Bug 167652 – imported/w3c/web-platform-tests/shadow-dom/slotchange.html is a flaky failure

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Ryan Haddad]

LayoutTest imported/w3c/web-platform-tests/shadow-dom/slotchange.html is a flaky failure

https://build.webkit.org/results/Apple%20Sierra%20Debug%20WK2%20(Tests)/r211432%20(1844)/results.html

https://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=imported%2Fw3c%2Fweb-platform-tests%2Fshadow-dom%2Fslotchange.html

--- /Volumes/Data/slave/sierra-debug-tests-wk2/build/layout-test-results/imported/w3c/web-platform-tests/shadow-dom/slotchange-expected.txt
+++ /Volumes/Data/slave/sierra-debug-tests-wk2/build/layout-test-results/imported/w3c/web-platform-tests/shadow-dom/slotchange-actual.txt
@@ -1,9 +1,9 @@
 
 Harness Error (TIMEOUT), message = null
 
-PASS slotchange event: Append a child to a host. 
-PASS slotchange event: Remove a child from a host. 
-PASS slotchange event: Remove a child before adding an event listener. 
+TIMEOUT slotchange event: Append a child to a host. Test timed out
+TIMEOUT slotchange event: Remove a child from a host. Test timed out
+TIMEOUT slotchange event: Remove a child before adding an event listener. Test timed out
 PASS slotchange event: Change slot= attribute to make it un-assigned. 
 TIMEOUT slotchange event: Change slot's name= attribute so that none is assigned. Test timed out
 PASS slotchange event: Change slot= attribute to make it assigned.

[Ryan Haddad]

Marked test as flaky in http://trac.webkit.org/projects/webkit/changeset/211593

[Matt Lewis]

This has since started to be a flaky timeout on Sierra as well:

https://build.webkit.org/results/Apple%20Sierra%20Debug%20WK2%20(Tests)/r222625%20(3254)/results.html
https://build.webkit.org/builders/Apple%20Sierra%20Debug%20WK2%20(Tests)/builds/3254

[Radar WebKit Bug Importer]

<rdar://problem/42841417>

[Ryosuke Niwa]

Created attachment 350479 [details]
Reduction

The issue here is that the JS wrapper of a HTMLSlotElement can be prematurely collected. We need to use the GC retention mechanism we introduced in https://bugs.webkit.org/show_bug.cgi?id=184307 to keep these slot elements alive.

[Ryosuke Niwa]

Created attachment 350486 [details]
Fixes the bug

[Ryosuke Niwa]

Comment on attachment 350486 [details]
Fixes the bug

View in context: https://bugs.webkit.org/attachment.cgi?id=350486&action=review

> LayoutTests/fast/shadow-dom/signal-slot-list-retains-js-wrappers.html:27
> +            const x = new Array(100);

This path is used for in-browser testing. This test also catches the same bug in Chrome sometimes LOL

[EWS Watchlist]

Attachment 350486 [details] did not pass style-queue:


ERROR: Source/WebCore/dom/MutationObserver.cpp:193:  More than one command on the same line  [whitespace/newline] [4]
Total errors found: 1 in 5 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Saam Barati]

Comment on attachment 350486 [details]
Fixes the bug

r=me seems reasonable

[Ryosuke Niwa]

Comment on attachment 350486 [details]
Fixes the bug

Clearing flags on attachment: 350486

Committed r236440: <https://trac.webkit.org/changeset/236440>

[Ryosuke Niwa]

All reviewed patches have been landed.  Closing bug.

[Ryosuke Niwa]

Removed the failing test expectation in https://trac.webkit.org/r236710 now that I observe the test isn't failing on the flakiness dashboard.