A veteran developer with 35 years of experience concludes a series on when automated tests help and when other approaches are better. The final installment examines how LLMs affect testing, noting they can generate code quickly but quality cannot be tested in after the fact. The author shares an LLM-generated list of 85 strategies to reduce testing needs, covering everything from property-based testing to formal verification. The key takeaway: replacing some tests with better design, compilers, linters, and language features leads to faster development and potentially higher confidence in code quality.
Sort: