[Tagging] My "weirdly unnatural aversion to relations"

Kevin Kenny kevin.b.kenny at gmail.com
Tue Oct 2 20:45:27 UTC 2018


On Tue, Oct 2, 2018 at 12:03 PM Paul Allen <pla16021 at gmail.com> wrote:
> Even if a multipolygon can have many disconnected outers, it seems I'd have to make each university
> building an outer.  And then there are no inners.  So even if it can be done that way, it seems like
> an abuse of the concept, which I thought was to be able to punch exclusionary holes in areas.

The one that I pointed you at has a great many disconnected outers.
It's called a MULTIpolygon precisely because it can represent multiple
polygons.

Most GIS software calls your idea of a multipolygon simply a
'polygon', and the OSM concept of a 'closed way' is a 'ring'.  Hence,
in ordinary GIS terminology, a 'ring' must be a simple closed region,
a 'polygon' has an outer ring and zero or more inner rings, and a
'multipolygon' may have multiple outer rings.

If mapping a complex area like
https://www.openstreetmap.org/relation/6360587 as a multipolygon with
disconnected outer rings is an 'abuse of the concept', that's news to
me - and I'd really like to know what the accepted concept for how to
map those is, because I've mapped hundreds of cases like it (well. all
right, it's an outlier at the high end of complexity, but I'm sure
that I've got hundreds of multipolygons with disconnected components).

I don't see adding the lot of an isolated building to a multipolygon
as being any more difficult than adding it to a site.

If the university's holding is an office or floor, represented by a
node in OSM, in addition to land area elsewhere, that might be a call
for a different type of relation, since only ways make sense as
components of a multipolygon. If 'site' is the correct description of
that, so be it.  I haven't yet confronted mapping that particular sort
of beast, but I concede that I can see wanting to map, "the university
comprises this lot of buildings and grounds, as well as operating out
of these offices and storefronts' - and a multipoygon can't do that.

Still, in any case, when a particular region or facility comprises
disconnected parcels of land, a multipolygon with disjoint outer rings
is a natural way to map it, and all the tools cope with that topology
quite well indeed.



More information about the Tagging mailing list