How Developers Implement Property-Based Tests

Property-based testing (PBT) is an interesting alternative to example-based testing where the inputs are randomly generated by the testing tool. In PBT, we check properties that always hold for any input. Despite being a promising testing category, to the best of our knowledge, we still lack studies that investigate in the wild how developers are using PBT in practice. In this paper, we report the preliminary results of a study we are conducting on the usage of PBT. We created a dataset of 30 popular Python repositories using Hypothesis (a PBT tool) and selected a random sample of 86 tests. We manually analyzed these tests to understand the most commonly implemented properties and also to reveal the most used features to create them.