There is just no excuse for making C function calls for a branch, since the inline code is probably about the same size as a callsite.
Created attachment 253890 [details] the patch
Comment on attachment 253890 [details] the patch View in context: https://bugs.webkit.org/attachment.cgi?id=253890&action=review r=me Very nice, I've definitely seen operationConvertJSValueToBoolean sitting around ~2% in various profiles, including Speedometer. > LayoutTests/ChangeLog:11 > + * js/regress/cse-new-array-buffer-expected.txt: Added. > + * js/regress/cse-new-array-buffer.html: Added. > + * js/regress/cse-new-array-expected.txt: Added. > + * js/regress/cse-new-array.html: Added. These files don't belong in this patch.
(In reply to comment #2) > Comment on attachment 253890 [details] > the patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=253890&action=review > > r=me > > Very nice, I've definitely seen operationConvertJSValueToBoolean sitting > around ~2% in various profiles, including Speedometer. > > > LayoutTests/ChangeLog:11 > > + * js/regress/cse-new-array-buffer-expected.txt: Added. > > + * js/regress/cse-new-array-buffer.html: Added. > > + * js/regress/cse-new-array-expected.txt: Added. > > + * js/regress/cse-new-array.html: Added. > > These files don't belong in this patch. That's true. Do you really want me to land these separately?
Landed in http://trac.webkit.org/changeset/185002