On Thu, Nov 07, 2013 at 03:22:52PM +0100, Christoph Hormann wrote: > On Thursday 07 November 2013, Frederik Ramm wrote: > > > > http://www.openstreetmap.org/browse/way/239992362 > > > > and similar ways? I am leaning towards at least stripping them of the > > natural=coastline (because they aren't) and of the name tag. If not > > delete them altogether. > > As far as I know osmcoastline is currently only able to close the > Antarctica coastline on its own and not other open ends further north. > So these are currently needed for correct processing. The Antarctica > closure segments exist as well (being split into several ways recently > and thereby complicating things a bit) but they are tagged > coastline=bogus to allow closing at different limits depending on the > coordinate system used. > > Ideally none of these would be in the database (because as you said > there is no coastline there).
Yes, osmcoastline currently does need the ways at the antimeridian, but I can fix osmcoastline. I'd rather have correct data and do the fixes in the postprocessing. Is anybody still using the old coastline processing software? > On the other hand you could of course also argue the same way that > multipolygons extending across the 180°-Meridian should also be > possible... That's a bigger proposition. There are only very few people who postprocess coastline data themselves, but there are many different programs doing the multipolygon-processing themselves. Jochen -- Jochen Topf [email protected] http://www.jochentopf.com/ +49-721-388298 _______________________________________________ talk mailing list [email protected] https://lists.openstreetmap.org/listinfo/talk
