[OSRM-talk] Zlevels and Restrictions

[Robert Scott]

On Thursday 28 November 2013, Stephen Woodbridge wrote:
> Hi,
> 
> I'm read over the documentation again and I am confused by the "Graph 
> representation" wiki page.
> 
> I have a correctly noded graph in pgRouting. By this I mean that all 
> intersections have a node(s) at them and there are no unnoded intersections.
> 
> One of the characteristics of this is the fact that segments at an 
> overpass/underpass intersection are not really connected because with 
> zlevel information we have assigned separate nodes for the over pass 
> node from the underpass node, in spite of the fact that they both have 
> the same 2D x-y locations (ie: they are separated in Z).
> 
> The "Graph representation" page seems to imply that all segments with 
> shape points are exploded into multiple straight line edges with the 
> idea that crossing polylines with a common x-y point will become noded. 
> Is this true? 

No.

OSM has a topological data model, remember. We don't rely on points happening to share coordinates to tell whether they're connected or not. "layer" tags in OSM are purely cosmetic hints, disregarded from a routing point of view.


robert.