Do Judge a Test by its Cover

Abstract Property-based testing uses randomly generated inputs to validate high-level program specifications. It can be shockingly effective at finding bugs, but it often requires generating a very large number of do so. In this paper, we apply ideas from combinatorial , powerful and widely studied methodology, modify the distributions our random generators so as find bugs with fewer tests. The key concept is coverage which measures degree given set tests exercises every possible choice values for small combination input features. its “classical” form, only applies programs whose have particular shape—essentially, Cartesian product finite sets. We generalize richer world algebraic data types by formalizing class sparse test descriptions based on regular tree expressions. This new definition inspires novel thinning algorithm improving generators, requiring many catch bugs. evaluate two case studies, typed evaluator System F terms Haskell compiler, showing significant improvements in both.