Lessons learned from refactors and rearchitectures
summary
A major refactor rolls back after a disastrously buggy release because unit and end-to-end tests got punted, proving manual verification fails; one module conversion shows five files costs far more time, about 12% more bugs and 72.25% more developer frustration in the example migrating Rails + custom webpack to Next.js. Knowledge concentrates in a bus factor of one, mob programming and scheduled pairings spread it, and incremental refactors (convert one page, measure benefit) plus release checklists prevent repeats.
AI-written summary · about
chapters (42)
- 0:00Intro
- 0:30Hydrate
- 1:23Why do devs make big mistakes
- 1:54Fivewise
- 2:58Systems to prevent big mistakes
- 4:28Show of hands
- 4:52Test
- 5:22Lesson Learned
- 6:58Release Checklist
- 7:48The Checklist Manifesto
- 8:08The Next Big Project
- 8:34Bus Factor
- 9:34Knowledge Sharing
- 12:19Documentation
- 14:37Complexity
- 14:58Im embarrassed
- 15:44Dont do it
- 16:10Lessons learned
- 16:43What went wrong
- 18:26Learning
- 19:56Collaboration
- 20:13Mistakes
- 20:37Experience
- 20:58Support
- 21:20Rule of 3
- 22:25Seniors are too busy
- 23:31Overload
- 24:07Unrealistic Expectations
- 24:31Aggressive Goals
- 24:48Slack
- 25:06Last Situation
- 25:34Celebrate
- 26:36Why is this important
- 26:55Presentation slides
- 27:25Quantifying penalties
- 28:09Incremental victories
- 28:39Incremental refactors
- 29:01Dont touch everything
- 29:30Work with Legacy Code
- 29:43Conclusion
- 30:04Learning TypeScript
- 30:14Outro
