Well shut my mouth! I did some testing this past weekend, as I said I would, and results are better than expected. First, leaving things the way I had them configured (X pointing to /dev/gpmdata and gpm pointing to /dev/psaux, I unplugged the PS/2 mouse from the mouse port. The mouse became dead in both X and gpm (duh!). I then plugged it back in again. It worked again, both in X and in gpm! Then I configured X and gpm both to use /dev/input/mice and turned off gpm's repeater function. After restarting both daemons and verifying that both X and gpm could use the mouse, I again unplugged the mouse. Again it was dead in both X and gpm (duh!). And again I plugged it back in and it started working again, both in X and in gpm, with no action on my part. I didn't have to restart the gpm daemon, I didn't have to restart X, I didn't have to unload and reload a kernel module, etc. This is better than I expected.
Of course, this is on a different machine (a Dell Dimension 4400) than I last tried this on (an IBM Thinkpad 600). But I'm impressed. Theoretically, one is not supposed to be able to hot swap a PS/2 mouse. But it works. Kudos to the kernel folks. The repeater function of gpm now appears to be obsolete, as you say. I would still like to see gpm installed by the Debian installer whenever a mouse is detected on the system in order to allow copy and paste in a virtual console. But I'm not going to flog a dead horse. The powers that be obviously don't like that idea. Thanks to all contributors to this thread. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]