-
-
Notifications
You must be signed in to change notification settings - Fork 32k
lib: add a diagnostic channel for observing AsyncContextFrame.set #58229
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
base: main
Are you sure you want to change the base?
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
You should also add the channel to the diagnostics_channel docs here: https://github.com/nodejs/node/blob/main/doc/api/diagnostics_channel.md?plain=1#L1104
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #58229 +/- ##
=======================================
Coverage 90.23% 90.23%
=======================================
Files 633 633
Lines 186818 186822 +4
Branches 36668 36667 -1
=======================================
+ Hits 168578 168587 +9
- Misses 11036 11040 +4
+ Partials 7204 7195 -9
🚀 New features to boost your workflow:
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Might also be a good idea to have the tests validate that the frame can be used to look up the state for a given store. The store is the key so it should just be frame.get(store) to get the value.
@@ -23,6 +26,7 @@ class ActiveAsyncContextFrame extends SafeMap { | |||
|
|||
static set(frame) { | |||
setContinuationPreservedEmbedderData(frame); | |||
onSet.publish(frame); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Should we expose/document the AsyncContextFrame
class via public API?
With doing just this users get some undocumented, internal thing into their hand and all they can do is to guess what it is, which API it exposes.
Any internal change on AsyncContextFrame
would be breaking for all users of this channel.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I discussed this with @Qard; he thinks (but can of course speak for himself here 😁) that since all channels are by default experimental, we aren't promising stability in what's returned here. As far as I'm concerned, I would've been fine with not even exposing the parameter at all.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
One of the probles to never stabilize async_hooks was because it leaks internal.
We should avoid to add again experimental APIs leaking internal as they can't be stabilized.
If the frame is not needed it's likely best to remove it. Alternative could be to replace it by a well documented wrapper object exposing only parts of AsyncContextFrame or other data like the cause for the set,...
|
||
* `frame` {Object} | ||
|
||
Emitted when the async context frame for the current execution context is set. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Does this hold true for await
?
As far as I remember await
or more concrete for Promise
s the activation/setting happens by v8.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No, an event won't be published on await
continuation. If we were to implement that, we'd be right back to using Isolate::SetPromiseHook
. Fortunately, it is not necessary. V8 appropriately propagates whatever is set as the Continuation Preserved Embedder Data (CPED) into the "then" callbacks. I was a bit vague when saying "native state is kept in sync with change in the active frame" but the idea really is that the derived native state itself should be stored in and retrieved from the AsyncContextFrame (ACF) object, e.g. by defining a private symbol-keyed property on it. This way when V8 changes current ACF as its CPED, the native state automatically tracks and we "only" need to observe when Node sets ACF in CPED for the current execution to ensure the native state in the current ACF object is up to date.
Unsurprisingly, I'm working on doing exactly this in the Datadog's profiler if you want to see a C++ example :-)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Are you really interested in the set()
calls?
If I understand this correct it's actually the creation of a new AsyncContextFrame
- so the ALS.enterWith()
and run()
calls which are of your interest but not the ACF.exchange()
calls.
exchange()
is called by e.g. timers, nextTick,... - to mimic what v8 does automatically for promises.
Adds an
async_context_frame.set
diagnostic channel for observingAsyncContextFrame.set
.This is helpful for functionality in extensions written in C++ that need to keep some derived native state from one or more
AsyncLocalStorage
s in sync with changes in active frame. It is possible that such an extension needs to use the state when it can not enter V8 to obtain the data throughAsyncLocalStorage
JS API. A subscriber to this diagnostic channel can be used instead to eagerly keep the native state updated.