Garbage collection via dynamic type inference - A formal treatment -

Abstract

A garbage collection (GC) scheme — what we call type inference GC — that dynamically performs type inference is studied. In contrast to conventional garbage collection that can collect only unreachable objects, this scheme can collect objects that are reachable yet semantically garbage. The idea is to exploit ML-like polymorphic types that can tell whether or not each object may be used in the rest of computation. There has been some work studying algorithms of the GC scheme. However, their descriptions had some obscurity in details of their methods, and did not give any formal correctness proof that captures the details. These facts, we believe, make their descriptions still unconvincing for implementors.

This paper aims to present a trustworthy specification of the GC scheme. In this specification, we first consider an underlying language that suitably reflects implementation details, on top of which we then formulate an algorithm of type inference GC, and formally prove its correctness. A significant point in our formulation is that we specify how to deal with Hindley-Milner polymorphism. Furthermore, showing our experimental results, we discuss in what cases this GC scheme is beneficial.