Strictness Analysis in UHC

Abstract

This thesis will investigate how strictness analysis can be done using a type-based approach in a real life compiler. It is inspired by the work of Holdermans et al. [9]. They used relevance typing as an approach to strictness analysis. The work is done in the context of the Utrecht Haskell Compiler (UHC). For the UHC Core language, an implementation is defined for the analysis and the transformation. In this thesis a monovariant analysis is chosen since there is no support for polyvariance in the code generation. The result is that there is no good support for higher order functions. Recursion is implemented using fixed point iteration. There is no support for datatypes yet.