On Fri, Jan 05, 2001 at 04:26:42PM -0500, Mithras wrote: > My mouse has been working fine, but perhaps something more subtle > could be wrong. Learning something new's always valuable.
Yes, it is only working because your errors zero'd out:) > On Fri, 5 Jan 2001, Carel Fellinger wrote: ... > > This doesn't sound right! (maybe it's just me failing to understand you) > > You should not touch /dev/gpmdata, let alone rename it. Let me recap: > > > > In XF86Config you should use: > > > > Protocol "whatever suits your mouse" > > Device "/dev/mouse" > > Currently I have: > > Protocol "Microsoft" > Device "/dev/ttyS0" If it works, fine; however X and gpm tent to give raise to problems when both try to read the mouse. IIRQ this applies only to busmouse, like PS/2, but it's good to be prepared. So to be on the safe side you better do as adviced:) Besides, if you ever deside to replace the mouse then you'll only have to change this symlink to have it working again (yeah right:) For busmouse the workaround is to use gpm in `repeater' mode, and have gpm read the mouse all the time, writing what it reads to /dev/gpmdata for other programs (like X) to read. By default gpm will translate the raw mouse data to the `msc' protocol before writing it out to /dev/gpmdata. This behaviour is triggered by the `-R' or `--raw' option. If you don't specify a protocol, or an empty protocol, then `msc' is used, but you're normally better of using `raw' as protocol as this leaves all the mouse data as is. If you setup gpm to do its repeating thingy, make sure that gpm is the *only* program reading the mouse directly! Have other programs read /dev/gpmdata instead. But be advised to let those other programs read from /dev/mouse and symlink /dev/gpmdata to /dev/mouse. The nice thing about using the /dev/mouse symlink and have gpm repeat in `raw' mode is, that you don't have to reconfigure those other programs whenever you decide to dump gpm (or the repeater mode). Just dump gmp and adjust the symlink /dev/ttyS0 to /dev/mouse and your done! If you don't have a busmouse (like /dev/ttyS0) then you don't really need this repeater stuff. Simply commenting the `raw=' line out in /etc/gpm.conf and the repeater stops: no /dev/gpmdata, and X needs to read from /dev/ttS0 it self. Be advised, however, to still let X read from /dev/mouse, and symlink /dev/ttS0 to /dev/mouse. > I haven't used anything like the repeat_type or append settings you > have there. Gpm's device is /dev/ttyS0, and the type is ms. > repeat_type is ''. So repeating is on, and translated into `msc' protocol. > Still, these settings did not work until I moved /dev/gpmdata. The I bet that after restartin gpm (eg. after a reboot) you'll have problems again. I think gpm creates this file each time it's running, so you'll have to remove it each time. Don't. Either setup gpm to do no repeating or do it properly. > mouse did not move, yet I'd set the pointer to use /dev/ttyS0. I > remembered reading here or on the web that someone else couldn't get > their mouse to work on X until they removed gpmdata, so I thought I'd > try that. I'd read it was a named pipe that the gpm process populated > with mouse input so X could get its data from it, so I figured it > wouldn't hurt anything in the console if I took it away. Maybe X was Taking it away was the error that evened out the previous errors:) In effect it prohibited gpm from doing its repeating thing, so it stopped reading from /dev/ttS0 and X no longer had a conflict on who was reading the mouse:) Oh and yes, gpm is smart enough to know when someone else (X) controls a virtual console, so it starts reading at the moment you switch back to a text oriented console. > This might not be ideal, but I'm a practical guy. To me this is not practical, but making a mess out of your system. A few of those and debugging odd behaviour gets very complicated, better leave that to the windows world:) -- groetjes, carel
