Simplify should_dispatch_immediately

ipykernel/inprocess/client.py

@@ -171,7 +171,10 @@ def _dispatch_to_kernel(self, msg):
             raise RuntimeError('Cannot send request. No kernel exists.')
 
         stream = kernel.shell_stream
-        kernel.dispatch_shell(msg)
+        self.session.send(stream, msg)
+        msg_parts = stream.recv_multipart()
+        kernel.dispatch_shell(msg_parts)
+
         idents, reply_msg = self.session.recv(stream, copy=False)
         self.shell_channel.call_handlers_later(reply_msg)
 

ipykernel/ipkernel.py

@@ -556,7 +556,7 @@ def do_clear(self):
         self.shell.reset(False)
         return dict(status='ok')
 
-    def should_dispatch_immediately(self, msg, *args):
+    def should_dispatch_immediately(self, msg):
         try:
             msg_type = msg['header']['msg_type']
             if msg_type in self.comm_msg_types:

ipykernel/kernelbase.py

@@ -399,9 +399,7 @@ def dispatch_queue(self):
     def _message_counter_default(self):
         return itertools.count()
 
-    def should_dispatch_immediately(
-        self, msg, idents, dispatch
-    ):
+    def should_dispatch_immediately(self, msg):
         """
         This provides a hook for dispatching incoming messages
         from the frontend immediately, and out of order.
@@ -423,9 +421,7 @@ def schedule_dispatch(self, msg, dispatch):
 
         new_args = (msg, idents)
 
-        if self.should_dispatch_immediately(
-            msg, idents, dispatch
-        ):
+        if self.should_dispatch_immediately(msg):
             return self.io_loop.add_callback(dispatch, *new_args)
 
         idx = next(self._message_counter)