/type-systems

Implementations of various type systems in OCaml.

Primary LanguageOCamlThe UnlicenseUnlicense

Grow Your Own Type System

This repository contains implementations of different type systems in OCaml.

It is meant to help out anyone who wants to learn more about advanced type systems and type inference or experiment by extending or implementing their own. The implementations are minimal and contain code that is (hopefully) simple and clear.

  • algorithm_w contains one of the most basic yet efficient implementation of Damas-Hindley-Milner type inference algorithm (used in functional languages such as OCaml, Haskell and Elm) called Algorithm W. Uses references to simulate type substitutions and assigns ranks/levels to type variables to simplify let-generalization.

  • extensible_rows extends algorithm_w with type inference for extensible records/rows with scoped labels, based on Daan Leijen's excellent paper. Although this is just one way of implementing extensible records, it's extremly simple and surprisingly useful, and was incorporated into the programming language Elm.

  • extensible_rows2 is an optimized implementation of extensible_rows.

  • first_class_polymorphism extends algorithm_w with type checking and partial type inference for first-class and higher-rank polymorphism, based on another one of Daan Leijen's papers. This system requires slightly more type annotations than other attempts at type inference for first-class polymorphism, such as MLF, but is considerably simpler to implements and use.

  • gradual_typing is another simple extension of algorithm_w based on a paper by Jeremy G. Siek and Manish Vachharajani. Gradual typing combines the benefits of static and dynamic typing, allowing programmers to make dynamic programs safer by adding static type information, and make static programs more flexible by delaying type-checking until runtime when necessary.

  • refined_types is an experiment that extends the HM type system with dependent types in the form of function contracts. It uses an external automatic theorem prover to verify that function contracts are satisfied, to prevent many of the most common software errors, such as division by zero and out-of-bounds array access.