I tried propagating neqo_transport::Connection and calling conn.stream_keep_alive(stream_id, true)?. This caused panic for the following tests in streams.rs.

wt_client_stream_uni
wt_server_stream_uni

The panic is originating from RecvStreams::keep_alive:

 pub fn keep_alive(&mut self, id: StreamId, k: bool) -> Res<()> {             
  65         let self_ka = &mut self.keep_alive;                                      
  66         let s = self.streams.get_mut(&id).ok_or(Error::InvalidStreamId)?;                                                                                                                                                                                            
  

FWIU, we shouldnt enable the keep alive flags for local unidirectional streams as they do not have recv streams associated with the session. Hence, we should enable keep alive flag for the following streams:

• Bidirectional - local and remote
• Unidirectional - remote

@KershawChang, @martinthomson : Please confirm if my understanding is correct

@MayyaSunil you should only need to enable a keep-alive for the session stream, that is the "request" that is still open. If that stream closes, the whole WebTransport session ends, so - that way - you will be keeping the connection alive as long as there is any WebTransport session active.