I finally understood (I think !) the reason why some mappers add a node where a riverbank way ends and another begins:
This is, as far as I can figure out, a way to avoid JOSM validator's "ovelapping ways" warning. Of course, the warning is yet another anoying false-positive, but to me it seems like the workaround is an actual semantic error (overlapping areas as opposed to areas sharing an edge). I should probably file a bug to the validator, but I wonder what people think. Is this workaround a good thing ? Wouldn't validators other than the JOSM one complain about this ? asked 02 Oct '13, 12:47  Vincent de P... ♦ |
One Answer:
I would file a bug against JOSM's validator because the fix displayed in your image should be avoided. It is not very clean and has no purpose except avoiding a warning of a single editor. Other validators handle this case correctly, for example keepright never reports such issues but is able to identify other overlapping ways. answered 02 Oct '13, 14:25  scai ♦ 3 Thanks. Ticket created : https://josm.openstreetmap.de/ticket/9140 (02 Oct '13, 17:16) Vincent de P... ♦ 1 Fixed in JOSM 6300. (06 Oct '13, 16:48) don-vip |
In a way, using a single multipolygon relation for the whole river fixes the issue cleanly, but that has even more annoying problems of its own. And it doesn't ultimately solve the issue of "touching riverbanks" once we reach a tributary.