👋 Thank you for your draft pull request! Do you know that you can use [ci skip] or [skip ci] in your commit messages to skip running continuous integration tests until you are ready?

@mhvk , do you know what is going on here in pyinstaller? If not, I guess we still have to pin numpy for the pyinstaller job.

  File "astropy/units/quantity_helper/function_helpers.py", line 42, in <module>
  File "<frozen importlib._bootstrap>", line 991, in _find_and_load
  File "<frozen importlib._bootstrap>", line 975, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 671, in _load_unlocked
  File "PyInstaller/loader/pyimod03_importers.py", line 476, in exec_module
  File "numpy/lib/recfunctions.py", line 16, in <module>
  File "<frozen importlib._bootstrap>", line 991, in _find_and_load
  File "<frozen importlib._bootstrap>", line 975, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 671, in _load_unlocked
  File "PyInstaller/loader/pyimod03_importers.py", line 476, in exec_module
  File "numpy/testing/__init__.py", line 17, in <module>
NameError: name '_private' is not defined

Actually, looking at https://github.com/astropy/astropy/runs/4704960655?check_suite_focus=true , I am not sure if this env var is doing anything for us. @dstansby , what do you think?

Looks like maybe the env var did render the skipping of test_copy_vla unnecessary? 🤷

There's a line in the failing test that says:

:228: RuntimeWarning: During parsing environment variable 'NPY_DISABLE_CPU_FEATURES':
You cannot disable CPU features ("AVX512F AVX512VL AVX512BW AVX512DQ AVX512_SKX"), since they are not part of the dispatched optimizations

I don't know exactly what this means, but disabling doesn't seem to have an effect.

This actually seems to be happening on all builds, with the extra line below that I missed in my previous comment:

(SSSE3 SSE41 POPCNT SSE42 AVX F16C FMA3 AVX2 AVX512F AVX512CD AVX512_KNL AVX512_KNM AVX512_SKX AVX512_CLX AVX512_CNL AVX512_ICL).

I'm guessing this is the list of available CPU features, so maybe try disabling all the ones in this list that start with AVX512? I have no idea why this is different to the list we disabled in MPL though...

    <frozen importlib._bootstrap>:241: RuntimeWarning: During parsing environment variable 'NPY_DISABLE_CPU_FEATURES':
    You cannot disable CPU features ("AVX512_SKX"), since they are not part of the dispatched optimizations
    (SSSE3 SSE41 POPCNT SSE42 AVX F16C FMA3 AVX2 AVX512F AVX512CD AVX512_KNL AVX512_KNM AVX512_SKX AVX512_CLX AVX512_CNL AVX512_ICL).

This makes no sense to me. 🤯

Hmm. Looks like we don't have to touch most of the tests if we go with #12685 . What do you think?

@WilliamJamieson , do you think your modeling patch is still worth going in despite numpy issues?

@plim I suggest that we go with #12685 and add part of my modeling patch in a separate PR (without the statement which excludes the Gaussian2D model). This way we can isolate future failures to "equivalent" parameters rather than broken model fits.