byteflow-actors 0.5.1

Embeddable flow runtime: M:N scheduler, Atomic Hop (Message+Cap), supervisor — use byteflow::
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

# Error model (engineering rules)


**Panic is a runtime bug, not an error-handling mechanism.**

## Taxonomy


| Kind | Example | Surface |
|------|---------|---------|
| **A — user / API** | `spawn` bad function index | `Result<T, SpawnError>` |
| **B — flow** | native fault, bad Atomic Hop | `FlowOutcome::Failed` → Supervisor |
| **C — infrastructure** | mutex poison | `RuntimeError` + fail-closed (`report_fault`) |
| **D — invariant** | empty VM frame stack | types / `debug_assert` — never `unwrap` to hide design |

## Mutex policy


Never:

```text
lock.lock().unwrap()
lock.lock().unwrap_or_else(|e| e.into_inner())
```

Use [`sync_lock::lock`](../src/scheduler/sync_lock.rs) → `Result<_, RuntimeError::PoisonedLock>`.
In worker loops: `report_fault` + `return`. At API boundaries: propagate `Result`.

## Clippy (core crate)


```toml
unwrap_used = "deny"
expect_used = "deny"
```

Tests are allowed unwrap/expect via `cfg_attr(test, allow(...))` on the lib crate.

## Host API (category A)


```text
let rt = Runtime::new(chunk)?;                         // SpawnError::VerifyFailed | ThreadSpawnFailed
let handle = rt.spawn(fn_idx, &args)?;                 // SpawnError::BadFunction | …
let sup = Supervisor::new(rt.spawner())?;
let child = sup.start_child(spec)?;
```

Never panic on bad function index, failed verify, or OS thread spawn failure.