Closed titusjan closed 3 years ago
The problem was that, apparently, ma.copy
doesn't copy all the data. After copying the OWNDATA
and WRITEABLE
flags where still False.
Starting from Numpy 1.19.0, np.copy(..., subok=True)
can be used to copy masked arrays as well. This fixes the issue.
When cross hair plots are enabled in the Image Plot inspector, and the user hovers above plot, and the selected item is a NetCDF scalar set to the fill value, the application exits with the following error:
This can be reproduced with the
scalars.nc/masked_scalar
item that can be generated bycreate_test_data.py