[Sync] Move sync/{internal_api,syncable} into syncer namespace

sync/sessions/ordered_commit_set_unittest.cc

Index: sync/sessions/ordered_commit_set_unittest.cc
diff --git a/sync/sessions/ordered_commit_set_unittest.cc b/sync/sessions/ordered_commit_set_unittest.cc
index fcdc891d0c949da8654513cd3eeb4a4336353f1f..8bf420a30c4190c16694bdd17669fbd8ba3966ab 100644
--- a/sync/sessions/ordered_commit_set_unittest.cc
+++ b/sync/sessions/ordered_commit_set_unittest.cc
@@ -8,6 +8,10 @@
 
 using std::vector;
 
+namespace syncer {
+namespace sessions {
+namespace {

[rlarocque, 2012/07/02 18:42:02]

This (anonymous namespace) looks like a good idea.  Why don't we do this for all
tests?

(You don't have to do it now, I'm just wondering if we should aim to put all
tests in anonymous namespaces eventually.)

[akalin, 2012/07/02 18:44:53]

We should.  ODR violations in test helper class names are a real possibility
(e.g., I believe it's happened multiple times that a class named 'FakeFoo' or
something was defined in more than one test file).

I think it's just an awareness issue.  The anon namespace pattern is in the
gmock documentation.  Maybe it should be mentioned in the style guide or
something.

+
 class OrderedCommitSetTest : public testing::Test {
  public:
   OrderedCommitSetTest() {
@@ -21,9 +25,6 @@ class OrderedCommitSetTest : public testing::Test {
   syncer::ModelSafeRoutingInfo routes_;
 };
 
-namespace syncer {
-namespace sessions {
-
 TEST_F(OrderedCommitSetTest, Projections) {
   vector<syncable::Id> expected;
   for (int i = 0; i < 8; i++)
@@ -127,6 +128,6 @@ TEST_F(OrderedCommitSetTest, AddAndRemoveEntries) {
   ASSERT_TRUE(commit_set.Empty());
 }
 
+}  // namespace
 }  // namespace sessions
 }  // namespace syncer
-