I've found that this article was really illuminating with respect to some of the issues around `libinput`: https://lwn.net/Articles/801767/
Example from the lwn article:
> The xf86-input-evdev driver is in maintenance mode at this point. The last commit was in May 2018 and, since the 2.10.0 release four years ago, there have been a total of 19 commits.
Maybe that's actually a sign that it's good enough and doesn't need major adjustments? But here it is presented as if it is a problem.
I say this not to criticize, but only to point out that the driver is not "done". I am incredibly impressed with Peter Hutterer's skills that things work as well as they do, given that he's essentially the only person working on it. It could use more eyes and hands.
I'm curious what hardware and drivers you've tried. I'm using a fairly large Synaptics touchpad on a Dell laptop with the xf86-input-synaptics driver. It's so good that I literally had to test it just now to see if it was possible to fool it at all. Even very light touches with the edge of my palm are unable to convince the driver to click. I can't move the cursor at all. At least in my case, it works perfectly... maybe I'm just lucky. (Or maybe the new libinput driver is still not up to snuff.)
Unless he's planning some way to get that context and implement it later, that pretty much makes libinput unusable for me. Relying on every individual program to implement their own kinetic scrolling guarantees it'll act differently across all of them - or just be nonexistent, as is the case now.
IIRC, the one bug introduced by the synaptics way of handling it is if the mouse is moved from one window to another during the period while it's still scrolling, it then scrolls the newly-focused window instead. This annoyance is so extremely minor I simply do not care.