I'm agree with most of the others that removing the buondaries out isn't a good approach to solve current problems. I am agains layering in general, as this will only lead to more inconsistencies, in the end, most if not all stuff is somehow interlinked directly or indirectly. Even underground lines often follow roads or other feature on the ground.
2013/11/5 Pieren <pier...@gmail.com> > Like in real world. In my country, when the road or the river is > changing, the admin border may or may not change. And it can take time > until the administration reacts. This depends on > local/regional/national authorities who decide if the admin border has > to be updated or not and when. So, if the river changes, move the > river and keep the admin border at its current definition. This might > require some "unglueing" nodes or ways. > It surely augments complexity (inevitably) when boundaries are involved. Ideally there should be a distinction already in the OSM database between boundaries that ARE defined by a feature (like a river, coastline, peak, road, railway, ...) and those that only "coincidently" share the same position with them. If things were clean like this, you'd know how to operate when a feature changes that also is part of a boundary. In the current situation you also often don't know if perceived irregularities or offsets derive from official data or if the importer simply introduced errors in the conversion (or if the original data already was wrong). IMHO we should recommend to only reuse existing geometry for boundaries if it IS the boundary, not if it just happens to be at the right location. cheers, Martin
_______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
