actor_lifecycle

Generic teardown helpers for actors that follow the SDK’s “polite stop message” convention. Used by credentials_cache and retry/rate_limiter — both opaque types wrap a Subject whose message variant set includes a Stop constructor that triggers actor.stop() next dispatch.

Lives in a separate module so the same fire-and-forget + monitor- based synchronous teardown lands in exactly one place. Adding a third long-lived actor later (a request-rate limiter, an event- stream demuxer) reuses these directly.

Values

shutdown_via_stop

</>

pub fn shutdown_via_stop(
  subject: process.Subject(msg),
  stop_message: msg,
) -> Nil

Fire-and-forget teardown: send the supplied Stop message and return immediately. The actor exits on its next dispatch. Safe to call against a dead actor — Erlang silently drops sends to a terminated Pid.

shutdown_via_stop_sync

</>

pub fn shutdown_via_stop_sync(
  subject: process.Subject(msg),
  stop_message: msg,
  timeout_ms: Int,
) -> Result(Nil, Nil)

Synchronous teardown: monitor the owning Pid, send Stop, then wait for the DOWN signal up to timeout_ms. Returns Ok(Nil) on clean exit (already-dead actor short-circuits here too — the subject_owner lookup fails fast). Returns Error(Nil) only on real timeout — i.e. the actor is alive but didn’t exit within the window. The monitor is demonitored on the timeout path so the caller’s mailbox doesn’t accumulate stray DOWN messages.