Vector ICs: platform support for vector-based stores.

src/ast.cc

Index: src/ast.cc
diff --git a/src/ast.cc b/src/ast.cc
index c61c29bf524ba300573958b13b96b1921e3fc1cd..19747d8b767be294e3a819db95f7beb14a3c68b2 100644
--- a/src/ast.cc
+++ b/src/ast.cc
@@ -343,12 +343,14 @@ FeedbackVectorRequirements ObjectLiteral::ComputeFeedbackRequirements(
   // This logic that computes the number of slots needed for vector store
   // ics must mirror FullCodeGenerator::VisitObjectLiteral.
   int ic_slots = 0;
+  bool saw_computed_name = false;
   for (int i = 0; i < properties()->length(); i++) {
     ObjectLiteral::Property* property = properties()->at(i);
     if (property->IsCompileTimeValue()) continue;
+    saw_computed_name |= property->is_computed_name();
 
     Expression* value = property->value();
-    if (property->is_computed_name() &&
+    if (saw_computed_name &&

[Jakob Kummerow, 2015/09/03 11:27:39]

This is surprising: having seen a computed name previously affects how the
current property is handled? Are you sure this is correct? (I bet we lack test
coverage for this; NeedsHomeObject() is fairly arcane.)

[mvstanton, 2015/09/03 15:25:27]

You are totally right. It is true though. As we discussed offline, I will be
refactoring this so that each ObjectLiteral::Property knows how many slots it
needs, and the code will be less fragile (also, it'll be tractable from the
point of view of the oracle).

         property->kind() != ObjectLiteral::Property::PROTOTYPE) {
       if (FunctionLiteral::NeedsHomeObject(value)) ic_slots++;
     } else if (property->emit_store()) {