radio-astro-tools / spectral-cube

Library for reading and analyzing astrophysical spectral data cubes
http://spectral-cube.rtfd.org
BSD 3-Clause "New" or "Revised" License
95 stars 62 forks source link

Restore `test_apply_everywhere_floordivide` #788

Closed dhomeier closed 2 years ago

dhomeier commented 2 years ago

Started this as experimental follow-up to #783 to restore the floordivide test, which seems to have failed due to unrelated open file exceptions. Additional open-file failures reported in https://github.com/radio-astro-tools/spectral-cube/pull/783#pullrequestreview-832819170 are fixed in a5f218f, although I could reproduce these only in local to runs under macOS. Both 6c9a4dc and a5f218f are passing all tests except the unrelated doctest failure; moving the pytest.skip() before any file read operations are started in the 2nd commit seems less error-prone in general, but since this did not appear to cause problems in CI, feel free to pick only the 1st one.

codecov-commenter commented 2 years ago

Codecov Report

Merging #788 (a5f218f) into master (7452dce) will not change coverage. The diff coverage is n/a.

Impacted file tree graph

@@           Coverage Diff           @@
##           master     #788   +/-   ##
=======================================
  Coverage   77.90%   77.90%           
=======================================
  Files          24       24           
  Lines        5826     5826           
=======================================
  Hits         4539     4539           
  Misses       1287     1287           

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 7452dce...a5f218f. Read the comment docs.

dhomeier commented 2 years ago

One complication was probably that open_files seems to have everything run in a sort of sandbox, making it almost impossible to track the origins of the failures to the respective tests. The twod_numpy tests, as far as I could see, were starting e.g. with test_twod_numpy[True-sum-slice-0-data_advs], which is skipped, but with the file already opened, and that probably interfered with the open-files check for the next test_twod_numpy[True-sum-cube-0-data_advs] in my local installation.