Fix vk::TransformMatrixKHR

[0xbeefd1ed]

This is a regression caused by #951.

matrix is a 3x4 row-major affine transformation matrix.

So the matrix should have 3 rows, each row with 4 elements.

After #951, to convert a glam::Affine3A into vk::TransformMatrixKHR you'd write

pub fn glam_to_vk_transform(affine: Affine3A) -> vk::TransformMatrixKHR {
    // affine.matrix3 is column-major
    let x = &affine.matrix3.x_axis; // first column
    let y = &affine.matrix3.y_axis; // second column
    let z = &affine.matrix3.z_axis; // third column
    let w = &affine.translation;
    vk::TransformMatrixKHR {
        // row major
        matrix: [[x.x, y.x, z.x], [w.x, x.y, y.y], [z.y, w.y, x.z], [y.z, z.z, w.z]], // This is clearly wrong
    }
}

This is clearly wrong. This PR fixes it by correcting the array dimensions.

• Alternatively, just keep the flat array structure as before.

[Ralith]

Thanks for catching this! I concur that the dimensions are reversed. However, we cannot accept this PR as written because this is generated code; what needs to be fixed is the generator, which is probably misinterpeting nested C array syntax in the XML.

[0xbeefd1ed]

@Ralith thanks for calling that out. I've fixed the generator as well. This part was entirely hand-written.

[Ralith]

Thanks!

[0xbeefd1ed]

Closing, as the maintainer has fixed this in 69afb6b without a PR.