[OSM-dev] osm-history-splitter quirk

[Peter Körner]

Hi

using softcut gives better results in most cases which is why I made id 
default some months ago.

The split-all-clipbounds script was hack which I don't use at all. It 
tries to optimize the splitting-process by creating country-extracts, 
then state extracts from them and so one. This fails if the inner 
polygons are not fully containes by the outer ones.

Does the same problem occur when you cut your state-polygon straight 
from the source file? If yes, could you tell me which poly-file you are 
running in which source?

Peter


Am 26.05.2012 18:29, schrieb Martijn van Exel:
> Hi,
>
> I'm using the split-all-clipbounds script often to generate US state
> and county sized pbf files using osm-history-splitter.
> I used to use the hardcut method, but this generates false positives
> in the analysis I am running currently (detecting disconnected _link
> roads) so I switched to the softcut method. I did a file size check to
> see if all went well.
> It turns out four state PBF files were empty (or rather, 133 bytes). I
> checked the POLY files used for the cut but they seem fine (no
> self-intersection, closed polygons).
> I am running the four states again now so I can have a closer look at
> what's going on. Any ideas up front about where to look for the cause
> of this?
>