Add a ScopedNSAutoreleasePool to TabModelTest.PersistSelectionChange

ios/chrome/browser/tabs/tab_model_unittest.mm

 // Copyright 2012 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 <objc/runtime.h>
 
 #include "base/files/file_path.h"
 #include "base/mac/scoped_nsautorelease_pool.h"
 #include "base/memory/ptr_util.h"
 #include "base/run_loop.h"
@@ skipping 813 matching lines @@
   EXPECT_TRUE([tab parentTabModel] == nil);
   [tab_model_ insertTab:tab atIndex:0];
   [tab setParentTabModel:tab_model_.get()];
   EXPECT_FALSE([tab parentTabModel] == nil);
   [tab_model_ closeTabAtIndex:0];
 }
 
 TEST_F(TabModelTest, PersistSelectionChange) {
   TestChromeBrowserState::Builder test_cbs_builder;
   auto chrome_browser_state = test_cbs_builder.Build();
+  base::mac::ScopedNSAutoreleasePool pool;

[rohitrao (ping after 24h), 2017/02/06 17:31:46]

Can you please add a comment explaining why this pool is necessary?

[sdefresne, 2017/02/06 17:46:43]

Done.

 
   NSString* stashPath =
       base::SysUTF8ToNSString(chrome_browser_state->GetStatePath().value());
 
   base::scoped_nsobject<TabModel> model([[TabModel alloc]
       initWithSessionWindow:session_window_.get()
              sessionService:[SessionServiceIOS sharedService]
                browserState:chrome_browser_state.get()]);
 
   [model addTabWithURL:kURL referrer:kReferrer windowName:@"window 1"];
@@ skipping 17 matching lines @@
   // Create tab model from saved session.
   base::scoped_nsobject<TestSessionService> test_service(
       [[TestSessionService alloc] init]);
 
   model.reset([[TabModel alloc]
       initWithSessionWindow:sessionWindow
              sessionService:test_service
                browserState:chrome_browser_state.get()]);
   EXPECT_EQ(model.get().currentTab, [model tabAtIndex:1]);
   [model browserStateDestroyed];
+  model.reset();
+  base::RunLoop().RunUntilIdle();

[rohitrao (ping after 24h), 2017/02/06 17:31:46]

Why is this necessary?  Does this drain the pool?  Could you just delete the
pool here instead?

[sdefresne, 2017/02/06 17:46:43]

This is unrelated to draining the pool.

It is there because restoring a session sends a task to the UI thread with a
callback on the UI thread (see https://codereview.chromium.org/2674293002/ where
I make this callback thread-safe). So this fixes another possible source of
flaky-ness.

I've put a comment why this is there.

 
   // Clean up.
   EXPECT_TRUE([[NSFileManager defaultManager] removeItemAtPath:stashPath
                                                          error:nullptr]);
 }
 
 }  // anonymous namespace