2013/5/28 Tac Tacelosky <[email protected]> > I thought the best practice was to define a building as a way, then > put a bank node within the building, and possibly at ATM node as well. > I guess it could be defined as a relation. >
Yes, one option is to do it like this, but generally if something is bigger than "punctual" you better represent it as an area, as this will allow for more precise positioning of the relative positions of the things in respect to each other. It also makes sense to use different objects for the building and the user of the building, as this allows to reduce ambiguity (e.g. otherwise you couldn't tell whether "start_date", "name", "wikipedia", "architect" and many other tags refer to the building or the user (the bank in this case). To avoid having to duplicate the geometry you can for example create a multipolygon relation for the shop/bank, which has the building as outer member (in the case it occupies the whole building footprint, otherwise you'd have to make it smaller). This technique is also useful in other cases, e.g. if you have a fenced "something" you could tag barrier=fence on the outline, and make a multipolygon relation for the "something" which is in the inside, hence differentiating between the area and the linear object. > > It doesn't appear that there are many ways with amenity=bank, they're > almost always nodes, even though they're almost always in buildings > big enough to justify a way. I guess it's mostly a matter of how much > data there is on the building footprint. > yes, it is easier and quicker to map a node, that's why in the first iteration banks are mostly mapped as nodes, but there are more than 10% ways as well: http://taginfo.openstreetmap.org/tags/amenity=bank cheers, Martin
_______________________________________________ Tagging mailing list [email protected] http://lists.openstreetmap.org/listinfo/tagging
