Create network change notifier.

webrtc/base/networkmonitor.cc

Index: webrtc/base/networkmonitor.cc
diff --git a/webrtc/base/networkmonitor.cc b/webrtc/base/networkmonitor.cc
new file mode 100644
index 0000000000000000000000000000000000000000..6bdaf8656995b38dc2157976c170c881913a4367
--- /dev/null
+++ b/webrtc/base/networkmonitor.cc
@@ -0,0 +1,47 @@
+/*
+ *  Copyright 2015 The WebRTC Project Authors. All rights reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "webrtc/base/networkmonitor.h"
+
+#include "webrtc/base/common.h"
+
+namespace {
+const uint32_t UPDATE_NETWORKS_MESSAGE = 1;
+
+// A global network monitor factory injected by the peer connection client.

[pthatcher1, 2015/10/15 08:11:43]

I don't think you need to say "by the peer connection client" here.  Code not
related to PeerConnection, such as ORTC, might use this.

In fact, I think you could replace the whole line with "This is set by
NetworkMonitor::SetFactory"

[honghaiz3, 2015/10/15 19:02:41]

Done.

+// This can be NULL if the client has not injected any factory. In that case,
+// no network monitor will be created. If a client injected a factory, it is

[pthatcher1, 2015/10/15 08:11:43]

Isn't it up to callers to decided what to do if there is no factory set?  Why
even comment on that here in this file?  I think you could remove these whole
comment with "The caller of NetworkMonitor::SetFactory must is responsible for
calling SetFactory(nullptr) when the factory is destroyed".

[honghaiz3, 2015/10/15 19:02:41]

Done.

+// up to the client to release the resource by setting it to nullptr.
+rtc::NetworkMonitorFactory* network_monitor_factory = nullptr;
+}  // namespace
+
+namespace rtc {
+void NetworkMonitor::OnNetworksChanged() {
+  LOG(LS_VERBOSE) << "Network change is received at the network monitor";
+  thread_->Post(this, UPDATE_NETWORKS_MESSAGE);
+}
+
+void NetworkMonitor::OnMessage(Message* msg) {
+  ASSERT(msg->message_id == UPDATE_NETWORKS_MESSAGE);
+  SignalNetworksChanged();
+}
+
+void NetworkMonitorFactory::SetFactory(NetworkMonitorFactory* factory) {
+  if (network_monitor_factory != nullptr) {
+    delete network_monitor_factory;
+  }
+  network_monitor_factory = factory;

[pthatcher1, 2015/10/15 08:11:43]

What's the thread safety on this?  Do we have to say it's thread unsafe?

[honghaiz3, 2015/10/15 19:02:41]

Added comments. 

+}
+
+NetworkMonitorFactory* NetworkMonitorFactory::GetFactory() {
+  return network_monitor_factory;
+}
+
+}  // namespace rtc