Stop writing tests!
summary
Argues to write fewer hand-crafted cases and rely more on property-based tests that check general invariants and generate inputs automatically. Demonstrates Hypothesis to generate thousands of cases (one demo produces ~3,500 lines), finds concrete bugs like JSON float issues (NaN/infinity) and git branch-name constraints, and shows the ghostwriter that scaffolds tests from type hints and a continuous fuzzer that runs tests indefinitely, saving minimal failing examples for replay.
AI-written summary · about
chapters (7)
- 0:00Start
- 22:28How does ghost writer work with custom written methods?
- 24:50Does every test run a different set of test cases with hypothesis -I mean in the traditional sense of pytest?
- 25:20Does this work for keras ml models ?
- 26:40does it support code coverage?
- 26:53Do you have metrics that define the success of the library?
- 27:45What does this mean for TDD?
