Hangup timeout handling #24
Conversation
| @@ -175,6 +178,7 @@ def datagram_received(self, data, addr): | |||
| if not self._is_connected: | |||
| return | |||
|
|
|||
| _LOGGER.debug("Datagram received on addr %s, current addr is %s", addr, self.addr) | |||
| if self.addr is None: | |||
There was a problem hiding this comment.
Should we just always set address?
There was a problem hiding this comment.
That would probably work as a stopgap fix to get out quickly, but I think a more technically correct solution would require some broader changes.
I suspect The VoipAssistSatellite should not be extending RtpDatagramProtocol as a new RtpDatagramProtocol should be getting created for each call, but the VoipAssistSatellite object has a much longer lifetime. It seems we can not assume a VOIP device is going to reuse the same RTP address/port. Based on this comment home-assistant/core#128372 (comment) I think we should be initializing the RtpDatagramProtocol with the host and port from the call_info instead of waiting to set it in the datagram_received method.
|
Thanks! Closing this in favor of: 5472d4c |
Reset the RtpDatagramProtocol addr attribute to None on disconnect. This will make sure the RTP port gets properly set for subsequent calls after hanging up.