[OSM-talk] Neat routing demo

[Jon Bright]

bvh wrote:
> On Wed, Sep 26, 2007 at 09:04:51PM +0200, Jon Bright wrote:
>> Interesting side point: the most inefficient thing about the OSM 
>> database for routing is, at the moment, the fact that a new way can 
>> start anywhere on some other way.  Instead of going from the start of a 
> 
> All this is going to be irrelevant very soon, but if you had coded your
> algo to route with segments instead of ways it wouldn't have been a
> problem.

That's effectively what it ends up doing - and exactly that is the 
problem.  It's much less efficient to examine all of a way's segments 
than it is to just examine the segment at the end and determine if 
there's a fork there.

-- 
Jon Bright
Silicon Circus Ltd.
http://www.siliconcircus.com