Offload functions to worker threads with shared memory primitives for Node.js.
chore: version 1.0.0
+57
-74
-10
.changeset/async-worker-cleanup.md
-10
.changeset/async-worker-cleanup.md
···
1
-
---
2
-
'moroutine': minor
3
-
---
4
-
5
-
Graceful async worker pool shutdown
6
-
7
-
- `await using run = workers(4)` waits for in-flight tasks to settle before terminating workers
8
-
- `run.signal` is an `AbortSignal` that fires when the pool starts disposing — thread it into tasks for cooperative cancellation
9
-
- `workers(size, { shutdownTimeout: ms })` force-terminates workers if graceful shutdown exceeds the timeout
10
-
- Existing `using run` (sync dispose) still terminates immediately
-20
.changeset/error-stack-traces.md
-20
.changeset/error-stack-traces.md
···
1
-
---
2
-
'moroutine': minor
3
-
---
4
-
5
-
Include main-thread call site in error stack traces
6
-
7
-
Errors thrown during task dispatch now have stack traces that show where `run()`, `exec()`, or `await task` was called. The original worker-side error is preserved as `err.cause` with its own stack pointing to the moroutine source.
8
-
9
-
```
10
-
Error: boom
11
-
at trackValue (worker-pool.ts:52:15)
12
-
at async loadUser (user-code.ts:7:3)
13
-
at async main (user-code.ts:11:3) {
14
-
[cause]: Error: boom
15
-
at fixtures/math.ts:6:9 // original throw site on the worker
16
-
at MessagePort.<anonymous> (worker-entry.ts:173:25)
17
-
}
18
-
```
19
-
20
-
Built-in error subclass identity (`TypeError`, `RangeError`, etc.) is preserved on the outer wrapper.
-7
.changeset/error-transfer.md
-7
.changeset/error-transfer.md
···
1
-
---
2
-
'moroutine': minor
3
-
---
4
-
5
-
Preserve error details across worker boundary
6
-
7
-
Errors thrown in moroutines now transfer with `message`, `stack`, `cause`, and built-in subclass identity (`TypeError`, `RangeError`, etc.) preserved via structured clone. Previously only the message string was kept.
-9
.changeset/load-balancing.md
-9
.changeset/load-balancing.md
···
1
-
---
2
-
'moroutine': minor
3
-
---
4
-
5
-
Configurable load balancing for worker pools
6
-
7
-
- `workers(size, { balance: leastBusy() })` routes tasks to the worker with the fewest in-flight tasks
8
-
- `roundRobin()` cycles through workers in order (default, same as before)
9
-
- Custom balancers implement `Balancer.select(workers, task)` for full control over scheduling
-10
.changeset/per-worker-dispatch.md
-10
.changeset/per-worker-dispatch.md
···
1
-
---
2
-
'moroutine': minor
3
-
---
4
-
5
-
Per-worker dispatch with `assign()` and `run.workers`
6
-
7
-
- `run.workers` exposes a read-only array of `WorkerHandle`s, one per pool worker
8
-
- `assign(worker, task)` returns a copy of the task pinned to a specific worker
9
-
- `worker.exec(task)` dispatches directly to a specific worker
10
-
- Channel fan-out no longer requires knowing the worker count
-7
.changeset/pool-worker-ref.md
-7
.changeset/pool-worker-ref.md
···
1
-
---
2
-
'moroutine': patch
3
-
---
4
-
5
-
Fix pool workers exiting early with top-level await
6
-
7
-
Pool workers are now ref'd for the lifetime of the pool, preventing the Node.js event loop from exiting prematurely when using `using run = workers()` with top-level `await`. Dedicated workers continue to ref only while tasks are in-flight.
-10
.changeset/task-type-refactor.md
-10
.changeset/task-type-refactor.md
···
1
-
---
2
-
'moroutine': major
3
-
---
4
-
5
-
Unify Task and StreamTask under a single `Task<T>` type
6
-
7
-
- `Task<T>` is `PromiseLike<T>` for value tasks, `AsyncIterable<T>` for streaming tasks, or just the base dispatch shape when unparameterized
8
-
- `Balancer.select()` receives `Task` instead of `Task<any> | StreamTask<any>`
9
-
- `Arg<T>` simplifies to `T | Task<T>`
10
-
- Classes renamed to `PromiseLikeTask<T>` and `AsyncIterableTask<T>` (internal, prefer `Task<T>` in type annotations)
+56
CHANGELOG.md
+56
CHANGELOG.md
···
1
1
# moroutine
2
2
3
+
## 1.0.0
4
+
5
+
### Major Changes
6
+
7
+
- e46aca0: Unify Task and StreamTask under a single `Task<T>` type
8
+
- `Task<T>` is `PromiseLike<T>` for value tasks, `AsyncIterable<T>` for streaming tasks, or just the base dispatch shape when unparameterized
9
+
- `Balancer.select()` receives `Task` instead of `Task<any> | StreamTask<any>`
10
+
- `Arg<T>` simplifies to `T | Task<T>`
11
+
- Classes renamed to `PromiseLikeTask<T>` and `AsyncIterableTask<T>` (internal, prefer `Task<T>` in type annotations)
12
+
13
+
### Minor Changes
14
+
15
+
- 9ab6016: Graceful async worker pool shutdown
16
+
- `await using run = workers(4)` waits for in-flight tasks to settle before terminating workers
17
+
- `run.signal` is an `AbortSignal` that fires when the pool starts disposing — thread it into tasks for cooperative cancellation
18
+
- `workers(size, { shutdownTimeout: ms })` force-terminates workers if graceful shutdown exceeds the timeout
19
+
- Existing `using run` (sync dispose) still terminates immediately
20
+
21
+
- 3612d52: Include main-thread call site in error stack traces
22
+
23
+
Errors thrown during task dispatch now have stack traces that show where `run()`, `exec()`, or `await task` was called. The original worker-side error is preserved as `err.cause` with its own stack pointing to the moroutine source.
24
+
25
+
```
26
+
Error: boom
27
+
at trackValue (worker-pool.ts:52:15)
28
+
at async loadUser (user-code.ts:7:3)
29
+
at async main (user-code.ts:11:3) {
30
+
[cause]: Error: boom
31
+
at fixtures/math.ts:6:9 // original throw site on the worker
32
+
at MessagePort.<anonymous> (worker-entry.ts:173:25)
33
+
}
34
+
```
35
+
36
+
Built-in error subclass identity (`TypeError`, `RangeError`, etc.) is preserved on the outer wrapper.
37
+
38
+
- 5137201: Preserve error details across worker boundary
39
+
40
+
Errors thrown in moroutines now transfer with `message`, `stack`, `cause`, and built-in subclass identity (`TypeError`, `RangeError`, etc.) preserved via structured clone. Previously only the message string was kept.
41
+
42
+
- d13cd40: Configurable load balancing for worker pools
43
+
- `workers(size, { balance: leastBusy() })` routes tasks to the worker with the fewest in-flight tasks
44
+
- `roundRobin()` cycles through workers in order (default, same as before)
45
+
- Custom balancers implement `Balancer.select(workers, task)` for full control over scheduling
46
+
47
+
- cdccfdb: Per-worker dispatch with `assign()` and `run.workers`
48
+
- `run.workers` exposes a read-only array of `WorkerHandle`s, one per pool worker
49
+
- `assign(worker, task)` returns a copy of the task pinned to a specific worker
50
+
- `worker.exec(task)` dispatches directly to a specific worker
51
+
- Channel fan-out no longer requires knowing the worker count
52
+
53
+
### Patch Changes
54
+
55
+
- e2dabc1: Fix pool workers exiting early with top-level await
56
+
57
+
Pool workers are now ref'd for the lifetime of the pool, preventing the Node.js event loop from exiting prematurely when using `using run = workers()` with top-level `await`. Dedicated workers continue to ref only while tasks are in-flight.
58
+
3
59
## 0.1.1
4
60
5
61
### Patch Changes