Here's the resulting WGSL (if I don't validate):

struct type_6 {
    member: f32,
}

var<private> global: vec3<u32>;
@group(0) @binding(0) 
var<storage, read_write> global_1: type_6;
@group(0) @binding(1) 
var global_2: texture_depth_2d;

fn function() {
    let _e5 = global;
    let _e9 = textureLoad(global_2, vec2<u32>(_e5.x, _e5.y));
    global_1.member = vec4(_e9).x;
    return;
}

@compute @workgroup_size(32, 1, 1) 
fn culltest_copy_depth(@builtin(global_invocation_id) param: vec3<u32>) {
    global = param;
    function();
}

This is transpiled from a rust-gpu shader here:

#[spirv(compute(threads(32)))]
pub fn test_copy_depth(
    #[spirv(descriptor_set = 0, binding = 0, storage_buffer)] depth: &mut f32,
    #[spirv(descriptor_set = 0, binding = 1)] depth_texture: &Image!(2D, type=f32, sampled=true, depth=true),
    #[spirv(global_invocation_id)] global_id: UVec3,
) {
    let fetched_depth: Vec4 = depth_texture.fetch(global_id.xy());
    *depth = fetched_depth.x;
}

naga is choking on WGSL also produced by naga... am I being stupid here

    Shader 'fragment' parsing error: expected assignment or increment/decrement, found '_e795'
         ┌─ wgsl:2070:11
         │
    2070 │     const _e795: vec2<f32> = vec2(2f);
         │           ^^^^^ expected assignment or increment/decrement

🤔

only thing I could think of is i'm using trunk naga to generate wgsl, did the syntax change since wgpu 22?

naga is choking on WGSL also produced by naga... am I being stupid here

```
🤔
only thing I could think of is i'm using trunk naga to generate the input wgsl, did the syntax change since wgpu 22?

naga is choking on WGSL also produced by naga... am I being stupid here

    Shader 'fragment' parsing error: expected assignment or increment/decrement, found '_e795'
         ┌─ wgsl:2070:11
         │
    2070 │     const _e795: vec2<f32> = vec2(2f);
         │           ^^^^^ expected assignment or increment/decrement


🤔
only thing I could think of is i'm using trunk naga to generate the input wgsl, did the syntax change since wgpu 22?

In reply to @chyyran:matrix.org
switched everything to trunk wgpu and it works so I'm assuming it's a syntax update 🙃