Patch forthcoming.
Created attachment 271391 [details] the patch
Comment on attachment 271391 [details] the patch Do you have perf numbers?
Comment on attachment 271391 [details] the patch r=me if perf numbers are good.
*** Bug 153069 has been marked as a duplicate of this bug. ***
(In reply to comment #3) > Comment on attachment 271391 [details] > the patch > > r=me if perf numbers are good. Thanks! I'm going to run them now.
<rdar://problem/24192536>
Created attachment 271442 [details] initial performance report
Because string-from-char-code.js ran for much shorter than I wanted, I made it run 10x longer, and reran just those benchmarks. Results: TipOfTree FromCharCode string-from-char-code 16.8306+-0.6567 ^ 6.7301+-0.4895 ^ definitely 2.5008x faster untyped-string-from-char-code 17.6565+-0.5887 ^ 10.9026+-2.4457 ^ definitely 1.6195x faster <geometric> 17.2385+-0.6145 ^ 8.5458+-0.8659 ^ definitely 2.0172x faster It's pretty cool that even in the untyped case, the FTL provides a significant win.
Landed in http://trac.webkit.org/changeset/196642