Avoiding a Shared Library Nightmare
summary
A large corporate Jenkins shared library grows into 20,000 lines over two years, becomes untestable, and is scrapped and rebuilt from scratch. The rebuild enforces standards (noun-verb naming, small methods), uses scripted pipelines with VARs only, adopts Spock tests and JaCoCo coverage, and gates PRs with a 13 builds regression pipeline before merging. Expect one monorepo, unit-testable tiny methods, enforced docs, and a choke-point prod job that verifies tagging and deploy history.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:56About me
- 1:26About Liberty Mutual
- 2:06Pipeline story
- 5:13Delivery date
- 10:16How to improve
- 15:12Planning
- 23:26Standards
- 29:35Testing
- 31:07Documentation
- 32:53Questions
- 40:12Giant Hero Method
- 43:19How to Roll Back
- 45:29Adding New Code
- 47:04Using VARs
- 47:47Using new products
- 50:31Code
related talks
34mTDD for jenkins pipelines
21mLessons from integrating third party library scanning in DevOps workflow
52mEscaping Developer Nightmares
35mOptimizing Your CI: Lessons Learned from a Successful Jenkins Rebuild
36mCI/CD BoF: Delivering Quality Quickly
42mSeven Habits of Highly Effective Jenkins Users
47mWhy your shared libraries are ill - and how to cure them
13mEvolution of containers in continuous integration
