|
What is the story with the following multipolygons? JOSM and Mapnik seem to say they are legal, complete loops. What's actually going on? |
|
The OSM Inspector sometimes claims that a ring is not closed when instead the polygon has an inner ring touching an outer ring. This may occasionally confuse the ring building logic in OSMI, leading to the "ring not closed" complaint instead of correctly saying "inner touches outer". I checked one of your examples and found exactly that: http://www.openstreetmap.org/relation/143464#map=16/33.2530/-84.5717 (right edge of the "hole" in the forest). The data also contains a ton of useless tags that should never have been imported in the first place. That solves the problem for the Georgia multipolygons. But I can't find a similar problem with the New Jersey multipolygons. The most I can find is an inner ring touching another inner ring. Do you have any other ideas?
(23 Jan '14, 20:31)
eric22
|
|
What is going one is that these (imported) multipolygons are bigger than you (or anyone else) can manage easily. You should split these at bigger roads (forest does not grow on highways) and deal with much smaller relations. Otherwise you will waste some hours to find the mistake. Some seconds to fix it. And in a month or two you can do it again. In the meantime only few advanced contributors have the courage to modify such big relations. Meaning this place will evolve much slower than it could. Not sure I agree. If you open JOSM and choose "Download object" and the multipolygon number, you can work on a multipolygon easily without being confused by other objects. (You just have to be aware that your changes might affect other objects, and download those as necessary before editing, to avoid conflicts.) I think Frederik is right that the problem is inner ways touching outer ways. The issue is some kind of multipolygon complexity, not the absolute number/length of ways...
(19 Jan '14, 20:15)
eric22
|
