Merging SMT solvers and programming languages

Galois is in the business of building trustworthy software. Such software will have well-defined behavior, and that behavior is assured in some way, whether via model checking, testing, or formal verification. SMT solvers — extensions to SAT solvers with support for variables of non-boolean type — offer powerful automation for solving a variety of assurance problems in software. We use them, for example, in Cryptol, to prove the equivalence (or otherwise) of algorithm implementations.

For a while now, Galois has been interested in connecting automated solvers to our programming language of choice, Haskell, to make it possible to prove automatically some properties of our functions (rather than just testing, e.g. with QuickCheck). We’ve pushed two efforts out this week, as previews of what we’re thinking in this space:

Both are embedded DSLs for representing propositions to an SMT solver via Haskell functions and values. They take different approaches (a compiler from Haskell to the SMT-LIB format, versus an interpreter for the Yices SMT solver). SBV is the more mature package, while yices-painless emphasizes a type-preserving translation from a minimal core language. SBV was built by Levent Erkok, yices-painless by Don Stewart. Documentation for the design of yices-painless is available, as is documentation on SBV.

Both are ready for experimentation and feedback, and we welcome your comments.