[OSM-dev] Various OSM troubles
Ludwig Max Brinckmann
ludwigbrinckmann at gmail.com
Mon Jan 8 18:16:09 GMT 2007
Thanks for pointing me there, it is interesting to see what experiments have
been made to address this issue.
It largely confirms my view that limiting requests to the API to a sensible
tile size (to be determined) together with a meaningful error message would
improve the current situation, because
1. people would know why their requests fail, rather than receiving the 500
2. no time would be wasted in trying to compute results that cannot be
computed with the resources available.
3. less repeated (and repeatedly failing requests) to the OSM server
=> a net gain of successfully served requests, with the same resources,
little programming effort and no risk to existing data.
On 1/8/07, Nick Black <nickblack1 at gmail.com> wrote:
> The reasons are well documented by Nick Hill in his posts to the osm
> lists, eg:
> On 1/8/07, Ludwig Max Brinckmann < ludwigbrinckmann at gmail.com > wrote:
> > > On the other hand there is a limit in size, but you probably will only
> > > see it
> > > if you are in a not so well filled area like Africa, ...
> > True. Maybe we should first find the bottleneck before looking for
> > solutions. Maybe someone who has access to the server logs could fill us in
> > what the actual error message on this error 500 is.
> > In the meantime have just been a bit browsing through the DB code and
> > one thing I noticed is -- correct me if I am wrong -- that OSM does not use
> > the MySql spatial extensions to represent the geometries, and thus cannot
> > use spatial indices on the data.
> > My personal experience is with PostGIS and there adding spatial indices
> > to large geometry tables speeds up retrieval by several magnitudes (*1000).
> > Is there any reason, apart from historic, that spatial extensions and
> > spatial indices are not used for OSM? Are there Ruby limitations in using
> > spatial extensions (good for prototyping, not so good for speed AFAIK)?
> > Ludwig
> > _______________________________________________
> > dev mailing list
> > dev at openstreetmap.org
> > http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/dev
> Nick Black
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the dev