Equal Rights for Tests
summary
Argues that tests must be equal to code and gives concrete practices: instrument tests for coverage and measure cyclomatic complexity of test functions with the same tools used on production code. Cites cyber-dojo experience where instrumenting tests exposed dead helpers and removed dead test code in ~1/8 commits, and warns of the NightTrading $450M loss when undefined C++ behavior jumped into untested dead code. Shows pytest tricks: UUID test names plus a t.id fixture, and uses coverage, linters and mutation ideas.
AI-written summary · about
chapters (11)
- 0:00Start
- 4:10Blood Sugar
- 7:02The Equilibrium Law
- 36:14Summary
- 42:06What Does a Function Do
- 49:29How Do We Name It a Function
- 49:58Is It a Global Function or a Member Function
- 55:53Side Effects
- 56:21Running Your Tests in a Random Order
- 1:07:08Not Using Long Test Names and Embedding It Somewhere Else Translate Equally Well into Other Languages beyond Python
- 1:24:34Static and Dynamic Analyzers
