2014-02-21 5:26 GMT+06:00 Peter Hutterer <[email protected]>: > Touchpad data is always subjective :) it depends on finger size, > conductivity and a whole bunch of other factors. what does help though is > having evemu recordings of certain events. They may still be subjective but > they're reproducible, allowing us to fine-tune based on a specific event > sequence and then go from there. > > e.g. you can take a recording fo an example one-finger tap and then make > sure that that always produces a tap event. if you build up a database of > these events, you can cover events quite well for a multitude of devices and > users.
OK. Attached are some sample recordings from my touchpad, while the
synaptics driver is in use (for visual feedback).
left-clicks.rec: 10 left clicks. Any pointer movement is not intended
and should be rejected by the driver.
move-and-double-tap{,2}.rec: contains finger movement to a certain
word on the screen, followed by double-tapping on that word. The
synaptics driver, however, converted one of these into a tap-and-drag.
move-tap-and-drag{,2,3}.rec: contains finger movement to a certain
word on the screen, followed by attempting to select a certain part of
it by tapping and dragging. In one case, contains a non-deliberate
vertical-movement error that led to selecting a line of text.
pretend-close-pidgin.rec: just an attempt to move the pointer to the
center f the close button of Pidgin.
pretend-close-pidgin-bad.rec: the same, but, upon removing the finger,
the synaptics driver picks up unwanted motion. Basically, the last
reported point before removing the finger should have been discarded
as invalid.
scroll-up.rec: scrolling the text up, 10 times.
swipe3-right: some three-finger swipes to the right on a touchpad that
can track only two fingers. Contains BTN_TOOL_TRIPLETAP.
--
Alexander E. Patrakov
rec.tar.xz
Description: application/xz
_______________________________________________ wayland-devel mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/wayland-devel
