Ensure CopyElementsImpl is always executed so it fills in holes even if from_size is 0. Allow Fixed…

src/objects-inl.h

Index: src/objects-inl.h
diff --git a/src/objects-inl.h b/src/objects-inl.h
index 7db9175d41d6604d0f1f12ee2b92753fb412a618..e23adf2c2452b137361ea9078fda18977e3ebe93 100644
--- a/src/objects-inl.h
+++ b/src/objects-inl.h
@@ -2342,8 +2342,14 @@ void SeededNumberDictionary::set_requires_slow_elements() {
 // Cast operations
 
 
+FixedDoubleArray* FixedDoubleArray::cast(Object* object) {

[Michael Starzinger (Google), 2012/11/19 12:55:46]

As discussed offline, I don't like this new casting function. It is lying about
the return type. Can we instead have a second
FixedDoubleArray::castButAllowEmptyFixedArray() that does this? That one would
still be lying about the return type, but at least the function name is scary
enough to let people know what's going on.

BTW, I am flexible about the name as long as it sounds complicated and scary. :)

+  ASSERT((object->IsFixedArray() && FixedArray::cast(object)->length() == 0) ||
+         object->IsFixedDoubleArray());
+  return reinterpret_cast<FixedDoubleArray*>(object);
+}
+
+
 CAST_ACCESSOR(FixedArray)
-CAST_ACCESSOR(FixedDoubleArray)
 CAST_ACCESSOR(DescriptorArray)
 CAST_ACCESSOR(DeoptimizationInputData)
 CAST_ACCESSOR(DeoptimizationOutputData)