Currently we only support the for-in construct in the LLInt and baseline JIT. We should add support for for-in to the DFG and FTL.
Created attachment 235649 [details] Patch
(In reply to comment #1) > Created an attachment (id=235649) [details] > Patch Initial cut. Still running perf tests.
(In reply to comment #2) > (In reply to comment #1) > > Created an attachment (id=235649) [details] [details] > > Patch > > Initial cut. Still running perf tests. Still a few bugs lurking.
Created attachment 235653 [details] Patch
Created attachment 235655 [details] Patch
Created attachment 235696 [details] benchmark results
Comment on attachment 235655 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=235655&action=review > Source/JavaScriptCore/ftl/FTLAbstractHeapRepository.h:79 > + macro(Structure_structureID, Structure::structureIDOffset()) \ > + macro(JSPropertyNameEnumerator_cachedStructureID, JSPropertyNameEnumerator::cachedStructureIDOffset()) \ > + macro(JSPropertyNameEnumerator_cachedInlineCapacity, JSPropertyNameEnumerator::cachedInlineCapacityOffset()) \ > + macro(JSPropertyNameEnumerator_cachedPropertyNamesLength, JSPropertyNameEnumerator::cachedPropertyNamesLengthOffset()) \ > + macro(JSPropertyNameEnumerator_cachedPropertyNamesVector, JSPropertyNameEnumerator::cachedPropertyNamesVectorOffset()) Sort in alphabetical order.
Landed in http://trac.webkit.org/changeset/171763