Node 26 and enable Scala.js stress tests#178
Merged
Merged
Conversation
Update to Node 26 that updated to 14.6.202.33, that fixed V8 issue discussed in lampepfl#165.
`1L to total` produce `NumericRange[Long]` instead of an Int `Range`. `NumericRange[Long]` iterates through boxed values, and that makes async-heavy loop even slower.
Contributor
|
Looks like there's an OOM error 🤔 |
Contributor
Author
|
Oh, |
Previously, a `WasmAsyncSupport.scheduleBoundary` created
2-nested `js.async` where outer `js.async` does nothing.
`WasmAsyncSupport.scheduleBoundary` ran schelduled
bodies by `execute(() => boundary(body))` inherited from
the implementation from `AsyncSupport`.
`JSAsyncScheduler` implements `execute` using `js.async`:
```scala
object JsAsyncScheduler extends Scheduler:
def execute(body: Runnable) = js.async(body.run())
```
and, on the other hand, boundary also call `js.async`
```scala
override def boundary[T](body: Label[T] ?=> T): T =
val label = WasmLabel[T]()
js.async:
val r = body(using label)
label.resolve(r) // see [[WasmLabel]]
js.await(label.promise)
```
As a result, each `Future` body was started as something like:
```scala
js.async {
val label = WasmLabel[T]()
js.async {
val r = body(using label)
label.resolve(r)
}
js.await(label.promise)
}
```
The outer async block almost do nothing: it only waits until the
inner body first suspends or completes, and then exists.
In stress tests that create thousands of `Future`s, and this
async boundary creates many Promise/JSPI continuations and JS-Wasm
boundary, which makes significant slow-down and memory consumption.
**fix**
This commit fixes the problem by removing the unnecessary outer
`js.async` block. Now `scheduleBoundary` is implemented like:
```scala
val label = WasmLabel[Unit]()
s.execute: () =>
body(using label)
label.resolve(())
```
so the `Future` would be
```scala
val label = WasmLabel[Unit]()
js.async {
body(using label)
label.resolve(())
}
```
Now, stress test consumes around 900MB of memory (previously it takes
around 3.5GB~).
Contributor
Author
|
I found that |
tanishiking
commented
May 7, 2026
| * this assumes that the root context is **already** under `js.async`. | ||
| */ | ||
| trait WasmJSPISuspend(using AsyncToken) extends SuspendSupport: | ||
| trait WasmJSPISuspend(using AsyncToken) extends AsyncSupport: |
Contributor
Author
There was a problem hiding this comment.
I guess it makes sense, because JSPI anyway supports Async wether you like it or not ?
Contributor
There was a problem hiding this comment.
I think it makes sense yeah! No reason to split this in the case of JSPI
Contributor
Author
|
Now, c220e96 is negligible on my local environment, maybe we can revert that. |
Contributor
I think we can keep it, it makes sense to use ints for these loops anyway |
Contributor
Author
|
Wow, now test-js finishes in 1m30s, which previously took 3m30s~4m 🎉 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Update to Node 26 that updated to 14.6.202.33, that fixed V8 issue discussed in #165.
edit: now StressTest doesn't fail with stack overflow, but it is still extremely slow on V8 🤔 -> It looks like loop over Long generates
NumericRange[Long]that takes so long. Using Integer instead, and now tests finishes approximately 6min. maybe worth investigating on scala-js side.