Hi, I want to write a gui application (wxPython) that runs on MSW and talks to a Linux server that has an application that only runs on Posix based systems. The current gui app runs on the Linux server and executes the commands directly using popen, but I want to make it flexible enough so that MSW users can run a local gui which will execute the commands on the Linux server and display the results locally. My question is one of architectural design. How is it best to structure the program so that it can use a remote shell connect (eg. ssh, or possibly other transports such as rsh, or rsh over a vpn, etc) as well as using direct calls to the utilities when running directly on the server ??? I've managed to get the sshsimpleclient.py going and talking to my linux server with my login and rsa keys :) I notice that this sample is different to the documentation mentioned here: http://twistedmatrix.com/projects/conch/documentation/howto/conch_client.htm... The sshsimpleclient.py does not have the following statements: factory = protocol.ClientFactory() factory.protocol = ClientTransport reactor.connectTCP('localhost', 22, factory) reactor.run() Is one better than the other? Is one method better going forward into the future? Also, the sshsimpleclient.py seems to have the closed() method called when issuing a command (eg. CatChannel, TrueChannel, etc). How can I keep a channel open to run commands when I want ?? Cheers, Brendan.