On Thu, Oct 16, 2008 at 5:28 PM, Andy Allan <[EMAIL PROTECTED]> wrote: > On the other hand, putting the information directly on the street > limits the ability to produce useful things like maps with numbers on > the building outlines. So I'd say we should go for "numbers on houses" > (e.g. Karlsruhe scheme), and downgrade using post-processing to > "numbers on streets" whenever there's such desire / technological > limitations.
since the house number (or flat number within a block) is an attribute related to both the house location *and* the containing street or block, it seems natural to use a relation to gather the data. one problem is that the accuracy of our tagging is exceeding the granularity of openstreetmap's data model (roads are not one-dimensional) so things like tagging pavements (sidewalks), houses set back from the street and bus stops are bound to be problematic. the dilemma is whether to tag locations accurately (bus stops off the road, houses are polys, train stations are polys / have multiple exits) or "topologically" (bus stops part of way, house numbers on the way, train stations are points which all the lines go through). the karlsruhe schema is clearly in the "accurate" camp :-) cheers, matt _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk