Actor-Web

Appearance

Runtimes ​

A topology is inert until it runs. There are two ways to run one.

startRuntime(topology, opts?) ​

ts
import { startRuntime } from '@actor-web/runtime';

const runtime = await startRuntime(topology, { tools });

Starts a single-node, in-process runtime — ideal for the browser and tests. Returns a started runtime with:

MemberPurpose
runtime.actors.<id>Source factories (commands(), readModel(), session(), …).
runtime.requireActor(id) / getActor(id)The live ActorRef.
runtime.nodes.<node>.systemThe underlying ActorSystem (e.g. for subscribe).
runtime.stop()Graceful shutdown.

serveNode(topology, opts) node ​

ts
import { serveNode } from '@actor-web/runtime/node';

const node = await serveNode(topology, {
  node: 'worker',
  transport: true,
  peers: { server: serverUrl },
  connect: ['server'],
  gateway: { expose: ['taskBoard'] },
});

Serves one node of a topology with optional inter-node transport and a consumer gateway. Use this for multi-process / multi-machine deployments. Returns { system, transport, actors, requireActor, getGatewayUrl, getTransportUrl, stop, … }.

  • transport — enable runtime-to-runtime messaging (see Transport).
  • peers / connect — static peer wiring.
  • gateway.expose — which actors are reachable by UI consumers.

serveActorWebHttp(...) node ​

A REST ingress adapter that maps HTTP routes onto actor send/ask, for exposing actors over HTTP. See @actor-web/runtime/node.

See also ​