Can I stop testing now? (...)
summary
Line coverage from OTP cover proves only that code runs; it misses combinational faults. The talk demonstrates MC/DC via Smother giving a decision/condition tree and colorized HTML that pinpoints the untested combination that causes a divide by zero. It shows mutation testing with mu2 that mutates the AST, compiles hundreds of mutants and measures how many are “killed” (eg 500 → 50% killed, 2000 → 90%) and mentions model inference as a complementary approach.
AI-written summary · about
chapters (22)
- 0:00Introduction
- 0:31What does test adequacy mean
- 2:26What is a test adequacy metric
- 3:26Code coverage
- 5:07The dumbest airline program
- 8:46Line coverage
- 13:28Pattern matching
- 16:47Receiving statements
- 17:12Code coverage limitations
- 18:12Mutation testing
- 18:39What is mutation testing
- 20:47Mutation testing workflow
- 21:42Semantic equivalence
- 24:01Mutation operators
- 25:10Basic mutant operator
- 26:02Swapping order of cases
- 27:40Limitations of mutation testing
- 30:07Model inference
- 32:46Conclusions
- 34:21Prototypes
- 35:26Questions
- 35:53Cover
related talks
41mMutation testing in practice
36mMutation Testing: Did my test break my code ? 🤔
38mMutants to the rescue: How effective are your unit tests?
01h 06mKill All Mutants! (Intro to Mutation Testing)
37mAre your tests good enough?
35mKill All Mutants! (Intro to Mutation Testing)
50mFrom jUnit to Mutation-Testing
29mKill All Mutants! (Intro to Mutation Testing)