gfx-rs/wgpu · error

not implemented

Error message

not implemented

What it means

The SPIR-V backend of naga does not implement the `ImageClass::External` texture class (external images, e.g. VK_EXT_external_memory / GL_TEXTURE_EXTERNAL_OES style images). When write_type_declaration_arena processes a type whose inner type is an external image, request_type_capabilities hits an unimplemented!() and panics, aborting shader translation.

Source

Thrown at naga/src/back/spv/writer.rs:2017

            }
        }
    }

    fn request_type_capabilities(&mut self, inner: &crate::TypeInner) -> Result<(), Error> {
        match *inner {
            crate::TypeInner::Image {
                dim,
                arrayed,
                class,
            } => {
                let sampled = match class {
                    crate::ImageClass::Sampled { .. } => true,
                    crate::ImageClass::Depth { .. } => true,
                    crate::ImageClass::Storage { format, .. } => {
                        self.request_image_format_capabilities(format.into())?;
                        false
                    }
                    crate::ImageClass::External => unimplemented!(),
                };

                match dim {
                    crate::ImageDimension::D1 => {
                        if sampled {
                            self.require_any("sampled 1D images", &[spirv::Capability::Sampled1D])?;
                        } else {
                            self.require_any("1D storage images", &[spirv::Capability::Image1D])?;
                        }
                    }
                    crate::ImageDimension::Cube if arrayed => {
                        if sampled {
                            self.require_any(
                                "sampled cube array images",
                                &[spirv::Capability::SampledCubeArray],
                            )?;
                        } else {
                            self.require_any(

View on GitHub (pinned to 3e11ff59bf)

Solutions

1. Replace the external image type with a standard Sampled or Storage image type in the shader/IR before SPIR-V export
2. Use a backend that supports external images (e.g. the GLSL backend, or handle the texture in the app via separate bind types)
3. Check naga releases/issues for external image SPIR-V support and upgrade when available
4. Pre-validate the module: walk crate::Module types and reject ImageClass::External before calling the spv backend

Example fix

// before: IR contains TypeInner::Image { class: ImageClass::External, .. }
// after: rewrite to a sampled image
// let inner = TypeInner::Image { dim, arrayed, class: crate::ImageClass::Sampled { kind: ScalarKind::Float, multi: false } };

fn uses_external_image(module: &naga::Module) -> bool {
    module.types.iter().any(|(_, t)| matches!(t.inner, naga::TypeInner::Image { class: naga::ImageClass::External, .. }))
}
// if uses_external_image(&module) { fall back to another backend or rewrite the type }

fn is_external_image(inner: &naga::TypeInner) -> bool {
    matches!(inner, naga::TypeInner::Image { class: naga::ImageClass::External, .. })
}

// naga returns Result for translation; the panic here aborts the process, so pre-validate instead of catching
let result = naga::back::spv::write_vec(&module, &header, &options, &pipeline);
match result { Ok(spv) => ..., Err(e) => eprintln!("spv export failed: {e}") }

When it happens

Trigger: Compiling a WGSL/GLSL/SPIR-V shader that declares a texture whose naga IR type has ImageClass::External, then translating it to SPIR-V via naga's spv backend.

Common situations: Frontends producing naga IR with external image types (e.g. Android/HAL camera or video surface textures, GL external textures) being fed into the SPIR-V writer; naga's SPIR-V export path simply lacks support for this feature.

Related errors

• not implemented
• Unexpected saturate({:?})
• Unexpected sign({:?})
• Unexpected firstLeadingBit({:?})
• Unexpected pointer expression {:?}

AI-assisted analysis of gfx-rs/wgpu@3e11ff59bf (2026-09-03). Data as JSON: /api/errors/a64f02e5ab4a0655. Report an issue: GitHub.