I find that too, ATM, there doesn't seem to be a reconnection made, when moving from 4G/5G to wifi, the WHIP connection will completely fail & need to be remade, this is because the status on the server changes from "connected" to "disconnected" & "client_disconnect", same for the WHEP. This includes if the stream input on the server side has the same UID. This can be fixed on the input client side by checking if the stream input goes down, to redo the /publish endpoint & restart the stream, also it's being done currently in the WHEP client by the looks of it, I haven't taken a longer look. There could be quite an issue with recordings for this, as if the stream goes down, does the video stop recording and make a new recording? Or continue on the old recording? I have been polling the API from when the WHIP connects to the server, when the server gets it and when the WHEP starts, for me it's very long. RTMP/HLS is much faster. But it's all beta, I am just noting what I see.