[OSM-dev] Topologicial correctness in OSM

[Nicola Ranaldo]

On Wednesday 17 January 2007 23:25, 80n wrote:
[...]
> >
> > You are right, howewer we lack:
> >
> > 1) hierarchycal ways
> > 2) advanced support for areas
> >
> > As the first may be a long term goal, the second is necessary to elevate
> > osm
> > to a full featured model. Actually we cannot define areas with inner
> > holes while the other formats do.
>
> I think maybe you can.  Because ways can be non-contiguous, you can create
> an area that comprises an outer circle and an inner circle.  If the outer
> circle is clockwise and then inner anti-clockwise then using something like
> the SVG non-zero rule, you get a definition of areas with holes.
>
> If you use the SVG even-odd rule, which is the other way of doing it, then
> you don't even need to ensure that the inner circle is anti-clockwise.
>
> The osmarender rule files currently implement the non-zero rule.  See:
> http://dev.openstreetmap.org/~ojw/Tiles/tile.php/16/32674/21824.png for an
> example of some lakes containing islands.

You are right, and this is sufficient, we should talk about to have this 
officially in the osm-model. But be carefull in this way the model will not 
be able to evolve in a ordered-node-way model (it could still evolve in a 
hierarchycal way model).
Assuming SVG even-odd rule, how to distinguish the outer from the inner?
Is the outer the absolute first subway and the inners all the others?

	Niko