Data handling:
- Fix an edge case for exclusively empty views/clones fragments (issue #15).
- Expect an edge case where
fetch.py
exits without having generated new snapshot files: in that case, regenerate the report using the most recent set of data (issue #17).
Documentation:
- Fix cron syntax in README (issue #20).
- Document elegant method for a multi-repo workflow using the
matrix
approach (PR #26). Thanks to David Farrell and to Egil Hansen.
Job robustness and performance:
- Use a pre-built Docker container image (jgehrcke/github-repo-stats-base) to base this action on. This image includes heavy Python and browser dependencies. This approach significantly reduces the probability for an action run to fail as of one of the many potential transient issues affecting a complex Docker image build. This also significantly reduces the time it takes for completing the
Build container for action use
build step (from ~2 minutes to ~30 seconds). Context: issue #24 and PR #25. If you have security concerns please open an issue and let's talk it through.