The Most Important API Design Guideline - No, It's Not That One
summary
Argues that the top API design rule is testability over ease-of-use or raw performance, supported by concrete checks: examine IG, the Standard Library and Boost guidelines; prefer strong types, compile-time checks and the compiler as the first line of defense; use property-based and fuzz-style tests and commit generated code so diffs show generator regressions. Shows a production disaster where a reused enum/bit plus a missed deployment on one of eight servers and a blind rollback causes repeated child orders and a ~$460M loss, urging relentless decoupling and message-passing modularity.
AI-written summary · about
chapters (15)
- 0:00Introduction to Design
- 2:58Library Design Principles
- 7:28Importance of Testability
- 12:12Continuous Performance Testing
- 18:03Tools for Testability
- 22:19Donald Knuth on Testing
- 27:33Extreme Programming Critique
- 34:04Torture Testing Techniques
- 39:15Compiler as a Testing Tool
- 44:24Case Study: Tomahawk Missile
- 48:06The Night Capital Failure
- 1:02:07Avoiding Team Effort Bugs
- 1:07:28Reviewing Generated Code
- 1:14:03Industry Best Practices
- 1:16:55True Object-Oriented Design
