libinput_event_tablet_tool_get_dx/y are documented as returning accelerated values if the tool employs those. It maybe doesn't make much sense to accelerate all tools, but at least mouse/lens tools do, as those will usually have a relative mapping by default and are expected to feel similar to regular mice.
btw, have you tested this with a real device on your box? There's a vocal group of people that want to disable acceleration for mice anyway, I'm wondering if we even need to bother for acceleration here.
(In reply to Peter Hutterer from comment #1) > btw, have you tested this with a real device on your box? There's a vocal > group of people that want to disable acceleration for mice anyway, I'm > wondering if we even need to bother for acceleration here. I did, it felt generally on the slow side to me (having to move the mouse a considerable distance on the tablet to move it across the display). In my probably biased opinion, some acceleration seemed enough to give it some more kick, but then having no acceleration at all felt odd to me. I found myself trying to move the mouse faster to make better use of the tablet area, to no avail.
ok, checked the code and we have an accel method for tablet mice (see tablet_accelerator_filter_flat) but it's designed to provide effectively the same motion as the absolute tablet. So it makes sense that it feels slow atm, you wouldn't have to move a normal mouse ~20cm to move across the screen. I'll speed this up to match actual mouse movement, then we can go from there.
Created attachment 124633 [details] [review] 0001-tablet-make-the-tablet-tools-behave-like-a-1000dpi-m.patch
That feels a lot better on mice :), although... if I enable relative mapping on the tablet (so get_dx/dy is used on all tools), pen tools feel far too nervious now. Being two entirely different ergonomics (holding above vs letting rest), I guess it will be hard to find something that works for both... I guess it'd be best to have 2 separate acceleration filters for those?
the pen was ok before? I thought about leaving the current code for the pen but after I updated the mouse the pen felt a bit slow in relative mode. Having separate per-tool acceleration is no biggie, the infrastructure is there.
(In reply to Peter Hutterer from comment #6) > the pen was ok before? I thought about leaving the current code for the pen > but after I updated the mouse the pen felt a bit slow in relative mode. Hmm, It's true that it might be slow, the same distance considerations definitely apply to those, but they felt just as precise as you can be in absolute mode. With this acceleration even little pen shakes move the pointer noticeably, which makes accurate aiming sensibly harder. But then, I perhaps have some expectations already on how does each device feel :). > > Having separate per-tool acceleration is no biggie, the infrastructure is > there. Good :).
https://lists.freedesktop.org/archives/wayland-devel/2016-June/029697.html
commit 0926f570c448b4d321e23e075f0ad3032c56b92a Author: Peter Hutterer <peter.hutterer@who-t.net> Date: Tue Jun 21 15:20:08 2016 +1000 tablet: make the cursor/lens tool behave like a 1000dpi mouse
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.