Finding common ground: choose, assert, and assume

At present, the “testing community” is on good speaking terms, but typically lacks a common language for expressing some computational ideas, even in cases where such a language would be both useful and plausible. In particular, a large body of testing systems define a testing problem in the language of the system under test, extended with operations for choosing inputs, asserting properties, and constraining the domain of executions considered. While the underlying algorithms used for “testing” include symbolic execution, explicit-state model checking, machine learning, and “old fashioned” random testing, there seems to be a common core of expressive need. We propose that the dynamic analysis community could benefit from working with some common syntactic (and to some extent semantic) mechanisms for expressing a body of testing problems. Such a shared language would have immediate practical uses and make cross-tool comparisons and research into identifying appropriate tools for different testing activities easier. We also suspect that considering the more abstract testing problem arising from this minimalist common ground could serve as a basis for thinking about the design of usable embedded domain-specific languages for testing and might help identify computational patterns that have escaped the notice of the community.