Move ListImplementation from coreimpl to core, as a private member.

runtime/vm/intrinsifier_ia32.cc

Index: runtime/vm/intrinsifier_ia32.cc
diff --git a/runtime/vm/intrinsifier_ia32.cc b/runtime/vm/intrinsifier_ia32.cc
index afdfff734709cfe9111f03ddd6a427edbf3aca36..cf47309dd930476c722d1b9015697e29352cfd68 100644
--- a/runtime/vm/intrinsifier_ia32.cc
+++ b/runtime/vm/intrinsifier_ia32.cc
@@ -175,9 +175,9 @@ bool Intrinsifier::ImmutableArray_getIndexed(Assembler* assembler) {
 
 static intptr_t ComputeObjectArrayTypeArgumentsOffset() {
   const String& class_name = String::Handle(Symbols::New("_ObjectArray"));
-  const Library& coreimpl_lib = Library::Handle(Library::CoreImplLibrary());
+  const Library& core_lib = Library::Handle(Library::CoreLibrary());

[Mads Ager (google), 2012/10/30 10:37:47]

Are these changes only needed on ia32? Aren't these intrinsified on x64?

[Anders Johnsen, 2012/10/30 11:53:38]

Well spotted, found a case in x64! Passes all tests on both platforms now.

   const Class& cls =
-      Class::Handle(coreimpl_lib.LookupClassAllowPrivate(class_name));
+      Class::Handle(core_lib.LookupClassAllowPrivate(class_name));
   ASSERT(!cls.IsNull());
   ASSERT(cls.HasTypeArguments());
   ASSERT(cls.NumTypeArguments() == 1);
@@ -252,9 +252,9 @@ static intptr_t GetOffsetForField(const char* class_name_p,
                                   const char* field_name_p) {
   const String& class_name = String::Handle(Symbols::New(class_name_p));
   const String& field_name = String::Handle(Symbols::New(field_name_p));
-  const Library& coreimpl_lib = Library::Handle(Library::CoreImplLibrary());
+  const Library& core_lib = Library::Handle(Library::CoreLibrary());
   const Class& cls =
-      Class::Handle(coreimpl_lib.LookupClassAllowPrivate(class_name));
+      Class::Handle(core_lib.LookupClassAllowPrivate(class_name));
   ASSERT(!cls.IsNull());
   const Field& field = Field::ZoneHandle(cls.LookupInstanceField(field_name));
   ASSERT(!field.IsNull());