Change preference APIs to take std::string instead of const char*.

base/values.cc

Index: base/values.cc
diff --git a/base/values.cc b/base/values.cc
index 5d45ec36c5551b6e72083d1f5f5736d69dd5bebc..d60b6484586a829357eb9af3302f99e21e42491e 100644
--- a/base/values.cc
+++ b/base/values.cc
@@ -421,12 +421,13 @@ void DictionaryValue::SetWithoutPathExpansion(const std::string& key,
                                               Value* in_value) {
   // If there's an existing value here, we need to delete it, because
   // we own all our children.
-  std::pair<ValueMap::iterator, bool> ins_res =
-      dictionary_.insert(std::make_pair(key, in_value));

[Nico, 2014/12/01 16:57:02]

Isn't the lhs better here? It requires on-stack allocation of a pair (which,
since it's a string pair, needs to go to the heap), but it saves a second
traversal of the tree to find the right insertion position.

(I guess it doesn't matter too much either way, unless there's a benchmark
saying otherwise.)

[Georges Khalil, 2014/12/01 20:34:11]

I'm removing this code from the CL, because it doesn't fit with the description
anymore.

I'll have some bench-marking done and resubmit this as a new CL.

-  if (!ins_res.second) {
-    DCHECK_NE(ins_res.first->second, in_value);  // This would be bogus
-    delete ins_res.first->second;
-    ins_res.first->second = in_value;
+  ValueMap::iterator it = dictionary_.find(key);
+  if (it == dictionary_.end()) {
+    dictionary_.insert(std::make_pair(key, in_value));
+  } else {
+    DCHECK_NE(it->second, in_value);  // This would be bogus
+    delete it->second;
+    it->second = in_value;
   }
 }