Bug 311032 – REGRESSION(310087@main): [GStreamer] Introduced 3 new layout test failures/timeouts

• Status: RESOLVED
• Resolution: FIXED
• Component: Platform

[Carlos Alberto Lopez Perez]

The following layout tests are timing out on WPE and GTK since 310087@main 



Regressions: Unexpected text-only failures (1)
  imported/w3c/web-platform-tests/mediacapture-streams/MediaStream-removetrack.https.html [ Failure ]

Regressions: Unexpected timeouts (2)
  media/media-fragments/TC0009.html [ Timeout ]
  media/media-fragments/TC0014.html [ Timeout ]


The diff on the failure is:

--- /sdk/webkit/layout-test-results/imported/w3c/web-platform-tests/mediacapture-streams/MediaStream-removetrack.https-expected.txt
+++ /sdk/webkit/layout-test-results/imported/w3c/web-platform-tests/mediacapture-streams/MediaStream-removetrack.https-actual.txt
@@ -7,6 +7,6 @@
 
 
 PASS Tests that a removal from a MediaStream works as expected
-PASS Test that removal from a MediaStream fires ended on media elements (video first)
+FAIL Test that removal from a MediaStream fires ended on media elements (video first) assert_equals: got video ended event expected true but got false
 FAIL Test that removal from a MediaStream fires ended on media elements (audio first) assert_equals: audio element ended because no more audio tracks expected true but got false
 


Note: I have double checked this: reverting 310087@main locally makes this test pass back again

[Philippe Normand]

Why was this not detected by EWS?

[Philippe Normand]

Pull request: https://github.com/WebKit/WebKit/pull/61682

[Alicia Boya García]

(In reply to Philippe Normand from comment #1)
> Why was this not detected by EWS?

I had a conversation with Claudio about this. Currently the EWS is very conservative about reporting errors and will not report any tests that don't fail consistently in both runs. When the EWS ran, the tests were reported as flaky and therefore ignored. https://ews-build.webkit.org/#/builders/34/builds/117975

Changes just landed to try to be less conservative, with the hope that that is feasible now that the bots are greener, but this is still being currently experimented. https://github.com/WebKit/WebKit/commit/471453201b4f4770ce4eb2edb0539c8ff2cb4066

[EWS]

Committed 310318@main (84afeac1a2e5): <https://commits.webkit.org/310318@main>

Reviewed commits have been landed. Closing PR #61682 and removing active labels.

[Radar WebKit Bug Importer]

<rdar://problem/173789590>