1209163 - [Meta] Make RTCDataChannels transferable (to workers)

Status: RESOLVED FIXED

(Core :: WebRTC: Networking, enhancement)

Description

[Randell Jesup [:jesup] (needinfo me)]

From my posting to the W3 public-webrtc list.  The PR is 317

Use-case:
A number of applications using RTCDataChannels would benefit from being able to process the data and send data from a Worker.  An example of this is a game which handles state updates and other processing in a Worker, keeping the mainthread application free to handle rendering and UI interactions, and avoiding or minimizing GC pauses.

Basic Mechanism:
The RTCDataChannel would be a transferrable object, which can be sent via postMessage:

  channel = createDataChannel("MyChannel", {});
  // or we're in an ondatachannel callback/event triggered by the other
  // side creating an RTCDataChannel
  worker.postMessage(channel, [channel]);
  // channel is now owned by the worker; no more events will occur on the
  // sender's version of 'channel'

The worker will receive the object, and can set event handlers on it.

Handling of Events:
After a transfer, all events occur in the worker, and the worker must set up event handlers to process them.  Any events that occur during the transfer, or that were queued to be fired in the sender, will be held until the worker gets the transferred object.  This means the worker should set EventHandlers on the channel immediately on receiving the channel.

Close:
The worker can call channel.close(), which will operate normally. If the PeerConnection is shut down via peerconnection.close(), or if the RTCDataChannel m=application line is renegotiated 'off', then the worker will receive a close event for the channel, as normal.

An RTCDataChannel may be transferred at any time, and may be transferred more than once.

Comment 1

[Bulk Bug Changes for mreavy's org]

Mass change P2->P3 to align with new Mozilla triage process.

Comment 2

[sideshowbarker [:sideshowbarker]]

Hi Byron, is this on the roadmap at all?

• WebKit has implemented it, and it shipped in Safari about a year ago — https://trac.webkit.org/changeset/274758/webkit/
• Blink hasn’t yet implemented it; feature bug is https://bugs.chromium.org/p/chromium/issues/detail?id=1244165

Comment 3

[Byron Campen [:bwc]]

We do not have any near-term plan to implement this at the moment, since we have an excess of P1/P2 work.

Comment 4

[Jimmy Warting]

Something like this can't be polyfilled nicely. would be great if this could be supported natively 🙏

Comment 5

[rpm]

Besides Safari this now also has been implemented by Blink (will be available in Chrome v130).

Comment 6

[fideltian]

@jim and all.

How about the features. the features is critical to Zoom's web client. Which could make audio/video media data transfer be more stable without the impact of UI. Could you prioritize the features. Thanks

Comment 7

[Byron Campen [:bwc]]

Attachment: Bug 1209163: Use MozPromise for datachannel stats gathering. r?ng

This will be necessary once it is possible for an RTCDataChannel to be
on a worker.

Comment 8

[Byron Campen [:bwc]]

Attachment: WIP: Bug 1209163: (WIP) Implement worker transfer for RTCDataChannel.

Depends on D260562

Comment 9

[Byron Campen [:bwc]]

Attachment: Bug 1209163: Re-enable and fix existing test-cases for RTCDataChannel transfer. r?jib

Depends on D261658

Comment 10

[Byron Campen [:bwc]]

Attachment: Bug 1209163: Expand test-cases for RTCDataChannel transfer. r?jib

Includes a shim and corresponding worker code that allows mainthread test code
to use a transferred RTCDataChannel very similarly to a mainthread
RTCDataChannel, to make it easy to write tests quickly, and adapt existing test
code to the transferred RTCDataChannel case.

Depends on D261659

Comment 11

[Byron Campen [:bwc]]

Attachment: Bug 1209163: StopGapEventTarget, a wrapper whose underlying target can be set after events have been dispatched. r?#xpcom-reviewers

We want this because RTCDataChannels often have events dispatched to them
before we know what thread they will actually be operating on.

This class allows us to dispatch events as normal while we are in this
indeterminate state, and when we learn what the event target will be, connect
it and allow the events to flow.

For example, it is possible for an RTCDataChannel to be created on an already
active SCTP transport that is already firing events. We do not want those
events to go to main until we know that is where the datachannel will live. As
another example, it is possible for JS to create a data channel, make a
getStats call, and then transfer that data channel; we do not want the event
dispatch for that getStats call to be sent to main, we want it to go to the
worker, but the worker thread has not even been created yet.

Depends on D261660

Comment 12

[Byron Campen [:bwc]]

Attachment: Bug 1209163: Calculate the origin string once in PeerConnection init, instead of separately for every datachannel. r?ng

We don't want to try to do this on a worker when we create an RTCDataChannel there.

Depends on D261661

Comment 13

[Byron Campen [:bwc]]

Attachment: Bug 1209163: This can be const. r?ng

Depends on D261662

Comment 14

[Byron Campen [:bwc]]

Attachment: Bug 1209163: Tighten up thread assertions. r?ng

Depends on D261663

Comment 15

[Byron Campen [:bwc]]

Attachment: Bug 1209163: Implement RTCDataChannel transfer. r?ng!,#webidl!,#dom-core!

Depends on D261665

Comment 17

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/6ff3d441b593
https://hg.mozilla.org/integration/autoland/rev/9ff70fdb5764
Use MozPromise for datachannel stats gathering. r=ng
https://github.com/mozilla-firefox/firefox/commit/ead4f7d733bf
https://hg.mozilla.org/integration/autoland/rev/2216cbf444b6
Re-enable and fix existing test-cases for RTCDataChannel transfer. r=jib
https://github.com/mozilla-firefox/firefox/commit/9c92e0b999e5
https://hg.mozilla.org/integration/autoland/rev/ab7d3e0055c5
Expand test-cases for RTCDataChannel transfer. r=jib
https://github.com/mozilla-firefox/firefox/commit/5c3ceb9d3330
https://hg.mozilla.org/integration/autoland/rev/2e3ec5c1c5b8
StopGapEventTarget, a wrapper whose underlying target can be set after events have been dispatched. r=xpcom-reviewers,nika
https://github.com/mozilla-firefox/firefox/commit/593ae753b71c
https://hg.mozilla.org/integration/autoland/rev/562dcf9a1628
Calculate the origin string once in PeerConnection init, instead of separately for every datachannel. r=ng
https://github.com/mozilla-firefox/firefox/commit/19478ecafd24
https://hg.mozilla.org/integration/autoland/rev/329ebbc61718
This can be const. r=ng
https://github.com/mozilla-firefox/firefox/commit/422443faa1d7
https://hg.mozilla.org/integration/autoland/rev/4bd86004f4c5
Tighten up thread assertions. r=ng
https://github.com/mozilla-firefox/firefox/commit/439d0a878078
https://hg.mozilla.org/integration/autoland/rev/0ca166528417
Implement RTCDataChannel transfer. r=ng,webidl,smaug,asuth
https://github.com/mozilla-firefox/firefox/commit/46d8b7408303
https://hg.mozilla.org/integration/autoland/rev/d6d506d58c91
1611953, 1884458, 1981202, 1983831: apply code formatting via Lando

Comment 18

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Created web-platform-tests PR https://github.com/web-platform-tests/wpt/pull/54706 for changes under testing/web-platform/tests

Comment 19

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/a2b09d5a7a6e
https://hg.mozilla.org/integration/autoland/rev/5db718a758de
Revert "Bug 1209163, Bug 1611953, Bug 1884458, Bug 1981202, Bug 1983831: apply code formatting via Lando"

Comment 20

[Atila Butkovits]

Backed out for causing failures at test_worker_interfaces_secureContext.html.

Backout link: https://hg-edge.mozilla.org/integration/autoland/rev/5db718a758debc4f6ebd8f04b27bfd2b406b8823

Push with failures: https://treeherder.mozilla.org/jobs?repo=autoland&resultStatus=testfailed%2Cbusted%2Cexception%2Cretry%2Cusercancel&revision=d6d506d58c9126587cddc8a47f10d2784bf8a56b&selectedTaskRun=WKwvrTx3Rl2y_FaorXLx4g.0

Failure logs:

https://treeherder.mozilla.org/logviewer?job_id=525771908&repo=autoland&task=JKnsAyA4SBSczZ26x4JMMQ.0&lineNumber=152194
https://treeherder.mozilla.org/logviewer?job_id=525783176&repo=autoland&task=LWLwDZHSSFq5Dg_yLvQ-jA.0&lineNumber=8114
https://treeherder.mozilla.org/logviewer?job_id=525771868&repo=autoland&task=CrcZPImbQGOggcThlbi41w.0&lineNumber=20926

Comment 21

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/04ab0ce9a292
https://hg.mozilla.org/integration/autoland/rev/4e3d560c9d7c
Use MozPromise for datachannel stats gathering. r=ng
https://github.com/mozilla-firefox/firefox/commit/51567d99ad76
https://hg.mozilla.org/integration/autoland/rev/e0dbd4eb11c0
Re-enable and fix existing test-cases for RTCDataChannel transfer. r=jib
https://github.com/mozilla-firefox/firefox/commit/441057e0f4e1
https://hg.mozilla.org/integration/autoland/rev/beb3cb7b25d6
Expand test-cases for RTCDataChannel transfer. r=jib
https://github.com/mozilla-firefox/firefox/commit/310807b93ff3
https://hg.mozilla.org/integration/autoland/rev/c5a258eed227
StopGapEventTarget, a wrapper whose underlying target can be set after events have been dispatched. r=xpcom-reviewers,nika
https://github.com/mozilla-firefox/firefox/commit/2e797c543417
https://hg.mozilla.org/integration/autoland/rev/509df70895ea
Calculate the origin string once in PeerConnection init, instead of separately for every datachannel. r=ng
https://github.com/mozilla-firefox/firefox/commit/98fc8a3084d9
https://hg.mozilla.org/integration/autoland/rev/cc55cc118195
This can be const. r=ng
https://github.com/mozilla-firefox/firefox/commit/71684c97f7c3
https://hg.mozilla.org/integration/autoland/rev/9844ec340530
Tighten up thread assertions. r=ng
https://github.com/mozilla-firefox/firefox/commit/2b36bae0880f
https://hg.mozilla.org/integration/autoland/rev/77533c2f3136
Implement RTCDataChannel transfer. r=ng,webidl,smaug,asuth
https://github.com/mozilla-firefox/firefox/commit/bfcca3129599
https://hg.mozilla.org/integration/autoland/rev/a821550ed296
1611953, 1884458, 1981202, 1983831: apply code formatting via Lando

Comment 22

[Sandor Molnar[:smolnar]]

Backed out for causing build bustages @ TestStopGapEventTarget.cpp

Backout link: https://github.com/mozilla-firefox/firefox/commit/9a817cb4a482fcbc946173c9d42ec2d530dbad72
https://hg.mozilla.org/integration/autoland/rev/de651b960db05a5da3784a1b1293311a05c789c5

Push with failures

Failure log -> ERROR - /builds/worker/checkouts/gecko/xpcom/tests/gtest/TestStopGapEventTarget.cpp:25:37: error: unknown class name 'CancelableRunnable'; did you mean 'mozilla::CancelableRunnable'?

Comment 23

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/9a817cb4a482
https://hg.mozilla.org/integration/autoland/rev/593c7df571d0
Revert "Bug 1209163, 1611953, 1884458, 1981202, 1983831: apply code formatting via Lando" for causing build bustages @ TestStopGapEventTarget.cpp

Comment 24

[Pulsebot]

Pushed by [e-mail]:
https://github.com/mozilla-firefox/firefox/commit/e292e126ba6b
https://hg.mozilla.org/integration/autoland/rev/bcb61012ba9d
Use MozPromise for datachannel stats gathering. r=ng
https://github.com/mozilla-firefox/firefox/commit/80dcee8bd7b2
https://hg.mozilla.org/integration/autoland/rev/aa46fdf155fd
Re-enable and fix existing test-cases for RTCDataChannel transfer. r=jib
https://github.com/mozilla-firefox/firefox/commit/4d73afe4c92a
https://hg.mozilla.org/integration/autoland/rev/a3a6ad306482
Expand test-cases for RTCDataChannel transfer. r=jib
https://github.com/mozilla-firefox/firefox/commit/05f20b632e13
https://hg.mozilla.org/integration/autoland/rev/6ee9daf54357
StopGapEventTarget, a wrapper whose underlying target can be set after events have been dispatched. r=xpcom-reviewers,nika
https://github.com/mozilla-firefox/firefox/commit/ab19147db3f8
https://hg.mozilla.org/integration/autoland/rev/8423f51240fd
Calculate the origin string once in PeerConnection init, instead of separately for every datachannel. r=ng
https://github.com/mozilla-firefox/firefox/commit/18ef532d3ce2
https://hg.mozilla.org/integration/autoland/rev/8efee5d1c515
This can be const. r=ng
https://github.com/mozilla-firefox/firefox/commit/13424b57e915
https://hg.mozilla.org/integration/autoland/rev/0e6b64411c08
Tighten up thread assertions. r=ng
https://github.com/mozilla-firefox/firefox/commit/965e644b8534
https://hg.mozilla.org/integration/autoland/rev/2965cdb041d3
Implement RTCDataChannel transfer. r=ng,webidl,smaug,asuth
https://github.com/mozilla-firefox/firefox/commit/bd1ff89aa136
https://hg.mozilla.org/integration/autoland/rev/78b06c570470
1611953, 1884458, 1981202, 1983831: apply code formatting via Lando

Comment 25

[amarc]

https://hg.mozilla.org/mozilla-central/rev/bcb61012ba9d
https://hg.mozilla.org/mozilla-central/rev/aa46fdf155fd
https://hg.mozilla.org/mozilla-central/rev/a3a6ad306482
https://hg.mozilla.org/mozilla-central/rev/6ee9daf54357
https://hg.mozilla.org/mozilla-central/rev/8423f51240fd
https://hg.mozilla.org/mozilla-central/rev/8efee5d1c515
https://hg.mozilla.org/mozilla-central/rev/0e6b64411c08
https://hg.mozilla.org/mozilla-central/rev/2965cdb041d3
https://hg.mozilla.org/mozilla-central/rev/78b06c570470

Comment 26

[Byron Campen [:bwc]]

Release Note Request (optional, but appreciated)
[Why is this notable]:
The ability to transfer RTCDataChannel to a worker is a long awaited feature in Firefox.

[Suggested wording]:
Implemented worker transfer for RTCDataChannel.

[Links (documentation, blog post, etc)]:
I guess https://w3c.github.io/webrtc-pc/#transfering-a-data-channel ?

Comment 27

[Donal Meehan [:dmeehan]]

Thanks, added to the Fx144 nightly release notes, please allow 30 minutes for the site to update.
Keeping the relnote-firefox flag as ? to keep it on the radar for inclusion in the final Fx144 release notes.

Comment 28

[Hamish Willee]

This isn't in the FF144 MDN release project because no one added dev-doc-needed.

My assumption is that all this needs is an update to the MDN release note, addition of RTCDataChannel to the list in Transferable objects, and a fix to the compatibility data in https://developer.mozilla.org/en-US/docs/Web/API/RTCDataChannel#browser_compatibility for the transferrable key.

@Donal FYI for MDN the docs link we use for information about this kind of thing is Transferable objects. You might choose to use that instead of a spec link.

Comment 29

[Donal Meehan [:dmeehan]]

Thanks Hamis. I changed the URL in the release notes to use https://developer.mozilla.org/docs/Web/API/Web_Workers_API/Transferable_objects.

Comment 30

[Donal Meehan [:dmeehan]]

Added to the Fx144 release notes

Comment 31

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Upstream PR was closed without merging

Comment 32

[Web Platform Test Sync Bot [:wpt-sync] (Matrix: #interop:mozilla.org)]

Upstream PR merged by moz-wptsync-bot