Self Improving Software
summary
Argues that splitting large Node codebases into small modules plus rigorous tests lets upgrades be automated and safe. Demonstrates using exact dependencies, runs dependency experiments with next update to measure upgrade success rates, and wires CI to publish with semantic release guarded by a don't break step that runs dependents’ tests. Shows a real snapshot/schema example, explains green keeper pull‑request automation, and reports that the pipeline successfully prevents most accidental breakages.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:09Outline
- 4:12Scope
- 10:13Versioning
- 16:38Dependencies
- 22:14Semantic Release
- 24:32Dont Break
- 29:29Conclusion
related talks
28mKill all humans
42mRevisit Old Problems with New Eyes!
40mSecure by Default: Building an AI-Augmented, OSS-Powered Reusable CI/CD Pipeline
16mDevin 2.0 and the Future of SWE
22mIncreasing code habitability by using AI coding assistants
27mThe State of Large-Scale Code Modernization
57mRise of the Machines - Automate Your Development
52mAI-Powered Impact-Based Testing for Faster Safer Releases