1531812 - Remove remaining mozRTC* stuff (mozRTCPeerConnection, mozRTCIceCandidate, mozRTCSessionDescription)

Status: RESOLVED FIXED

(Core :: WebRTC: Signaling, task, P3)

Comment 1

[Chris Mills [:cmills]]

It doesn't look like we mention that these are available in the first place. Almost pointless to add them considering they are now being removed, but hey ;-)

Comment 2

[Byron Campen [:bwc]]

Do you think we're ready to remove this stuff? webrtc-landing still uses some of this:

https://mozilla.github.io/webrtc-landing/data_test.html
https://mozilla.github.io/webrtc-landing/data_latency_test.html
https://mozilla.github.io/webrtc-landing/canvas_demo.html
https://mozilla.github.io/webrtc-landing/canvas_filter_demo.html

Comment 3

[Jan-Ivar Bruaroey [:jib] (needinfo? me)]

I'd like to say yes—with the exception of navigator.mozGetUserMedia (see bug 1279936)—but from bug 1155923 comment 5 I'm not sure. That was 4 years ago, but unfortunately we don't have telemetry on this.

The safe way to go about it might be to gather telemetry on mozRTCPeerConnection usage first, to know what breakage to expect.

That said, even having a telemetry number on mozRTCPeerConnection usage would not tell us about breakage, because it wouldn't distinguish

var RTCPeerConnection = window.webkitRTCPeerConnection || window.mozRTCPeerConnection || window.RTCPeerConnection; // OK!

from

var RTCPeerConnection = window.webkitRTCPeerConnection || window.mozRTCPeerConnection; // Would break!

I don't see a way to detect actual breakage numbers. That said, a ballpark mozRTCPeerConnection usage number would be instructive if it's really low.

Nico, is this something we could add to bug 1528078 which you'd expressed interest in tackling?

Comment 4

[Nico Grunbaum [:ng, @chew:mozilla.org]]

Yes, I will be tackling it in bug 1528078.

Comment 5

[Jan-Ivar Bruaroey [:jib] (needinfo? me)]

Turns out we do have telemetry probe for this: 2.7% of peer connections use moz-prefix. Breakage may be even lower based on comment 3.

• 0.218% of pageloads called any of the moz-prefixed peer connection methods
• 8% of pageloads call the non-deprecated RTCPeerConnection constructor (from chromestats, since we don't have counters for that)

Comment 6

[Byron Campen [:bwc]]

Taking this; it is about time, and low-hanging fruit.

Comment 7

[Byron Campen [:bwc]]

Attachment: Bug 1531812: Remove remaining mozRTC stuff. r?jib

Depends on D168415

Comment 8

[Byron Campen [:bwc]]

https://treeherder.mozilla.org/jobs?repo=try&revision=3cd05f17a7fc08fae22df095f70fd1be79d44edd
https://treeherder.mozilla.org/jobs?repo=try&revision=3fd07279480f4929f595fd756f34c0969bb103a3

Comment 9

[Byron Campen [:bwc]]

Try looks fine.

Comment 10

[Byron Campen [:bwc]]

Need to make a pull request here to get rid of this mozRTC stuff:

https://github.com/mozilla/webrtc-landing

Comment 11

[Byron Campen [:bwc]]

See https://github.com/mozilla/webrtc-landing/pull/35

Comment 12

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/cc4c2ac428b0
Remove remaining mozRTC stuff. r=jib,webidl,smaug

Comment 13

[Cristian Tuns]

Backed out for causing build bustages on PeerConnectionImpl.cpp

• Backout link
• Push with failures
• Failure Log
• Failure line: /builds/worker/checkouts/gecko/dom/media/webrtc/jsapi/PeerConnectionImpl.cpp:2024:38: error: unused variable 'id' [-Werror=unused-variable]

mochitest log: https://treeherder.mozilla.org/logviewer?job_id=405685822&repo=autoland

Comment 14

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/6f031686043e
Remove remaining mozRTC stuff. r=jib,webidl,smaug

Comment 15

[Sandor Molnar]

Backed out for causing wpt failures in

Backout link: https://hg.mozilla.org/integration/autoland/rev/19e0bc283894a3bf801c7edb6c73879c3aeec836

Push with failures

Failure log #1 | TEST-UNEXPECTED-TIMEOUT | /webrtc/RTCPeerConnection-remote-track-mute.https.html | pc.close() on one side causes mute events on the other - Test timed out

Failure log #2 | TEST-FAIL | /webrtc/RTCPeerConnection-getStats.https.html | getStats() audio and video validate all mandatory stats - assert_equals: Expect dictionary.transportId to be string expected "string" but got "undefined"

Failure log #3 | TEST-UNEXPECTED-FAIL | /webrtc/RTCPeerConnection-getStats.https.html | getStats() on track associated with RTCRtpReceiver should return stats report containing inbound-rtp stats - assert_unreached: required stats <...>

Comment 16

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/e8e892e3e742
Remove remaining mozRTC stuff. r=jib,webidl,smaug

Comment 17

[Natalia Csoregi [:nataliaCs]]

https://hg.mozilla.org/mozilla-central/rev/e8e892e3e742

Comment 18

[Byron Campen [:bwc]]

Release Note Request (optional, but appreciated)
[Why is this notable]:
Sadly, there are still services that are using this.
[Affects Firefox for Android]:
Yes.
[Suggested wording]:
Removed the long-deprecated mozRTCPeerConnection, mozRTCIceCandidate, and mozRTCSessionDescription types.
[Links (documentation, blog post, etc)]:

Comment 19

[Ryan VanderMeulen [:RyanVM]]

Added to the Fx113 Nightly relnotes. I added an additional line saying that sites should use the non-prefixed variants instead.

Comment 20

[Hamish Willee]

FYI MDN docs work for this can be tracked in https://github.com/mdn/content/issues/26146. There wasn't much to do in this case other than a release note mentioning the removal.