[OSM-talk] colour fills in osmarender

[TM]

Raphaël Jacquot wrote:
> TM wrote:
>
>   
>> But what if roundabouts were defined (or treated by the routing
>> algorithm) as points, albeit of a special kind? Would this make sense?
>> (If they were regular points, displaying the data  would mean
>> visualising the data structure rather than describing actual features of
>> the physical road network. Surely, this would not make sense, as we'd
>> need separate datasets for rendering and routing.)
>>     
>
> One idea I had was to have a topology layer above the graphic
> representation layers that would represent things like roundabouts and
> flat intersections as one node.
>
>   
What I was suggesting was not too different. But this layer would merely
be an internal representation within the routing software, derived from
the points, segments and/or ways of the OSM data. This would not need to
be defined in the database, provided the roundabouts could be identified
as such by the software. It could be considered a trade-off between
additional data in the database and additional computation in the
routing engine. I think it would make sense to keep the underlying data
as "clean" as possible and leave the work of sorting it out to the
special-purpose programs that do things like route planning. Otherwise
we'd have to support all kinds of odd applications within the database
itself.

    /    TM


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20060609/c48a0200/attachment.html>