| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Sorry, something went wrong.
With libopenblas 0.3.34 we see intermittent failures on windows. Let's see if using an older version makes them go away.
|
@lesteve want to have a look? It seems to pass/only fail less than 1 in 10 times. |
Sorry, something went wrong.
|
AFAICT from this AI-generated script all the Windows CI build succeeded repo="scikit-learn/scikit-learn"
job="Windows x64 pymin_conda_forge_openblas"
gh api "repos/$repo/pulls/34778/commits" --paginate --jq '.[].sha' |
while read -r sha; do
gh api "repos/$repo/commits/$sha/check-runs" --paginate \
--jq ".check_runs[]
| select(.name == \"$job\")
| [\"$sha\", .status, (.conclusion // \"pending\"), .html_url]
| @tsv"
| cat
done Thanks for testing this! It's hard to know whether the OpenBLAS issue is related but empirically it seems to fix the problem. We'll keep an eye out for the OpenBLAS 0.3.35 release and hope the problem doesn't come back. |
Sorry, something went wrong.
|
One less issue to look at for my week on Triage (next week), really appreciated @betatim 🙏! |
Sorry, something went wrong.
|
🤝 |
Sorry, something went wrong.
| Back | FazBrowse Home | New Git URL |
Reference Issues/PRs
Fixes #34717
What does this implement/fix? Explain your changes.
With libopenblas 0.3.34 we see intermittent failures on windows. Let's see if using an older version makes them go away.
We need to re-run the windows CI job a few times to see if the failure is really gone or not.
AI thinks OpenMathLib/OpenBLAS#5954 is related/the same problem.
AI usage disclosure
I used AI assistance for:
First, second, third, fourth, fifth, sixth, seventh, eight and ninth run succeeded. Commits with a red cross can be the result of stopping the CI before all jobs ran. The important thing is to check the status of the windows job.