Bug 197207 – REGRESSION: (r244182) Layout Test http/wpt/resource-timing/rt-resources-per-frame.html is failing

• Component: Tools / Tests
• Status: RESOLVED CONFIGURATION CHANGED

[Truitt Savell]

The following layout test is failing on Mac

http/wpt/resource-timing/rt-resources-per-frame.html

Probable cause:

This failure started with 244182. I was able to confirm locally with:

run-webkit-tests --root debug-244182 http/wpt/resource-timing/rt-resources-per-frame.html --iterations 1000 -f --debug

Flakiness Dashboard:

https://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=http%2Fwpt%2Fresource-timing%2Frt-resources-per-frame.html

Diff:
--- /Volumes/Data/slave/highsierra-debug-tests-wk2/build/layout-test-results/http/wpt/resource-timing/rt-resources-per-frame-expected.txt
+++ /Volumes/Data/slave/highsierra-debug-tests-wk2/build/layout-test-results/http/wpt/resource-timing/rt-resources-per-frame-actual.txt
@@ -1,7 +1,6 @@
 Resource Timing: Entries per Frame
 
 
-PASS main frame resources differ from other frames 
-PASS iframe1 resources differ from other frames 
-PASS iframe2 resources differ from other frames 
+Harness Error (TIMEOUT), message = null
+

[Radar WebKit Bug Importer]

<rdar://problem/50140085>

[Ryosuke Niwa]

That's very strange. This test is nothing to do with rAF...

[Shawn Roberts]

Was able to reproduce in iOS Simulator in r244182, but not r244181.

Marked flaky in https://trac.webkit.org/changeset/244650/webkit

[Said Abou-Hallawa]

(In reply to Ryosuke Niwa from comment #2)
> That's very strange. This test is nothing to do with rAF...

The test load three resources through the windows of the main and the two sub-frames. But it has to wait before requesting the load. Do you know why we have to:

   wait(); // Ensure a DOMHighResTimeStamp change.

before making the load request?

[Ryosuke Niwa]

(In reply to Said Abou-Hallawa from comment #4)
> (In reply to Ryosuke Niwa from comment #2)
> > That's very strange. This test is nothing to do with rAF...
> 
> The test load three resources through the windows of the main and the two
> sub-frames. But it has to wait before requesting the load. Do you know why
> we have to:
> 
>    wait(); // Ensure a DOMHighResTimeStamp change.
> 
> before making the load request?

We want to make sure performance.now() changes because the test depends on performance.now value being different beyond that point.

[Said Abou-Hallawa]

Flakiness dashboard shows that the test has been passing for the last week and half. With the trunk WebKit I can't reproduce the test flakiness.