Remove tracing's nesting logic since noone is using it.

tools/telemetry/telemetry/core/backends/chrome/tracing_backend.py

 # Copyright 2013 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 logging
 
 from telemetry.core.backends.chrome import inspector_websocket
 from telemetry.core.platform import tracing_category_filter
 from telemetry.core.platform import tracing_options
 
 
 class TracingUnsupportedException(Exception):
   pass
 
 
 class TracingTimeoutException(Exception):
   pass
 
 
 class TracingBackend(object):
   def __init__(self, devtools_port, chrome_browser_backend):
     self._inspector_websocket = inspector_websocket.InspectorWebsocket(
         self._NotificationHandler,
         self._ErrorHandler)
 
     self._inspector_websocket.Connect(
         'ws://127.0.0.1:%i/devtools/browser' % devtools_port)
+    self._tracing_data = []
     self._category_filter = None
-    self._nesting = 0
-    self._tracing_data = []
     self._is_tracing_running = False
     self._chrome_browser_backend = chrome_browser_backend
 
   @property
   def is_tracing_running(self):
     return self._is_tracing_running
 
   def StartTracing(self, trace_options, custom_categories=None, timeout=10):
-    """ Starts tracing on the first nested call and returns True. Returns False
+    """ Starts tracing on the first call and returns True. Returns False
         and does nothing on subsequent nested calls.
     """
-    self._nesting += 1
     if self.is_tracing_running:

[dtu, 2014/09/15 22:06:33]

Maybe we don't need this check and/or this variable?
If we remove this we can also remove _category_filter?

[nednguyen, 2014/09/15 22:09:47]

No, if we allow StartTracing to be called multiple times. Or else, this need to
raise Exception to let caller knows that tracing is already running.

[chrishenry, 2014/09/23 17:49:26]

Should we throw exception? The return value from StopTracing is a bit weird if
it's nested.

[nednguyen, 2014/10/03 17:13:47]

done.

[nednguyen, 2014/10/03 17:13:47]

tracing_controller_backend return False in case it's started. Let's make it
consistent.

       new_category_filter = tracing_category_filter.TracingCategoryFilter(
           filter_string=custom_categories)
       is_subset = new_category_filter.IsSubset(self._category_filter)
       assert(is_subset != False)
       if is_subset == None:
         logging.warning('Cannot determine if category filter of nested ' +
                         'StartTracing call is subset of current filter.')
       return False
     self._CheckNotificationSupported()
     #TODO(nednguyen): remove this when the stable branch pass 2118.
@@ skipping 15 matching lines @@
     req['params']['options'] = m[trace_options.record_mode]
     self._category_filter = tracing_category_filter.TracingCategoryFilter(
         filter_string=custom_categories)
     if custom_categories:
       req['params']['categories'] = custom_categories
     self._inspector_websocket.SyncRequest(req, timeout)
     self._is_tracing_running = True
     return True
 
   def StopTracing(self, timeout=30):
-    """ Stops tracing on the innermost (!) nested call, because we cannot get
-        results otherwise. Resets _tracing_data on the outermost nested call.
-        Returns the result of the trace, as TracingTimelineData object.
+    """ Stops tracing and returns the raw json trace result. It this is called
+    after tracing has been stopped, empty trace data is returned.
     """
-    self._nesting -= 1
-    assert self._nesting >= 0
     if self.is_tracing_running:
       req = {'method': 'Tracing.end'}
       self._inspector_websocket.SendAndIgnoreResponse(req)
       # After Tracing.end, chrome browser will send asynchronous notifications
       # containing trace data. This is until Tracing.tracingComplete is sent,
       # which means there is no trace buffers pending flush.
       try:
         self._inspector_websocket.DispatchNotificationsUntilDone(timeout)
       except \
           inspector_websocket.DispatchNotificationsUntilDoneTimeoutException \
           as err:
         raise TracingTimeoutException(
             'Trace data was not fully received due to timeout after %s '
             'seconds. If the trace data is big, you may want to increase the '
             'time out amount.' % err.elapsed_time)
       self._is_tracing_running = False
-    if self._nesting == 0:
-      self._category_filter = None
-      return self._GetTraceResultAndReset()
-    else:
-      return self._GetTraceResult()
-
-  def _GetTraceResult(self):
-    assert not self.is_tracing_running
-    return self._tracing_data
+    return self._GetTraceResultAndReset()
 
   def _GetTraceResultAndReset(self):
-    result = self._GetTraceResult()
-
+    result = self._tracing_data
     self._tracing_data = []
     return result
 
   def _ErrorHandler(self, elapsed):
     logging.error('Unrecoverable error after %ds reading tracing response.',
                   elapsed)
     raise
 
   def _NotificationHandler(self, res):
     if 'Tracing.dataCollected' == res.get('method'):
@@ skipping 12 matching lines @@
 
   def _CheckNotificationSupported(self):
     """Ensures we're running against a compatible version of chrome."""
     req = {'method': 'Tracing.hasCompleted'}
     res = self._inspector_websocket.SyncRequest(req)
     if res.get('response'):
       raise TracingUnsupportedException(
           'Tracing not supported for this browser')
     elif 'error' in res:
       return