Touchpad accel, spirals and xset

Paul Fox pgf at
Mon Jan 18 10:11:53 EST 2010

martin wrote:
 > Late last year Gary and others volunteered to test my 802b1 build,
 > with a nice test practice from Gary of drawing spirals. It took me a
 > while to get back onto this (apologies - real life intruded in a big
 > way), but I have been working on this again, hoping to close it.
 > I took 4 laptops (2 with ALPS, 2 w Synaptics) and did my own 802 vs
 > 802B1 bakeoff, following the spiral trail. Results -
 > * No jumpiness for me in the least -- probably because it's winter here.
 > * No observable regressions in the touchpad code -- maybe 802B1 is
 > marginally better, but the more I test the less conclusive it is.
 > Specially when I undo the xset changes...
 > * Going from xset 7/4 0 to 7/4 1 did not make a noticeable improvement
 > -- and does bring the drawback that a fast finger swipe no longer gets
 > you across the screen. This is a usability regression, and I don't
 > have much evidence to show to counter it.
 > I did read in Gary's email that there was a reference to a big
 > sensitivity difference between 802 on one hand and the F11 build and
 > 802B1 on the other. I cannot find this big sensitivity difference.
 > Happy to try again if anyone can suggest what to attempt.
 > So for the 8.2.2 release, current plan is:
 > * The new kernel (with modular psmouse, with the new code) stays.
 > * xset goes back to 7/4 0

could other people running build 8.2 on XO-1 please try various
xset commands, before martin makes this final?  in my testing, setting
the final argument to 0 resulted in extremely jerky performance.  if
setting the last arg to 4 is too slow, i think using 1, 2, or 3, would be
a better choice than 0.

 paul fox, pgf at

More information about the Devel mailing list