-
Notifications
You must be signed in to change notification settings - Fork 29.8k
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
child_process: create proper public API for channel
#30165
Conversation
CITGM: https://ci.nodejs.org/view/Node.js-citgm/job/citgm-smoker/2081/ @nodejs/child_process |
@vsemozhetbyt If it makes things easier for you, you can feel free to push to my branches directly – your suggestions are always helpful and I trust you to know how the docs should ideally look like :) |
@addaleax Thank you) Personally, I am not so sure about my nitpicking myself, and recently I usually review in occasions not so convenient for git operations. But I am grateful for your lingering encouragement) |
// in progress. Once the user has explicitly requested a certain state, these | ||
// methods become no-ops in order to not interfere with the user's intentions. | ||
refCounted() { | ||
if (++this.#refs === 1 && !this.#refExplicitlySet) { |
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.
Is refExplicitlySet
necessary? Are these *Counted
methods? It's the user's responsibility to pair ref and unref calls, balancing internal ref/unref calls is our responsibility. I'm not sure I see why they're needed.
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.
@bnoordhuis Yeah, I’m not a fan of this either, but without this tests are failing and I wouldn’t know a better solution … the issue isn’t so much that the calls might end up unbalanced, it’s that currently, a process.channel.unref()
call for example from userland is expected to unref the channel even if there are event listeners that would otherwise keep it ref()
’ed.
Oh, and for a bit of meta discussion: I added the control object as a fix for a bug, I didn't otherwise put a lot of thought into it. Since you're thinking of promoting it to public API: do you think it's the best possible interface? |
I think it’s fine … or, rather, I think having an API of the shape |
a1a949b
to
8e5f915
Compare
edit: although it could also be #25988. |
f3b2b3b
to
b47e13e
Compare
@lundibundi Addressed your comments and rebased, PTAL |
This needs a rebase. |
Ping @addaleax |
Instead of exposing the C++ bindings object as `subprocess.channel` or `process.channel`, provide the “control” object that was previously used internally as the public-facing variant of it. This should be better than returning the raw pipe object, and matches the original intention (when the `channel` property was first added) of providing a proper way to `.ref()` or `.unref()` the channel. Refs: nodejs#9322 Refs: nodejs#9313
b47e13e
to
ef418e1
Compare
This comment has been minimized.
This comment has been minimized.
Instead of exposing the C++ bindings object as `subprocess.channel` or `process.channel`, provide the “control” object that was previously used internally as the public-facing variant of it. This should be better than returning the raw pipe object, and matches the original intention (when the `channel` property was first added) of providing a proper way to `.ref()` or `.unref()` the channel. PR-URL: #30165 Refs: #9322 Refs: #9313 Reviewed-By: Jeremiah Senkpiel <fishrock123@rocketmail.com> Reviewed-By: Denys Otrishko <shishugi@gmail.com> Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl> Reviewed-By: James M Snell <jasnell@gmail.com>
Landed in e65bed1 🎉 |
Instead of exposing the C++ bindings object as
subprocess.channel
or
process.channel
, provide the “control” object that waspreviously used internally as the public-facing variant of it.
This should be better than returning the raw pipe object, and
matches the original intention (when the
channel
property wasfirst added) of providing a proper way to
.ref()
or.unref()
the channel.
Refs: #9322
Refs: #9313
Checklist
make -j4 test
(UNIX), orvcbuild test
(Windows) passes