Kevin Kreiser

ok so i was able to reproduce the problem with a danish extract. basically the algorithm i had written assumed that the input was good. as we all know, with...

stil segfaulted on the planet, i guess i have to debug the planet sadly..

its failing, of course, on the netherlands relation near the baarle hertog area. looks like the line merge algorithm needs to get a bit more complicated :smile:

ok actually im stupid. i should have actually thought even more about what this executable does before starting to port out the geos stuff. honestly i doubt we even need...

ok i went on a bit of a saga. nils was of course right, just because some relations behave doesnt mean we can trust them all. so i have a...

Oh i havent given up. I'll get the change to the c API done and we can forget about this until next time

Another option is we could try to do it all in spatialite with geos in there. Problem is osm isn't built to really tell you which outer an inner belongs...

This is ready for review. I converted the planet from... 2 years ago. i compared it to the admin db i keep around (even older) and the results are pretty...

ok so the rewrite, here are the important parts - i switched to using boost geometry because its ever so slightly easier to work with than geos' c api. especially...

@nilsnolde yep ill do all the nits after i figure out why the hole in my NL test is being flagged as degenerate