I have a problem with the mall that I have fiddled with in this changeset. The problem is that the two building parts 370434881 and 370434882 cross a road. They are first-floor parts of the mall where the road passes under. You can see them in for example this picture: I suppose I could try tagging them like the nearby passage through the building part 370411677, but I think of these parts of the mall as less of a tunnel through a building than smaller building parts crossing the road. I have currently tried tagging the building parts as How would you suggest tagging this situation to get the road correctly rendered as covered by the buildings? asked 25 Oct '17, 12:31 ThomasA |
This is not the question you should be asking. Instead, you should be asking a question like, "How would you suggest tagging this situation to best represent the reality on-the-ground?" The rendering will vary depending on the data consumer, so don't worry about that part. Just make sure the objects are tagged as accurately as you can. As for how I would tag it, I would tag the building part as layer=1 (because it physically crosses above the implicit layer=0 highway) and the highway underneath as covered=yes. At the time of my writing this, it appears that this is how these objects are already tagged, so I don't see any reason to change anything. The "Standard" rendering at osm.org renders highways differently when tagged with covered=yes, so there's already an indication that a portion of the highway is covered. answered 25 Oct '17, 20:54 alester I guess my question here is then: Should the mall be represented by a multipolygon (as it is now)? The entire multipolygon is implicitly
(26 Oct '17, 09:18)
ThomasA
Yes, the mall should be represented as a multipolygon because it has a hole. The way https://www.openstreetmap.org/way/370434882 is tagged with building:part=yes and layer=1. As far as I understand the building:part spec that's the correct tagging: the building part is a separate way that overlaps the base building (which, in this case, happens to be a multipolygon).
(27 Oct '17, 20:14)
dsh4
|