STR: Execute the following code in JSC: 'abcdefghijk'.match('.?e.?') Expected result: def Actual result: abcdefghijk This also happens in SpiderMonkey, so it's almost guaranteed to be a Yarr issue.
For SpiderMonkey, this regressed between two Yarr imports we did: 86639 was good, 130234 is bad.
Created attachment 344084 [details] Patch
The benchmark results are at https://gist.github.com/sukolsak/da11c94998776a4dfb3340f12f7b7688. The patch doesn't seem to cause a significant slowdown. I can think of a way to optimize these cases (/.?abc.*/, /.*abc.?/, /.{0,4}abc.*/, etc), which involves storing the quantifiers of the leading and trailing dots in the DotStarEnclosure. But I'm not sure if we should do it.
Comment on attachment 344084 [details] Patch Attachment 344084 [details] did not pass win-ews (win): Output: https://webkit-queues.webkit.org/results/8410946 New failing tests: http/tests/security/canvas-remote-read-remote-video-redirect.html http/tests/security/video-poster-cross-origin-crash2.html
Created attachment 344094 [details] Archive of layout-test-results from ews205 for win-future The attached test failures were seen while running run-webkit-tests on the win-ews. Bot: ews205 Port: win-future Platform: CYGWIN_NT-6.1-2.9.0-0.318-5-3-x86_64-64bit
Comment on attachment 344084 [details] Patch r=me
Thanks! I ran some additional benchmark tests. (Thanks rniwa and yusukesuzuki for the help!). Here are the results. Without the patch: RexBench 23.00 +- 1.71ms JetStream Score 203.41 +- 37.830 With the patch: RexBench 23.19 +- 3.13ms JetStream Score 200.63 +- 41.824
Created attachment 344156 [details] Patch for landing
Comment on attachment 344156 [details] Patch for landing Clearing flags on attachment: 344156 Committed r233453: <https://trac.webkit.org/changeset/233453>
All reviewed patches have been landed. Closing bug.
<rdar://problem/41750698>