Modernizing legacy through thousands of contextual tools … per system
summary
Argues for building thousands of contextual tools per system that answer tiny reading questions instead of sprawling manual inspection. Shows LifeWare (≈35M LOC, 150k tests) running tests in ~18 minutes, grouping 454 failures into 36 identical comparison failures, batch-accepting a refactor that updates signatures, and tracing a slow AWS worker to an injected 5-minute wait which is edited and rerun in-place. Demonstrates automating targeted translation changes for an admin UI and tracks progress with time-to-question / time-to-answer metrics.
AI-written summary · about
chapters (10)
- 0:00Legacy system challenges
- 1:03Thousands of contextual tools
- 7:45Manual testing to engineering
- 10:36Code reading as frontier
- 13:09LifeWare case study
- 27:14IDE and tool design
- 31:25Ephemeral tools for problems
- 37:50Moldable development
- 39:56Legacy to business value
- 41:01Q&A
related talks
54mTesting in Legacy: From Rags to Riches
40mScaling testing & test automation without losing control
38mScaling Automated Testing & Architecture from Startup to Enterprise
21mScaling Test Automation: Building Reusability for 10x Speed and Efficiency
01h 35mRedesigning Legacy Systems - Keys to Success / Lessons Learned
43mTransformative Culture
43mHow to Innovate in a Legacy
28mManaging App Complexity