We can use references for ExecState and thisObject.
Created attachment 291042 [details] Patch
Comment on attachment 291042 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=291042&action=review Seems OK for now; makes things better in some ways, but worse in others with those extra conditions and variables. > Source/WebCore/bindings/scripts/CodeGeneratorJS.pm:3903 > + my $useReferences = shift; This boolean is really ugly, all the mysterious "1", all the extra variables and cases. Is there some way to avoid this? > Source/WebCore/bindings/scripts/CodeGeneratorJS.pm:4938 > + my ($signature, $inFunctionCall, $interface, $value, $thisValue, $useReferences) = @_; This boolean is really ugly, all the mysterious "1", all the extra variables and cases. Is there some way to avoid this? > Source/WebCore/bindings/scripts/CodeGeneratorJS.pm:4945 > + # FIXNE: We should move to using references only FIXME misspelled here
Thanks for the review. > > Source/WebCore/bindings/scripts/CodeGeneratorJS.pm:3903 > > + my $useReferences = shift; > > This boolean is really ugly, all the mysterious "1", all the extra variables > and cases. Is there some way to avoid this? This is temporary, since it should use references once attribute setters and methods are also moved to using references. That said, in the meantime, it might be best to introduce a NativeToJSValueWithReferences and NativeToJSValueWithPointers, that would call a sub NativeToJSValue routine. > > Source/WebCore/bindings/scripts/CodeGeneratorJS.pm:4945 > > + # FIXNE: We should move to using references only > > FIXME misspelled here OK
Created attachment 291077 [details] Patch for landing
Created attachment 291078 [details] Patch for landing
Comment on attachment 291078 [details] Patch for landing Clearing flags on attachment: 291078 Committed r206992: <http://trac.webkit.org/changeset/206992>
All reviewed patches have been landed. Closing bug.