[OSM-talk] area topology

Stephen Gower socks-openstreetmap.org at earth.li
Tue May 13 16:41:29 BST 2008


On Tue, May 13, 2008 at 02:31:32PM +0100, Andrew Chadwick (email lists) wrote:
> 
> I subscribe to the view that areas should correspond to the real area on
> the ground and mostly be kept clear of roadways. Placing an Area's Nodes
> near the adjacent Way's nodes helps make the map easier to maintain. I
> will often abut adjacent areas that are separated only by something thin
> and make their nodes share, however.

  Putting the other side of the argument, as Andrew I'm sure knew I
  would:
  
  A road is represented by a single way.  Although the way has zero
  width in the database, it represents the whole width of the
  carriageway (pavement) and well as the pavement (sidewalk).  If a
  minor road meets a more important way at a T-junction, we do not
  put the last node where the minor road ends, instead we extend it
  to the centre of the more important one.  In the same way, if an
  area comes right up to the edge of a road (including its pavement,
  etc), we should extend the area to use the same defining nodes.
  
  If we do not do this, we have an undefined space between the area
  and the road.  This undefined space is of variable width and,
  without knowing how every renderer is going to treat the highway,
  there is no way of knowing if it will appear or not, unless it is
  arterially small (aka 0!).
  
  There is some merit to the argument that seperation would help with
  routing. We could have a convention that if an area is accessible
  from any point on the highway they should share segments, but if
  that's not the case (there's a fence between, for example) they
  should be seperated. While I can see how this would work, it feels
  like an ugly hack.  It's not my itch, but there's got to be a
  better way of expressing the boundary between highway and area - I
  guess with a relation.

> Rectilinear buildings in particular should be kept rectilinear: there's
> no excuse for trapezoidal buildings with the new extrusion stuff now in
> JOSM :)

  I agree with that as a potential stumbling block, and was concerned
  about this until I actually started mapping buildings.  In
  practice, the resolution of accuracy in OSM is such that you can
  make a fair representation of the shape of the building and still
  share nodes with the highway it abuts.
 
> However, rivers are Interesting: quite often an Area whose edge is
> defined by a river may change over time as the river meanders... In that
> case, it probably does make sense to abut a Way to an Area.

  It should be noted that roads also change position sometimes,
  affecting the areas that are defined by them!.
  
  s




More information about the talk mailing list