LayoutTest tiled-drawing/scrolling/latched-div-with-scroll-snap.html is flaky This test is frequently failing or timing out on mac-wk2. It has been flaky for a while, but as of 5/28/16 it fails much more frequently. Most recent failure: <https://build.webkit.org/builders/Apple%20El%20Capitan%20Debug%20WK2%20(Tests)/builds/5581> <https://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=tiled-drawing%2Fscrolling%2Flatched-div-with-scroll-snap.html> --- /Volumes/Data/slave/elcapitan-debug-tests-wk2/build/layout-test-results/tiled-drawing/scrolling/latched-div-with-scroll-snap-expected.txt +++ /Volumes/Data/slave/elcapitan-debug-tests-wk2/build/layout-test-results/tiled-drawing/scrolling/latched-div-with-scroll-snap-actual.txt @@ -3,7 +3,7 @@ PASS div honored snap points. Testing that latch moves to bottom div: PASS top div did not move. -PASS div scrolled to next window. +FAIL div did not honor snap points. Expected 200, but got 50 PASS successfullyParsed is true TEST COMPLETE
El Capitan only, seems to fail 100% of the time https://build.webkit.org/builders/Apple%20El%20Capitan%2032-bit%20JSC%20(BuildAndTest)?numbuilds=50
(In reply to comment #1) > El Capitan only, seems to fail 100% of the time > > https://build.webkit.org/builders/Apple%20El%20Capitan%2032- > bit%20JSC%20(BuildAndTest)?numbuilds=50 This comment was intended for a different bug. Please disregard. However, tiled-drawing/scrolling/latched-div-with-scroll-snap.html is still frequently failing or timing out on Mac, especially on El Capitan Debug WK2
(In reply to comment #2) > (In reply to comment #1) > > El Capitan only, seems to fail 100% of the time > > > > https://build.webkit.org/builders/Apple%20El%20Capitan%2032- > > bit%20JSC%20(BuildAndTest)?numbuilds=50 > > This comment was intended for a different bug. Please disregard. > > However, tiled-drawing/scrolling/latched-div-with-scroll-snap.html is still > frequently failing or timing out on Mac, especially on El Capitan Debug WK2 We have general flakiness in a lot of these scroll-snap/latching tests. Much of this is due to generating 'fake' wheel gestures to drive the tests. Maybe some partial event state is getting held over from test-to-test and causing things to behave in weird ways. Simon and I spent a lot of time trying to get this more stable, but clearly more effort is needed.
> It has been flaky for a while, but as of 5/28/16 it fails much more frequently. Is there any way to more precisely isolate when this started? There were sweeping changes to callOnMainThread and NoncopyableFunction on 5/28 and 5/27, and while I don't see anything wrong with those patches, a mistake there could easily cause such symptoms.
It started early on 5/28, roughly around 1am. <http://trac.webkit.org/projects/webkit/changeset/201482> seems to fit.
Or looking at revision numbers and not at times, <http://trac.webkit.org/projects/webkit/changeset/201476>.
Marked as flaky in <http://trac.webkit.org/r202269>. Zalan, do you think that you could look into this?
Created attachment 281771 [details] Patch
Comment on attachment 281771 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=281771&action=review r=me pending EWS > LayoutTests/tiled-drawing/scrolling/latched-div-with-scroll-snap.html:30 > - <script src="../../resources/js-test.js"></script> > + <script src="../../resources/js-test-pre.js"></script> Why this change? I think that js-test.js is slightly preferable in general.
Comment on attachment 281771 [details] Patch Please remove the test from TestExpectations, now that it's expected to pass reliably.
Created attachment 281787 [details] Patch
Comment on attachment 281787 [details] Patch Clearing flags on attachment: 281787 Committed r202304: <http://trac.webkit.org/changeset/202304>
All reviewed patches have been landed. Closing bug.