https://trac.webkit.org/changeset/235917 introduced ~27% regression in nth-child(~) tests and 2% regression for the total Dromaeo css selector tests.
<rdar://problem/44595527>
Created attachment 350099 [details] Fixes the bug
Comment on attachment 350099 [details] Fixes the bug r=me with one more fix. I think generateNthLastChildParentCheckAndRelationUpdate has the same issue too.
(In reply to Yusuke Suzuki from comment #3) > Comment on attachment 350099 [details] > Fixes the bug > > r=me with one more fix. > I think generateNthLastChildParentCheckAndRelationUpdate has the same issue > too. Good catch! Thanks for the review.
Created attachment 350143 [details] Patch for landing
Comment on attachment 350143 [details] Patch for landing Wait for EWS.
Committed r236228: <https://trac.webkit.org/changeset/236228>
Nice, it seems the regression is recovered. https://perf.webkit.org/v3/#/charts?since=1536816937999&paneList=((19-31)-(19-39)-(19-36))