OPENGL_TO_WGPU_MATRIX constant for NDC conversion is incorrect about learn-wgpu HOT 2 CLOSED

I originally got the matrix from the wgpu repo, though I can't find it anymore, but it makes sense to change it.

from learn-wgpu.

The matrix was actually correct like it was. cgmath::Matrix4::new() has a HUGE pitfall! That is, its arguments are ordered in column-major order. So essentially, it creates a transpose version of the matrix you see, when formatting like this. I don't know why anyone thought this to be a good idea, but this is how it is...

I suggest changing it to this version, because its semantic meaning unambiguously conveys the real meaning of it. (There is no from_rows())

#[rustfmt::skip]
pub const OPENGL_TO_WGPU_MATRIX: cgmath::Matrix4<f32> = cgmath::Matrix4::from_cols(
    cgmath::Vector4::new(1.0, 0.0, 0.0, 0.0),
    cgmath::Vector4::new(0.0, 1.0, 0.0, 0.0),
    cgmath::Vector4::new(0.0, 0.0, 0.5, 0.0),
    cgmath::Vector4::new(0.0, 0.0, 0.5, 1.0),
);

See: https://docs.rs/cgmath/latest/src/cgmath/matrix.rs.html#285-297

from learn-wgpu.