[OSM-dev] osm2pgsql and only-named multipolygons

Jochen Topf jochen at remote.org
Wed Oct 5 20:58:03 BST 2011


On Wed, Oct 05, 2011 at 08:20:10AM -0700, Richard Fairhurst wrote:
> Jochen Topf wrote:
> > Thats sounds rather optimistic to me. As far as I know everybody who 
> > has thought about a proper area type has given up, because nobody 
> > could find a way how it was to be implemented solving all the different 
> > design problems with it.
> 
> That sounds rather un-OSM-like to me. :) We've never been dissuaded from
> solving 95% of the problems just because the remaining 5% are hard.
> 
> There's some good stuff on 'The Future of Areas' page. I could see either
> 'areas on nodes' or 'areas on ways' working. But we shouldn't get tied up
> with an all-singing, all-dancing solution for the 5% (Lake Geneva or the
> British coastline), if a simple solution exists for the 95% and a
> coastline-like workaround can be found for the edge cases.

I'd agree with you that if there is a 95%/5% solution we should pursue it.
Unfortunately I think its more a 60%/40% solution or so. Nobody has run
the numbers yet. And lets not forget that with the size of OSM 5% can be
a lot.

To be a bit more concrete: We have huge amounts of tiny areas (buildings
for instance) and we have huge amounts of really big areas (countries,
forests, ...). So we need some solution that works for both.

Jochen
-- 
Jochen Topf  jochen at remote.org  http://www.remote.org/jochen/  +49-721-388298




More information about the dev mailing list