runact 1.0.0

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
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
# Runact Actor Communication Principles

This document defines the communication model for Runact actors. It is the reference for how actors interact, how messages flow, and what constraints the runtime enforces.

## 1. Actors Never Share Mutable State

An actor owns its state.

```
Actor A                    Actor B
┌───────────┐              ┌───────────┐
│ State A   │              │ State B   │
└───────────┘              └───────────┘
      │                          │
      └────── messages ──────────┘
```

No actor should directly mutate another actor's state.

Rust's ownership system enforces this wherever possible.

## 2. Message Sending is Asynchronous

The fundamental operation is:

```rust
actor.send(message);
```

It means:

> "Put this message into the recipient's mailbox and return."

It does **not** mean:

> "Wait for the recipient to process this message."

Therefore:

```rust
a.send(Message::Something);
```

should normally return immediately.

## 3. `send()` Must Not Wait for the Actor

This is an important distinction.

**Bad:**

```rust
a.send(msg); // blocks until A processes msg
```

**Correct:**

```rust
a.send(msg); // enqueue and return
```

The sender should not depend on the recipient's execution speed.

## 4. Request/Reply is Still Asynchronous

Sometimes an actor needs an answer.

Don't turn that into synchronous actor blocking.

Instead:

```rust
let handle = actor.request(GetUser { id });
```

Conceptually:

```
A
│
│ Request
▼
B
│
│ Reply
▼
A
```

The request returns a handle, not the result itself.

For example:

```rust
let result = actor.request(message);
result.on_reply(|reply| {
    // handle reply
});
```

Or, if Runact eventually has an actor-native async mechanism:

```rust
let result = request.await;
```

But the crucial question is:

> What happens to the actor while it waits?

It must remain schedulable.

## 5. An Actor Must Never Block Its Scheduler Worker

This is probably the most important runtime rule.

**Bad:**

```
Worker 1
   │
   └── Actor A
          │
          └── waiting for B
```

Worker 1 should not become:

```
Worker 1
   │
   └── BLOCKED
```

**Correct:**

```
Actor A
   │
   └── waiting for reply
          │
          ↓
      scheduler
          │
          ├── Actor B
          ├── Actor C
          └── Actor D
```

A waiting actor becomes Waiting/Suspended, while the worker executes something else.

## 6. Waiting is a State, Not Blocking

Think of an actor state machine:

```
              ┌───────────┐
              │   Ready   │
              └─────┬─────┘
                    │
                    ▼
              ┌───────────┐
              │ Running   │
              └─────┬─────┘
                    │
          ┌─────────┼─────────┐
          │         │         │
          ▼         ▼         ▼
       Ready      Waiting   Sleeping
                    │
                    │ reply
                    ▼
                  Ready
```

The actor can wait for:

- reply
- timer
- I/O
- resource
- external event

without blocking an OS thread.

## 7. Actor-to-Actor Calls Should Not Hold Locks

**Avoid:**

```
A
│
├── lock state
│
└── call B
      │
      └── B calls A
```

This creates traditional lock-style deadlocks.

Instead, actor state should normally be accessed only while that actor is executing.

This gives us:

- No locks across actor boundaries.
- Locks can still exist internally when genuinely necessary, but they should not be the normal communication mechanism.

## 8. Request/Reply Must Have Correlation IDs

A request should carry an ID.

**Request:**

```
Request
├── request_id
├── sender
└── payload
```

**Reply:**

```
Reply
├── request_id
├── sender
└── result
```

**Example:**

```
A ── request #42 ──> B
A ── request #43 ──> B

B ── reply #43 ──> A
B ── reply #42 ──> A
```

The runtime can match replies correctly.

This becomes important when an actor has many outstanding requests.

## 9. Timeouts are Optional, Not Implicit

A request may specify:

```rust
request.timeout(Duration::from_secs(5));
```

If the timeout expires:

```
Request
   │
   ├── Reply → success
   │
   └── Timeout → failure
```

But Runact should not automatically impose arbitrary timeouts.

A local computation might legitimately take 30 seconds.

Timeouts are application-level policy.

## 10. Cancellation Should Be Cooperative

Never try to forcibly kill arbitrary Rust code.

**Bad model:**

```
cancel()
   ↓
kill thread
```

**Instead:**

```
cancel()
   ↓
CancellationToken
   ↓
computation checks token
   ↓
stops safely
```

For example:

```rust
if ctx.cancelled() {
    return Err(Error::Cancelled);
}
```

This matters particularly for the Runact compute pool.

## 11. Fire-and-Forget is a First-Class Pattern

Sometimes no response is required.

```rust
logger.send(Log(message));
```

The sender doesn't care about a reply.

This should be extremely cheap.

## 12. Request/Reply is Not the Default

This distinction is important.

**Prefer:**

- event
- notification
- command
- message

**Over:**

- call
- wait
- reply

when possible.

**Example:**

```
FileActor
   │
   └── FileChanged
           │
           ├── BufferActor
           ├── UIActor
           └── SearchActor
```

rather than having everyone synchronously query the file actor.

This encourages loose coupling.

## 13. Don't Make Every Operation Request/Reply

**Bad architecture:**

```
UI
 ↓
Buffer.call()
 ↓
File.call()
 ↓
Git.call()
 ↓
Network.call()
 ↓
AI.call()
```

You end up recreating RPC inside one process.

**Better:**

```
             ┌── BufferActor
             │
UIActor ──────┼── FileActor
             │
             ├── GitActor
             │
             ├── AIActor
             │
             └── LSPActor
```

Communication remains message-oriented.

## 14. Mailbox Backpressure Must Be Explicit

Asynchronous doesn't mean unlimited queues.

Consider:

```
Producer
   │
   │ 1,000,000 messages/sec
   ▼
Mailbox
   │
   │ 100 messages/sec
   ▼
Actor
```

Eventually memory explodes.

Therefore Runact needs mailbox policies.

**Potential policies:**

- Bounded
- Drop
- Reject
- Block external producer
- Coalesce
- Priority

But blocking an actor because its mailbox is full should not be the default.

For v0.1:

> Use bounded mailboxes where appropriate and return an explicit error when capacity is exhausted.

## 15. External Threads Are Different

There is an important distinction between:

- Actor → Actor
- External OS thread → Actor

A normal OS thread can potentially block.

For example:

```rust
let result = actor.request(msg).wait();
```

could be acceptable from an external thread.

But:

```rust
actor_context.call(other_actor, msg).wait();
```

inside an actor should be discouraged or prohibited.

So Runact can have two APIs:

```rust
// Actor context
request(actor, message)

// External/blocking context
request_blocking(actor, message)
```

This distinction makes the semantics clear.

## 16. Actor Code Should Be Mostly Deterministic

An actor should conceptually behave like:

```
State + Message → New State + Effects
```

For example:

```rust
fn handle(
    state: &mut State,
    message: Message,
) -> Effects {
    // ...
}
```

The actor owns the state transition.

External effects should go through runtime services.

This makes actors easier to test.

## 17. Effects Should Be Explicit

An actor shouldn't directly do everything.

**Good:**

```
BufferActor
   │
   ├── state mutation
   │
   └── request FileActor to save
```

**Bad:**

```
BufferActor
   │
   └── directly manipulates filesystem
```

This produces clearer architecture.

## 18. Long Computations Are Not Actor Work

Suppose an actor receives:

```
CompileProject
```

It shouldn't perform a 30-second compilation inside its actor execution.

Instead:

```
CompileActor
     │
     └── submit
          ↓
     Compute Pool
          │
          ↓
       Result
          │
          ↓
     CompileActor
```

This preserves responsiveness.

## 19. Failure is a Message/Runtime Event

An actor failing should not mean:

```
panic
 ↓
Runact dies
```

**Instead:**

```
Actor
  │
  X
  │
  ▼
Runtime
  │
  ▼
Supervisor
  │
  ├── restart
  ├── stop
  └── escalate
```

This is one of Runact's major BEAM-inspired properties.

## 20. Actor Lifecycle Must Be Explicit

An actor should have a lifecycle such as:

```
Created
   ↓
Starting
   ↓
Running
   ↓
Waiting
   ↓
Running
   ↓
Stopping
   ↓
Stopped
```

**Failure:**

```
Running
   ↓
Failed
   ↓
Supervisor
   ↓
Restart / Stop / Escalate
```

This becomes important for editor services.

## 21. No Actor Should Depend on Another Actor Staying Alive Forever

If:

```
A → B
```

A must be able to handle:

- B stopped
- B restarted
- B unavailable
- B timed out

This encourages resilient systems.

## 22. Backpressure Belongs at Boundaries

Suppose:

```
AIActor
   ↓
10,000 requests
   ↓
ComputePool
```

The runtime should prevent unlimited work from accumulating.

Therefore:

```
Actor
 ↓
bounded queue
 ↓
Compute
```

If overloaded:

- Rejected
- Busy
- Deferred
- Dropped

depending on the API.

## 23. The Runtime Must Distinguish Three Kinds of Waiting

This is especially important for Runact.

| Kind | Behavior |
|---|---|
| **Actor waiting** | Waiting for reply, timer, I/O — must not block a worker |
| **Compute waiting** | A CPU worker may legitimately be occupied doing CPU work |
| **External blocking** | An external OS thread may block if the API explicitly allows it |

So:

- Actor waiting → suspend actor
- CPU computation → occupy compute worker
- External blocking → allowed at boundary

## 24. The Fundamental Runact Invariant

> **An actor must never synchronously wait for another actor while occupying a Runact scheduler worker.**

And alongside it:

> **All actor-to-actor communication is asynchronous at the runtime level.**

Then:

> Synchronous request/reply is a convenience abstraction implemented through asynchronous messaging, suspension, correlation, and optional timeout — not through blocking the scheduler.

That is the principle that differentiates Runact from a simple thread/channel library.

---

## The Resulting Model

### Runact Actor Communication

```
                     RUNACT ACTOR
                          │
             ┌────────────┼────────────┐
             │            │            │
          Message      Request       Event
             │            │            │
             ▼            ▼            ▼
         Mailbox      Request ID     Mailbox
                          │
                          ▼
                       Reply
                          │
                          ▼
                    Resume Actor
```

### Scheduler State Machine

```
READY
  ↓
RUNNING
  ↓
WAITING ───────────────┐
  │                    │
  │ reply/timer/I/O     │
  └────────────────────┘
           ↓
         READY
```

**No actor-level blocking. No lock-based actor communication. No forced cancellation. Asynchronous messaging first; synchronous-looking APIs only as a safe abstraction on top.**