Split rx continues after tx is closed

[huntc]

My expectation when splitting the rx and tx was that closing the tx would cause the rx.read call to fail the read or generate a close message. Neither happens, and the receive function continuous. Is this the expected behaviour? If so then should we periodically timeout the read to test whether the connection remains open?

    println!("starting");
    loop {
        match rx.read(&mut buf).await? {
            Message::Text => {
                let _s = String::from_utf8(buf.to_vec())?;
                tx.write(&mut buf, PayloadType::Text).await?;
                tx.flush().await?;
                buf.clear();
                println!("closing tx");
                tx.close(ratchet_core::CloseReason {
                    code: ratchet_core::CloseCode::GoingAway,
                    description: None,
                })
                .await?;
            }
            Message::Binary => {
                tx.write(&mut buf, PayloadType::Binary).await?;
                tx.flush().await?;
                buf.clear();
            }
            Message::Ping(_) | Message::Pong(_) => {}
            Message::Close(_) => {
                println!("ending");
                return Ok(());
            }
        }
    }

[huntc]

Howdy. Just wondering if there are any thoughts on this? I can imagine this being problematic for any use of splitting the socket. Thanks.

[SirCipher]

Hey @huntc, in theory the read half should terminate automatically once it receives the echoed close frame. I'll have to have a look into this to see if it's not happening as expected

[huntc]

Thanks @SirCipher ! Hopefully, the above code modifications to the autobahn-split-server.rs help you reproduce the problem, as they did for me.

[huntc]

Hi @SirCipher - is there anything I can do here? Happy to push a PR if you can guide me on what area I should be looking at. Thanks.

[huntc]

Were you able to reproduce it @SirCipher ?