4 ms·

Are the tests testing?

• Wewoc 6mo ago
  Mostly yes — they were written by Claude/Gemini/ChatGPT, but they've already caught several regressions introduced by the models during later refactors; there’s a fascinating loop in letting one iteration of an LLM audit the next via a static test suite.
  • vrighter 6mo ago
    WDYM mostly? Did you scrutinize the tests even harder than the code?
    • Wewoc 6mo ago
      [dead]