Mouse wheel is too fast in wasm builds: unit Pixel vs Line confusion

[strowk]

Bevy version and features

[dependencies.bevy]
version = "0.16.1"

Relevant system information

I made two builds - one for Windows 11 and one for wasm, which I use with Edge.

What you did

I was trying to work out how to zoom out camera. Here is the system that reproduces the issue:

fn zoom_camera(
    mut scroll_evr: EventReader<MouseWheel>,
    camera_query: Single<&mut Projection, With<Camera2d>>,
) {
    match *camera_query.into_inner() {
        Projection::Orthographic(ref mut orthographic) => {
            for event in scroll_evr.read() {
                info!("Mouse wheel: {event:?}");
                //Adding to scale zooms out (the -= is there since it is reversed -> adding to scale zooms out)
                orthographic.scale = (orthographic.scale - event.y * 0.03).clamp(0.1, 3.0);
            }
        }
        _ => (),
    }
}

Note the log info!("Mouse wheel: {event:?}");

What went wrong

Moving mouse wheel on windows build and on web build results in two very different changes and two different logs. Web build produces much bigger change for some reason.

Here is log from browser (after one minimal wheel rotation):

Mouse wheel: MouseWheel { unit: Pixel, x: 0.0, y: 100.0, window: 0v1#4294967296 }

and same for Windows build:

 Mouse wheel: MouseWheel { unit: Line, x: 0.0, y: 1.0, window: 0v1#4294967296 }

As you can see the change in "y" is 100 times different!

Additional information

Nothing I could find so far.

[alice-i-cecile]

Note that the unit, Lines vs Pixel is different! You should match on the unit provided, and provided some sort of scale correction.

We should add docs to point people in the right direction here though.

[strowk]

Ooh, I did not notice this. So Line has 100 pixels in it? I know what pixel is but line is a new one. Is there any place where line to pixel ratio specified?
Asking for the case when I would want to make a game that runs similarly in windows and in web.

[alice-i-cecile]

My understanding is that the ratio is not fixed: traditionally in web content this varies based on the line height of the selected content. In the absence of content though, it seems to default to 100? I would do more testing, but please report back: 100 on web might be stable enough for us to document and recommend?

Alternatively, we could explain how to grab the conversion factor by doing some Javascript analysis of the context of your project?