Jannik Kissinger

The numerical stability (linalg solve) should be investigated. probably linked to #64

use integers instead of floats for a higher accuracy for the angle representations (fixed value range: 0.0 - 4.0)

For future reference: the [scripts folder](https://github.com/jannikmi/timezonefinder/tree/master/scripts) already contains scripts for time benchmarks

The regular `copy()` and `deepcopy()` methods can be used for this. In my custom node class i have added these functions: ``` python def __copy__(self): return self.__class__( self.attributes, ) def...

Sure! Any input is more than welcome. Let me know if you need any input.

Sure! Any input is more than welcome. Let me know if you need any input.

The first useful thing would be to use https://docs.pytest.org/en/6.2.x/parametrize.html#pytest-mark-parametrize-parametrizing-test-functions for testing different timezone locations (each location should be it's own "test" case)

FIXME: the polygon lists are empty

Thanks for bringing this up. It seems like v4 introduces a lot of breaking changes: ``` FAILED tests/shortcut_test.py::test_resolutions - AttributeError: module 'h3.api.numpy_int' has no attribute 'h3_get_resolution' FAILED tests/shortcut_test.py::test_unique_pole_cells - AttributeError:...

Great! Ideally you could also add tests for both h3 versions (ensuring to test backward compatibility)