A Coq Framework For Verified Property-Based Testing ( Extended Abstract )

We introduce a novel methodology for formally verified property-based testing and implement it as a framework on top of the QuickChick testing plugin for Coq.1 Our verification framework is aimed at proving the correctness of executable testing code with respect to a high-level specification, which captures the conjecture under test in a more direct way. To this end, we provide a systematic way for reasoning about the set of outcomes a random data generator can produce with non-zero probability. We have used our methodology to prove the correctness of most QuickChick combinators, with respect to the axiomatic semantics of a small number of primitive ones. We have also applied our methodology on a red-black tree example and made good progress on a more complex noninterference one. These encouraging preliminary results indicate that this verification methodology is modular, scalable, and requires minimal changes to existing code.