WebSocketTransport::close method does not make sense

[bsergean]

We have 3 states, OPEN, CLOSED, CLOSING.

        // We start by doing an early out if we are not OPENED essentially
        if (_readyState == ReadyState::CLOSING || _readyState == ReadyState::CLOSED) return;


        // connection is opened, so close without sending close frame
        if (_readyState == ReadyState::OPEN)
        {
            {
                std::lock_guard<std::mutex> lock(_closeDataMutex);
                _closeCode = code;
                _closeReason = reason;
                _closeWireSize = closeWireSize;
                _closeRemote = remote;
            }
            {
                std::lock_guard<std::mutex> lock(_closingTimePointMutex);
                _closingTimePoint = std::chrono::steady_clock::now();
            }
            setReadyState(ReadyState::CLOSING);

            sendCloseFrame(code, reason);
            // wake up the poll, but do not close yet
            _socket->wakeUpFromPoll(Socket::kSendRequest);
        }

        // Why would we ever come back here ? If that function is called twice ?
        else
        {

[bsergean]

@Kumamon38 can you explain how this method work ? This code does not make sense to me.

[Dimon4eg]

comment connection is opened, so close without sending close frame is not correct because below I see sendCloseFrame(code, reason);

[bsergean]

I like the new close unittest which looks sensible and which pass, there used to be bugs in that code, but I have to say that this code is now much more complex than it used to be and that it doesn't make much sense anymore.

[AlexandreK38]

There is also the OPENING state, but I wasn’t sure about if we can be in. I think we can remove the else part and the function

[AlexandreK38]

*and the comment

[AlexandreK38]

Cleaned 🧹

[bsergean]

This is much better now, closing. Thanks for the fix.