On Jan 30, 2011, at 5:27 PM, Steve Bennett wrote: > On Mon, Jan 31, 2011 at 11:39 AM, Daniel Sabo <[email protected]> wrote: >> This is a really bad idea. Drawing collinear features by sharing nodes is >> NEVER a good idea beyond 1 or 2 shared corners, that's what multipolygons >> are for. > > Does that apply for coastlines as well? Or do coastlines not even need > the multipolygon?
I'm honestly not sure about coastlines, because I'm not sure how osm2pgsql's coastline removal works, but inland lakes (natural=water) definitely work with multipolygons. >> When the ways get attached to large objects (like an administrative boundary >> or national park) it becomes impossible to edit them from an extract without >> unintended side effects. > > Good to know. I wonder how we can make multipolygons more user > friendly? Maybe a magic "convert to multipolygon" tool, that starts > with one way (say the golf course in the previous example), converts > it to a multipolygon, and converts any other affected ways (the > landuse=industrial) as well. > > Steve I've been thinking about this too, because multipolygons still a lot are harder in Potlatch than JOSM. I would think we could make a tool smart enough to handle it but I don't have any experience with ActionScript programming. e.g. Select two collinear polygons , click "convert to multipolygon", tool make a new way for each collinear segment, then make a new multipolygon for each with all the new segments as role=outer. Right now Potlatch doesn't even render landuse multipolygons though, so there's not much incentive for people to click a button like that :). I do appreciate that my pro-multipolygon perspective comes from working primarily with JOSM where they've gotten very easy to use, Potlatch users probably are as annoyed by working with my multipolygons as much as I am by their collinear ways. - Daniel _______________________________________________ talk mailing list [email protected] http://lists.openstreetmap.org/listinfo/talk
