This paper presents a new closure conversion algorithm for simply-typed languages. We have have implemented the algorithm as part of MLton, a whole-program compiler for Standard ML (SML). MLton first applies all functors and eliminates polymorphism by code duplication to produce a simply-typed program. MLton then performs closure conversion to produce a first-order, simply-typed program. In contrast to typical functional language implementations, MLton performs most optimizations on the first-order language, after closure conversion. There are two notable contributions of our work: 1. The translation uses a general flow-analysis framework which includes OCFA. The types in the target language fully capture the results of the analysis. MLton uses the analysis to insert coercions to translate between different representations of a closure to preserve type correctness of the target language program. 2. The translation is practical. Experimental results over a range of benchmarks...

This thesis is about a framework for elaborating and interpreting module language constructs at compile time in such a way that (1) arbitrary compile time information about declared identifiers of a module may be propagated to other modules and (2) no code is generated for module language constructs. The framework for interpreting module language constructs is called static interpretation. More information about referenced identifiers than can be obtained from programmer provided interfaces is necessary for analyses such as region inference. Further, many other analyses improve significantly from availability of analysis specific information about referenced identifiers. Static interpretation facilitates intermodule optimisation, yet, it still supports a variant of separate compilation called cut-off incremental recompilation.

We present a Java-like language in which objects are explicitly put in regions. The language has constructs for allocating, updating and deallocating regions, as well as region types for objects. For this language we present a static semantics ensuring that well-typed programs use regions safely, and we present a dynamic semantics that is intentional with respect to a region-based store. We formulate and prove a soundness theorem stating that well-typed programs do not go wrong. Finally, we develop a concrete model for implementing regions, and we compare this model to garbage collection for small examples.