You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I just pushed a small commit which should only have affected the code, not anything about the environment setup.
Repeated attempts to run tests fail at the Setup Conda stage, for example here
To verify that my changes did not trigger this, I went to the github actions page from the previous commit (where all tests passed) here
And then requested to re-run the tests (for python 3.8). The previously passing test now fails here
This suggests that no recent update to MTH5 repo is causing this, but rather that some behaviour on github has changed which makes the tests.yaml invalid.
Two possible hints (or red herrings:/)
Examining the logs on github, the previously passing tests, when "Installing conda" summarized with:
@kujaku11 For the record, retriggering tests on aurora features branch (which inherit the changes on mth5's features branch) are passing. One difference is that aurora's tests.yaml use miniconda instead of conda, not sure if that matters.
The text was updated successfully, but these errors were encountered:
Searching the issues in s-weigand/setup-conda, I found the same error referenced here. This issue refers to an old, merged, pull request, but one of the suggestions was to use setup-conda@v1.0.7. If I make that change, all tests except python 3.8 pass.
- use s-weigand/setup-conda@v1.0.7
- this change allows 3.9, 3.10, 3.11 to pass
- 3.8 does not pass so removed it for now
- tried adding 3.12 but no joy.
- ModuleNotFoundError: No module named 'attr'
- usr/share/miniconda/envs/mth5-test/lib/python3.12/site-packages/pytest_subtests.py
@kkappler I think there are some new issues with using conda directly because of their business model, so going with miniconda is likely the better option. I adjusted the test.yml in PR #276 which should fix the issue for now.
I just pushed a small commit which should only have affected the code, not anything about the environment setup.
Repeated attempts to run tests fail at the
Setup Conda
stage, for example hereTo verify that my changes did not trigger this, I went to the github actions page from the previous commit (where all tests passed) here
And then requested to re-run the tests (for python 3.8). The previously passing test now fails here
This suggests that no recent update to MTH5 repo is causing this, but rather that some behaviour on github has changed which makes the tests.yaml invalid.
Two possible hints (or red herrings:/)
Examining the logs on github, the previously passing tests, when "Installing conda" summarized with:
The failing tests:
But in both cases the same hash of ca-certificates get's installed. nevertheless, it points at a change in github's runner.
@kujaku11 For the record, retriggering tests on aurora
features
branch (which inherit the changes on mth5'sfeatures
branch) are passing. One difference is that aurora's tests.yaml use miniconda instead of conda, not sure if that matters.The text was updated successfully, but these errors were encountered: