Messages in this thread | | | Date | Wed, 18 Dec 2013 09:12:45 +0100 | Subject | Re: [PATCH 0/4] Input: ABS2 and friends | From | David Herrmann <> |
| |
Hi
On Tue, Dec 17, 2013 at 10:28 PM, <simon@mungewell.org> wrote: >> Hi >> >> This implements the recently discussed ABS2 API. It's working fine on my >> machine with libevdev. Comments welcome! > > Just looking at the documentation file, I have a couple of suggestions. > > 1). Make a note on the direction of gravity wrt the example image. Ie what > which axis/value will be positive with the device laid flat/stationary on > a desk. Might just prevent some confusion later.
Yepp, can add that to the Accelerometer section.
> 2). Calibration, Accels and Gyros are noisy in different ways. Gyros tend > to be wrong in the long term, giving some consistant bias which should be > calibrated out. Accels are noisy in the short term, but generally right > over a long period.
I'm not entirely sure where to put calibration/normalization. Letting user-space deal with that should simplify things a lot. We could even add it to libevdev. Comments welcome..
> Am I correct that the event framework will provide time-stamping of data, > thus making computation of relative movements possible.
Yes, part of the evdev API.
Thanks David
| |