The following layout test started failing with significantly increased frequency after the changes landed in r251262: scrollingcoordinator/ios/non-stable-viewport-scroll.html Steps to Reproduce: run-webkit-tests --ios-simulator --no-retry --no-sample-on-timeout scrollingcoordinator/ios/non-stable-viewport-scroll.html --iter 100 --force Test History: https://results.webkit.org/?suite=layout-tests&test=scrollingcoordinator%2Fios%2Fnon-stable-viewport-scroll.html Diff: --- /Volumes/Data/slave/ios-simulator-13-release-tests-wk2/build/layout-test-results/scrollingcoordinator/ios/non-stable-viewport-scroll-expected.txt +++ /Volumes/Data/slave/ios-simulator-13-release-tests-wk2/build/layout-test-results/scrollingcoordinator/ios/non-stable-viewport-scroll-actual.txt @@ -2,30 +2,29 @@ (GraphicsLayer (anchor 0.00 0.00) (bounds 320.00 5013.00) - (visible rect 0.00, 1000.00 106.67 x 182.67) - (coverage rect 0.00, 1000.00 106.67 x 182.67) + (visible rect 0.00, 817.50 320.00 x 548.00) + (coverage rect 0.00, 817.50 320.00 x 548.00) (intersects coverage rect 1) - (contentsScale 6.00) + (contentsScale 2.00) (children 1 (GraphicsLayer (bounds 320.00 5013.00) (contentsOpaque 1) - (visible rect 0.00, 1000.00 106.67 x 182.67) - (coverage rect 0.00, 957.33 192.00 x 268.00) + (visible rect 0.00, 817.50 320.00 x 548.00) + (coverage rect 0.00, 689.50 576.00 x 804.00) (intersects coverage rect 1) - (contentsScale 6.00) + (contentsScale 2.00) (children 1 (GraphicsLayer (position 10.00 0.00) - (approximate position 10.00 634.66) + (approximate position 10.00 817.50) (bounds 100.00 548.00) - (usingTiledLayer 1) (contentsOpaque 1) (drawsContent 1) - (visible rect 0.00, 365.34 96.67 x 182.66) - (coverage rect -10.00, 322.68 192.00 x 268.00) + (visible rect 0.00, 0.00 100.00 x 548.00) + (coverage rect -10.00, -128.00 576.00 x 804.00) (intersects coverage rect 1) - (contentsScale 6.00) + (contentsScale 2.00) ) ) )
<rdar://problem/56410921>
Marked test as flaky in r251288
This is surprising given than r251262 should have only affected macOS.
I think that r251261 is the more likely culprit.
Not happening any more now that r251261 was reverted.