Skip to content

Conversation

@vtavana
Copy link
Contributor

@vtavana vtavana commented May 21, 2025

When running tests on Linux in the CI with python 3.9 and 3.10, mkl_fft package was installed from https://software.repos.intel.com/python/conda and not the local channel. This PR fixes it to always install mkl_fft from local channel.

In addition, the only scipy available for python 3.9 and 3.10 from https://software.repos.intel.com/python/conda is scipy=1.10.1 which requires mkl>=2024.2,<2025. Therfore, mkl-devel is pinned for python 3.9 and 3.10 in meta.yaml file.

@vtavana vtavana self-assigned this May 21, 2025
@vtavana vtavana force-pushed the fix-CI branch 9 times, most recently from 7948c2e to 196b67e Compare May 21, 2025 01:52
@vtavana vtavana changed the title add specific version add version and build of mkl_fft package when running test on linux in CI May 21, 2025
@vtavana vtavana changed the title add version and build of mkl_fft package when running test on linux in CI add version and build of mkl_fft package when running test on Linux in CI May 21, 2025
@vtavana vtavana marked this pull request as ready for review May 21, 2025 02:22
Copilot AI review requested due to automatic review settings May 21, 2025 02:22
Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull Request Overview

This PR ensures that tests on Linux in CI install the mkl_fft package from the local channel and pins the mkl-devel dependency correctly for Python 3.9 and 3.10.

  • Pins mkl-devel version to 2024.2.* for Python 3.9 and 3.10 in meta.yaml.
  • Adds steps in the conda workflow to test the local channel and extract the package version for installation.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File Description
conda-recipe/meta.yaml Added a conditional dependency line to pin mkl-devel for specific Python versions.
.github/workflows/conda-package.yml Introduced new workflow steps to retrieve and utilize the package version in the CI process.
Comments suppressed due to low confidence (1)

.github/workflows/conda-package.yml:143

  • Using PACKAGE_VERSION in the package installation command is a good approach; however, ensure that the version extraction step reliably produces a valid version string to prevent installation mismatches during the CI process.
conda create -n ${{ env.TEST_ENV_NAME }} $PACKAGE_NAME=${{ env.PACKAGE_VERSION }} python=${{ matrix.python }} pytest "scipy>=1.10" $CHANNELS

Copy link
Collaborator

@ndgrigorian ndgrigorian left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@vtavana vtavana merged commit 93ef7f2 into master Jun 3, 2025
57 checks passed
@vtavana vtavana deleted the fix-CI branch June 3, 2025 18:31
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants