Pointer Acceleration Code Being Cleaned Up For Libinput

[phoronix]

Phoronix: Pointer Acceleration Code Being Cleaned Up For Libinput

Yesterday when writing about libinput 0.21 being released and libinput 1.0 appearing around the corner, the forum discussion quickly turned into a matter of libinput's pointer acceleration support...

http://www.phoronix.com/scan.php?page=news_item&px=Libinout-Point-Acceleration

[duby229]

Good. It really sucks right now. I mean really at this day and age that mouse cursers can be so laggy and jittery is just plain wrong. It only requires a few thousand cycles out of billions available to process. I just don't get it. It doesn't seem possible in todays age to have a buggy CPU bound curser. And yet that's what it is....

[swoorup]

And development takes so slow for something that's already implemented into X

[magika]

And development takes so slow for something that's already implemented into X

I fail to comprehend how presence of xf86-input-libinput slows upstream developement. However pointer acceleration was already implemented in libinput so we get a clean up, no reason to complain whatsoever.

[Ericg]

And development takes so slow for something that's already implemented into X

No one who is actually knowledgable about X's input stack can look at it think its better. Even if libinput DOES have pointer accel issues the very fact that its a singular stack rather than THREE stacks makes it night and day better. Scrapping X isn't about copying X, it's about rewriting the entire ideas and methodologies and muttering "Never again..." as we set the old one's on fire deep in the woods-- that takes time.

[boxie]

And development takes so slow for something that's already implemented into X

Clearly you are not a developer. I would rather them take their time and make the right decisions.

[swoorup]

Clearly you are not a developer. I would rather them take their time and make the right decisions.

You are right. I don't do low-level programming. I can't grasp how different it would be to evdev?

[hubick]

Yeah, when I hotplug an external monitor to my Thinkpad, the touchpad acceleration doubles along that axis alone (ie, with second monitor above primary, vertical acceleration doubles, while horizontal stays the same).

I confirmed this regression at https://bugs.freedesktop.org/show_bug.cgi?id=31636 ("Greater horizontal than vertical pointer acceleration with synaptics"), and was kinda wondering about it lately.

So, it looks like this should either fix it, or certainly lay the ground work for fixing it soon. I'm very glad to see this work, and appreciate Peter and all the other dev's work on this stuff. It's not glamorous, but it's certainly important!