Efficient Strictness Analysis of Haskell

Strictness analysis has been a living field of investigation since Mycroft’s original work in 1980, and is getting increasingly significant with the still wider use of lazy functional programming languages. This paper focuses on an actual implementation of a strictness analyser for Haskell. The analyser uses abstract interpretation with chaotic fixpoint iteration. The demand-driven nature of this iteration technique allows us to use large domains including function domains in the style of Burn et al. [BHA86] and Wadler [Wad87] and retain reasonable efficiency. The implementation, furthermore, allows us to introduce a new way of handling polymorphism by parameterizing the strictness functions with domainvariables. Finally we present some results of efficiency and precision and compare them to other works.