I am more likely to use your option 1: Each shopping center a separate landuse=retail, name=* covering the entire area including parking but not the land associated with the roadway right of way. As an example there are two named shopping centers at this intersection: http://www.openstreetmap.org/#map=18/37.36843/-122.03593
I would really avoid using locality=* as I associate using that for a populated
place.
I think that relations could work in theory but it seems that they are poorly
supported/understood by the average editor/mapper and the complexity is not
needed to capture the information.
Not sure what your item 2 was going to be. :)
Cheers
Tod
On Dec 23, 2014, at 6:40 PM, Doug Hembry wrote:
> I'm a relative newbie, and here's a question I've been puzzling over for a
> while: What's the best practice for tagging a north American outdoor shopping
> center? For example, often, on an intersection between major suburban
> streets, there are collections of stores, in one or multiple buildings,
> grouped around one or more shared car parks. And they have names
> ("Cornerstone", "Kings Court",... or whatever). Sometimes there are four
> groups of stores, one on each quadrant of the intersection, with four
> different names. In the past, someone may have tagged the whole general area
> with landuse=retail (or landuse=commercial - not sure why the difference),
> but the map doesn't know of, nor display, the distinct identities (which
> are frequently used locally in ads, etc). How to incorporate these distinct
> names, and if possible have mapnik display something? I have considered or
> seen several ways:
>
> 1. Split a big generic landuse=retail area into multiple smaller
> landuse=retail polygons, one for each shopping area. Then there are issues
> about whether adjacent areas should share boundary nodes with each other, or
> with separating roads. It gets complicated, and tedious to implement.
>
> 3. I've seen place=locality used on a single node with a name=*. It displays,
> but place=locality is supposed to describe an uninhabited region, according
> to the wiki.
>
> 4. Is this a legitimate use of the site relation? Buildings, shops, car park
> areas, gas stations, etc, could be grouped together and named, perhaps with a
> label tag, and no explicit boundary way required. The boundary of a shopping
> center is usually fairly obvious when viewing the map - a drawn boundary
> might not be considered essential. This is attractive, but are site relations
> approved at this point, and will Mapnik display their names (I know... don't
> map for the renderer...)? Plus, I've never seen this used.
>
> Breaking up a big landuse=retail area seems clumsy and problematic. And I
> suspect the usage of landuse=retail is supposed to be a generic, "broad
> brush" classification of a whole region rather than a way of identifying
> smallish distinct contiguous areas, identical except for their names. What I
> think I need is a shop=shopping_center tag (or shopping_centre, if our
> European colleagues insist :) ), applied to either a strategically placed
> node or a newly defined boundary way. But it doesn't exist, strangely. Note
> that shop=mall isn't right, because malls are explicitly indoors. Maybe it's
> only here in California, where it never rains ( dark humor. At least until
> very recently) that we have this phenomenon of outdoor shoping areas, but I
> don't think so. Note also that single isolated shopping areas are not a
> problem - the landuse=retail area can simply be given a name=* tag. But for
> the more complicated cases - any suggestions?
>
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ Talk-us mailing list [email protected] https://lists.openstreetmap.org/listinfo/talk-us
