On Tue, Jun 2, 2009 at 12:33 PM, Jonathan Bennett < [email protected]> wrote:
> > "Relations are not Categories" uses the argument that you can query for > > all objects that carry a tag. This isn't enough, however, if the > > combined information from tags and coordinates doesn't qualify as an > > unique identifier. The situation for architects, imo, resembles the > > route relations (also in that there can be more than one architect), > > thus there is a case for relations. > > I disagree. The reason we use relations for routes is that each section > of the route are part of the same "thing" -- a signed route that uses > many different sections of highway. While each individual section may > have some other identity, the route exists as a separate abstract > entity. A route is also a geographic entity. > > Buildings designed by the same architect don't form part of some greater > whole -- it's just metadata about the buildings themselves, and the > relationship between them certainly isn't geographical. That's why using > a relation to group them would simply be classification. > I agree - think there's a broad agreement on this now. Whilst having to deal with ambiguities and synonyms isn't ideal from a perfect data perspective, I think it fits in with the OSM practise of using unrestricted tagging in the first place, and making it as easy as possible to contribute data. Incidentally, the same problem exists for operator=*, and I notice that there's a proposal to use relations for that: http://wiki.openstreetmap.org/wiki/Relations/Proposed/Operators Frankie -- Frankie Roberto Experience Designer, Rattle 0114 2706977 http://www.rattlecentral.com
_______________________________________________ talk mailing list [email protected] http://lists.openstreetmap.org/listinfo/talk
