoirs out in the middle of nowhere, representing planned lakeside developments marketed at retirees.
Sure, but still a lot more used then residential=suburb. Maybe not residential=single_family, but I donāt think they are analogous. Although, honestly I donāt really care that much about it. Except that I feel like a āsubdivisionā is something that is more easily mappable, verifiable, and would lead to less issues in the long-term. Like we almost always know where a subdivision starts and ends, but with a suburb the boundary is essentially where ever the mapper feels like ending it. I donāt think suburbs are clearly defined, agreed on concept in landuse or urban planning either. Whereas, subdivisions clearly are.
To be clear, I was only reacting to the suggestion about place=subdivision as opposed to a named landuse=residential.
Iām also not in favor of residential=urban/suburban/rural. These values donāt reliably add any additional meaning to the landuse area. Itās entirely possible that one mapper may use one of these tags to imply housing density, building construction style, or some vague notion of proximity to an urban center (but we store coordinates for that purpose), all the while another mapper or data consumer interprets the tag completely differently.
The residential values I consider most useful are apartments and trailer_park. These values not only describe the predominant form of building construction but are also typically applied to planned developments large and small, distinguishing them from the informal landuse areas that mappers blanket the map with as a first pass. These values correspond to the English terms āapartment complexā and ātrailer parkā (or more polite synonyms like āmobile home parkā).
Iāve occasionally used single_family in this manner as well, just because it seems to be common. However, I would prefer something more general, because single_family excludes otherwise identical subdivisions that consist of duplexes, condominium row houses, or some mix of configurations within the same planned development. I could get behind residential=subdivision, but it risks being misinterpreted as a legal subdivision. (Typically a planned residential subdivision consists of multiple legal subdivisions on a plat map, one per development phase.)
In some parts of the U.S., entire subdivisions consist of duplex homes. A duplex consists of two conjoined single-family dwellings but is otherwise constructed in the same manner as a single-family home (detached house).
You could think of this as the minimal base case of a terrace in many cases, but there isnāt necessarily any symmetry, and an over-and-under configuration is also common in some places.
it is less a question of the number and more about the type of shop and their size. Is it small shops for the residents, is the character of the lot commercial or residential? If there is a single shop with noone living on the plot, I would tag it as landuse=retail and not as residential.
But these are usually mapped as building=apartments anyway. I donāt think anyone would argue that these should be in a landuse=retail. On the other hand: some probably would
For an āeither orā decision, e.g. in the mentioned German city centres, a distinction could be made whether these upper floors are indeed apartments, or maybe offices, studios etc.
Last year, the community rejected a proposal for indicating additional land uses besides the one that lends the area its overall character.
In the specific case of an urban shopping district that occupies the first floors of apartment buildings that back out into residential quarters, perhaps there could be a landuse=retail area that covers the street, storefronts, and part of these buildings, with the landuse=residential areas covering the rest of the same buildings.
In urban areas where I map, there seems to be a preference for keeping things simple, letting the use of the ground floor dictate the landuse tagging, unless the storefronts are only there to take up space (typical of a luxury apartment complex that fronts the noisy street). I donāt think thereās a rule saying landuse areas have to be consistent with the airspace above. This approach is plausible in these areas because the residences above the storefronts tend to be rather anonymous.
Iāve always assumed thatās mainly due to the main type of traffic the area receives. Like with a multi-story building that has residential buildings on the second floor and retail on the bottom in the Bay Area they will often get large freight trucks delivering to the stores, but the people in the apartments either donāt drive or park their vehicles off site in parking garages. Plus they arenāt really their during the day either. The roads arenāt really tagged as highway=residential either due to the heavier traffic they usually get. So it doesnāt make as much sense to call it a residential landuse area as it does to call it a retail one.
Mixed use tagging is something Iāve always thought would be interesting, but it likely muddy the waters way to much. IMO OpenStreetMap is at itās strongest when things are being mapped/tagged as single, clearly defined concepts or whatever. Even in cases where thatās not always the case on the ground, itās still better that we operate like it is.
I could see same thing happening with these tags. Like someone deleting a residential=single_family area just because thereās a few apartment complexes mixed in with the single family houses or whatever.