runact 1.2.2

A Rust-native actor runtime with BEAM-style processes
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
# Runact Runtime

## Overview

The Runact runtime is the execution kernel. It manages processes (actors), message delivery, scheduling, supervision, and timers. Uses a custom BEAM-style scheduler with work stealing. No Tokio types leak through public APIs.

## Core Concepts

### Actor

An Actor is an independent unit of computation with:

- Stable `ActorId`
- Owns mutable state
- Receives messages sequentially
- Processes one message at a time
- Can send messages to other actors
- Can spawn child actors (via supervisor)
- Can be supervised
- Can schedule timers
- Can submit compute tasks
- Can reply to requests

```rust
pub trait Actor: Send + 'static {
    type Message: Send + 'static;

    fn handle(
        &mut self,
        message: Self::Message,
        context: &mut ActorContext,
    ) -> Result<(), ActorError>;
}
```

### ActorId

A stable, unique identifier for an actor.

- Created at spawn time
- Never reused within a runtime instance
- Serializable (via `serde`)
- Orderable (for debugging)

```rust
#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord, Serialize, Deserialize)]
pub struct ActorId(u64);
```

### ActorContext

Provided to actor handlers.

```rust
pub struct ActorContext {
    actor_id: ActorId,
    _sender: crossbeam_channel::Sender<MessageEnvelope>,
    senders: Option<SenderMap>,
    reply_sender: Option<ReplySender>,
    compute_sender: Option<crossbeam_channel::Sender<Task>>,
    timer_handle: Option<TimerHandle>,
}
```

### Capabilities

- `actor_id()` — Get own ID
- `send_to(target, msg)` — Send fire-and-forget message to another actor
- `reply(msg)` — Reply to a request
- `is_request()` — Check if current message expects a reply
- `spawn_compute(job)` — Submit CPU-intensive task
- `schedule_timer(dur, msg)` — One-shot timer
- `schedule_interval(interval, msg)` — Periodic timer
- `cancel_timer(timer_id)` — Cancel a timer

### Lifecycle

```text
spawned → running → stopping → stopped
    │                  │
    │                  └──→ terminated
    │
    └──→ failed → restarting → running
```

States:

- **spawned** — Actor created, not yet scheduled
- **running** — Actor is executing
- **stopping** — Graceful shutdown initiated
- **stopped** — Actor completed normally
- **terminated** — Actor was forcibly stopped
- **failed** — Actor crashed or returned error
- **restarting** — Supervisor is restarting the actor

## Mailbox

Each actor has a mailbox with bounded capacity.

### Requirements

- FIFO ordering within an actor
- Configurable capacity (default: 1000)
- Backpressure policy: Reject (default) or Block
- Shutdown behavior: drain remaining messages

### Backpressure Strategies

When the mailbox is full:

| Strategy | Behavior |
|----------|----------|
| `Reject` | Return `RuntimeError::MailboxFull` to sender |
| `Block` | Sender waits until space available (`send_blocking`) |

### BackpressurePolicy (Internal)

```rust
pub(crate) enum BackpressurePolicy {
    Reject,
    Block,
    DropLowPriority,
}
```

### Mailbox Interface

The actual mailbox used by the runtime is `crossbeam_channel::bounded`:

```rust
let (tx, rx) = crossbeam_channel::bounded::<MessageEnvelope>(self.mailbox_capacity);
```

### Metrics

Mailbox depth is exposed via `ActorInfo`:

```rust
pub struct ActorInfo {
    pub id: ActorId,
    pub name: Option<String>,
    pub mailbox_depth: usize,
}
```

## Scheduler

BEAM-style scheduler with work stealing.

### Architecture

```text
┌─────────────────────────────────────────────────────────┐
│                    Scheduler                            │
├─────────────┬─────────────┬─────────────┬──────────────┤
│  Worker 0   │  Worker 1   │  Worker 2   │  Worker 3    │
│  ┌───────┐  │  ┌───────┐  │  ┌───────┐  │  ┌───────┐  │
│  │ RunQ  │  │  │ RunQ  │  │  │ RunQ  │  │  │ RunQ  │  │
│  │[A,B,C]│  │  │[D,E,F]│  │  │[G,H,I]│  │  │[J,K,L]│  │
│  └───────┘  │  └───────┘  │  └───────┘  │  └───────┘  │
│      │      │      │      │      │      │      │      │
│      ▼      │      ▼      │      ▼      │      ▼      │
│  Actor A    │  Actor D    │  Actor G    │  Actor J    │
└─────────────┴─────────────┴─────────────┴──────────────┘
                    │
            Work Stealing
```

### Requirements

- Lightweight actor creation
- Work stealing across cores
- Per-worker run queues
- Cooperative scheduling with reduction counting
- Cancellation (via stop flag)
- Timers (via TimerService)
- Process lifecycle management

### Scheduler Interface

```rust
pub struct Scheduler {
    queues: Vec<RunQueue>,
    next_worker: AtomicUsize,
    handles: Vec<JoinHandle<()>>,
    stop: Arc<AtomicBool>,
}

impl Scheduler {
    pub fn new() -> Self;
    pub fn spawn<F>(&self, f: F) where F: FnOnce() + Send + 'static;
    pub fn shutdown(&mut self);
    pub fn worker_count(&self) -> usize;
}
```

Workers spawn at `new()` and default to `available_parallelism()` count. The round-robin dispatch uses `next_worker.fetch_add(1, Ordering::Relaxed) % queues.len()`.

### ReductionCounter

```rust
pub const MAX_REDUCTIONS: u64 = 4000;

pub struct ReductionCounter {
    count: AtomicU64,
}

impl ReductionCounter {
    pub fn new() -> Self;
    pub fn increment(&self) -> u64;
    pub fn reset(&self);
}
```

After each message is processed, the counter is incremented. When it reaches `MAX_REDUCTIONS`, the counter is reset and `thread::yield_now()` is called.

## Compute Scheduler

For CPU-intensive tasks that would starve actors.

### Architecture

```text
┌─────────────────────────────────────────────────────────┐
│                 Compute Scheduler                       │
├─────────────┬─────────────┬─────────────┬──────────────┤
│  Worker 0   │  Worker 1   │  Worker 2   │  Worker 3    │
│  ┌───────┐  │  ┌───────┐  │  ┌───────┐  │  ┌───────┐  │
│  │ Task  │  │  │ Task  │  │  │ Task  │  │  │ Task  │  │
│  │[X,Y,Z]│  │  │[P,Q,R]│  │  │[S,T,U]│  │  │[V,W,X]│  │
│  └───────┘  │  └───────┘  │  └───────┘  │  └───────┘  │
│      │      │      │      │      │      │      │      │
│      ▼      │      ▼      │      ▼      │      ▼      │
│  Run to     │  Run to     │  Run to     │  Run to     │
│  completion │  completion │  completion │  completion │
└─────────────┴─────────────┴─────────────┴──────────────┘
```

### Requirements

- CPU-intensive tasks run to completion
- Work stealing across cores
- Bounded queue for backpressure (soft limit via `queue_capacity` config)
- Panic isolation at worker boundary
- Oneshot result channel
- Never block actors

### Compute Scheduler Interface

```rust
pub struct ComputeScheduler {
    sender: Option<crossbeam_channel::Sender<Task>>,
    handles: Vec<JoinHandle<()>>,
    _config: ComputeConfig,
    stop: Arc<AtomicBool>,
}

pub struct ComputeConfig {
    pub max_workers: usize,        // Defaults to available parallelism
    pub queue_capacity: usize,     // Soft limit (default: 1024)
    pub task_timeout: Option<Duration>,  // Reserved for future use
}

impl ComputeScheduler {
    pub fn new(config: ComputeConfig) -> Result<Self, ComputeError>;
    pub fn spawn<F, T>(&self, job: F) -> Result<ComputeHandle<T>, ComputeError>
    where
        F: FnOnce() -> T + Send + 'static,
        T: Send + 'static;
    pub fn sender(&self) -> Option<crossbeam_channel::Sender<Task>>;
    pub fn shutdown(&mut self);
}
```

### Task

```rust
pub struct Task {
    pub id: TaskId,
    pub job: Box<dyn FnOnce() -> Box<dyn Any + Send> + Send>,
    pub result_sender: crossbeam_channel::Sender<ComputeResult>,
    pub created_at: Instant,
    pub cancelled: Arc<AtomicBool>,
}
```

### ComputeResult

```rust
pub enum ComputeResult {
    Ok(Box<dyn Any + Send>),
    Err(ComputeError),
    Panic(String),
    Timeout,        // Reserved for future use
    Cancelled,
}
```

### ComputeHandle

```rust
pub struct ComputeHandle<T> {
    receiver: crossbeam_channel::Receiver<ComputeResult>,
    cancelled: Arc<AtomicBool>,
    _phantom: PhantomData<T>,
}

impl<T> ComputeHandle<T> {
    pub fn cancel(&self);
    pub fn is_cancelled(&self) -> bool;
    pub fn try_recv(&self) -> Option<Result<T, ComputeError>> where T: 'static;
    pub fn recv(self) -> Result<T, ComputeError> where T: 'static;
    pub fn recv_timeout(self, timeout: Duration) -> Result<T, ComputeError> where T: 'static;
}
```

### Panic Handling

Workers catch panics at the boundary:

```rust
let result = std::panic::catch_unwind(std::panic::AssertUnwindSafe(|| (task.job)()));
```

Panics are surfaced as `ComputeResult::Panic(msg)` and then as `ComputeError::WorkerPanic(msg)`.

### Backpressure

When the compute queue is full (unbounded channel in current implementation — queue_capacity is a soft limit not yet enforced), tasks are still accepted. The `ComputeError::QueueFull` error variant exists for future bounded queue support.

## Timer

Scheduled message delivery.

### Types

- **One-shot** — Fire once after delay
- **Periodic** — Fire repeatedly at interval (with drift correction)

### Timer Interface

```rust
pub struct TimerService {
    timers: Arc<Mutex<Vec<TimerEntry>>>,
    stop: Arc<AtomicBool>,
    handle: Option<JoinHandle<()>>,
}

pub struct TimerHandle {
    timers: Arc<Mutex<Vec<TimerEntry>>>,
}

impl TimerHandle {
    pub fn schedule_timer<M: Send + 'static>(&self, duration: Duration, actor_id: ActorId, message: M) -> TimerId;
    pub fn schedule_interval<M: Clone + Send + Sync + 'static>(&self, interval: Duration, actor_id: ActorId, message: M) -> TimerId;
    pub fn cancel_timer(&self, timer_id: TimerId);
}
```

### Drift Correction

Periodic timers use scheduled-at-based rescheduling:

```rust
// Reschedule based on previous scheduled time
scheduled_at: entry_scheduled_at + interval
```

This prevents drift accumulation.

## Cancellation

Cooperative cancellation via tokens.

### CancellationToken (via ComputeHandle)

```rust
pub fn cancel(&self) {
    self.cancelled.store(true, Ordering::Relaxed);
}

pub fn is_cancelled(&self) -> bool {
    self.cancelled.load(Ordering::Relaxed)
}
```

The compute worker checks the flag before and after execution. If set before, the task is skipped. If set after, the result is discarded.

## Runtime

Top-level coordinator that owns all components.

### Runtime Interface

```rust
pub struct Runtime {
    scheduler: Scheduler,
    compute: ComputeScheduler,
    timers: TimerService,
    actors: Arc<RwLock<HashMap<ActorId, ActorInfo>>>,
    senders: Arc<RwLock<HashMap<ActorId, crossbeam_channel::Sender<MessageEnvelope>>>>,
    request_counter: AtomicU64,
    stop: Arc<AtomicBool>,
    mailbox_capacity: usize,
    shutdown_timeout: Duration,
}

impl Runtime {
    pub fn new() -> Result<Self, RuntimeError>;
    pub fn with_config(config: RuntimeConfig) -> Result<Self, RuntimeError>;
    pub fn spawn<A: Actor>(&mut self, actor: A) -> Result<ActorId, RuntimeError>;
    pub fn send<M: Send + 'static>(&self, target: ActorId, message: M) -> Result<(), RuntimeError>;
    pub fn send_blocking<M: Send + 'static>(&self, target: ActorId, message: M) -> Result<(), RuntimeError>;
    pub fn request<M: Send + 'static>(&self, target: ActorId, message: M) -> Result<RequestHandle, RuntimeError>;
    pub fn request_blocking<M: Send + 'static, R: Send + 'static>(&self, target: ActorId, message: M, timeout: Option<Duration>) -> Result<R, RuntimeError>;
    pub fn spawn_supervisor(&self, strategy: RestartStrategy, children: Vec<ChildSpec>) -> Result<ActorId, RuntimeError>;
    pub fn list_actors(&self) -> Result<Vec<ActorInfo>, RuntimeError>;
    pub fn inspect_actor(&self, id: ActorId) -> Result<Option<ActorInfo>, RuntimeError>;
    pub fn compute(&self) -> &ComputeScheduler;
    pub fn schedule_timer<M: Send + 'static>(&self, duration: Duration, target: ActorId, message: M) -> TimerId;
    pub fn schedule_interval<M: Clone + Send + Sync + 'static>(&self, interval: Duration, target: ActorId, message: M) -> TimerId;
    pub fn cancel_timer(&self, timer_id: TimerId);
    pub fn stats(&self) -> RuntimeStats;
    pub fn shutdown(&mut self) -> Result<(), RuntimeError>;
}
```

### RequestHandle

```rust
pub struct RequestHandle {
    id: u64,
    receiver: crossbeam_channel::Receiver<Box<dyn std::any::Any + Send>>,
}

impl RequestHandle {
    pub fn id(&self) -> u64;
    pub fn recv(self) -> Result<Box<dyn std::any::Any + Send>, RuntimeError>;
    pub fn try_recv(&self) -> Result<Box<dyn std::any::Any + Send>, RuntimeError>;
    pub fn recv_timeout(&self, timeout: Duration) -> Result<Box<dyn std::any::Any + Send>, RuntimeError>;
}
```

### RuntimeConfig

```rust
pub struct RuntimeConfig {
    pub compute: ComputeConfig,
    pub mailbox_capacity: usize,       // Default: 1000
    pub shutdown_timeout: Duration,   // Default: 5 seconds
}
```

### RuntimeStats

```rust
pub struct RuntimeStats {
    pub actor_count: usize,
    pub request_count: u64,
}
```

### Shutdown

Graceful shutdown:

1. Sets the stop flag (actors stop accepting new messages and drain remaining)
2. Waits up to `shutdown_timeout` for actors to finish
3. Clears all internal state
4. Shuts down scheduler, timers, and compute pool

```rust
impl Drop for Runtime {
    fn drop(&mut self) {
        let _ = self.shutdown();
    }
}
```

## Supervisor

Manages child actor lifecycle and restart policies.

### Restart Strategy

```rust
pub enum RestartStrategy {
    OneForOne {
        max_restarts: usize,
        within: Duration,
        base_backoff: Duration,
    },
}
```

### Child Specification

```rust
pub struct ChildSpec {
    pub name: String,
    pub restart_policy: RestartPolicy,
    pub shutdown_timeout: Duration,
}

pub enum RestartPolicy {
    Permanent,   // Always restart
    Temporary,   // Never restart
    Transient,   // Restart only on abnormal exit
}
```

## Exit Criterion

Phase 1 (v1.0.0) is complete when:

- 100K+ concurrent actors communicate reliably
- Actor creation is lightweight
- Message delivery is reliable
- Work stealing across cores
- No starvation (reduction counting works)
- Timers fire correctly
- Cancellation works
- Shutdown is graceful
- Metrics are available