Procs#

A Proc is a container for actors and provides the runtime context for actor execution, message routing, and lifecycle management. In hyperactor’s architecture, procs serve as the fundamental unit of deployment and isolation.

Overview#

Procs bridge the gap between actors (local computation) and the distributed system (remote communication). The three types in this section fit together as follows:

A Proc hosts actors and manages their lifecycle, but is not itself a network endpoint.
A Gateway is the connectivity layer a proc attaches to: it multiplexes inbound traffic to the right local proc and routes outbound traffic on the proc’s behalf.
A Host manages the procs on one machine and owns the single gateway through which they are reached.

Key Components#

Proc#

The Proc struct represents a single actor runtime. It owns actor lifecycle and identity (all actors within a proc share its ProcId as part of their ActorId) and reaches the outside world only through its gateway.

See Proc for details on construction and usage.

Gateway#

The Gateway is the connectivity layer for procs. It has one table for local in-process procs, one table for peer gateways keyed by Via(uid) hops, and a forwarder for everything else. The same routing decision handles local delivery, spawned children, attached clients, and ordinary egress.

See Gateway for the type and its routing decision.

Host#

The Host manages the lifecycle of the procs on one machine. It owns exactly one gateway and delegates all connectivity to it: service and local attach as in-process procs, while spawned children attach as gateway peers.

See Host for the complete hosting architecture.