I have found another edge case while dividing up OpenStreetMap country boundaries into QuadTiles.

I have attached the geojson in the "genericTestCase" format that @bluenote10 put together.

I am intersecting a polygon of Spain's border, which has an enclave cut out for Gibraltar, with a square tile. The Gibraltar hole is completely contained within the square tile, so should be a straight forward intersection.

This is what I expect:

This is what I get. Note the internal hole is coloured red, as it is now a part of the multipolygon

Here are is the generic test case geojson:

issue128.geojson.txt

And here is the actual geojson result I get from the intersection:

actual-output.geojson.txt

Perhaps the winding order of the interior ring is reversing?

I am using the master branch of this repo, and I have also applied the #125 pull request too.

@bluenote10 If you get a chance, would you mind running this through your Rust implementation to see how it handles it?