Best Practices for Continuous Integration in Python V02
summary
CI focus narrows to suggested-patch builds: most CPU time and feedback cycles live there. Claims: evaluate accuracy, actionability, promptness, and cost; use GitHub Actions/GitLab/Jenkins, run linters (flake8/black) and pytest, test in pinned container images to avoid incidental failures, and pin dependencies so upgrades are separate patches. Built solutions include container images with cached layers and test-parallelization; aggressive logging of env + high verbosity makes failures reproducible. Also: cache aggressively to cut wall time and runner cost.
AI-written summary · about
chapters (18)
- 0:00Intro
- 1:47Traditional steps to run
- 3:45When to run
- 5:23Suggested patches
- 7:49Suggested patch build
- 8:37CI goal
- 9:35CI system
- 12:26Actionability
- 13:10Attainable
- 16:34Dependencies
- 18:01Regressions
- 18:53Test Quality
- 20:56Details
- 22:43Improving Promptness
- 24:45Improving Cost
- 25:57Decision Making
- 26:44Conclusion
- 27:12Outro
related talks
32mBuilding Confidence Together: Continuous Integration & Testing
01h 02mContinuous Integration: Tips & Tricks
41mContinuous Integration Tips & Tricks
01h 15mThe Practices That Make Continuous Integration
32mContinuous Integration and Deployment: Five Lessons Learned
30mContinuous Integration; Tips & Tricks
45mRunning Tests in Continuous Integration
38mContinuous Integration May Have Negative Effects