Hi all!
I'd like to continue with the Shape extraction for Wikipedia articles, like Kolossos has begun with the http://wiki.openstreetmap.org/wiki/Query-to-map Query-to-Map-project. There is pretty much already there, even the complete endresult, like I mentioned in https://wiki.toolserver.org/view/GeoShape#Frontend_Examples
What I'm thinking about is: I don't think the Osm2pgsql database schema which we have (I think) is suitable for the long-term for the reasons mentioned in the same article further below, https://wiki.toolserver.org/view/GeoShape#Comparasion_osm2pgsql_and_osmosis mainly because I cant find a good changemanagement for derivate tables, which can be worked around maybe or which can be omitted creating a new Shape table every now and then.
Personally I like the lonvia approach best, it's a Osmosis snapshot schema which is quite up to date to my opinion and its a ready GPL package. On the other hand, it's in phyton which may not be 'everybodys darling' [PK], and it's rendering functions are not needed for this purpose here. In one thing I tend to agree with Stephan Plepelits:
Unfortunately we have to import the database twice ... once with osm2pgsql to have a database structure which is usable for mapnik. A second time with osmosis, because we want to have all possibly available data. http://gitorious.org/openstreetbrowser/openstreetbrowser/blobs/master/DOCU
What do you think? Guverneu/Bernhard