New issue
Advanced search Search tips

Issue 845149 link

Starred by 3 users

Issue metadata

Status: Verified
Owner:
Closed: Aug 21
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug

Blocked on:
issue 777617

Blocking:
issue 828866
issue 843979



Sign in to add a comment

virtual/webrtc-wpt-unified-plan/external/wpt/webrtc/RTCPeerConnection-setRemoteDescription-tracks.https.html times out

Project Member Reported by hbos@chromium.org, May 21 2018

Issue description

This is the same as running external/wpt/webrtc/RTCPeerConnection-setRemoteDescription-tracks.https.html with RTCConfiguration sdpSemantics:"unified-plan".

While Unified Plan will not be fully supported until we have transceivers ( https://crbug.com/777617 ), these spec compliant tests are Plan B compatible and should not time out when sdpSemantics:"unified-plan" is used, that would be a regression.

Most likely some event is not firing, investigate before the cut.
 

Comment 1 by hbos@chromium.org, May 21 2018

Blocking: 843979

Comment 2 by hbos@chromium.org, May 21 2018

Labels: M-68

Comment 3 by hbos@chromium.org, May 21 2018

NextAction set to tomorrow, if I haven't gotten around to it by then me or you can take it on.

Comment 4 by hta@chromium.org, May 21 2018

Reproducing it from the command line:

out/Default/content_shell --enable-blink-features=RTCUnifiedPlanByDefault --run-layout-test third_party/WebKit/LayoutTests/external/wpt/webrtc/RTCPeerConnection-setRemoteDescription-tracks.https.html

This hangs before producing any FAIL or PASS messages.

Comment 5 by hbos@chromium.org, May 22 2018

That is expected, it will only produce the FAIL/PASS messages at the end upon completing all tests. Even if it completes all but the last test it will not produce any FAIL/PASS messages.

Commenting out individual tests is a good way to determine what test(s) time out.

Comment 6 by hbos@chromium.org, May 22 2018

console.log and printf works too, separate from FAIL/PASS lines.
The NextAction date has arrived: 2018-05-22

Comment 8 by hbos@chromium.org, May 24 2018

Labels: -Pri-1 Pri-2
NextAction: ----
Lowering the priority because other testing (of real applications) is positive and the M68 cut is today. This should still be investigated though.
I'll be OOO for a few days.

Comment 9 by hta@chromium.org, May 27 2018

Spent a little time doing printf debugging.
One problem is this test:

'addTrack() with a track and two streams makes ontrack fire with a track and two streams.'

The function makes the name "performOffer" do double duty. Can be fixed in the test.

This test:

'removeTrack() makes stream.onremovetrack fire and the track to be removed from the stream.'

gets to the addTrack event, but not the removeTrack event.
 

Comment 10 by hbos@chromium.org, May 30 2018

Blockedon: 777617
Oh okay, that makes sense, probably that since receivers are no longer removed the plumbing does not work correctly. The code path needs to be updated for "transceiver's direction changed, it is inactive so remove tracks from streams".

This will all naturally be fixed with transceivers.

Comment 11 by hbos@chromium.org, May 30 2018

Owner: hbos@chromium.org
Status: Started (was: Available)
We won't fix this separately from transceivers I don't think, so I'm marking it as "started" since I'm working on transcievers.

Sign in to add a comment