Invert bit position in quality flags #91
Conversation
|
Ouch. I think this means this code isn't tested properly? |
|
Yes, I will add a more realistic test that really fills 32 bit big endian integers as expected from the files and get the corrupt mask for them. |
pygac/tests/test_pod.py
Outdated
| self.assertEqual(bits[0][-1], 1) # the last bit is filled | ||
| # The fatal flag fills only the first bit | ||
| self.assertEqual(bits[1].sum(), 1) # only one bit is filled | ||
| self.assertEqual(bits[1][0], 1) # only one bit is filled |
There was a problem hiding this comment.
Not sure if this is required, but elements 3 and 4 of the quality_indicators array are not tested for byte order here
There was a problem hiding this comment.
Typo in comment: I guess it should be "only the first bit is filled"
There was a problem hiding this comment.
Not sure if this is required, but elements 3 and 4 of the
quality_indicatorsarray are not tested for byte order here
My idea is to check only the bit representation for the trivial cases (fist and last bit set). This should be sufficient to be sure that the bits look as expected for this test.
mraspaud
changed the title
This PR corrects the quality flags which leads to differences in the masked scanlines as presented in #90.