ANONYMOUS wrote: Yes - both 'ends' need to support the sending of their own data, receiving data from the other end, receiving acknowledgements (of your data) from the other end, and sending acknowledgements (about the data from the other end). We term this a bi-directional protocol, where both ends perform exactly the same functions. Each end acts as both a sender and a receiver. In cnet, you only need to write the 'enhanced' stop-and-wait code once to support piggybacking. Each node will run the same code, reading and writing its own copy of any variables. I'm afraid I don't understand your question; hopefully it was answered, above.