On Mon, 2008-01-14 at 12:49 +1300, Robin Paulson wrote: > On 14/01/2008, Jon Burgess <[EMAIL PROTECTED]> wrote: > > > areas must form complete loops, otherwise they are treated as ways by > > > the renderer > > > > > > > The problem is the opposite: You've got a way forming a circular shape, > > how do you know it is circular loop (e.g. like a moat circling a castle) > > and not an area? > > by drawing it as an area: given that the moat has a width, it can be > drawn as an area, forming a loop. imagine it as a rectangle, where one > end of the rectangle is extruded in a curve, to butt up to the other > end
Yes, but this does not answer the question about how a renderer is meant to interpret a closed way. If you are forcing all rings to render as areas then all roundabouts and ring roads might end up being treated as areas. In the past, none of the items in Map_Features applied the same key to both linear features and areas. Recently highway=pedestrian introduced the idea that area=yes should force a closed way to become a pedestrianised area instead of a simple loop. Now I see a few more keys being added like waterway=dam which apply to both linear features and areas these will cause ambiguity for the renderers unless an extra key like area=yes is used. The Mapnik osm2pgsql import code has hard coded knowledge about each key to indicate whether a closed way should be interpreted as an area or as a loop. My main concern is that the more complex it is to interpret the keys the longer it will take to get the correct support in the renderers and the more likely you'll see inconsistent rendering on the different map layers. Jon _______________________________________________ talk mailing list talk@openstreetmap.org http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/talk