info-vax@ucbvax.ARPA (07/12/85)
From: *Hobbit* <AWalker@RUTGERS.ARPA> I found the bit in the I/O users guide that sez that a setmode QIO with the IO$M_HANGUP modifier will disconnect a virtual terminal. I have been playing around with this a bit and find that it ain't necessarily so. First of all, if you make that a qiow, you sit there forever until the person you are trying to zap types something on his terminal with a line terminator. Doing $set term/hang tqxn: has the same effect. If the terminal is not set /modem, you'll never be able to hang it up, and likewise /disconnect. My question is, how do I reliably bash dialups without having to wait for the terminal driver to decide that I *really* want that line disconnected? This waiting around is a pain, and doesn't work all the time, and a straight qio with no wait doesn't work at all. What *should* the documentation have said? As an aside, while I was playing with this using RTAn: connections, I crashed the machine twice. The foreign end said something about data set hangup, while the local machine [where the forced-disconnect had taken place] got very bent out of shape and eventually lost its lunch [like when I went into NCP and tried setting the circuit off]. Since RTDRIVER doesn't support disconnecting, I wouldn't really expect this to work, but Decnet should be able to recover from what it sees as a dataset hangup much more gracefully than this. _H* -------