Mouse settings: add a button debounce timeout next to double-click (worn mice / hardware longevity)

[Egzoset]

Using:
Linux Mint 22.3 (Zena)
Cinnamon 6.6.9

Cinnamon already exposes a double-click timeout in Mouse settings (org.cinnamon.desktop.peripherals.mouse double-click, currently 270 ms here). There is no matching control for button debounce — the minimum time that filters a worn switch bouncing after release.

That gap is easy to miss. libinput already implements bounce + spurious debounce, but the timeouts are hardcoded (historically ~25 ms / ~12 ms). There is no public per-device setting, unlike pointer speed. After a distribution update, a Microsoft Wireless Mouse 1000 (model 1454) that had stayed usable became unusable: stock xf86-input-evdev ignores Xorg Option "DebounceDelay", and forcing evdev also bypasses libinput’s built-in debounce. The same class of mice often remains usable under Windows, which applies a more forgiving software debounce.

This is not a request to slow down gaming mice, and it is not a complaint that Cinnamon “broke” clicks. It is a request for a user-visible, per-device timeout next to the existing double-click slider, defaulting to current behaviour. Desktop settings already expose a maximum inter-click interval for classifying a double-click; they do not expose a minimum release-hold / bounce filter for hardware that still works but chatters with age.

On this machine, 90 ms debounce plus the 270 ms GTK double-click timeout restored reliable single clicks without forbidding intentional doubles. A userspace uinput filter should not be required for a timeout the stack already has internally.

A local MIT workaround exists: https://github.com/Egzoset/mouse-debounce

Please consider adding that slider in Mouse settings (and, if the delay must come from libinput, relaying the same request upstream). I will file a libinput issue on GitLab Freedesktop when that instance is reachable again (it currently returns 504). The goal is hardware longevity: keep working switches usable instead of treating them as disposable.

[rphair]

thanks @Egzoset — I've looked over the code & conception of your utility and believe it should be considered essential: perhaps across the whole Linux ecosystem. For years that ecosystem has abandoned this issue for the argument expressed here: https://bugs.freedesktop.org/show_bug.cgi?id=11833

Basically: this won't get fixed because the cost of adding and maintaining the code forever is a lot higher than just replacing a buggy mouse, sorry.

If this evasive & bureaucratic rationale were valid, Linux would have no support on older computers because the cost of maintaining / documenting / improving it would only be "worth it" for new computers: and therefore the old ones should just be "replaced". Given the scientific focus of Linux, it's ironic that Windows can perfectly debounce a mouse in software while the ability to do this on Linux went out with the evdev update years ago.

Of course I would be happy to see this in Mint — which I use 1 or 2 days a year — but our user communities would benefit hugely if this could work more generally across GTK environments all through the Debian ecosystem: especially our huge user base on Ubuntu.

I don't know what it would take to generalise your mouse-debounce tool beyond Mint... but I know it would be wonderful to have something on Google search results besides 15-year-old articles about an obsolete solution and 10-year old threads showing that nobody cares. 🧐 Egzoset/mouse-debounce#1

[Egzoset]

No need to post in French, please use the language that serves you best.

It's true there are fundamental notions remaining to explain better, the present documentation is only a first step. In the meantime here's what's on my mind so far - no promises made, no strings attached:

https://www.transfernow.net/dl/mouse_debounce

MD5:  d7234562d805600f973accee90b0e707
SHA256:  00082016912897b7b005961c7959ae05a401bc7ff49fbe9ad4262eef7781a54b

(This is short lived, less than a week)

Tests using LUbuntu v24.04.5 "LIVE" went fine, but not so much with v26.04 though. This .DEB package warned about its need for python3-evdev, python3-psutil and python3-xapp; after that the trick was to only rely on its "WebLab" alternate GUI, evoked as follows from the terminal once installed:

mouse-debounce --beep --weblab

Then, you may also wish to experience its sound feature in your web browser:

http://127.0.0.1:8765/?beep

I'm afraid the GTK GUI is still necessary to access a few default settings conveniently, this is little more than work in progress!

...

About the rest.

Indeed, Peter Hutterer did justify hard-coded magic numbers out of fear that debouncer code would cause "recurring costs" and a "maintenance burden", although already in use... His perfect "open-source" solution was then to transfer this entire issue onto a broad user base, suddenly sentenced to arbitrary hardware obsolescence.

Today, such a questionable subtlety finally relies on some hypothetical level of automation that can never materialize unless everyone yields to such consumerist drive. Yet there's an opportunity to cease in between which he still fails to see: to address one big real-world weakness of the 2.4 GHz spectrum: its overlooked, saturated crow on top of mechanical wear, plus defaults carved in stone.

Time has come to transcend this debate far beyond material longevity, the return of wired pointing devices just ain't appealing to me. That's why my attention shifted to ways of acknowledging our dynamic reality, by putting together a tool aimed at showing external influences like radio interference, eventually expressed via a "confidence score" assigned to crude peripheral data being collected, for example. Lets face it, wireless environments can become unpredictable and that's still not included in the big picture at FreeDesktop.Org it seems.

Well, at least it's clear there's a real need to investigate signal reliability assessments IMO.

[rphair]

I'm happy to have something to try out: also with appreciation for the continuing historical / technological perspective. My daily mouse driver — hosting a mouse just recently "failing" from hardware click bounce, which I still use 80 hours per week — is on Ubuntu GNOME 24.04.5 and I've collected your software package to hopefully test in the next few days.