Plugging-in proof development environments usingLocksinLF

Abstract

We present two extensions of theLFconstructive type theory featuring monadiclocks. A lock is a monadic type construct that captures the effect of anexternal call to an oracle. Such calls are the basic tool forplugging-inand gluing together, different metalanguages and proof development environments. Oracles can be invoked either to check that a constraint holds or to provide a witness. The systems are presented in thecanonical styledeveloped by the ‘CMU School.’ The first system,CLLF𝒫, is the canonical version of the systemLLF𝒫, presented earlier by the authors. The second system,CLLF𝒫?, features the possibility of invoking the oracle to obtain also a witness satisfying a given constraint. In order to illustrate the advantages of our new frameworks, we show how to encode logical systems featuring rules that deeply constrain the shape of proofs. The locks mechanisms ofCLLF𝒫andCLLF𝒫?permit to factor out naturally the complexities arising from enforcing these ‘side conditions,’ which severely obscure standardLFencodings. We discuss Girard's Elementary Affine Logic, Fitch–Prawitz set theory, call-by-value λ-calculi and functions, both total and even partial.