Refactor sync notifier out of sync api.

chrome/browser/sync/notifier/sync_notifier_callback.h

+// Copyright (c) 2011 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.
+
+
+#ifndef CHROME_BROWSER_SYNC_NOTIFIER_SYNC_NOTIFIER_CALLBACK_H_
+#define CHROME_BROWSER_SYNC_NOTIFIER_SYNC_NOTIFIER_CALLBACK_H_
+
+#include "jingle/notifier/listener/notification_defines.h"
+
+namespace sync_notifier {
+
+struct NotificationData {
+  std::string service_url;
+  std::string service_specific_data;
+};
+
+class SyncNotifierCallback {

[akalin, 2011/03/08 02:48:30]

I think it would be better to rename this SyncNotifierObserver.  In fact, the
usual pattern would be to make this an inner class of SyncNotifier named
Observer, but having it outside is fine, too.

[Agrawal, 2011/03/09 01:57:04]

Done.

+ public:
+  SyncNotifierCallback() {}
+  virtual ~SyncNotifierCallback() {}
+
+  virtual void OnIncomingNotification(
+      const IncomingNotificationData& notification_data) = 0;

[akalin, 2011/03/08 02:48:30]

I think this should take a browser_sync::sessions::TypePayloadMap instead; the
details of marshaling the data from the string should be an implementation
detail.

[Agrawal, 2011/03/09 01:57:04]

The data parsing also needs ModelSafeWorkerRegistrar for p2p notifications. p2p
notifications logic looks at the service_url and then it just generates the map
for ALL enabled datatypes. If service_url check is not important we can just
override the type_map in syncapi.cc (for p2p notifications).

+  virtual void OnNotificationStateChange(bool notifications_enabled) = 0;
+  virtual void OnOutgoingNotification() = 0;

[akalin, 2011/03/08 02:48:30]

No need for an outgoing notification.

[Agrawal, 2011/03/09 01:57:04]

SyncInternal::OnOutgoingNotification increments some stats on this. Is this no
longer used?

+  virtual void StoreCookie(const std::string& cookie) = 0;

[akalin, 2011/03/08 02:48:30]

I don't think this should be part of the observer interface; a separate object
should be passed to SyncNotifier.

[Agrawal, 2011/03/09 01:57:04]

This is not clear to me.
The state/cookie is currently just a string in the notifier interface, and an
implementation may decide to not use it at all. What are the benefits for
removing this from here?

+};
+}  // namespace sync_notifier
+#endif  // CHROME_BROWSER_SYNC_NOTIFIER_SYNC_NOTIFIER_CALLBACK_H_