[OSM-dev] Simplifying data for routing
Christof Dallermassl
cdaller.hw at gmx.at
Sun Jul 29 19:44:28 BST 2007
Hi Flo,
looks great. What I wanted to do lately: provide a web api where you can
ask "please route from node 123425 to node 98345 (or from latlon start
to latlon end) and return the ids of nodes, segments, ways to use and a
description which is human readable". It seems that you are faster than
I am :-)
How large are the graphs you tried? Munich only. Germany or the whole
planet? Is it possible to put the complete planet file into the db and
provide the service I described above?
code is always welcome (best in svn).
regards
christof
flo detig schrieb:
>>> Since I'm dropping most of the OSM nodes, I'm thinking that the
>>> vertices on the graph should be intersections.
>>> describe nearby intersections in terms of how many "ways" they
>
>> ...on the whole, A* is preferred for routing.
>>
>> I believe there are some routing functions available for PostGIS (google
>> for "pgRouting");
>
>
> Hi,
> my name is flo - I'm newbie but very happy / eager to see you discussing this
> stuff :)
>
> this is what I tried:
> - rails port
> - postgis
> - http://pgrouting.postlbs.org
> (implemented in c with boost graph lib -> very fast)
> - google maps
>
> screenshots of results (compared to routing via googlemaps api):
> http://www.flickr.com/photos/detig/
>
>
> I used a ruby-script to build network-graph-topology appropriate for routing
> - created two new tables (edges, vertices) required by pgrouting
> - removed all vertices (nodes between way-segments) with degree less than three
> (non-intersections)
> (took one and a half hours just for the city of munich)
>
> anybody interested?
> Can I commit the code?
>
>
>
>
>
> _______________________________________________
> dev mailing list
> dev at openstreetmap.org
> http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/dev
--
---------------------------------------------------------------------------
Christof Dallermassl
cdaller dot hw at gmx dot at
---------------------------------------------------------------------------
More information about the dev
mailing list