[dev build] very fast mouse scroll speed on client

[markmandel]

I just noticed this on the updated dev build today.

When I scroll with my mouse wheel, it tends to scroll way further than it used to on the client (i.e. the non primary computer).

I'm 96.7% sure that it wasn't this aggressive before - I would have noticed - I use this every day 👍🏻

Is this just a me thing?

Running:
Debian Testing
Hyprland

[markmandel]

Was just looking through history - I'm assuming #325 is now why I'm finding scrolling SUPER fast in Chrome.

This is one click of my scrollwheel back and forth:

On my host machine, it only moves slightly down the page smoothly.

Update:

I check wev on host vs client machine - and a scroll is the same value across both:

wl_pointer] axis: time: -212050813; axis: 0 (vertical), value: 15.000000
[        15:      wl_pointer] axis_source: 0 (wheel)

[feschber]

Yeah this is a regression in 39b79d8. axis_descrete takes tiks, not value120

[feschber]

@selckin could you check, if setting axis_source is sufficient for your case? It should technically fix the scrolling regardless.

[selckin]

Without it and only axis_source it still only scrolls a tiny amount for me

lan-mouse
[        15:      wl_pointer] axis_source: 0 (wheel)
[        15:      wl_pointer] axis_value120: axis: 0 (vertical), value120: -1
[        15:      wl_pointer] axis_relative_direction: axis: 0 (vertical), direction: 0
[        15:      wl_pointer] axis: time: -142902005; axis: 0 (vertical), value: -20.000000
[        15:      wl_pointer] frame
real-mouse
[        15:      wl_pointer] axis_source: 0 (wheel)
[        15:      wl_pointer] axis_value120: axis: 0 (vertical), value120: 120
[        15:      wl_pointer] axis_relative_direction: axis: 0 (vertical), direction: 0
[        15:      wl_pointer] axis: time: 1454827716; axis: 0 (vertical), value: 15.000000

Do you have an axis_value120 event ?
Maybe its capturing a wrong input value then?

[selckin]

I install hyprland on my the client & server, and did some testing:
Server side niri or hyprland have same behavior
Client being hyprland then chrome scrolls normally and the events also have 120 value:
EDIT: ^ with my change/patch reverted

lan-mouse
[        15:      wl_pointer] axis: time: -138830402; axis: 0 (vertical), value: 20.000000
[        15:      wl_pointer] axis_source: 0 (wheel)
[        15:      wl_pointer] axis_relative_direction: axis: 0 (vertical), direction: 0
[        15:      wl_pointer] axis_value120: axis: 0 (vertical), value120: 120
real-mouse
[        15:      wl_pointer] axis: time: 279190; axis: 0 (vertical), value: 15.000000
[        15:      wl_pointer] axis_source: 0 (wheel)
[        15:      wl_pointer] axis_relative_direction: axis: 0 (vertical), direction: 0
[        15:      wl_pointer] axis_value120: axis: 0 (vertical), value120: 120

So difference in behavior between niri/smithay and hyprland, events also in different order not sure if that matters (yet)

[selckin]

With or without my change, hyprland+chrome scrolls the same amount in my tests

I'll try and investigating it in more in details when I have time and look at what smithay and hyprland do; if you want to revert it till then that makes sense.

[markmandel]

Appreciate y'all looking into this!

Bit busy at the moment, but happy to run tests or provide whatever information I can on my end in the next couple of days if needed or wanted.

[feschber]

@selckin you are using niri on both sides, right? I think I will do a partial revert for now and test this myself.

[selckin]

@selckin you are using niri on both sides, right? I think I will do a partial revert for now and test this myself.

yes

[feschber]

Alright, I'm pretty sure this is a niri bug: https://github.com/YaLTeR/niri/blob/e6f3c538da0c646bda43fcde7ef7dc3b771e0c8b/src/protocols/virtual_pointer.rs#L522

they are directly using the discrete (=steps) value here as the v120. This should be multiplied by 120. I'm going to open a PR.

[feschber]

YaLTeR/niri#2684

[selckin]

Thank you very much for looking into this, can confirm that everything works for me with the niri patch applied

[markmandel]

New dev build is working much better for me! Thank you!