On regions and linear types (extended abstract)

Abstract

We explore how two different mechanisms for reasoning about state, linear typing and the type, region and effect discipline, complement one another in the design of a strongly typed functional programming language. The basis for our language is a simple lambda calculus containing first-class memory regions, which are explicitly passed as arguments to functions, returned as results and stored in user-defined data structures. In order to ensure appropriate memory safety properties, we draw upon the literature on linear type systems to help control access to and deallocation of regions. In fact, we use two different interpretations of linear types, one in which multiple-use values are freely copied and discarded and one in which multiple-use values are explicitly reference-counted, and show that both interpretations give rise to interesting invariants for manipulating regions. We also explore new programming paradigms that arise by mixing first-class regions and conventional linear data structures.

Context-Free Session Type Inference

“…What we need is a type-level mechanism to reason about the identity of endpoints. Similar requirements have already arisen in different contexts, to identify regions [27,2] and to associate resources with capabilities [1,25,23]. Reframing the techniques used in these works to our setting, the idea is to refine endpoint types to a form [T ] ρ where ρ is a variable that represents the abstract identity of the endpoint at the type level.…”

“…The use of type variables abstracting over the identity of endpoints has been inspired by works on regions and linear types [27,2], by L 3 [1], a language with locations supporting strong updates, and Alms [25,23], an experimental generalpurpose programming language with affine types. In these works, abstract identities are used to associate an object with the region it belongs to [27,2], or to link the (non-linear) reference to a mutable object with the (linear or affine) capability for accessing it.…”

“…In these works, abstract identities are used to associate an object with the region it belongs to [27,2], or to link the (non-linear) reference to a mutable object with the (linear or affine) capability for accessing it. Interestingly, in these works separating the reference from the capability (hence the use of abstract identities) is not really a necessity, but rather a technique that results in increased flexibility: the reference can be aliased without restrictions to create cyclic graphs [1] or to support "dirty" operations on shared data structures [25].…”

A Retrospective on Region-Based Memory Management

“…Other region-based frameworks for reasoning about memory reuse have been proposed, including a system for programs in continuation-passing style [61], one based on linear types [62], and one based on an imperative region sublanguage [32,44]. Another line of related work investigates techniques for improving region-based memory management [2].…”

Type Checking Systems Code

“…We are currently working to expand the region type and effects system to support (a) early reclamation of regions and (b) first-class regions in a style similar to what Walker and Watkins suggest [9]. We are also working on limited support for dependent types in the style of Hongwei Xi's Xanadu [10] so that programmers may better control the placement of bounds information or dynamic type tags.…”