Section: Application Domains

High-assurance software

A large part of our work on programming languages and tools focuses on improving the reliability of software. Functional programming, program proof, and static type-checking contribute significantly to this goal.

Because of its proximity with mathematical specifications, pure functional programming is well suited to program proof. Moreover, functional programming languages such as OCaml are eminently suitable to develop the code generators and verification tools that participate in the construction and qualification of high-assurance software. Examples include Esterel Technologies's KCG 6 code generator, the Astrée static analyzer, the Caduceus/Jessie program prover, and the Frama-C platform. Our own work on compiler verification combines these two aspects of functional programming: writing a compiler in a pure functional language and mechanically proving its correctness.

Static typing detects programming errors early, prevents a number of common sources of program crashes (null dereferences, out-of bound array accesses, etc), and helps tremendously to enforce the integrity of data structures. Judicious uses of generalized abstract data types (GADTs), phantom types, type abstraction and other encapsulation mechanisms also allow static type checking to enforce program invariants.