Specification-guided component-based synthesis from effectful libraries

Component-based synthesis seeks to build programs using the APIs provided by a set of libraries. Oftentimes, these have effects, which make it challenging reason about correctness potential candidates. This is because changes global state made effectful library procedures affect how they may be composed together, yielding an intractably large search space that can confound typical enumerative techniques. If nature effects are exposed as part their specification, however, deductive approaches used help guide for components. In this paper, we present new specification-guided procedure uses Hoare-style pre- and post-conditions express fine-grained component candidates drive bi-directional strategy. The alternates between forward process larger terms given existing context but otherwise unaware actual goal, alongside backward mechanism consistent with desired goal from must synthesized. To further improve efficiency scalability, integrate conflict-driven learning into algorithm provides semantic characterization previously encountered unsuccessful paths prune possible proceeds. We implemented our ideas in tool called demonstrate its effectiveness on number problems defined over OCaml libraries equipped specifications.