[OSM-dev] Relations, landuse, and gml2osm mess

Frederik Ramm frederik at remote.org
Sun Mar 30 02:53:50 BST 2008


> If I use a relation of ways (making up a ring) as the outer (or inner) hull of 
> a multipolygon relation[2], will it work?

Don't think it would currently, but it sure could be arranged. Some
editors (*cough*) still have problems with nested relations though.[*]

I recently suggested that, for the outer hull, you could directly use
a number of ways that make up a ring, because there's only ever one
"outer" ring anyway - this would yield a multipolygon relation with
"n" ways in role outer, together making up the outer hull, and "m"
ways in role inner, each being a hole. The idea was critizised because
it rather half-bakedly allows more than one way for "outer" but the
inner rings would still have to be one way. 

While it is true that this would introduce a kind of asymmetry, I
think it would be rather pragmatic as you'd seldom need multiple ways
on the "inner" ring.

About the GML:

Out of interest about the whole GML stuff - if you apply the XSLT from


to your OSM file, does the result even remotely look like the GML that
you originally put in? 

I have heard people say that GML supports topology, so is it possible
to take an OSM file where, say, a road and a forest share some nodes
along the forest border, convert it to GML with topology support, open
it in a GIS application, move the forest border, and the road moves as
well? Until now I thought that any OSM to GML conversion would make
two distinct objects, one forest, one road, and they'd lose their


[*] JOSM will give you trouble if you create, within one session, two
new relation, where one is referenced from the other. It is not smart
enough to know that it must first upload the referenced one, then the
referring one, to avoid "precondition failed" errors. So this will
fail in 50% of cases.

Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09" E008°23'33"

More information about the dev mailing list