On Wednesday, February 19, 2014 at 4:55:03 PM UTC+1, Guido van Rossum wrote: > > Because that would require duplicating the logic in > create_datagram_endpoint(), which is totally separate (all the way down to > the I/O callbacks). > > Feel free to add a feature request to the tracker for this though. Be sure > to indicate an actual use case (rather than just abstract interest). E.g. > do you expect to need this in OpenStack? How? >
Hi everybody, I would wake-up this thread because I have a concrete use case to have that in AsyncIO, even if it would the only one concrete use case from a production environment: systemd has a watchdog mechanism and a way to communicate between a daemon and systemd. It uses a UNIX socket with DGRAM, the mechanism is named sd_notify. For example, a Python that implements sd_notify: https://github.com/bb4242/sdnotify/blob/master/sdnotify/__init__.py#L35 Another example in Python: https://gist.github.com/Spindel/1d07533ef94a4589d348#file-watchdogged-py-L78 More technical documentation about sd_notify protocol: https://www.freedesktop.org/software/systemd/man/sd_notify.html In fact, if somebody has a workaround or a way to implement that gently with AsyncIO, it would be awesome, because I need to ping-pong with systemd to say that my daemon is still alive. Thanks for your help. I have tried to find a way with loop.create_datagram_endpoint() that would be the most closer what I want and a custom socket, I have this error: File "/usr/local/pythonz/pythons/CPython-3.6.2/lib/python3.6/asyncio/base_events.py", line 951, in create_datagram_endpoint yield from waiter AssertionError: yield from wasn't used with future My piece of code: sock = socket.socket(socket.AF_UNIX, socket.SOCK_DGRAM) sock.setblocking(False) sock.connect(os.environ["NOTIFY_SOCKET"]) transport, protocol = loop.create_datagram_endpoint( lambda: WatchDogClientProtocol(loop), sock=sock)
