Closed sixy6e closed 3 years ago
@sixy6e Will these updates still work with 2.x?
it should as it is being explicit on casting to NumPy's Unicode (U) type, which has been around for a while.
When NCI comes back online I can run the test there. Or get @truth-quark to do it.
@ga-ellis
Results from a build I just created on NCI:
collected 340 items
tests/test_acquisition.py ...................................................................................................................................................................... [ 48%]
tests/test_acquisition_sentinel.py ....................... [ 55%]
tests/test_blrb.py ......... [ 58%]
tests/test_geobox.py ............................. [ 66%]
tests/test_hdf5.py ......................... [ 74%]
tests/test_interpolation.py ...... [ 75%]
tests/test_land_sea_masking.py .. [ 76%]
tests/test_lon_lat.py .. [ 77%]
tests/test_ls7_gap_mask.py .................. [ 82%]
tests/test_metadata.py ... [ 83%]
tests/test_mtl.py .................... [ 89%]
tests/test_random_locations.py .. [ 89%]
tests/test_read_subset.py ssss..........s [ 94%]
tests/test_reformat_tp5.py .. [ 94%]
tests/test_tiling.py ..... [ 96%]
tests/test_tle.py ... [ 97%]
tests/test_tp5.py ......... [ 99%]
tests/test_vincenty.py . [100%]
=============================================================================================== warnings summary ===============================================================================================
/g/data/v10/public/modules/dea-env/20200612/lib/python3.6/site-packages/osgeo/gdal.py:107
/g/data/v10/public/modules/dea-env/20200612/lib/python3.6/site-packages/osgeo/gdal.py:107: DeprecationWarning:
gdal.py was placed in a namespace, it is now available as osgeo.gdal
-- Docs: https://docs.pytest.org/en/latest/warnings.html
============================================================================ 335 passed, 5 skipped, 1 warning in 130.92s (0:02:10) =============================================================================
Thanks for all the help @sixy6e! You are awesome!
For h5py > 3.0.0, string behaviour changed (see here causing string arrays to be loaded as byte strings, eg
This causes an issue for string comparison such as here during interpolation of the atmospheric coefficients, resulting in a zero-length array of point samples used for driving the interpolation.
The existing test, does pick this up.