remtypes

Remtypes are a concept in type theory and programming language design that annotate data with information about how values are moved or accessed across computation boundaries, such as between processes, machines, or security contexts. They attach movement semantics to traditional types, enabling the compiler and runtime to enforce safe cross-boundary interaction.

Origins and motivation: In distributed and capability-based systems, passing references or values across boundaries can lead to safety violations if references become invalid, if shared mutable state is uncontrolled, or if data is not properly serialized. Remtypes provide a framework to describe whether a value can be copied, whether only a serializable form can cross the boundary, and how lifetimes are managed.

Structure and semantics: A remtype annotation attaches to a type to express its remote behavior. Typical categories include local (opaque within the local boundary), remote-copy (the value can be copied and sent across), remote-reference (a reference to a remote object), and detachable (the value can be moved and relocated). Type-checking rules ensure that operations on remtypes obey boundary constraints, and runtime support may serialize data or recreate remote references safely.

Usage and examples: Remtypes appear in theoretical discussions of distributed type systems and in experimental languages and frameworks that implement safe cross-boundary communication, such as RPC systems, dataflow engines, or microservice architectures. They help prevent aliasing hazards, ensure consistent serialization, and clarify ownership semantics.

Relation to related ideas: Remtypes relate to serialization, marshalling, remote references, and ownership or capability-based type systems. They are typically orthogonal to the base data types and work as annotations or qualifiers.