[OSM-dev] segfault in osm2pgsql

Stephan Knauss osm at stephans-server.de
Sun Feb 20 21:00:24 GMT 2011

On 20.02.2011 18:26, M∡rtin Koppenhoefer wrote:
> I also found another problem in recent osm2pgsql: the southpole
> problem seems not to be correctly fixed. Before recent fix it used to
> render on 0,0 (and also connect the antarctic national borders to this
> point), now it renders at lon 0, lat 85.xxxxx (border of our
> rendering) which looks weird for the boundaries. It should instead not
> render the point at all, but render the boundaries as if they were
> going to lat 89.99999999999999999999999 (because I guess that 0,90
> doesn't exist in Spherical Mercator).
Is this possible with our projection? The closer to the pole the more 
infinite the values get.

You might reproject, calculate the intersection of the borders with lat 
85 and create a new end point of the borders instead of Southpole.
That way they could be correctly displayed.

Maybe insert the mentioned points as nodes into OSM and have osm2pgsql 
filter away all nodes with lat >85. So at least the display of the 
remaining lines should work.


More information about the dev mailing list