CLAUDE/notes/progress.md

@@ -2,7 +2,7 @@
 
 ## Current Stage: Phase 1 — Research + First Improvement Cycle
 
-### Status: Cycle 9 COMPLETE
+### Status: Cycle 10 COMPLETE
 
 ## Plan Overview
 1. **Phase 0**: Codebase audit — understand current swactor architecture, existing tests, benchmarks ✅
@@ -124,6 +124,37 @@
   - `stop_nonexistent_actor_returns_error` — stop on bad address returns Err
 - **Result**: 82 tests pass, all workspace compiles
 
+### Cycle 10: Actor Timers (Tick-Counting)
+- **Research**: Studied timer/scheduling patterns across Erlang (timer:send_after, erlang:start_timer),
+  Akka (scheduleOnce, scheduler), Actix (ctx.run_later, ctx.run_interval), Kameo (tokio::time::sleep),
+  Tokio (tokio::time), Go (time.After, time.NewTicker)
+  - Also researched priority messages (REJECTED: lifecycle hooks cover 95% of use cases)
+  - Also researched SmallBox optimization (DEFERRED: measure allocation cost first)
+  - Key finding: per-worker tick-counting is ideal for swactor's synchronous model (deterministic)
+- **Implementation**: Per-worker `TimerWheel` with deterministic tick-based scheduling
+  - `OnceTimer`: fire once at `fire_at` tick, consumed after firing
+  - `IntervalTimer`: fire every `period` ticks, message cloned via `CloneMsg` trait
+  - `CloneMsg` trait: type-erased clone for interval timer messages (blanket impl for `Message`)
+  - `TimerRequest` enum: `Once { dest, msg, ticks }` | `Interval { dest, msg, period }`
+  - `ctx.send_after_ticks(addr, msg, ticks)` — one-shot timer API
+  - `ctx.send_interval_ticks(addr, msg, period)` — interval timer API
+  - Phase 2.5 in tick_once: fire due timers, route through full delivery system
+    (pool.deliver for local actors, transfer_txs for cross-worker, inbox_registry for inboxes)
+  - Phase 5.5: drain timer requests from handler buffer into TimerWheel
+  - GC: interval timers for removed actors cleaned up after cleanup_dead
+  - `schedule_timer` on Runtime's ContextInner: no-op with warning (timers are per-worker only)
+- **Bug fixed**: `gc_dead_intervals` was over-aggressive — removed timers for ANY address not
+  in the local pool, including inboxes and cross-worker actors. Fixed to only GC timers for
+  addresses in the `dead` set from cleanup_dead.
+- **Tests**: 6 new tests
+  - `one_shot_timer_fires_after_n_ticks` — timer with delay=3 fires on tick 4
+  - `handler_can_schedule_one_shot_timer` — timer scheduled from handler, fires correctly
+  - `one_shot_timer_fires_only_once` — consumed after firing, doesn't repeat
+  - `interval_timer_fires_repeatedly` — period=2, fires every 2 ticks (3 fires verified)
+  - `interval_timer_cleaned_up_when_actor_dies` — GC removes orphaned timers
+  - `timer_with_zero_delay_fires_next_tick` — delay=0 fires on next tick
+- **Result**: 88 tests pass, all workspace compiles, zero warnings
+
 ### Cycle 8: Dead Actor Cleanup (Memory Leak Fix)
 - **Research**: Audited remaining improvement gaps, found ActorPool and AddressMap both leak
   permanently poisoned actors. Known class of bug in Akka (#22990), CAF (#420).
@@ -194,8 +225,10 @@
 - [x] **Cycle 7: Actor recovery (factory restart)** ✅
 - [x] **Cycle 8: Dead actor cleanup** ✅
 - [x] **Cycle 9: Lifecycle hooks + graceful stop** ✅
-- [ ] **Cycle 10: Next improvement**
-  - Candidates: priority messages, actor timers, SmallBox optimization, property-based tests
+- [x] **Cycle 10: Actor timers (tick-counting)** ✅
+- [ ] **Cycle 11: Next improvement**
+  - Candidates: SmallBox optimization, property-based tests, named actors/registry, actor groups
+  - Priority messages REJECTED (lifecycle hooks cover 95% of cases)
   - LIFO slot rejected (0-5% benefit for typical workloads, not worth complexity)
 
 ## Open Questions

src/actor.rs

@@ -141,12 +141,43 @@ where
 /// Not a `Message` — intercepted in `tick_all` before reaching `handle_any`.
 pub(crate) struct StopSignal;
 
+/// Type-erased cloneable message for interval timers.
+/// Since `Message: Clone`, all actor messages can implement this.
+pub(crate) trait CloneMsg: Send {
+    fn clone_boxed(&self) -> Box<dyn Any + Send>;
+}
+
+impl<M: Message> CloneMsg for M {
+    fn clone_boxed(&self) -> Box<dyn Any + Send> {
+        Box::new(self.clone())
+    }
+}
+
+/// Timer request from a handler, queued for processing after tick_all.
+pub(crate) enum TimerRequest {
+    /// One-shot: deliver `msg` to `dest` after `ticks` worker ticks.
+    Once {
+        dest: ActorAddress,
+        msg: Box<dyn Any + Send>,
+        ticks: u64,
+    },
+    /// Repeating: deliver a clone of `msg` to `dest` every `period` ticks.
+    Interval {
+        dest: ActorAddress,
+        msg: Box<dyn CloneMsg>,
+        period: u64,
+    },
+}
+
 /// Object-safe inner trait for sending type-erased messages.
+#[allow(private_interfaces)]
 pub trait ContextInner {
     fn send_any(&self, addr: ActorAddress, msg: Box<dyn Any + Send>) -> Result<(), Error>;
     fn spawn_any(&self, addr: ActorAddress, actor: Box<dyn AnyActor>);
     /// Request graceful stop for an actor. Takes effect after the current message.
     fn request_stop(&self, addr: ActorAddress);
+    /// Schedule a timer (one-shot or interval).
+    fn schedule_timer(&self, request: TimerRequest);
 }
 
 /// Actor syscall interface — passed to `ActorInterface::handle()`.
@@ -193,6 +224,31 @@ impl<'a> Ctx<'a> {
         self.inner.request_stop(self.self_addr);
     }
 
+    /// Schedule a one-shot timer: deliver `msg` to `addr` after `ticks` worker ticks.
+    ///
+    /// The message is delivered as a normal mailbox message during the fire tick,
+    /// before `tick_all` processes messages. The timer is tick-counted (deterministic),
+    /// not wall-clock based.
+    pub fn send_after_ticks<M: Message>(&self, addr: ActorAddress, msg: M, ticks: u64) {
+        self.inner.schedule_timer(TimerRequest::Once {
+            dest: addr,
+            msg: Box::new(msg),
+            ticks,
+        });
+    }
+
+    /// Schedule a repeating timer: deliver a clone of `msg` to `addr` every `period` ticks.
+    ///
+    /// The first delivery happens after `period` ticks. The message is cloned for each
+    /// delivery. The timer continues until the target actor is stopped/poisoned.
+    pub fn send_interval_ticks<M: Message>(&self, addr: ActorAddress, msg: M, period: u64) {
+        self.inner.schedule_timer(TimerRequest::Interval {
+            dest: addr,
+            msg: Box::new(msg),
+            period,
+        });
+    }
+
     /// Spawn a restartable actor. On panic, recreated via `factory` up to
     /// `max_restarts` times before permanent poisoning.
     pub fn spawn_restartable<A, F>(

src/delivery.rs

@@ -161,6 +161,7 @@ impl InboxRegistry {
     }
 
     /// Check if an address is registered without consuming a message.
+    #[cfg(feature = "transport")]
     pub fn contains(&self, addr: &ActorAddress) -> bool {
         self.senders.read().unwrap().contains_key(addr)
     }

src/runtime.rs

@@ -5,7 +5,7 @@ use std::sync::{Arc, OnceLock};
 use std::thread::{self, JoinHandle, Thread};
 use std::time::Instant;
 
-use crate::actor::{Actor, ActorAddress, ActorInterface, AnyActor, Message, StopSignal};
+use crate::actor::{Actor, ActorAddress, ActorInterface, AnyActor, Message, StopSignal, TimerRequest};
 use crate::channel::{Receiver, Sender};
 // Re-export config types so existing code using `runtime::RuntimeConfig` still works
 pub use crate::config::{BackoffPolicy, MailboxOverflow, RuntimeConfig};
@@ -419,6 +419,7 @@ pub(crate) fn notify_worker(threads: &[OnceLock<Thread>], wid: usize) {
     }
 }
 
+#[allow(private_interfaces)]
 impl ContextInner for Runtime {
     fn send_any(&self, addr: ActorAddress, msg: Box<dyn Any + Send>) -> Result<(), Error> {
         match self.address_map.lookup(&addr) {
@@ -448,4 +449,11 @@ impl ContextInner for Runtime {
             notify_worker(&self.worker_threads, wid.as_usize());
         }
     }
+
+    fn schedule_timer(&self, _request: TimerRequest) {
+        // Timers are per-worker and tick-counted; scheduling from outside
+        // a worker context (e.g., rt.spawn() callback) is not supported.
+        // Use rt.send_to() with a delay loop instead.
+        eprintln!("swactor: schedule_timer called outside worker context — ignored");
+    }
 }

src/worker.rs

@@ -6,13 +6,110 @@ use std::sync::Arc;
 use std::thread;
 use std::time::Instant;
 
-use crate::actor::{ActorAddress, AnyActor, ContextInner, Ctx, StopSignal};
+use crate::actor::{ActorAddress, AnyActor, CloneMsg, ContextInner, Ctx, StopSignal, TimerRequest};
 use crate::channel::Receiver;
 use crate::config::MailboxOverflow;
 use crate::delivery::{Envelope, TickContext, WorkerId};
 use crate::stats::{ActorSnapshot, TickTiming, WorkerStats};
 use crate::Error;
 
+// ─── Per-Worker Timer Wheel ─────────────────────────────────────────────────
+
+struct OnceTimer {
+    fire_at: u64,
+    dest: ActorAddress,
+    msg: Box<dyn Any + Send>,
+}
+
+struct IntervalTimer {
+    next_fire: u64,
+    period: u64,
+    dest: ActorAddress,
+    msg: Box<dyn CloneMsg>,
+}
+
+/// Per-worker tick-counting timer wheel.
+///
+/// Timers are deterministic (tick-counted, not wall-clock). One-shot timers
+/// fire once and are consumed; interval timers fire repeatedly every N ticks.
+struct TimerWheel {
+    current_tick: u64,
+    once_timers: Vec<OnceTimer>,
+    interval_timers: Vec<IntervalTimer>,
+}
+
+impl TimerWheel {
+    fn new() -> Self {
+        Self {
+            current_tick: 0,
+            once_timers: Vec::new(),
+            interval_timers: Vec::new(),
+        }
+    }
+
+    /// Advance the tick counter and collect all due timer messages.
+    /// Returns the messages to be routed by the caller (may target local or remote actors/inboxes).
+    fn fire(&mut self) -> Vec<(ActorAddress, Box<dyn Any + Send>)> {
+        self.current_tick += 1;
+        let tick = self.current_tick;
+        let mut result = Vec::new();
+
+        // Fire one-shot timers (swap-remove for O(1) removal)
+        let mut i = 0;
+        while i < self.once_timers.len() {
+            if self.once_timers[i].fire_at <= tick {
+                let timer = self.once_timers.swap_remove(i);
+                result.push((timer.dest, timer.msg));
+            } else {
+                i += 1;
+            }
+        }
+
+        // Fire interval timers
+        for timer in &mut self.interval_timers {
+            if timer.next_fire <= tick {
+                let msg = timer.msg.clone_boxed();
+                result.push((timer.dest, msg));
+                timer.next_fire = tick + timer.period;
+            }
+        }
+
+        result
+    }
+
+    /// Remove interval timers whose target was just removed from the worker.
+    /// Only GCs timers for addresses in `dead` — inboxes and cross-worker actors
+    /// are not in the local pool but are still valid targets.
+    fn gc_dead_intervals(&mut self, dead: &[ActorAddress]) {
+        if dead.is_empty() {
+            return;
+        }
+        self.interval_timers.retain(|t| !dead.iter().any(|d| *d == t.dest));
+    }
+
+    /// Add a one-shot timer.
+    fn add_once(&mut self, dest: ActorAddress, msg: Box<dyn Any + Send>, ticks: u64) {
+        self.once_timers.push(OnceTimer {
+            fire_at: self.current_tick + ticks,
+            dest,
+            msg,
+        });
+    }
+
+    /// Add an interval timer. First fire is after `period` ticks.
+    fn add_interval(&mut self, dest: ActorAddress, msg: Box<dyn CloneMsg>, period: u64) {
+        let period = period.max(1); // prevent zero-period infinite loop
+        self.interval_timers.push(IntervalTimer {
+            next_fire: self.current_tick + period,
+            period,
+            dest,
+            msg,
+        });
+    }
+}
+
+// ─── Worker ─────────────────────────────────────────────────────────────────
+
 /// A worker owns a set of actors and runs them in a loop.
 pub(crate) struct Worker {
     pub(crate) id: WorkerId,
@@ -22,6 +119,8 @@ pub(crate) struct Worker {
     stats: Arc<WorkerStats>,
     /// Reusable scratch buffer for building per-actor snapshots.
     snapshot_buf: Vec<ActorSnapshot>,
+    /// Per-worker tick-counting timer wheel.
+    timers: TimerWheel,
 }
 
 impl Worker {
@@ -40,6 +139,7 @@ impl Worker {
             spawn_rx,
             stats,
             snapshot_buf: Vec::new(),
+            timers: TimerWheel::new(),
         }
     }
 
@@ -75,10 +175,32 @@ impl Worker {
         }
         let t2 = Instant::now();
 
+        // 2.5. Fire due timers → deliver to mailboxes before tick_all
+        let timer_msgs = self.timers.fire();
+        for (dest, msg) in timer_msgs {
+            if self.pool.contains(&dest) {
+                // Same-worker: deliver directly to actor's mailbox
+                self.pool.deliver(&dest, msg);
+            } else {
+                // Inbox or cross-worker: route through address map / inbox registry
+                match tc.address_map.lookup(&dest) {
+                    Some(wid) => {
+                        tc.transfer_txs[wid.as_usize()].send(Envelope::new(dest, msg));
+                        crate::runtime::notify_worker(tc.worker_threads, wid.as_usize());
+                    }
+                    None => {
+                        let _ = tc.inbox_registry.try_deliver(dest, msg);
+                    }
+                }
+            }
+            did_work = true;
+        }
+
         // 3. Tick all actors with WorkerContext
         let pending_local: RefCell<Vec<(ActorAddress, Box<dyn Any + Send>)>> =
             RefCell::new(Vec::new());
         let stop_requests: RefCell<Vec<ActorAddress>> = RefCell::new(Vec::new());
+        let timer_requests: RefCell<Vec<TimerRequest>> = RefCell::new(Vec::new());
 
         let processed;
         {
@@ -87,6 +209,7 @@ impl Worker {
                 tc,
                 pending_local: &pending_local,
                 stop_requests: &stop_requests,
+                timer_requests: &timer_requests,
                 stats: &self.stats,
             };
             processed = self.pool.tick_all(&worker_ctx, &self.stats, tc.config.actor_message_budget, &stop_requests);
@@ -121,6 +244,18 @@ impl Worker {
         for (addr, msg) in pending {
             self.pool.deliver(&addr, msg);
         }
+
+        // 5.5. Process timer requests from handlers
+        for request in timer_requests.into_inner() {
+            match request {
+                TimerRequest::Once { dest, msg, ticks } => {
+                    self.timers.add_once(dest, msg, ticks);
+                }
+                TimerRequest::Interval { dest, msg, period } => {
+                    self.timers.add_interval(dest, msg, period);
+                }
+            }
+        }
         let t5 = Instant::now();
 
         // 6. Publish stats (skip entirely when idle to avoid allocation + mutex)
@@ -172,12 +307,14 @@ impl Worker {
         let cleanup_pending: RefCell<Vec<(ActorAddress, Box<dyn Any + Send>)>> =
             RefCell::new(Vec::new());
         let cleanup_stops: RefCell<Vec<ActorAddress>> = RefCell::new(Vec::new());
-        {
+        let cleanup_timers: RefCell<Vec<TimerRequest>> = RefCell::new(Vec::new());
+        let dead = {
             let cleanup_ctx = WorkerContext {
                 worker_id: self.id,
                 tc,
                 pending_local: &cleanup_pending,
                 stop_requests: &cleanup_stops,
+                timer_requests: &cleanup_timers,
                 stats: &self.stats,
             };
             let dead = self.pool.cleanup_dead(&cleanup_ctx);
@@ -189,12 +326,16 @@ impl Worker {
                 self.stats.num_actors.store(self.pool.len(), Ordering::Relaxed);
                 did_work = true;
             }
-        }
+            dead
+        };
         // Deliver any messages sent during on_stop callbacks
         for (addr, msg) in cleanup_pending.into_inner() {
             self.pool.deliver(&addr, msg);
         }
 
+        // GC orphaned interval timers for actors that were just removed
+        self.timers.gc_dead_intervals(&dead);
+
         did_work
     }
 
@@ -237,6 +378,7 @@ struct WorkerContext<'a> {
     tc: &'a TickContext<'a>,
     pending_local: &'a RefCell<Vec<(ActorAddress, Box<dyn Any + Send>)>>,
     stop_requests: &'a RefCell<Vec<ActorAddress>>,
+    timer_requests: &'a RefCell<Vec<TimerRequest>>,
     stats: &'a WorkerStats,
 }
 
@@ -272,6 +414,10 @@ impl ContextInner for WorkerContext<'_> {
     fn request_stop(&self, addr: ActorAddress) {
         self.stop_requests.borrow_mut().push(addr);
     }
+
+    fn schedule_timer(&self, request: TimerRequest) {
+        self.timer_requests.borrow_mut().push(request);
+    }
 }
 
 struct ActorSlot {
@@ -463,6 +609,10 @@ impl ActorPool {
         self.actors.len()
     }
 
+    pub fn contains(&self, addr: &ActorAddress) -> bool {
+        self.actors.contains_key(addr)
+    }
+
     pub fn total_mailbox_depth(&self) -> usize {
         self.actors.values().map(|slot| slot.mailbox.len()).sum()
     }

tests/runtime_api.rs

@@ -2663,3 +2663,216 @@ fn stop_nonexistent_actor_returns_error() {
     let result = rt.stop_actor(fake_addr);
     assert!(result.is_err(), "stop_actor on nonexistent address should return Err");
 }
+
+// ── Timer Helpers ─────────────────────────────────────────────────────────
+
+/// Actor that schedules a one-shot timer in on_start: sends a Ping to target after N ticks.
+struct TimerStartActor {
+    target: ActorAddress,
+    delay_ticks: u64,
+}
+
+impl ActorInterface for TimerStartActor {
+    type Incoming = Ping;
+    type Response = Pong;
+
+    fn on_start(&mut self, ctx: &Ctx) {
+        ctx.send_after_ticks(self.target, Ping { reply_to: ctx.self_addr() }, self.delay_ticks);
+    }
+
+    fn handle(&mut self, _ctx: &Ctx, _msg: Ping) {}
+}
+
+/// Actor that schedules a one-shot timer when it receives a Forward message.
+struct DelayEchoActor;
+
+impl ActorInterface for DelayEchoActor {
+    type Incoming = Forward;
+    type Response = Done;
+
+    fn handle(&mut self, ctx: &Ctx, msg: Forward) {
+        ctx.send_after_ticks(msg.reply_to, Done(msg.value), 3);
+    }
+}
+
+/// Actor that schedules an interval timer on start: sends Ping every N ticks.
+struct HeartbeatActor {
+    target: ActorAddress,
+    period: u64,
+}
+
+impl ActorInterface for HeartbeatActor {
+    type Incoming = Ping;
+    type Response = Pong;
+
+    fn on_start(&mut self, ctx: &Ctx) {
+        ctx.send_interval_ticks(self.target, Ping { reply_to: ctx.self_addr() }, self.period);
+    }
+
+    fn handle(&mut self, _ctx: &Ctx, _msg: Ping) {}
+}
+
+// ── Timer Tests ───────────────────────────────────────────────────────────
+
+/// Given an actor that schedules a one-shot timer in on_start,
+/// when enough ticks pass,
+/// then the timer message is delivered to the target.
+#[test]
+fn one_shot_timer_fires_after_n_ticks() {
+    let rt = Runtime::new(RuntimeConfig::default());
+    let inbox = rt.new_inbox::<Ping>().unwrap();
+
+    let _timer_actor = rt.spawn(TimerStartActor {
+        target: *inbox.addr(),
+        delay_ticks: 3,
+    }).unwrap();
+
+    // Tick 1: on_start schedules timer (fire_at = current_tick + 3 = 4)
+    // Timer fires when current_tick >= fire_at, so after tick 4 completes
+    rt.tick(); // tick 1: on_start, timer scheduled
+    assert!(inbox.try_recv().is_none(), "no delivery before delay");
+
+    rt.tick(); // tick 2
+    assert!(inbox.try_recv().is_none(), "no delivery on tick 2");
+
+    rt.tick(); // tick 3
+    assert!(inbox.try_recv().is_none(), "no delivery on tick 3");
+
+    rt.tick(); // tick 4: timer fires
+    let msg = inbox.try_recv();
+    assert!(msg.is_some(), "timer message delivered after 3-tick delay");
+}
+
+/// Given an actor that schedules a one-shot timer from a message handler,
+/// when enough ticks pass after the triggering message,
+/// then the delayed response arrives.
+#[test]
+fn handler_can_schedule_one_shot_timer() {
+    let rt = Runtime::new(RuntimeConfig::default());
+    let inbox = rt.new_inbox::<Done>().unwrap();
+
+    let addr = rt.spawn(DelayEchoActor).unwrap();
+
+    let _ = rt.send_to(addr, Forward { value: 42, reply_to: *inbox.addr() });
+    rt.tick(); // process Forward, schedule timer (delay=3)
+
+    assert!(inbox.try_recv().is_none(), "no immediate reply");
+
+    rt.tick(); // tick 2
+    rt.tick(); // tick 3
+    assert!(inbox.try_recv().is_none(), "not yet");
+
+    rt.tick(); // tick 4: timer fires
+    let reply = inbox.try_recv();
+    assert_eq!(reply, Some(Done(42)), "delayed reply arrives after 3 ticks");
+}
+
+/// Given a one-shot timer,
+/// when it fires,
+/// then it does NOT fire again on subsequent ticks (consumed).
+#[test]
+fn one_shot_timer_fires_only_once() {
+    let rt = Runtime::new(RuntimeConfig::default());
+    let inbox = rt.new_inbox::<Ping>().unwrap();
+
+    let _timer_actor = rt.spawn(TimerStartActor {
+        target: *inbox.addr(),
+        delay_ticks: 1,
+    }).unwrap();
+
+    rt.tick(); // on_start schedules timer
+    rt.tick(); // timer fires
+    assert!(inbox.try_recv().is_some(), "first fire");
+
+    // Subsequent ticks should NOT fire again
+    for _ in 0..5 { rt.tick(); }
+    assert!(inbox.try_recv().is_none(), "one-shot does not repeat");
+}
+
+/// Given an interval timer with period 2,
+/// when multiple ticks pass,
+/// then the timer fires repeatedly every 2 ticks.
+#[test]
+fn interval_timer_fires_repeatedly() {
+    let rt = Runtime::new(RuntimeConfig::default());
+    let inbox = rt.new_inbox::<Ping>().unwrap();
+
+    let _heartbeat = rt.spawn(HeartbeatActor {
+        target: *inbox.addr(),
+        period: 2,
+    }).unwrap();
+
+    rt.tick(); // tick 1: on_start, interval scheduled (next_fire = current + 2 = 3)
+    assert!(inbox.try_recv().is_none(), "no fire on tick 1");
+
+    rt.tick(); // tick 2
+    assert!(inbox.try_recv().is_none(), "no fire on tick 2");
+
+    rt.tick(); // tick 3: first fire
+    assert!(inbox.try_recv().is_some(), "fire on tick 3");
+
+    rt.tick(); // tick 4
+    assert!(inbox.try_recv().is_none(), "no fire on tick 4");
+
+    rt.tick(); // tick 5: second fire
+    assert!(inbox.try_recv().is_some(), "fire on tick 5");
+
+    rt.tick(); // tick 6
+    assert!(inbox.try_recv().is_none(), "no fire on tick 6");
+
+    rt.tick(); // tick 7: third fire
+    assert!(inbox.try_recv().is_some(), "fire on tick 7");
+}
+
+/// Given an interval timer targeting an actor that gets stopped,
+/// when the actor is removed,
+/// then the interval timer is cleaned up (no orphan timers).
+#[test]
+fn interval_timer_cleaned_up_when_actor_dies() {
+    let rt = Runtime::new(RuntimeConfig::default());
+    let _inbox = rt.new_inbox::<Ping>().unwrap();
+
+    // Heartbeat sends to a counter that we'll kill
+    let counter_addr = rt.spawn(CounterActor { count: 0 }).unwrap();
+
+    // HeartbeatActor sends Ping to counter every tick
+    let _hb = rt.spawn(HeartbeatActor {
+        target: counter_addr,
+        period: 1,
+    }).unwrap();
+
+    // Let it run a few ticks
+    for _ in 0..3 { rt.tick(); }
+
+    // Stop the counter
+    rt.stop_actor(counter_addr).unwrap();
+    for _ in 0..5 { rt.tick(); }
+
+    // Counter is gone, interval timer should be GC'd.
+    // No crash, no leak — just verifying it doesn't panic.
+    let stats = rt.stats();
+    // Only the heartbeat actor should remain
+    assert_eq!(stats.workers[0].num_actors, 1);
+}
+
+/// Given a timer with delay 0,
+/// when the next tick fires,
+/// then the message is delivered immediately on the next tick.
+#[test]
+fn timer_with_zero_delay_fires_next_tick() {
+    let rt = Runtime::new(RuntimeConfig::default());
+    let inbox = rt.new_inbox::<Ping>().unwrap();
+
+    let _timer_actor = rt.spawn(TimerStartActor {
+        target: *inbox.addr(),
+        delay_ticks: 0,
+    }).unwrap();
+
+    rt.tick(); // on_start schedules timer with delay=0
+    // Timer requests are processed after tick_all (phase 5.5)
+    // Timer fires on the NEXT tick (phase 2.5)
+    assert!(inbox.try_recv().is_none(), "not yet — timer fires next tick");
+
+    rt.tick(); // timer fires
+    assert!(inbox.try_recv().is_some(), "zero-delay timer fires on next tick");
+}