[spv-out] Matrix + Matrix operations not implemented

[parasyte]

WGSL test case:

[[stage(vertex)]]
fn vertex() {
    let model = mat4x4<f32>(
        vec4<f32>(1.0, 0.0, 0.0, 0.0),
        vec4<f32>(0.0, 1.0, 0.0, 0.0),
        vec4<f32>(0.0, 0.0, 1.0, 0.0),
        vec4<f32>(0.0, 0.0, 0.0, 1.0),
    );

    let foo = model + model;
}

Panics at:

naga/src/back/spv/block.rs

Lines 321 to 328 in e69a70b

	crate::BinaryOperator::Add => match *left_ty_inner {
	crate::TypeInner::Scalar { kind, .. }
	| crate::TypeInner::Vector { kind, .. } => match kind {
	crate::ScalarKind::Float => spirv::Op::FAdd,
	_ => spirv::Op::IAdd,
	},
	_ => unimplemented!(),
	},

Workaround by decomposing the matrices into their column vectors and constructing a new matrix after adding the columns separately:

[[stage(vertex)]]
fn vertex() {
    let model = mat4x4<f32>(
        vec4<f32>(1.0, 0.0, 0.0, 0.0),
        vec4<f32>(0.0, 1.0, 0.0, 0.0),
        vec4<f32>(0.0, 0.0, 1.0, 0.0),
        vec4<f32>(0.0, 0.0, 0.0, 1.0),
    );

    let foo = mat4x4<f32>(
        model.x + model.x,
        model.y + model.y,
        model.z + model.z,
        model.w + model.w,
    );
}

I believe this is the transformation required for SpirV.

[kvark]

I don't think WGSL supports that today, see the spec - https://gpuweb.github.io/gpuweb/wgsl/#arithmetic-expr

[parasyte]

Looks like it does, though. In the “Matrix arithmetic” table.

[kvark]

Sorry, you are correct. I'm surprised it's in a separate section.