Touchpad acceleration

[Hans de Goede]

<re-adding the list + Peter to the Cc, please keep the Cc intact when replying>

Hi,

On 11-04-15 11:24, Stefanos A. wrote:
> No, this is a Samsung RC520. After switching between Xorg and Wayland
> a few times, I'm fairly sure this is caused by the different
> acceleration curves. Xorg appears to ramp up to 1.0+ quicker than
> libinput, making it feel more responsive. Libinput remains in the <1.0
> region for larger finger movements, making it feel "laggy".
>
> Could it be that my touchpad is not yet part of the hardware database?

Yes that is definitely a possibility.

Can you please install "evemu" and then do:

sudo evemu-record
<press ctrl+c>

Then look which input node is your touchpad and do:

sudo evemu-record /dev/input/event# > touchpad-1fg-move.log

And then move your finger from left to right at normal mouse move
speed, then press ctrl+c and send us the generated log file.

Also please measure the physical size (width and height of usable
area in mm) of your touchpad.

Thanks & Regards,

Hans


>
> 2015-04-11 10:58 GMT+02:00, Hans de Goede <hdegoede at redhat.com>:
>> Hi,
>>
>> On 04/10/2015 11:15 AM, Stefanos A. wrote:
>>> Hello,
>>>
>>> I recently started testing Gnome 3.16 over Wayland 1.7.0 & libinput
>>> 0.13.0 and came across the following interesting behavior on my
>>> touchpad. In short, the pointer feels as if it lagging on the monitor
>>> compared to my finger movement. The result is that I consistently
>>> overshoot small targets, such as the "X" button to close a window.
>>>
>>> Now, I understand that there is a familiarity effect going on here,
>>> but I have to say that I am not seeing this behavior when using Xorg
>>> or Windows on the same hardware (or OS X on different hardware.)
>>>
>>> Moving the "touchpad acceleration" slider to the "slowest" setting
>>> appears to remove this lag, at the cost of a too-slow-to-be-usable
>>> pointer. Moving the slider to the "fastest" setting appears to make
>>> things worse, in the sense that I find it impossible to hit any screen
>>> target reliably.
>>>
>>> My theory is that the libinput acceleration curve starts with a <1.0
>>> acceleration for slow movements and moves to a >1.0 acceleration for
>>> faster movements and the lag is a placebo caused by the <1.0 part of
>>> the curve.
>>>
>>> Is there a way (configuration or tool) I could use to test this
>>> theory? My gut feeling is that I would prefer to remove or make the
>>> <1.0 part of the acceleration curve less aggressive and keep the >1.0
>>> part, if that makes sense.
>>>
>>> `cat /proc/bus/input/devices` returns the following:
>>>
>>> I: Bus=0011 Vendor=0002 Product=0007 Version=01b1
>>> N: Name="SynPS/2 Synaptics TouchPad"
>>> P: Phys=isa0060/serio1/input0
>>> S: Sysfs=/devices/platform/i8042/serio1/input/input9
>>> U: Uniq=
>>> H: Handlers=event12 mouse0
>>> B: PROP=9
>>> B: EV=b
>>> B: KEY=6420 30000 0 0 0 0
>>> B: ABS=260800011000003
>>
>> Do you happen to have a x230 laptop ? That one is know to have a couple
>> of issues with the libinput pointer accel code.
>>
>> Regards,
>>
>> Hans
>>