[OSM-talk] Statistics on planet.osm.gz

[Tom Carden]

On 4/13/06, Andy Robinson <Andy_J_Robinson at blueyonder.co.uk> wrote:
> I'd agree that gaps perhaps are not as ideal, however plenty of my ways have
> a split or a short "T" off segment so I think to confine only to contiguous
> may be impractical unless enforced.
>

By contiguous I meant connected - splits and Ts aren't a problem, gaps
are.  What I want to be able to do (for routing, like Joerg) is infer
that if two segments are in the same way then I can assume they are
connected.  I don't mean linear: there can be loops and branches. 
Roundabouts etc can be part of multiple ways, but the whole roundabout
should be considered part of each way.

Is there a solid definition of the current implementation of way?  It
sounds slightly incoherent to me as it stands - just a bag of segments
with the same properties.  The definition could be tighter (and more
useful) without complicating the interface or database.

Tom.