-
Notifications
You must be signed in to change notification settings - Fork 101
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add MacOS CI workflow using mamba #1865
Conversation
@JPGlaser : I am trying to add a MacOS environment using micromamba like the one you did in https://github.com/ipta/pulsar-env/blob/patch/py-3.11/.github/workflows/test_CondaEnv.yml |
OK, I think I got this working under |
Looks good to me. It's much easier to control these kinds of tests and package deployments with the micromamba action. ~ Joe G. |
I think the macos test using mamba is too much of a special case to belong in the matrix. It will be more readable if it's separate, IMO. |
@JPGlaser : I thought I was running the precision tests here before, but I guess I wasn't. When I try to run the whole test suite it fails on the |
I added some more verbosity to the tests, and it looks OK:
that value for |
And then when I add a check of the precision to the actual test, it gives:
so somehow the right python/numpy is not being used/loaded here. |
Yes, somehow within the
when it should be python 3.11. So it's not actually loading within the mamba environment. |
For some reason the CI has failed several times on: |
Also please update changelog. |
I'm not sure what the original requirement was for, but we had included |
This looks good. Shall I merge this? |
I think this is ready, but would appreciate a look-through from @JPGlaser |
Yep, everything looks good and things seem to be operating correctly on the tests. I'm good with it getting merged! |
This implements the full CI test on MacOS using mamba to ensure an x86 processor. It also removes the old MacOS tests from the testing matrix. Note that the new tests are not done through
tox
, becausetox
makes its own virtual environment, so they are just called directly. Hopefully this is sufficient. Otherwise more development (likely usingtox-conda
) would be needed.There was additionally one test that failed repeatedly. As far as I could tell this was just due to random chance, although I am not sure. But I fixed it by putting in a fixed random seed.