Switch to setup-micromamba action

[maresb]

Description

Related Issue

• [X] Closes #7046
• [ ] Related to #

Checklist

• [ ] Checked that the pre-commit linting/style checks pass
• [ ] Included tests that prove the fix is effective or that the new feature works
• [ ] Added necessary documentation (docstrings and/or example notebooks)
• [ ] If you are a pro: each commit corresponds to a relevant logical change

Type of change

• [ ] New feature / enhancement
• [ ] Bug fix
• [ ] Documentation
• [X] Maintenance
• [ ] Other (please specify):

---

📚 Documentation preview 📚: https://pymc--7213.org.readthedocs.build/en/7213/

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 75.92%. Comparing base (bd74474) to head (03bebb6).

:exclamation: Current head 03bebb6 differs from pull request most recent head bf491b6. Consider uploading reports for the commit bf491b6 to get more accurate results

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #7213       +/-   ##
===========================================
+ Coverage   55.50%   75.92%   +20.41%     
===========================================
  Files         101      100        -1     
  Lines       16862    16875       +13     
===========================================
+ Hits         9360    12813     +3453     
+ Misses       7502     4062     -3440     

see 68 files with indirect coverage changes

[maresb]

I think the error may be fixed here: https://github.com/pymc-devs/pytensor/pull/678

[maresb]

~Ah ya, thanks, that's the one!~ (Wrong PR)

[ricardoV94]

Do we need https://github.com/pymc-devs/pytensor/pull/678 here?

[twiecki]

Worth a try.

[maresb]

Ok, rebased.

[maresb]

That's weird, maybe it needs to be rerun?

I still find GH's UI here weird. I can't find a rerun button for that task. Maybe I need to wait until the whole thing finishes?

[ricardoV94]

Yeah, you can only rerun failed runs once everything is done

[maresb]

Just started the rerun. Any idea why I can't view logs (see screenshot)? Is this a GH outage?

[maresb]

Viewing the CI logs is still really flaky for me, but somehow I was able to get a glimpse of:

 _________________ ERROR collecting tests/sampling/test_jax.py __________________
tests/sampling/test_jax.py:30: in <module>
    from numpyro.infer import MCMC
../../../micromamba/envs/pymc-test/lib/python3.12/site-packages/numpyro/__init__.py:6: in <module>
    from numpyro import compat, diagnostics, distributions, handlers, infer, ops, optim
../../../micromamba/envs/pymc-test/lib/python3.12/site-packages/numpyro/distributions/__init__.py:4: in <module>
    from numpyro.distributions.conjugate import (
../../../micromamba/envs/pymc-test/lib/python3.12/site-packages/numpyro/distributions/conjugate.py:9: in <module>
    from numpyro.distributions.continuous import Beta, Dirichlet, Gamma
../../../micromamba/envs/pymc-test/lib/python3.12/site-packages/numpyro/distributions/continuous.py:36: in <module>
    from jax.scipy.linalg import cho_solve, solve_triangular
../../../micromamba/envs/pymc-test/lib/python3.12/site-packages/jax/scipy/linalg.py:18: in <module>
    from jax._src.scipy.linalg import (
../../../micromamba/envs/pymc-test/lib/python3.12/site-packages/jax/_src/scipy/linalg.py:403: in <module>
    @_wraps(scipy.linalg.tril)
E   AttributeError: module 'scipy.linalg' has no attribute 'tril'

Then for attempt #2, which I believe was post-rebase, I think I was still seeing the old "filename too long" error. :cry: But I can't access the logs in order to paste them here...

Am I the only one who is having trouble viewing the CI logs?

[maresb]

Regarding scipy.linalg.tril, see https://docs.scipy.org/doc/scipy-1.11.4/reference/generated/scipy.linalg.tril.html:

Deprecated since version 1.11.0: tril is deprecated in favour of numpy.tril and will be removed in SciPy 1.13.0.

[maresb]

Yeah, you can only rerun failed runs once everything is done

@ricardoV94, here I at least have a "Re-run" link, even though it doesn't appear to have actually triggered a rerun yet.