UI

I previously wrote on statistical and mutation testing for UIs. After additional thought, I don’t think UI mutation testing is feasible, at least in the traditional sense. Structure-based generative testing still could be. Lack of Deterministic Errors for Mutation Mutation testing is a tool to measure completeness. It tells us how well our other tests detect possible errors in our code. Mutation testing frameworks measure test suite completeness by crawling the code and injecting breaking changes.

I have a wide range of techniques for testing code behavior. I can reliably increase quality over time no matter how large the feature set grows and even measure the completeness of my test suite (mutation testing). The same can’t be said for UIs. I have long struggled to automate UI tests at all, and manual tests don’t sustain as the system grows. However, I think I may now have a technique to close the gap.