Reconnecting after a disconnect

[Terrance]

I'm attempting to write a long-running app with a Hangouts connection, so in the event of a disconnect I'd like to handle this quietly in the background. Currently I have something like the following:

client = hangups.Client(hangups.auth.get_auth_stdin("refresh_token.txt"))
loop = asyncio.get_event_loop()

def on_connect():
    # Some other initialisation stuff here.
    pass

def on_disconnect():
    try:
        loop.stop()
    except RuntimeError as e:
        # Couldn't stop the event loop?
        pass
    while loop.is_running():
        time.sleep(1)
    loop.run_until_complete(client.connect())

def run():
    client.on_connect.add_observer(on_connect)
    client.on_reconnect.add_observer(on_connect)
    client.on_disconnect.add_observer(on_disconnect)
    try:
        loop.run_until_complete(client.connect())
    except Exception as e:
        # Hangups exception?
        pass
    on_disconnect()

But on the second connect() call inside on_disconnect, I get:

ValueError: <function on_connect at 0x6ffffffffff> is already an observer of Event('Channel.on_connect')

Trying to restart a stopped event loop also gives errors like:

RuntimeError: Event loop stopped before Future completed.

This all feels rather messy -- I haven't used asyncio before so I'm not entirely sure if there's a better way to work with the event loop. Is there a correct way to handle reconnects?

[tdryer]

Client currently doesn't support connect being called more than once, so after a disconnect you will need to re-instantiate it.

I think to fix the event loop errors you have to do something like this:

task = asyncio.async(client.connect())
try:
    loop.run_until_complete(task)
except:
    task.cancel()
    loop.run_forever()
finally:
    loop.close()

[Terrance]

I've now got it wrapped up like that, and it recreates the client when the event loop fails or the client disconnects. Seems to work! 👍