If we consider it as a work in progress, is it a good idea then to "publish" it via wsconsctl immediately? Shouldn't we leave wsconsctl as it is until we have figured out what to do, or at least hide that new field? And, speaking of hiding, it is for a feature that's only useful for a subset of touchpads: MT-clickpads with MT-support in our kernel. How do you want to handle the cases where it is useless? And MT-clickpads without MT-support? It may raise wrong expectations. Without MT-data reasonable filtering won't be possible.
On 2/8/23 00:48, Patrick Wildt wrote: > On Tue, Feb 07, 2023 at 10:07:56PM +0100, Ulf Brosziewski wrote: >> I wouldn't mind seeing such a feature in the driver, but I think there's more >> to do than counting contacts. The start of a click-and-drag gesture may >> involve >> two contacts and a button-press event, or people who place a thumb in the >> lower >> clickpad area and use it for pressing the clickpad button might leave the >> index >> finger in the main area. In both cases you probably shouldn't generate a >> middle- >> button event, and I guess it doesn't happen on MacOS, or does it? >> >> There may be various means to distinguish the gestures. The driver might >> check >> the positions and the distance of the contacts, or identify their duration, >> their initial position or the current direction and speed of movement, etc. >> >> I don't know which strategies work well and can be implemented with >> reasonable >> effort, it might not be easy to figure that out. It seems that libinput uses >> distances (see >> >> https://wayland.freedesktop.org/libinput/doc/1.22.0/clickpad-softbuttons.html >> ) as well as additional means for identifying "thumbs", but I'm not familiar >> with the details. > > Sounds like this good be something one could improve upon in-tree? The > diff already feels much better than the current default. > > Cheers, > Patrick >
