Re: Mouse in command line
Rex Chan wrote:
- Kent West <westk@acu.edu> [2004-04-28 07:23:18 -0500]:
You'll need to configure it to repeat the mouse data (either as "raw" or
as "ms3" I believe; IIRC, the author of gpm or some other person in the
know says "ms3" should be the correct choice, although until I read that
post (search archives over the past year) I had always used "raw" and it
worked for me), and you'll need to reconfigure the mouse in X to read
mouse input from /dev/gpmdata instead of wherever it's currently reading
(/dev/input/mice? /dev/psaux?). Although in the past year or so, the
"/dev/input/mice" may have become "magic" enough to handle reading from
gpm and might now be the canonical mouse location for X.
Hmm... I just tried this out with my usb mouse.
I got gpm working pretty easily, using gpmconfig...
Set repeating type to ms3 and protocol to imps2
I put set the device under X as /dev/gpmdata but that didn't work out
So i set it back to /dev/input/mice and it was all good.
However, it seemed like from the posts that X should have worked
perfectly with the /dev/gpmdata line.
That's good to know; I'm glad you did the experiment. According to the
GPM maintainer, Zephaniah Hull
(http://lists.debian.org/debian-user/2004/debian-user-200403/msg00887.html):
I have said it MANY times, and I will say it again.
There is exactly ONE protocol which it is valid to talk to gpm over
/dev/gpmdata with, and that protocol is NOT PS/2.
No matter how much people scream and yell about how it works fine,
this is a perfect example of why trying to use a bi-directional
protocol over a fifo DOES NOT WORK.
The protocol you need to use is the one gpm calls ms3, X calls it
IntelliMouse, which is the serial version of the IMPS/2 protocol,
however it is NOT bi-directional, it has 3 buttons, a wheel, and it
is safe to use over the fifo.
I don't know the context of his rant (his term, not mine), nor if it
applies to /dev/input/mice as well as to /dev/gpmdata, but thought I'd
throw it in here for completeness.
--
Kent
Reply to: