Bug 162305 – REGRESSION(r206133): LayoutTest inspector/css/manager-preferredInspectorStyleSheetForFrame.html is a flaky failure

• Status: RESOLVED
• Resolution: FIXED
• Component: Web Inspector

[Ryan Haddad]

LayoutTest inspector/css/manager-preferredInspectorStyleSheetForFrame.html is a flaky failure

This test has started frequently failing after http://trac.webkit.org/changeset/206133

https://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=inspector%2Fcss%2Fmanager-preferredInspectorStyleSheetForFrame.html

--- /Volumes/Data/slave/elcapitan-debug-tests-wk2/build/layout-test-results/inspector/css/manager-preferredInspectorStyleSheetForFrame-expected.txt
+++ /Volumes/Data/slave/elcapitan-debug-tests-wk2/build/layout-test-results/inspector/css/manager-preferredInspectorStyleSheetForFrame-actual.txt
@@ -20,12 +20,12 @@
 PASS: StyleSheet.isInspectorStyleSheet() should be true.
 
 -- Running test case: CreateSubFrameInspectorStyleSheet
+PASS: Added StyleSheet origin should be 'inspector'.
+PASS: StyleSheet.isInspectorStyleSheet() should be true.
+PASS: Added StyleSheet frame should be a child frame.
 PASS: Should be two stylesheets.
 PASS: StyleSheet origin should be 'inspector'.
 PASS: StyleSheet.isInspectorStyleSheet() should be true.
-PASS: Added StyleSheet origin should be 'inspector'.
-PASS: StyleSheet.isInspectorStyleSheet() should be true.
-PASS: Added StyleSheet frame should be a child frame.
 
 -- Running test case: AnotherSubFrameRequestDoesNothing
 PASS: Should be two stylesheets.

[Ryan Haddad]

https://build.webkit.org/builders/Apple%20El%20Capitan%20Debug%20WK2%20(Tests)/builds/8283

[Radar WebKit Bug Importer]

<rdar://problem/28389364>

[Ryan Haddad]

Marked test as flaky in https://trac.webkit.org/r206232

[Alexey Proskuryakov]

Is there some reason why this change hasn't been rolled out?

[Alexey Proskuryakov]

Seriously, a test change that broke the test, and we are keeping it in the tree?

[Joseph Pecoraro]

I'll take a look. Reverting the original change, scoped to just this test, seems fine to me. I don't find the new awaitEvent clearer in these tests at all anyways, and they add an extra turn of the runloop which (in cases like we see in this test) may cause flakiness.

[Joseph Pecoraro]

Created attachment 298627 [details]
[PATCH] Proposed Fix

Before this change this test failed 30 out of 100 times when run with --iterations=100.
After this change I saw 0 failures.

[WebKit Commit Bot]

Comment on attachment 298627 [details]
[PATCH] Proposed Fix

Clearing flags on attachment: 298627

Committed r210622: <http://trac.webkit.org/changeset/210622>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.