如何在python 3.4中调用方法并使其在后台运行？

Question

I have implemented the Google Cloud Messaging server in python and I want that method to be Asynchronous. I do not expect any return values from that method. Is there a simple way to do this? I have tried using async from asyncio package:

...
loop = asyncio.get_event_loop()
 if(module_status=="Fail"):
      loop.run_until_complete(sendNotification(module_name, module_status))
... 

and here is my method sendNotification() :

async def sendNotification(module_name, module_status):
    gcm = GCM("API_Key")
    data ={"message":module_status, "moduleName":module_name}
    reg_ids = ["device_tokens"]
    response = gcm.json_request(registration_ids=reg_ids, data=data)
    print("GCM notification sent!")

Answer 1

You could use a ThreadPoolExecutor :

from concurrent.futures import ThreadPoolExecutor

def send_notification(module_name, module_status):
    [...]

with ThreadPoolExecutor() as executor:
    future = executor.submit(send_notification, module_name, module_status)

Answer 2

Since GCM is not async library compatible need to use an external event loop.

There are a few, simplest one IMO is probably gevent .

Note that gevent monkey patching may introduce dead locks if the underlying libraries used rely on blocking behaviour to operate.

import gevent
from gevent.greenlet import Greenlet
from gevent import monkey
monkey.patch_all()

def sendNotification(module_name, module_status):
    gcm = GCM("API_Key")
    data ={"message":module_status, "moduleName":module_name}
    reg_ids = ["device_tokens"]
    response = gcm.json_request(registration_ids=reg_ids, data=data)
    print("GCM notification sent!")

greenlet = Greenlet.spawn(sendNotification, 
                          args=(module_name, module_status,))
# Yield control to gevent's event loop without blocking
# to allow background tasks to run
gevent.sleep(0)
# 
# Other code, other greenlets etc here
# 
# Call get to get return value if needed
greenlet.get()

Answer 3

You can use asyncio's api: loop.run_in_executor(None, callable)

This will run the code using an executor (by default a ThreadPoolExecutor )

See the documentation