Session types provide a formal framework to enforce rich communication protocols, ensuring correctness properties such as type safety and deadlock freedom. However, the traditional API of functional session type systems with first-class channels often leads to problems with modularity and composability. This paper proposes a new, alternative session type API based on borrowing, embodied in the core calculus BGV. The borrowing-based API enables building modular and composable code for session type clients without imposing clutter or undue limitations. Its basis is a novel type system, founded on ordered linear typing, for functional session types with an explicit operation for splitting ownership of channels. We establish the semantics of BGV via a type-preserving translation to PGV, a deadlock-free functional session type calculus. We establish type safety and deadlock freedom for BGV by this translation. We also present an external version of BGV that supports use of borrow notation. We developed an algorithmic version of the type system that includes a mechanized verified translation from the external language to BGV. This part establishes decidable type checking.

Typestate systems are notoriously complex as they require sophisticated machinery for tracking aliasing. We propose a new, transition-oriented foundation for typestate in the setting of impure functional programming. Our approach relies on ordered types for simple alias tracking and its formalization draws on work on bunched implications. Yet, we support a flexible notion of borrowing in the presence of typestate. Our core calculus comes with a notion of resource types indexed by an ordered partial monoid that models abstract state transitions. We prove syntactic type soundness with respect to a resource-instrumented semantics. We give an algorithmic version of our type system and prove its soundness. Algorithmic typing facilitates a simple surface language that does not expose tedious details of ordered types. We implemented a typechecker for the surface language along with an interpreter for the core language.

Session types provide a principled approach to typed communication protocols that guarantee type safety and protocol fidelity. Formalizations of session-typed communication are typically based on process calculi, concurrent lambda calculi, or linear logic. An alternative model based on context-sensitive typing and typestate has not received much attention due to its apparent restrictions. However, this model is attractive because it does not force programmers into particular patterns like continuation-passing style or channel-passing style, but rather enables them to treat communication channels like mutable variables.

Imperative session types provide an imperative interface to session-typed communication. In such an interface, channel references are first-class objects with operations that change the typestate of the channel. Compared to functional session type APIs, the program structure is simpler at the surface, but typestate is required to model the current state of communication throughout. Following an early work that explored the imperative approach, a significant body of work on session types has neglected the imperative approach and opts for a functional approach that uses linear types to manage channel references soundly. We demonstrate that the functional approach subsumes the early work on imperative session types by exhibiting a typing and semantics preserving translation into a system of linear functional session types. We further show that the untyped backwards translation from the functional to the imperative calculus is semantics preserving. We restrict the type system of the functional calculus such that the backwards translation becomes type preserving. Thus, we precisely capture the difference in expressiveness of the two calculi and conclude that the lack of expressiveness in the imperative calculus is largely due to restrictions imposed by its type system.

Systems programming often requires the manipulation of resources like file handles, network connections, or dynamically allocated memory. Programmers need to follow certain protocols to handle these resources correctly. Violating these protocols causes bugs ranging from type mismatches over data races to use-after-free errors and memory leaks. These bugs often lead to security vulnerabilities. While statically typed programming languages guarantee type soundness and memory safety by design, most of them do not address issues arising from improper handling of resources. An important step towards handling resources is the adoption of linear and affine types that enforce single-threaded resource usage. However, the few languages supporting such types require heavy type annotations. We present Affe, an extension of ML that manages linearity and affinity properties using kinds and constrained types. In addition Affe supports the exclusive and shared borrowing of affine resources, inspired by features of Rust. Moreover, Affe retains the defining features of the ML family: it is an impure, strict, functional expression language with complete principal type inference and type abstraction. does not require any linearity annotations in expressions and supports common functional programming idioms.