Martin Koppenhoefer wrote: > 2012/10/15 Jukka Rahkonen <[email protected]>: >> Perhaps examples B and F could also contain at least two ways? If there is >> only one way, why to make a multipolygon relation at all? > > > there can be reasons. E.g. you want to distinguish linear features from areas. The way could be tagged barrier=fence/wall/etc. and the area landuse=* (this is a quite common situation).
Yes, that is understandable situation indeed. What do you think would be the best thing to do if an user wants to re-use a zig-zagging linestring for creating a multipolygon relation? In a raw format with vertices in the same order as they appear in a linestring the resulting polygon will be invalid as OGC polygon because of self-intersections. I guess it is pretty expensive to do topology checks in the main OSM database triggered automatically every time when relations are saved. Other alternatives I can imagine are - Make OSM editor to check the topology of multipolygon relations before saving and warn users about conflicts. - Run some kind of validating bots which harvest areas and multipolygons from OSM database, resolve simple cases and make topology error reports from the remaining. - Do nothing in the OSM db side and trust that programs which consume OSM data and need OGC style validity can resolve the conflicts in a reasonable way. - All of those together. What I have been doing myself in just to accept the results from osm2pgslq and ogr2ogr and in addition delete all the imported features which are reported invalid by IsValid function. -Jukka Rahkonen- _______________________________________________ dev mailing list [email protected] http://lists.openstreetmap.org/listinfo/dev
