So it seems like the "get admin boundaries out of core OSM" opinion has fairly decent support in the US but maybe not elsewhere. I think the reason for this is because we imported these borders a few years ago. Because the import process wasn't perfect, it created duplicate nodes wherever a border intersected a TIGER road even if there would normally be no reason for a node to exist on either one of the ways. These duplicate nodes then got merged together by a bot trying to clean up keepright "duplicate node" errors. But now because our boundaries are merged to roads where they have no business being merged, it makes them a nightmare to update. It also means that well meaning users see this and think that is the way it is supposed to be and make the problem worse.
If the boundaries were in their own space we could update them at will from authoritative sources on a regular basis. Of course "authoritative" doesn't always mean "accurate" but for admin boundaries I think it is much more often the case than for most other physical things we can easily survey with a GPS device. Sure, you can sometimes find evidence of an admin boundary on the ground but this is definitely the exception rather than the rule around here. And even if you *can* it doesn't mean this is going to happen for each city in the US on a regular basis. So yes, I think we could have a better and more up to date map if we made boundaries separate from most other features in the database. The idea of some filters being active by default in editors has also been suggested before but some people are opposed to this because "everyone can edit everything all the time" in OSM. This is a nice sentiment but I personally have no problem saying that a first time mapper has no business touching admin boundaries that have the potential to break geocoding for an entire country/state. As a last note, Paul's example of the Colorado River is hardly unique. My county border is defined by the course of a river - when the border was defined. Since then a dam has been built which means most of the eastern side of the border now runs through a lake. The part of the river that is not in the lake changed course during a massive flood in the 1950s but the border still follows the original course of the river. Maybe the different handling of borders following a natural feature is another regional difference? Toby On Tue, Nov 5, 2013 at 9:18 AM, Paul Johnson <ba...@ursamundi.org> wrote: > > On Tue, Nov 5, 2013 at 8:05 AM, Janko Mihelić <jan...@gmail.com> wrote: > >> The problem is, some admin borders are supposed to be glued to roads or >> rivers, and they change when the flow of a road or river changes. How do >> you deal with that? > > > Well, historically, the border doesn't move. This has caused an on-going > border dispute between California and Arizona (compounded by the fact that > the Colorado River no longer regularly flows at all that far south), and > some spits and islands only accessible from the neighboring state elsewhere > in the country. > > _______________________________________________ > talk mailing list > talk@openstreetmap.org > https://lists.openstreetmap.org/listinfo/talk > >
_______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
