Updated CRWSessionController interface to use NavigationItems.

ios/web/navigation/navigation_item_impl_list.mm

+// Copyright 2017 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#import "ios/web/navigation/navigation_item_impl_list.h"
+
+#import "ios/web/navigation/navigation_item_impl.h"
+
+namespace web {
+
+ScopedNavigationItemImplList CreateScopedNavigationItemImplList(
+    ScopedNavigationItemList scoped_item_list) {
+  ScopedNavigationItemImplList list;

[Eugene But (OOO till 7-30), 2017/02/03 23:33:33]

Construct with size?

[kkhorimoto, 2017/02/04 02:45:30]

Done.

+  for (size_t index = 0; index < scoped_item_list.size(); ++index) {
+    std::unique_ptr<NavigationItemImpl> scoped_item_impl(
+        static_cast<NavigationItemImpl*>(scoped_item_list[index].release()));
+    list.push_back(std::move(scoped_item_impl));

[Eugene But (OOO till 7-30), 2017/02/03 23:33:33]

emplace_back ?

[kkhorimoto, 2017/02/04 02:45:30]

Judging by C++ documentation, it seems that using std::move() is more
performant.  Using emplace_back() would require calling release() on
|scoped_item_impl| and calling the constructor for the inserted item, whereas
std::move() uses some C++ rvalue reference magic to avoid constructing a new
object.  Additionally, from a self-documenting memory management perspective,
personally I feel that it's easier to think about "moving" a unique_ptr rather
than releasing one unique_ptr's value and creating a new unique_ptr with the
unowned pointer value.

+  }
+  return list;
+}
+
+}  // namespace web