Standardize usage of virtual/override/final in sandbox/

sandbox/mac/xpc_message_server_unittest.cc

 // Copyright 2014 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.
 
 #include "sandbox/mac/xpc_message_server.h"
 
 #include <Block.h>
 #include <mach/mach.h>
 #include <servers/bootstrap.h>
 
@@ skipping 25 matching lines @@
 // A MessageDemuxer that manages a test server and executes a block for every
 // message.
 class BlockDemuxer : public MessageDemuxer {
  public:
   BlockDemuxer()
       : demux_block_(NULL),
         server_(this, MACH_PORT_NULL),
         pipe_(NULL) {
   }
 
-  virtual ~BlockDemuxer() {
+  ~BlockDemuxer() override {
     if (pipe_)
       xpc_release(pipe_);
     if (demux_block_)
       Block_release(demux_block_);
   }
 
   // Starts running the server, given a block to handle incoming IPC messages.
   bool Initialize(void (^demux_block)(IPCMessage request)) {
     if (!server_.Initialize())
       return false;
 
     // Create a send right on the port so that the XPC pipe can be created.
     if (mach_port_insert_right(mach_task_self(), server_.GetServerPort(),
             server_.GetServerPort(), MACH_MSG_TYPE_MAKE_SEND) != KERN_SUCCESS) {
       return false;
     }
     scoped_send_right_.reset(server_.GetServerPort());
 
     demux_block_ = Block_copy(demux_block);
     pipe_ = xpc_pipe_create_from_port(server_.GetServerPort(), 0);
 
     return true;
   }
 
-  virtual void DemuxMessage(IPCMessage request) override {
-    demux_block_(request);

[Robert Sesek, 2014/10/21 20:52:40]

nit: please leave on separate line

[dcheng, 2014/10/21 20:53:57]

How strongly do you feel about this? This is how clang-format prefers it, and
I'd prefer not to fight it, especially in giant patches like this.

[Robert Sesek, 2014/10/21 20:58:36]

This is less readable. I feel strongly about not caring about what clang-format
outputs.

[dcheng, 2014/10/21 21:01:11]

I'll revert this particular line. However, note it's generally impractical for
me to do this on a large scale--due to the nature of this patch, reflowing some
lines is almost always going to be required and reverting all instances of this
(as I was asked to do in ipc/) is time consuming.

[Robert Sesek, 2014/10/21 21:09:05]

Thanks. I appreciate that, but trading off readability for mechanized
convenience is not the right choice IMO.

-  }
+  void DemuxMessage(IPCMessage request) override { demux_block_(request); }
 
   xpc_pipe_t pipe() { return pipe_; }
 
   XPCMessageServer* server() { return &server_; }
 
  private:
   void (^demux_block_)(IPCMessage request);
 
   XPCMessageServer server_;
 
@@ skipping 126 matching lines @@
   ASSERT_EQ(0, xpc_pipe_routine(first.pipe(), request, &reply));
 
   EXPECT_EQ(1, xpc_dictionary_get_int64(reply, "seen_by_first"));
   EXPECT_EQ(2, xpc_dictionary_get_int64(reply, "seen_by_second"));
 
   xpc_release(request);
   xpc_release(reply);
 }
 
 }  // namespace sandbox