fix double free

shaders can be either a path or a string
chore: Release
2024-02-07 15:52:36 +11:00 · 2024-02-07 15:52:36 +11:00 · 2024-02-06 21:41:00 -05:00 · 2024-02-06 21:38:17 -05:00 · 2024-02-06 21:38:17 -05:00 · 2024-02-06 21:31:00 -05:00
105 changed files with 7641 additions and 773 deletions
--- a/.github/workflows/build-linux-arm64.yml
+++ b/.github/workflows/build-linux-arm64.yml
@ -29,10 +29,11 @@ jobs:
    - uses: actions/setup-python@v1
      name: Setup Python 3.11
      with:
-        python-version: '3.11.6'
+        python-version: '3.11'
    - name: Install ARM64 cross-compilation dependencies
      continue-on-error: true
      run: |
+        sudo apt-get update || true
        sudo dpkg --add-architecture arm64
        echo "deb [arch=arm64] http://ports.ubuntu.com/ubuntu-ports jammy main restricted" | sudo tee -a /etc/apt/sources.list
        echo "deb [arch=arm64] http://ports.ubuntu.com/ubuntu-ports jammy-updates main restricted" | sudo tee -a /etc/apt/sources.list
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@ -29,15 +29,12 @@ jobs:
    - uses: actions/setup-python@v1
      name: Setup Python 3.11
      with:
-        python-version: '3.11.6'
+        python-version: '3.11'
    - name: Install Vulkan SDK
      uses: humbletim/install-vulkan-sdk@v1.1.1
      with:
        version: latest
        cache: true
-    - if: runner.os == 'Windows'
-      run: pip install meson ninja mako
-      name: Install Meson for spirv-to-dxil-sys
    - name: Build dynamic library
      run: cargo run -p librashader-build-script -- --profile ${{ matrix.profile }}
    - name: Upload build artifacts
--- a/.github/workflows/package-obs.yml
+++ b/.github/workflows/package-obs.yml
@ -0,0 +1,48 @@
+name: build Linux packages with Open Build Service
+
+on:
+  pull_request:
+    branches: [ "master" ]
+env:
+  CARGO_TERM_COLOR: always
+
+jobs:
+  build-obs-binary:
+    strategy:
+      matrix:
+        include:
+          - repo: Fedora_39
+            spec: librashader.spec
+          - repo: xUbuntu_23.10
+            spec: librashader.spec
+          - repo: Arch
+            spec: PKGBUILD
+    runs-on: ubuntu-latest
+    container: 
+      image: fedora:39
+      options: --privileged
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+      - name: Install OSC and dependencies
+        env:
+          OBS_CONFIG: ${{ secrets.OBS_CONFIG }}
+        run: |
+          sudo dnf install -y osc obs-service-obs_scm obs-service-cargo_vendor cargo python3-zstandard
+          mkdir -p ~/.config/osc
+          echo "$OBS_CONFIG" > ~/.config/osc/oscrc
+      - name: Checkout Open Build Service repository
+        run: |
+          osc co home:chyyran:librashader/librashader
+      - name: Copy spec from repository
+        run: |
+          cp -r ./pkg/obs/ home:chyyran:librashader/librashader/
+          sed -r -i 's/(<param name="revision">)(.+)(<\/param>)/<param name="revision">${{ github.sha }}<\/param>/g' home:chyyran:librashader/librashader/_service
+      - name: Vendor sources for Open Build Service
+        run: |
+          cd home:chyyran:librashader/librashader/
+          osc service mr
+      - name: Build package
+        run: |
+          cd home:chyyran:librashader/librashader/
+          osc build --no-verify --trust-all-projects ${{ matrix.repo }} x86_64 ${{ matrix.spec }}
--- a/.github/workflows/publish-obs.yml
+++ b/.github/workflows/publish-obs.yml
@ -0,0 +1,80 @@
+name: publish packages using Open Build Service
+
+on:
+  push:
+    branches: [ "master" ]
+env:
+  CARGO_TERM_COLOR: always
+
+jobs:
+  build-obs-binary:
+    strategy:
+      matrix:
+        include:
+          - repo: Fedora_39
+            spec: librashader.spec
+          - repo: xUbuntu_23.10
+            spec: librashader.spec
+          - repo: Arch
+            spec: PKGBUILD
+    runs-on: ubuntu-latest
+    container: 
+      image: fedora:39
+      options: --privileged
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+      - name: Install OSC and dependencies
+        env:
+          OBS_CONFIG: ${{ secrets.OBS_CONFIG }}
+        run: |
+          sudo dnf install -y osc obs-service-obs_scm obs-service-cargo_vendor cargo python3-zstandard
+          mkdir -p ~/.config/osc
+          echo "$OBS_CONFIG" > ~/.config/osc/oscrc
+      - name: Checkout Open Build Service repository
+        run: |
+          osc co home:chyyran:librashader/librashader
+      - name: Copy spec from repository
+        run: |
+          cp -r ./pkg/obs/ home:chyyran:librashader/librashader/
+          sed -r -i 's/(<param name="revision">)(.+)(<\/param>)/<param name="revision">${{ github.sha }}<\/param>/g' home:chyyran:librashader/librashader/_service
+      - name: Vendor sources for Open Build Service
+        run: |
+          cd home:chyyran:librashader/librashader/
+          rm *.obscpio
+          osc service mr
+      - name: Build package
+        run: |
+          cd home:chyyran:librashader/librashader/
+          osc build --no-verify --trust-all-projects ${{ matrix.repo }} x86_64 ${{ matrix.spec }}
+  publish-obs:
+    needs: build-obs-binary
+    runs-on: ubuntu-latest
+    container: fedora:39
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+      - name: Install OSC and dependencies
+        env:
+          OBS_CONFIG: ${{ secrets.OBS_CONFIG }}
+        run: |
+          sudo dnf install -y osc obs-service-obs_scm obs-service-cargo_vendor cargo python3-zstandard
+          mkdir -p ~/.config/osc
+          echo "$OBS_CONFIG" > ~/.config/osc/oscrc
+      - name: Checkout Open Build Service repository
+        run: |
+          osc co home:chyyran:librashader/librashader
+      - name: Copy spec from repository
+        run: |
+          cp -r ./pkg/obs/ home:chyyran:librashader/librashader/
+          sed -r -i 's/(<param name="revision">)(.+)(<\/param>)/<param name="revision">${{ github.sha }}<\/param>/g' home:chyyran:librashader/librashader/_service
+      - name: Vendor sources for Open Build Service
+        run: |
+          cd home:chyyran:librashader/librashader/
+          rm *.obscpio
+          osc service mr
+      - name: Commit source artifacts to Open Build Service
+        run: |
+          cd home:chyyran:librashader/librashader/
+          osc ar
+          osc commit -f -m "git-rev ${{ github.sha }}"
--- a/.gitignore
+++ b/.gitignore
@ -5,6 +5,7 @@
 *.rdc
 *.cap
 /.vs/
+.idea/
 librashader_runtime_*.exe
 /test/capi-tests/librashader-capi-tests/.vs/
 /test/capi-tests/librashader-capi-tests/x64/
--- a/.idea/.gitignore
+++ b/.idea/.gitignore
@ -1,8 +0,0 @@
-# Default ignored files
-/shelf/
-/workspace.xml
-# Editor-based HTTP Client requests
-/httpRequests/
-# Datasource local storage ignored files
-/dataSources/
-/dataSources.local.xml
--- a/.idea/misc.xml
+++ b/.idea/misc.xml
@ -1,6 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<project version="4">
-  <component name="CMakeWorkspace">
-    <contentRoot DIR="$PROJECT_DIR$" />
-  </component>
-</project>
--- a/.idea/modules.xml
+++ b/.idea/modules.xml
@ -1,8 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<project version="4">
-  <component name="ProjectModuleManager">
-    <modules>
-      <module fileurl="file://$PROJECT_DIR$/.idea/src.iml" filepath="$PROJECT_DIR$/.idea/src.iml" />
-    </modules>
-  </component>
-</project>
--- a/.idea/src.iml
+++ b/.idea/src.iml
@ -1,2 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<module classpath="CMake" type="CPP_MODULE" version="4" />
--- a/.idea/vcs.xml
+++ b/.idea/vcs.xml
@ -1,7 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<project version="4">
-  <component name="VcsDirectoryMappings">
-    <mapping directory="$PROJECT_DIR$/.." vcs="Git" />
-    <mapping directory="$PROJECT_DIR$/test/shaders_slang" vcs="Git" />
-  </component>
-</project>
--- a/Cargo.lock
+++ b/Cargo.lock
--- a/Cargo.toml
+++ b/Cargo.toml
@ -13,7 +13,7 @@ members = [
    "librashader-cache",
    "librashader-capi",
    "librashader-build-script"
-]
+, "librashader-runtime-wgpu"]
 resolver = "2"

 [workspace.metadata.release]
--- a/README.md
+++ b/README.md
@ -6,9 +6,13 @@

 librashader (*/ˈli:brəʃeɪdɚ/*) is a preprocessor, compiler, and runtime for RetroArch 'slang' shaders, rewritten in pure Rust.

-[![Latest Version](https://img.shields.io/crates/v/librashader.svg)](https://crates.io/crates/librashader) [![Docs](https://docs.rs/librashader/badge.svg)](https://docs.rs/librashader) ![License](https://img.shields.io/crates/l/librashader)
+[![Latest Version](https://img.shields.io/crates/v/librashader.svg)](https://crates.io/crates/librashader) [![Docs](https://docs.rs/librashader/badge.svg)](https://docs.rs/librashader) [![build result](https://build.opensuse.org/projects/home:chyyran:librashader/packages/librashader/badge.svg)](https://software.opensuse.org//download.html?project=home%3Achyyran%3Alibrashader&package=librashader)
+![License](https://img.shields.io/crates/l/librashader)
 ![Nightly rust](https://img.shields.io/badge/rust-nightly-orange.svg) 

+## Installation
+For end-users, librashader is available from the [Open Build Service](https://software.opensuse.org//download.html?project=home%3Achyyran%3Alibrashader&package=librashader) for a variety of Linux distributions and platforms. Windows users can grab the latest DLL from [GitHub Releases](https://github.com/SnowflakePowered/librashader/releases).
+
 ## Supported Render APIs
 librashader supports OpenGL 3, OpenGL 4.6, Vulkan, Direct3D 11, and Direct3D 12.  Metal and WebGPU
 are not currently supported (but pull-requests are welcome). librashader does not support legacy render
@ -208,6 +212,8 @@ will check to ensure that `LIBRASHADER_CURRENT_ABI` matches that of the loaded l
 librashader will not load. A value of `0` for `LIBRASHADER_CURRENT_ABI` indicates the "null" instance where every operation
 is a no-op, which occurs if no compatible librashader implementation could be found.

+The `SONAME` of `librashader.so` when installed via package manager is set to `LIBRASHADER_CURRENT_ABI`.
+
 The above does not apply to releases of librashader prior to `0.1.0`, which were allowed to break API and ABI compatibility
 in both the Rust and C API without an increase to either `LIBRASHADER_CURRENT_VERSION` or `LIBRASHADER_CURRENT_ABI`.

--- a/librashader-build-script/Cargo.toml
+++ b/librashader-build-script/Cargo.toml
@ -6,7 +6,7 @@ publish = false


 [dependencies]
-cbindgen = { git = "https://github.com/eqrion/cbindgen" }
+cbindgen = "0.26.0"
 clap = { version = "4.1.0", features = ["derive"] }


--- a/librashader-build-script/src/main.rs
+++ b/librashader-build-script/src/main.rs
@ -36,10 +36,7 @@ pub fn main() {
    ));

    if let Some(target) = &args.target {
-        cmd.arg(format!(
-            "--target={}",
-            &target
-        ));
+        cmd.arg(format!("--target={}", &target));
    }

    Some(cmd.status().expect("Failed to build librashader-capi"));
@ -53,7 +50,6 @@ pub fn main() {
        .canonicalize()
        .expect("Could not find output directory.");

-        
    println!("Generating C headers...");

    // Create headers.
@ -86,11 +82,20 @@ pub fn main() {
            "librashader_capi.d",
            "librashader_capi.dll.exp",
            "librashader_capi.dll.lib",
-            "librashader_capi.pdb",
        ];
        for artifact in artifacts {
            let ext = artifact.replace("_capi", "");
+            println!("Renaming {artifact} to {ext}");
            fs::rename(output_dir.join(artifact), output_dir.join(ext)).unwrap();
        }
+
+        if output_dir.join("librashader_capi.pdb").exists() {
+            println!("Renaming librashader_capi.pdb to librashader.pdb");
+            fs::rename(
+                output_dir.join("librashader_capi.pdb"),
+                output_dir.join("librashader.pdb"),
+            )
+            .unwrap();
+        }
    }
 }
--- a/librashader-cache/Cargo.toml
+++ b/librashader-cache/Cargo.toml
@ -2,7 +2,7 @@
 name = "librashader-cache"
 edition = "2021"
 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -12,8 +12,8 @@ description = "RetroArch shaders for all."

 [dependencies]
 serde = { version = "1.0" }
-librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.2", features = ["serialize"] }
-librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.2" }
+librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.5", features = ["serialize"] }
+librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.5" }
 platform-dirs = "0.3.0"
 blake3 = { version = "1.3.3" }
 thiserror = "1.0.38"
--- a/librashader-cache/src/compilation.rs
+++ b/librashader-cache/src/compilation.rs
@ -1,6 +1,9 @@
 //!  Cache helpers for `ShaderCompilation` objects to cache compiled SPIRV.
 use librashader_preprocess::ShaderSource;
-use librashader_reflect::back::targets::{DXIL, GLSL, HLSL, SPIRV};
+#[cfg(all(target_os = "windows", feature = "d3d"))]
+use librashader_reflect::back::targets::DXIL;
+use librashader_reflect::back::targets::{GLSL, HLSL, SPIRV};
+
 use librashader_reflect::back::{CompilerBackend, FromCompilation};
 use librashader_reflect::error::{ShaderCompileError, ShaderReflectError};
 use librashader_reflect::front::{GlslangCompilation, ShaderCompilation};
--- a/librashader-capi/Cargo.toml
+++ b/librashader-capi/Cargo.toml
@ -3,7 +3,7 @@ name = "librashader-capi"
 edition = "2021"

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -23,7 +23,7 @@ runtime-d3d12 = ["windows", "librashader/runtime-d3d12", "windows/Win32_Graphics
 runtime-vulkan = ["ash", "librashader/runtime-vk"]

 [dependencies]
-librashader = { path = "../librashader", version = "0.2.0-beta.2", features = ["internal"] }
+librashader = { path = "../librashader", version = "0.2.0-beta.5", features = ["internal"] }
 thiserror = "1.0.37"
 paste = "1.0.9"
 gl = { version = "0.14.0", optional = true }
--- a/librashader-common/Cargo.toml
+++ b/librashader-common/Cargo.toml
@ -3,7 +3,7 @@ name = "librashader-common"
 edition = "2021"

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -18,10 +18,12 @@ d3d11 = ["windows", "dxgi"]
 d3d12 = ["windows", "dxgi"]
 dxgi = ["windows"]
 vulkan = ["ash"]
+wgpu = ["wgpu-types"]

 [dependencies]
 gl = { version = "0.14.0", optional = true }
 ash = { version = "0.37", optional = true }
+wgpu-types = { version = "0.19.0", optional = true }

 num-traits = "0.2.15"

--- a/librashader-common/src/lib.rs
+++ b/librashader-common/src/lib.rs
@ -8,6 +8,10 @@ pub mod gl;
 #[cfg(feature = "vulkan")]
 pub mod vk;

+/// WGPU common conversions.
+#[cfg(feature = "wgpu")]
+pub mod wgpu;
+
 /// DXGI common conversions.
 #[cfg(all(target_os = "windows", feature = "dxgi"))]
 pub mod dxgi;
@ -21,6 +25,7 @@ pub mod d3d11;
 pub mod d3d12;

 mod viewport;
+
 pub use viewport::Viewport;

 use num_traits::AsPrimitive;
--- a/librashader-common/src/wgpu.rs
+++ b/librashader-common/src/wgpu.rs
@ -0,0 +1,177 @@
+use crate::{FilterMode, ImageFormat, Size, WrapMode};
+
+impl From<ImageFormat> for Option<wgpu_types::TextureFormat> {
+    fn from(format: ImageFormat) -> Self {
+        match format {
+            ImageFormat::Unknown => None,
+            ImageFormat::R8Unorm => Some(wgpu_types::TextureFormat::R8Unorm),
+            ImageFormat::R8Uint => Some(wgpu_types::TextureFormat::R8Uint),
+            ImageFormat::R8Sint => Some(wgpu_types::TextureFormat::R8Sint),
+            ImageFormat::R8G8Unorm => Some(wgpu_types::TextureFormat::Rg8Unorm),
+            ImageFormat::R8G8Uint => Some(wgpu_types::TextureFormat::Rg8Uint),
+            ImageFormat::R8G8Sint => Some(wgpu_types::TextureFormat::Rg8Sint),
+            ImageFormat::R8G8B8A8Unorm => Some(wgpu_types::TextureFormat::Rgba8Unorm),
+            ImageFormat::R8G8B8A8Uint => Some(wgpu_types::TextureFormat::Rgba8Uint),
+            ImageFormat::R8G8B8A8Sint => Some(wgpu_types::TextureFormat::Rgba8Sint),
+            ImageFormat::R8G8B8A8Srgb => Some(wgpu_types::TextureFormat::Rgba8UnormSrgb),
+            ImageFormat::A2B10G10R10UnormPack32 => Some(wgpu_types::TextureFormat::Rgb10a2Unorm),
+            ImageFormat::A2B10G10R10UintPack32 => Some(wgpu_types::TextureFormat::Rgb10a2Uint),
+            ImageFormat::R16Uint => Some(wgpu_types::TextureFormat::R16Uint),
+            ImageFormat::R16Sint => Some(wgpu_types::TextureFormat::R16Sint),
+            ImageFormat::R16Sfloat => Some(wgpu_types::TextureFormat::R16Float),
+            ImageFormat::R16G16Uint => Some(wgpu_types::TextureFormat::Rg16Uint),
+            ImageFormat::R16G16Sint => Some(wgpu_types::TextureFormat::Rg16Sint),
+            ImageFormat::R16G16Sfloat => Some(wgpu_types::TextureFormat::Rg16Float),
+            ImageFormat::R16G16B16A16Uint => Some(wgpu_types::TextureFormat::Rgba16Uint),
+            ImageFormat::R16G16B16A16Sint => Some(wgpu_types::TextureFormat::Rgba16Sint),
+            ImageFormat::R16G16B16A16Sfloat => Some(wgpu_types::TextureFormat::Rgba16Float),
+            ImageFormat::R32Uint => Some(wgpu_types::TextureFormat::R32Uint),
+            ImageFormat::R32Sint => Some(wgpu_types::TextureFormat::R32Sint),
+            ImageFormat::R32Sfloat => Some(wgpu_types::TextureFormat::R32Float),
+            ImageFormat::R32G32Uint => Some(wgpu_types::TextureFormat::Rg32Uint),
+            ImageFormat::R32G32Sint => Some(wgpu_types::TextureFormat::Rg32Sint),
+            ImageFormat::R32G32Sfloat => Some(wgpu_types::TextureFormat::Rg32Float),
+            ImageFormat::R32G32B32A32Uint => Some(wgpu_types::TextureFormat::Rgba32Uint),
+            ImageFormat::R32G32B32A32Sint => Some(wgpu_types::TextureFormat::Rgba32Sint),
+            ImageFormat::R32G32B32A32Sfloat => Some(wgpu_types::TextureFormat::Rgba32Float),
+        }
+    }
+}
+
+impl From<wgpu_types::TextureFormat> for ImageFormat {
+    fn from(format: wgpu_types::TextureFormat) -> Self {
+        match format {
+            wgpu_types::TextureFormat::R8Unorm => ImageFormat::R8Unorm,
+            wgpu_types::TextureFormat::R8Uint => ImageFormat::R8Uint,
+            wgpu_types::TextureFormat::R8Sint => ImageFormat::R8Sint,
+            wgpu_types::TextureFormat::Rg8Unorm => ImageFormat::R8G8Unorm,
+            wgpu_types::TextureFormat::Rg8Uint => ImageFormat::R8G8Uint,
+            wgpu_types::TextureFormat::Rg8Sint => ImageFormat::R8G8Sint,
+            wgpu_types::TextureFormat::Rgba8Unorm => ImageFormat::R8G8B8A8Unorm,
+            wgpu_types::TextureFormat::Rgba8Uint => ImageFormat::R8G8B8A8Uint,
+            wgpu_types::TextureFormat::Rgba8Sint => ImageFormat::R8G8B8A8Sint,
+            wgpu_types::TextureFormat::Rgba8UnormSrgb => ImageFormat::R8G8B8A8Srgb,
+            wgpu_types::TextureFormat::Rgb10a2Unorm => ImageFormat::A2B10G10R10UnormPack32,
+            wgpu_types::TextureFormat::Rgb10a2Uint => ImageFormat::A2B10G10R10UintPack32,
+            wgpu_types::TextureFormat::R16Uint => ImageFormat::R16Uint,
+            wgpu_types::TextureFormat::R16Sint => ImageFormat::R16Sint,
+            wgpu_types::TextureFormat::R16Float => ImageFormat::R16Sfloat,
+            wgpu_types::TextureFormat::Rg16Uint => ImageFormat::R16G16Uint,
+            wgpu_types::TextureFormat::Rg16Sint => ImageFormat::R16G16Sint,
+            wgpu_types::TextureFormat::Rg16Float => ImageFormat::R16G16Sfloat,
+            wgpu_types::TextureFormat::Rgba16Uint => ImageFormat::R16G16B16A16Uint,
+            wgpu_types::TextureFormat::Rgba16Sint => ImageFormat::R16G16B16A16Sint,
+            wgpu_types::TextureFormat::Rgba16Float => ImageFormat::R16G16B16A16Sfloat,
+            wgpu_types::TextureFormat::R32Uint => ImageFormat::R32Uint,
+            wgpu_types::TextureFormat::R32Sint => ImageFormat::R32Sint,
+            wgpu_types::TextureFormat::R32Float => ImageFormat::R32Sfloat,
+            wgpu_types::TextureFormat::Rg32Uint => ImageFormat::R32G32Uint,
+            wgpu_types::TextureFormat::Rg32Sint => ImageFormat::R32G32Sint,
+            wgpu_types::TextureFormat::Rg32Float => ImageFormat::R32G32Sfloat,
+            wgpu_types::TextureFormat::Rgba32Uint => ImageFormat::R32G32B32A32Uint,
+            wgpu_types::TextureFormat::Rgba32Sint => ImageFormat::R32G32B32A32Sint,
+            wgpu_types::TextureFormat::Rgba32Float => ImageFormat::R32G32B32A32Sfloat,
+            _ => ImageFormat::Unknown,
+        }
+    }
+}
+
+impl From<Option<wgpu_types::TextureFormat>> for ImageFormat {
+    fn from(format: Option<wgpu_types::TextureFormat>) -> Self {
+        let Some(format) = format else {
+            return ImageFormat::Unknown;
+        };
+        ImageFormat::from(format)
+    }
+}
+
+impl From<wgpu_types::Extent3d> for Size<u32> {
+    fn from(value: wgpu_types::Extent3d) -> Self {
+        Size {
+            width: value.width,
+            height: value.height,
+        }
+    }
+}
+
+//
+// impl From<Size<u32>> for vk::Extent2D {
+//     fn from(value: Size<u32>) -> Self {
+//         vk::Extent2D {
+//             width: value.width,
+//             height: value.height,
+//         }
+//     }
+// }
+//
+
+//
+// impl From<vk::Extent2D> for Size<u32> {
+//     fn from(value: vk::Extent2D) -> Self {
+//         Size {
+//             width: value.width,
+//             height: value.height,
+//         }
+//     }
+// }
+//
+// impl From<vk::Viewport> for Size<u32> {
+//     fn from(value: vk::Viewport) -> Self {
+//         Size {
+//             width: value.width as u32,
+//             height: value.height as u32,
+//         }
+//     }
+// }
+//
+// impl From<&vk::Viewport> for Size<u32> {
+//     fn from(value: &vk::Viewport) -> Self {
+//         Size {
+//             width: value.width as u32,
+//             height: value.height as u32,
+//         }
+//     }
+// }
+//
+// impl From<Size<u32>> for vk::Viewport {
+//     fn from(value: Size<u32>) -> Self {
+//         vk::Viewport {
+//             x: 0.0,
+//             y: 0.0,
+//             width: value.width as f32,
+//             height: value.height as f32,
+//             min_depth: 0.0,
+//             max_depth: 1.0,
+//         }
+//     }
+// }
+
+impl From<FilterMode> for wgpu_types::FilterMode {
+    fn from(value: FilterMode) -> Self {
+        match value {
+            FilterMode::Linear => wgpu_types::FilterMode::Linear,
+            FilterMode::Nearest => wgpu_types::FilterMode::Nearest,
+        }
+    }
+}
+
+impl From<WrapMode> for wgpu_types::AddressMode {
+    fn from(value: WrapMode) -> Self {
+        match value {
+            WrapMode::ClampToBorder => wgpu_types::AddressMode::ClampToBorder,
+            WrapMode::ClampToEdge => wgpu_types::AddressMode::ClampToEdge,
+            WrapMode::Repeat => wgpu_types::AddressMode::Repeat,
+            WrapMode::MirroredRepeat => wgpu_types::AddressMode::MirrorRepeat,
+        }
+    }
+}
+
+impl From<Size<u32>> for wgpu_types::Extent3d {
+    fn from(value: Size<u32>) -> Self {
+        wgpu_types::Extent3d {
+            width: value.width,
+            height: value.height,
+            depth_or_array_layers: 1,
+        }
+    }
+}
--- a/librashader-preprocess/Cargo.toml
+++ b/librashader-preprocess/Cargo.toml
@ -3,7 +3,7 @@ name = "librashader-preprocess"
 edition = "2021"

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -14,7 +14,7 @@ description = "RetroArch shaders for all."
 [dependencies]
 thiserror = "1.0.37"
 nom = "7.1.1"
-librashader-common = { path = "../librashader-common", version = "0.2.0-beta.2" }
+librashader-common = { path = "../librashader-common", version = "0.2.0-beta.5" }
 rustc-hash = "1.1.0"
 encoding_rs = "0.8.31"

--- a/librashader-presets/Cargo.toml
+++ b/librashader-presets/Cargo.toml
@ -3,7 +3,7 @@ name = "librashader-presets"
 edition = "2021"

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -15,7 +15,7 @@ description = "RetroArch shaders for all."
 thiserror = "1.0.37"
 nom = "7.1.1"
 nom_locate = "4.0.0"
-librashader-common = { path = "../librashader-common", version = "0.2.0-beta.2" }
+librashader-common = { path = "../librashader-common", version = "0.2.0-beta.5" }
 num-traits = "0.2"

 [features]
--- a/librashader-presets/src/extract_if.rs
+++ b/librashader-presets/src/extract_if.rs
@ -0,0 +1,312 @@
+//! This is a stable polyfill for [`Vec::extract_if`](https://github.com/rust-lang/rust/issues/43244).
+
+use core::ptr;
+use core::slice;
+
+/// Polyfill trait for [`Vec::extract_if`](https://github.com/rust-lang/rust/issues/43244).
+pub(crate) trait MakeExtractIf<T> {
+    /// Creates an iterator which uses a closure to determine if an element should be removed.
+    ///
+    /// If the closure returns true, then the element is removed and yielded.
+    /// If the closure returns false, the element will remain in the vector and will not be yielded
+    /// by the iterator.
+    ///
+    /// If the returned `ExtractIf` is not exhausted, e.g. because it is dropped without iterating
+    /// or the iteration short-circuits, then the remaining elements will be retained.
+    ///
+    /// Note that `extract_if` also lets you mutate every element in the filter closure,
+    /// regardless of whether you choose to keep or remove it.
+    ///
+    /// # Examples
+    ///
+    /// Splitting an array into evens and odds, reusing the original allocation:
+    ///
+    /// ```
+    /// use vec_extract_if_polyfill::MakeExtractIf;
+    /// let mut numbers = vec![1, 2, 3, 4, 5, 6, 8, 9, 11, 13, 14, 15];
+    ///
+    /// let evens = numbers.extract_if(|x| *x % 2 == 0).collect::<Vec<_>>();
+    /// let odds = numbers;
+    ///
+    /// assert_eq!(evens, vec![2, 4, 6, 8, 14]);
+    /// assert_eq!(odds, vec![1, 3, 5, 9, 11, 13, 15]);
+    /// ```
+    fn extract_if<F>(&mut self, filter: F) -> ExtractIf<T, F>
+    where
+        F: FnMut(&mut T) -> bool;
+}
+
+impl<T> MakeExtractIf<T> for Vec<T> {
+    fn extract_if<F>(&mut self, filter: F) -> ExtractIf<T, F>
+    where
+        F: FnMut(&mut T) -> bool,
+    {
+        let old_len = self.len();
+
+        // Guard against us getting leaked (leak amplification)
+        unsafe {
+            self.set_len(0);
+        }
+
+        ExtractIf {
+            vec: self,
+            idx: 0,
+            del: 0,
+            old_len,
+            pred: filter,
+        }
+    }
+}
+/// An iterator which uses a closure to determine if an element should be removed.
+///
+/// This struct is created by [`Vec::extract_if`].
+/// See its documentation for more.
+///
+/// # Example
+///
+/// ```
+/// use vec_extract_if_polyfill::MakeExtractIf;
+///
+/// let mut v = vec![0, 1, 2];
+/// let iter = v.extract_if(|x| *x % 2 == 0);
+/// ```
+#[derive(Debug)]
+#[must_use = "iterators are lazy and do nothing unless consumed"]
+pub struct ExtractIf<'a, T, F>
+where
+    F: FnMut(&mut T) -> bool,
+{
+    vec: &'a mut Vec<T>,
+    /// The index of the item that will be inspected by the next call to `next`.
+    idx: usize,
+    /// The number of items that have been drained (removed) thus far.
+    del: usize,
+    /// The original length of `vec` prior to draining.
+    old_len: usize,
+    /// The filter test predicate.
+    pred: F,
+}
+
+impl<T, F> Iterator for ExtractIf<'_, T, F>
+where
+    F: FnMut(&mut T) -> bool,
+{
+    type Item = T;
+
+    fn next(&mut self) -> Option<T> {
+        unsafe {
+            while self.idx < self.old_len {
+                let i = self.idx;
+                let v = slice::from_raw_parts_mut(self.vec.as_mut_ptr(), self.old_len);
+                let drained = (self.pred)(&mut v[i]);
+                // Update the index *after* the predicate is called. If the index
+                // is updated prior and the predicate panics, the element at this
+                // index would be leaked.
+                self.idx += 1;
+                if drained {
+                    self.del += 1;
+                    return Some(ptr::read(&v[i]));
+                } else if self.del > 0 {
+                    let del = self.del;
+                    let src: *const T = &v[i];
+                    let dst: *mut T = &mut v[i - del];
+                    ptr::copy_nonoverlapping(src, dst, 1);
+                }
+            }
+            None
+        }
+    }
+
+    fn size_hint(&self) -> (usize, Option<usize>) {
+        (0, Some(self.old_len - self.idx))
+    }
+}
+
+impl<T, F> Drop for ExtractIf<'_, T, F>
+where
+    F: FnMut(&mut T) -> bool,
+{
+    fn drop(&mut self) {
+        unsafe {
+            if self.idx < self.old_len && self.del > 0 {
+                // This is a pretty messed up state, and there isn't really an
+                // obviously right thing to do. We don't want to keep trying
+                // to execute `pred`, so we just backshift all the unprocessed
+                // elements and tell the vec that they still exist. The backshift
+                // is required to prevent a double-drop of the last successfully
+                // drained item prior to a panic in the predicate.
+                let ptr = self.vec.as_mut_ptr();
+                let src = ptr.add(self.idx);
+                let dst = src.sub(self.del);
+                let tail_len = self.old_len - self.idx;
+                src.copy_to(dst, tail_len);
+            }
+            self.vec.set_len(self.old_len - self.del);
+        }
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use crate::MakeExtractIf;
+    #[test]
+    fn drain_filter_empty() {
+        let mut vec: Vec<i32> = vec![];
+
+        {
+            let mut iter = vec.extract_if(|_| true);
+            assert_eq!(iter.size_hint(), (0, Some(0)));
+            assert_eq!(iter.next(), None);
+            assert_eq!(iter.size_hint(), (0, Some(0)));
+            assert_eq!(iter.next(), None);
+            assert_eq!(iter.size_hint(), (0, Some(0)));
+        }
+        assert_eq!(vec.len(), 0);
+        assert_eq!(vec, vec![]);
+    }
+
+    #[test]
+    fn drain_filter_zst() {
+        let mut vec = vec![(), (), (), (), ()];
+        let initial_len = vec.len();
+        let mut count = 0;
+        {
+            let mut iter = vec.extract_if(|_| true);
+            assert_eq!(iter.size_hint(), (0, Some(initial_len)));
+            while let Some(_) = iter.next() {
+                count += 1;
+                assert_eq!(iter.size_hint(), (0, Some(initial_len - count)));
+            }
+            assert_eq!(iter.size_hint(), (0, Some(0)));
+            assert_eq!(iter.next(), None);
+            assert_eq!(iter.size_hint(), (0, Some(0)));
+        }
+
+        assert_eq!(count, initial_len);
+        assert_eq!(vec.len(), 0);
+        assert_eq!(vec, vec![]);
+    }
+
+    #[test]
+    fn drain_filter_false() {
+        let mut vec = vec![1, 2, 3, 4, 5, 6, 7, 8, 9, 10];
+
+        let initial_len = vec.len();
+        let mut count = 0;
+        {
+            let mut iter = vec.extract_if(|_| false);
+            assert_eq!(iter.size_hint(), (0, Some(initial_len)));
+            for _ in iter.by_ref() {
+                count += 1;
+            }
+            assert_eq!(iter.size_hint(), (0, Some(0)));
+            assert_eq!(iter.next(), None);
+            assert_eq!(iter.size_hint(), (0, Some(0)));
+        }
+
+        assert_eq!(count, 0);
+        assert_eq!(vec.len(), initial_len);
+        assert_eq!(vec, vec![1, 2, 3, 4, 5, 6, 7, 8, 9, 10]);
+    }
+
+    #[test]
+    fn drain_filter_true() {
+        let mut vec = vec![1, 2, 3, 4, 5, 6, 7, 8, 9, 10];
+
+        let initial_len = vec.len();
+        let mut count = 0;
+        {
+            let mut iter = vec.extract_if(|_| true);
+            assert_eq!(iter.size_hint(), (0, Some(initial_len)));
+            while let Some(_) = iter.next() {
+                count += 1;
+                assert_eq!(iter.size_hint(), (0, Some(initial_len - count)));
+            }
+            assert_eq!(iter.size_hint(), (0, Some(0)));
+            assert_eq!(iter.next(), None);
+            assert_eq!(iter.size_hint(), (0, Some(0)));
+        }
+
+        assert_eq!(count, initial_len);
+        assert_eq!(vec.len(), 0);
+        assert_eq!(vec, vec![]);
+    }
+
+    #[test]
+    fn drain_filter_complex() {
+        {
+            //                [+xxx++++++xxxxx++++x+x++]
+            let mut vec = vec![
+                1, 2, 4, 6, 7, 9, 11, 13, 15, 17, 18, 20, 22, 24, 26, 27, 29, 31, 33, 34, 35, 36,
+                37, 39,
+            ];
+
+            let removed = vec.extract_if(|x| *x % 2 == 0).collect::<Vec<_>>();
+            assert_eq!(removed.len(), 10);
+            assert_eq!(removed, vec![2, 4, 6, 18, 20, 22, 24, 26, 34, 36]);
+
+            assert_eq!(vec.len(), 14);
+            assert_eq!(
+                vec,
+                vec![1, 7, 9, 11, 13, 15, 17, 27, 29, 31, 33, 35, 37, 39]
+            );
+        }
+
+        {
+            //                [xxx++++++xxxxx++++x+x++]
+            let mut vec = vec![
+                2, 4, 6, 7, 9, 11, 13, 15, 17, 18, 20, 22, 24, 26, 27, 29, 31, 33, 34, 35, 36, 37,
+                39,
+            ];
+
+            let removed = vec.extract_if(|x| *x % 2 == 0).collect::<Vec<_>>();
+            assert_eq!(removed.len(), 10);
+            assert_eq!(removed, vec![2, 4, 6, 18, 20, 22, 24, 26, 34, 36]);
+
+            assert_eq!(vec.len(), 13);
+            assert_eq!(vec, vec![7, 9, 11, 13, 15, 17, 27, 29, 31, 33, 35, 37, 39]);
+        }
+
+        {
+            //                [xxx++++++xxxxx++++x+x]
+            let mut vec = vec![
+                2, 4, 6, 7, 9, 11, 13, 15, 17, 18, 20, 22, 24, 26, 27, 29, 31, 33, 34, 35, 36,
+            ];
+
+            let removed = vec.extract_if(|x| *x % 2 == 0).collect::<Vec<_>>();
+            assert_eq!(removed.len(), 10);
+            assert_eq!(removed, vec![2, 4, 6, 18, 20, 22, 24, 26, 34, 36]);
+
+            assert_eq!(vec.len(), 11);
+            assert_eq!(vec, vec![7, 9, 11, 13, 15, 17, 27, 29, 31, 33, 35]);
+        }
+
+        {
+            //                [xxxxxxxxxx+++++++++++]
+            let mut vec = vec![
+                2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 1, 3, 5, 7, 9, 11, 13, 15, 17, 19,
+            ];
+
+            let removed = vec.extract_if(|x| *x % 2 == 0).collect::<Vec<_>>();
+            assert_eq!(removed.len(), 10);
+            assert_eq!(removed, vec![2, 4, 6, 8, 10, 12, 14, 16, 18, 20]);
+
+            assert_eq!(vec.len(), 10);
+            assert_eq!(vec, vec![1, 3, 5, 7, 9, 11, 13, 15, 17, 19]);
+        }
+
+        {
+            //                [+++++++++++xxxxxxxxxx]
+            let mut vec = vec![
+                1, 3, 5, 7, 9, 11, 13, 15, 17, 19, 2, 4, 6, 8, 10, 12, 14, 16, 18, 20,
+            ];
+
+            let removed = vec.extract_if(|x| *x % 2 == 0).collect::<Vec<_>>();
+            assert_eq!(removed.len(), 10);
+            assert_eq!(removed, vec![2, 4, 6, 8, 10, 12, 14, 16, 18, 20]);
+
+            assert_eq!(vec.len(), 10);
+            assert_eq!(vec, vec![1, 3, 5, 7, 9, 11, 13, 15, 17, 19]);
+        }
+    }
+}
--- a/librashader-presets/src/lib.rs
+++ b/librashader-presets/src/lib.rs
@ -7,10 +7,13 @@
 //! as input to create a filter chain.
 //!
 //! Re-exported as [`librashader::presets`](https://docs.rs/librashader/latest/librashader/presets/index.html).
-#![feature(extract_if)]
+
+#![allow(unstable_name_collisions)]

 mod error;
+mod extract_if;
 mod parse;
 mod preset;
+
 pub use error::*;
 pub use preset::*;
--- a/librashader-presets/src/parse/preset.rs
+++ b/librashader-presets/src/parse/preset.rs
@ -1,3 +1,4 @@
+use crate::extract_if::MakeExtractIf;
 use crate::parse::remove_if;
 use crate::parse::value::Value;
 use crate::{ParameterConfig, Scale2D, Scaling, ShaderPassConfig, ShaderPreset, TextureConfig};
--- a/librashader-presets/src/parse/value.rs
+++ b/librashader-presets/src/parse/value.rs
@ -16,6 +16,8 @@ use std::io::Read;
 use std::path::{Path, PathBuf};
 use std::str::FromStr;

+use crate::extract_if::MakeExtractIf;
+
 #[derive(Debug)]
 pub enum Value {
    ShaderCount(i32),
--- a/librashader-reflect/Cargo.toml
+++ b/librashader-reflect/Cargo.toml
@ -3,7 +3,7 @@ name = "librashader-reflect"
 edition = "2021"

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -19,25 +19,29 @@ thiserror = "1.0.37"
 bitflags = "1.3.2"
 rustc-hash = "1.1.0"

-librashader-common = { path = "../librashader-common", version = "0.2.0-beta.2" }
-librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.2" }
-librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.2" }
+librashader-common = { path = "../librashader-common", version = "0.2.0-beta.5" }
+librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.5" }
+librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.5" }

 spirv_cross = { package = "librashader-spirv-cross", version = "0.23", optional = true }
-naga = { version = "0.11.0", features = ["glsl-in", "spv-in", "spv-out", "glsl-out", "wgsl-out"], optional = true }

-rspirv = { version = "0.11.0+1.5.4", optional = true }
+naga = { version = "0.19.0", features = ["spv-in", "wgsl-out"], optional = true }
+rspirv = { version = "0.12.0+sdk-1.3.268.0", optional = true }
+spirv = { version = "0.3.0+sdk-1.3.268.0", optional = true}

 serde = { version = "1.0", features = ["derive"], optional = true }

+indexmap = { version = "2.1.0", features = [] }
+matches = { version = "0.1.10", features = [] }
+
 [target.'cfg(windows)'.dependencies.spirv-to-dxil]
 version = "0.4"
 optional = true

 [features]
-default = ["cross", "serialize"]
-unstable-naga = [ "naga", "rspirv" ]
+default = ["cross", "wgsl", "serialize"]
 standalone = ["shaderc/build-from-source", "shaderc/prefer-static-linking"]
 dxil = ["cross", "spirv-to-dxil"]
+wgsl = ["cross", "naga", "spirv", "rspirv"]
 cross = [ "spirv_cross", "spirv_cross/glsl", "spirv_cross/hlsl" ]
 serialize = [ "serde" ]
--- a/librashader-reflect/basic.spv
+++ b/librashader-reflect/basic.spv
--- a/librashader-reflect/src/back/mod.rs
+++ b/librashader-reflect/src/back/mod.rs
@ -3,6 +3,7 @@ pub mod cross;
 pub mod dxil;
 mod spirv;
 pub mod targets;
+pub mod wgsl;

 use crate::back::targets::OutputTarget;
 use crate::error::{ShaderCompileError, ShaderReflectError};
@ -114,3 +115,14 @@ where
        self.backend.reflect(pass_number, semantics)
    }
 }
+
+#[cfg(test)]
+mod test {
+    use crate::front::GlslangCompilation;
+    use librashader_preprocess::ShaderSource;
+
+    pub fn test() {
+        let result = ShaderSource::load("../test/basic.slang").unwrap();
+        let _cross = GlslangCompilation::compile(&result).unwrap();
+    }
+}
--- a/librashader-reflect/src/back/targets.rs
+++ b/librashader-reflect/src/back/targets.rs
@ -18,16 +18,27 @@ pub struct MSL;
 /// must be validated using platform APIs before usage.
 pub struct DXIL;

+/// Shader compiler target for WGSL.
+///
+/// The resulting WGSL will split sampler2Ds into
+/// split textures and shaders. Shaders for each texture binding
+/// will be in descriptor set 1.
+#[derive(Debug)]
+pub struct WGSL;
 impl OutputTarget for GLSL {
    type Output = String;
 }
 impl OutputTarget for HLSL {
    type Output = String;
 }
+impl OutputTarget for WGSL {
+    type Output = String;
+}
 impl OutputTarget for SPIRV {
    type Output = Vec<u32>;
 }

+#[cfg(test)]
 mod test {
    use crate::back::targets::GLSL;
    use crate::back::FromCompilation;
--- a/librashader-reflect/src/back/wgsl/lower_samplers.rs
+++ b/librashader-reflect/src/back/wgsl/lower_samplers.rs
--- a/librashader-reflect/src/back/wgsl/mod.rs
+++ b/librashader-reflect/src/back/wgsl/mod.rs
@ -0,0 +1,227 @@
+mod lower_samplers;
+
+use crate::back::targets::WGSL;
+use crate::back::wgsl::lower_samplers::LowerCombinedImageSamplerPass;
+use crate::back::{CompileShader, CompilerBackend, FromCompilation, ShaderCompilerOutput};
+use crate::error::{ShaderCompileError, ShaderReflectError};
+use crate::front::GlslangCompilation;
+use crate::reflect::naga::NagaReflect;
+use crate::reflect::ReflectShader;
+use naga::back::wgsl::WriterFlags;
+use naga::valid::{Capabilities, ValidationFlags};
+use naga::{AddressSpace, Module};
+use rspirv::binary::Assemble;
+use rspirv::dr::Builder;
+
+/// The context for a WGSL compilation via Naga
+pub struct NagaWgslContext {
+    pub fragment: Module,
+    pub vertex: Module,
+}
+
+/// Compiler options for WGSL
+#[derive(Debug, Default, Clone)]
+pub struct WgslCompileOptions {
+    pub write_pcb_as_ubo: bool,
+    pub sampler_bind_group: u32,
+}
+
+impl FromCompilation<GlslangCompilation> for WGSL {
+    type Target = WGSL;
+    type Options = WgslCompileOptions;
+    type Context = NagaWgslContext;
+    type Output = impl CompileShader<Self::Target, Options = Self::Options, Context = Self::Context>
+        + ReflectShader;
+
+    fn from_compilation(
+        compile: GlslangCompilation,
+    ) -> Result<CompilerBackend<Self::Output>, ShaderReflectError> {
+        fn lower_fragment_shader(words: &[u32]) -> Vec<u32> {
+            let mut loader = rspirv::dr::Loader::new();
+            rspirv::binary::parse_words(words, &mut loader).unwrap();
+            let module = loader.module();
+            let mut builder = Builder::new_from_module(module);
+
+            let mut pass = LowerCombinedImageSamplerPass::new(&mut builder);
+
+            pass.ensure_op_type_sampler();
+            pass.do_pass();
+
+            let module = builder.module();
+
+            module.assemble()
+        }
+
+        let options = naga::front::spv::Options {
+            adjust_coordinate_space: true,
+            strict_capabilities: false,
+            block_ctx_dump_prefix: None,
+        };
+
+        let vertex =
+            naga::front::spv::parse_u8_slice(bytemuck::cast_slice(&compile.vertex), &options)?;
+
+        let fragment = lower_fragment_shader(&compile.fragment);
+        let fragment = naga::front::spv::parse_u8_slice(bytemuck::cast_slice(&fragment), &options)?;
+
+        Ok(CompilerBackend {
+            backend: NagaReflect { vertex, fragment },
+        })
+    }
+}
+
+impl CompileShader<WGSL> for NagaReflect {
+    type Options = WgslCompileOptions;
+    type Context = NagaWgslContext;
+
+    fn compile(
+        mut self,
+        options: Self::Options,
+    ) -> Result<ShaderCompilerOutput<String, Self::Context>, ShaderCompileError> {
+        fn write_wgsl(module: &Module) -> Result<String, ShaderCompileError> {
+            let mut valid =
+                naga::valid::Validator::new(ValidationFlags::all(), Capabilities::empty());
+            let info = valid.validate(&module)?;
+
+            let wgsl = naga::back::wgsl::write_string(&module, &info, WriterFlags::EXPLICIT_TYPES)?;
+            Ok(wgsl)
+        }
+
+        if options.write_pcb_as_ubo {
+            for (_, gv) in self.fragment.global_variables.iter_mut() {
+                if gv.space == AddressSpace::PushConstant {
+                    gv.space = AddressSpace::Uniform;
+                }
+            }
+
+            for (_, gv) in self.vertex.global_variables.iter_mut() {
+                if gv.space == AddressSpace::PushConstant {
+                    gv.space = AddressSpace::Uniform;
+                }
+            }
+        } else {
+            for (_, gv) in self.fragment.global_variables.iter_mut() {
+                if gv.space == AddressSpace::PushConstant {
+                    gv.binding = None;
+                }
+            }
+        }
+
+        // Reassign shit.
+        let images = self
+            .fragment
+            .global_variables
+            .iter()
+            .filter(|&(_, gv)| {
+                let ty = &self.fragment.types[gv.ty];
+                match ty.inner {
+                    naga::TypeInner::Image { .. } => true,
+                    naga::TypeInner::BindingArray { base, .. } => {
+                        let ty = &self.fragment.types[base];
+                        matches!(ty.inner, naga::TypeInner::Image { .. })
+                    }
+                    _ => false,
+                }
+            })
+            .map(|(_, gv)| (gv.binding.clone(), gv.space))
+            .collect::<naga::FastHashSet<_>>();
+
+        self.fragment
+            .global_variables
+            .iter_mut()
+            .filter(|(_, gv)| {
+                let ty = &self.fragment.types[gv.ty];
+                match ty.inner {
+                    naga::TypeInner::Sampler { .. } => true,
+                    naga::TypeInner::BindingArray { base, .. } => {
+                        let ty = &self.fragment.types[base];
+                        matches!(ty.inner, naga::TypeInner::Sampler { .. })
+                    }
+                    _ => false,
+                }
+            })
+            .for_each(|(_, gv)| {
+                if images.contains(&(gv.binding.clone(), gv.space)) {
+                    if let Some(binding) = &mut gv.binding {
+                        binding.group = options.sampler_bind_group;
+                    }
+                }
+            });
+
+        let fragment = write_wgsl(&self.fragment)?;
+        let vertex = write_wgsl(&self.vertex)?;
+        Ok(ShaderCompilerOutput {
+            vertex,
+            fragment,
+            context: NagaWgslContext {
+                fragment: self.fragment,
+                vertex: self.vertex,
+            },
+        })
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use crate::back::targets::WGSL;
+    use crate::back::wgsl::WgslCompileOptions;
+    use crate::back::{CompileShader, FromCompilation};
+    use crate::reflect::semantics::{Semantic, ShaderSemantics, UniformSemantic, UniqueSemantics};
+    use crate::reflect::ReflectShader;
+    use librashader_preprocess::ShaderSource;
+    use rustc_hash::FxHashMap;
+
+    #[test]
+    pub fn test_into() {
+        // let result = ShaderSource::load("../test/shaders_slang/crt/shaders/crt-royale/src/crt-royale-scanlines-horizontal-apply-mask.slang").unwrap();
+        // let result = ShaderSource::load("../test/shaders_slang/crt/shaders/crt-royale/src/crt-royale-scanlines-horizontal-apply-mask.slang").unwrap();
+        let result = ShaderSource::load("../test/basic.slang").unwrap();
+
+        let mut uniform_semantics: FxHashMap<String, UniformSemantic> = Default::default();
+
+        for (_index, param) in result.parameters.iter().enumerate() {
+            uniform_semantics.insert(
+                param.1.id.clone(),
+                UniformSemantic::Unique(Semantic {
+                    semantics: UniqueSemantics::FloatParameter,
+                    index: (),
+                }),
+            );
+        }
+
+        let compilation = crate::front::GlslangCompilation::try_from(&result).unwrap();
+
+        let mut wgsl = WGSL::from_compilation(compilation).unwrap();
+
+        wgsl.reflect(
+            0,
+            &ShaderSemantics {
+                uniform_semantics,
+                texture_semantics: Default::default(),
+            },
+        )
+        .expect("");
+
+        let compiled = wgsl
+            .compile(WgslCompileOptions {
+                write_pcb_as_ubo: true,
+                sampler_bind_group: 1,
+            })
+            .unwrap();
+
+        println!("{}", compiled.fragment);
+
+        // println!("{}", compiled.fragment);
+        // let mut loader = rspirv::dr::Loader::new();
+        // rspirv::binary::parse_words(compilation.vertex.as_binary(), &mut loader).unwrap();
+        // let module = loader.module();
+        //
+        // let outputs: Vec<&Instruction> = module
+        //     .types_global_values
+        //     .iter()
+        //     .filter(|i| i.class.opcode == Op::Variable)
+        //     .collect();
+        //
+        // println!("{outputs:#?}");
+    }
+}
--- a/librashader-reflect/src/error.rs
+++ b/librashader-reflect/src/error.rs
@ -26,6 +26,15 @@ pub enum ShaderCompileError {
    #[cfg(all(target_os = "windows", feature = "dxil"))]
    #[error("spirv-to-dxil")]
    SpirvToDxilCompileError(#[from] spirv_to_dxil::SpirvToDxilError),
+
+    /// Error when transpiling from naga
+    #[cfg(feature = "wgsl")]
+    #[error("naga-wgsl")]
+    NagaWgslError(#[from] naga::back::wgsl::Error),
+    /// Error when transpiling from naga
+    #[cfg(feature = "wgsl")]
+    #[error("naga-wgsl")]
+    NagaValidationError(#[from] naga::WithSpan<naga::valid::ValidationError>),
 }

 /// The error kind encountered when reflecting shader semantics.
@ -34,6 +43,8 @@ pub enum SemanticsErrorKind {
    /// The number of uniform buffers was invalid. Only one UBO is permitted.
    InvalidUniformBufferCount(usize),
    /// The number of push constant blocks was invalid. Only one push constant block is permitted.
+    InvalidPushBufferCount(usize),
+    /// The size of the push constant block was invalid.
    InvalidPushBufferSize(u32),
    /// The location of a varying was invalid.
    InvalidLocation(u32),
@ -43,12 +54,16 @@ pub enum SemanticsErrorKind {
    InvalidInputCount(usize),
    /// The number of outputs declared was invalid.
    InvalidOutputCount(usize),
+    /// Expected a binding but there was none.
+    MissingBinding,
    /// The declared binding point was invalid.
    InvalidBinding(u32),
    /// The declared resource type was invalid.
    InvalidResourceType,
    /// The range of a struct member was invalid.
    InvalidRange(u32),
+    /// The number of entry points in the shader was invalid.
+    InvalidEntryPointCount(usize),
    /// The requested uniform or texture name was not provided semantics.
    UnknownSemantics(String),
    /// The type of the requested uniform was not compatible with the provided semantics.
@ -59,11 +74,6 @@ pub enum SemanticsErrorKind {
 #[non_exhaustive]
 #[derive(Error, Debug)]
 pub enum ShaderReflectError {
-    /// Compile error from naga.
-    #[cfg(feature = "unstable-naga")]
-    #[error("shader")]
-    NagaCompileError(#[from] naga::front::spv::Error),
-
    /// Reflection error from spirv-cross.
    #[error("spirv")]
    SpirvCrossError(#[from] spirv_cross::ErrorCode),
@ -100,6 +110,10 @@ pub enum ShaderReflectError {
    /// The binding number is already in use.
    #[error("the binding is already in use")]
    BindingInUse(u32),
+    /// Error when transpiling from naga
+    #[cfg(feature = "wgsl")]
+    #[error("naga-spv")]
+    NagaInputError(#[from] naga::front::spv::Error),
 }

 #[cfg(feature = "unstable-naga")]
--- a/librashader-reflect/src/front/mod.rs
+++ b/librashader-reflect/src/front/mod.rs
@ -1,16 +1,10 @@
 use crate::error::ShaderCompileError;
 use librashader_preprocess::ShaderSource;

-#[cfg(feature = "unstable-naga")]
-mod naga;
-
 mod shaderc;

 pub use crate::front::shaderc::GlslangCompilation;

-#[cfg(feature = "unstable-naga")]
-pub use crate::front::naga::NagaCompilation;
-
 /// Trait for types that can compile shader sources into a compilation unit.
 pub trait ShaderCompilation: Sized {
    /// Compile the input shader source file into a compilation unit.
--- a/librashader-reflect/src/front/naga.rs
+++ b/librashader-reflect/src/front/naga.rs
@ -1,152 +0,0 @@
-use crate::error::ShaderCompileError;
-use librashader_preprocess::ShaderSource;
-use naga::front::glsl::{Options, Parser};
-use naga::{Module, ShaderStage};
-
-/// A reflectable shader compilation via naga.
-#[derive(Debug)]
-pub struct NagaCompilation {
-    pub(crate) vertex: Module,
-    pub(crate) fragment: Module,
-}
-
-impl TryFrom<&ShaderSource> for NagaCompilation {
-    type Error = ShaderCompileError;
-
-    /// Tries to compile SPIR-V from the provided shader source.
-    fn try_from(source: &ShaderSource) -> Result<Self, Self::Error> {
-        compile_spirv(source)
-    }
-}
-
-fn compile_spirv(source: &ShaderSource) -> Result<NagaCompilation, ShaderCompileError> {
-    let mut parser = Parser::default();
-    let vertex = parser.parse(&Options::from(ShaderStage::Vertex), &source.vertex)?;
-    let fragment = parser.parse(&Options::from(ShaderStage::Fragment), &source.fragment)?;
-    Ok(NagaCompilation { vertex, fragment })
-}
-
-#[cfg(test)]
-mod test {
-    use crate::front::naga::compile_spirv;
-    use crate::front::shaderc::GlslangCompilation;
-    use librashader_preprocess::ShaderSource;
-    use naga::back::glsl::{PipelineOptions, Version};
-    use naga::back::spv::{Capability, WriterFlags};
-    use naga::front::glsl::{Options, Parser};
-    use naga::front::spv::Options as SpvOptions;
-    use naga::valid::{Capabilities, ValidationFlags};
-    use naga::{FastHashSet, ShaderStage};
-    use rspirv::binary::Disassemble;
-
-    #[test]
-    pub fn compile_naga_test() {
-        let result = ShaderSource::load(
-            "../test/slang-shaders/blurs/shaders/royale/blur3x3-last-pass.slang",
-        )
-        .unwrap();
-
-        let fragment_source = result.fragment;
-        let mut parser = Parser::default();
-        println!("{fragment_source}");
-        let _fragment = parser
-            .parse(&Options::from(ShaderStage::Fragment), &fragment_source)
-            .unwrap();
-    }
-
-    #[test]
-    pub fn compile_shader() {
-        let result = ShaderSource::load(
-            "../test/slang-shaders/blurs/shaders/royale/blur3x3-last-pass.slang",
-        )
-        .unwrap();
-        let spirv = compile_spirv(&result).unwrap();
-        eprintln!("{spirv:?}")
-    }
-
-    #[test]
-    pub fn compile_shader_roundtrip() {
-        let result = ShaderSource::load("../test/basic.slang").unwrap();
-        let cross = GlslangCompilation::compile(&result).unwrap();
-        let naga_fragment =
-            naga::front::spv::parse_u8_slice(cross.fragment.as_binary_u8(), &SpvOptions::default())
-                .unwrap();
-        println!("{:#?}", naga_fragment.constants);
-        println!("{:#?}", naga_fragment.global_variables);
-        println!("{:#?}", naga_fragment.types);
-    }
-
-    #[test]
-    pub fn naga_playground() {
-        let result = ShaderSource::load("../test/basic.slang").unwrap();
-        let spirv = GlslangCompilation::compile(&result).unwrap();
-
-        let module =
-            naga::front::spv::parse_u8_slice(spirv.fragment.as_binary_u8(), &SpvOptions::default())
-                .unwrap();
-
-        let capability = FastHashSet::from_iter([Capability::Shader]);
-        let mut writer = naga::back::spv::Writer::new(&naga::back::spv::Options {
-            lang_version: (1, 0),
-            flags: WriterFlags::all(),
-            binding_map: Default::default(),
-            capabilities: Some(capability),
-            bounds_check_policies: Default::default(),
-        })
-        .unwrap();
-
-        let mut validator =
-            naga::valid::Validator::new(ValidationFlags::empty(), Capabilities::all());
-        let info = validator.validate(&module).unwrap();
-        let mut spv_out = Vec::new();
-        let pipe = naga::back::spv::PipelineOptions {
-            shader_stage: ShaderStage::Fragment,
-            entry_point: "main".to_string(),
-        };
-        writer
-            .write(&module, &info, Some(&pipe), &mut spv_out)
-            .unwrap();
-
-        let mut glsl_out = String::new();
-        let opts = naga::back::glsl::Options {
-            version: Version::Desktop(330),
-            writer_flags: naga::back::glsl::WriterFlags::all(),
-            binding_map: Default::default(),
-        };
-        let pipe = PipelineOptions {
-            shader_stage: ShaderStage::Fragment,
-            entry_point: "main".to_string(),
-            multiview: None,
-        };
-        let mut glsl_naga = naga::back::glsl::Writer::new(
-            &mut glsl_out,
-            &module,
-            &info,
-            &opts,
-            &pipe,
-            Default::default(),
-        )
-        .unwrap();
-
-        glsl_naga.write().unwrap();
-
-        let wgsl =
-            naga::back::wgsl::write_string(&module, &info, naga::back::wgsl::WriterFlags::all())
-                .unwrap();
-
-        let mut loader = rspirv::dr::Loader::new();
-        rspirv::binary::parse_words(&spv_out, &mut loader).unwrap();
-        let module = loader.module();
-        println!("--- spirv --");
-        println!("{:#}", module.disassemble());
-        println!("--- cross glsl --");
-
-        let loaded = spirv_cross::spirv::Module::from_words(&spv_out);
-        let mut ast = spirv_cross::spirv::Ast::<spirv_cross::glsl::Target>::parse(&loaded).unwrap();
-        println!("{:#}", ast.compile().unwrap());
-        println!("--- naga glsl---");
-        println!("{glsl_out:#}");
-        println!("--- naga wgsl---");
-        println!("{wgsl:#}")
-    }
-}
--- a/librashader-reflect/src/reflect/cross.rs
+++ b/librashader-reflect/src/reflect/cross.rs
@ -1,8 +1,8 @@
 use crate::error::{SemanticsErrorKind, ShaderCompileError, ShaderReflectError};
 use crate::front::GlslangCompilation;
 use crate::reflect::semantics::{
-    BindingMeta, BindingStage, MemberOffset, PushReflection, ShaderReflection, ShaderSemantics,
-    TextureBinding, TextureSemanticMap, TextureSemantics, TextureSizeMeta, TypeInfo, UboReflection,
+    BindingMeta, BindingStage, BufferReflection, MemberOffset, ShaderReflection, ShaderSemantics,
+    TextureBinding, TextureSemanticMap, TextureSemantics, TextureSizeMeta, TypeInfo,
    UniformMemberBlock, UniqueSemanticMap, UniqueSemantics, ValidateTypeSemantics, VariableMeta,
    MAX_BINDINGS_COUNT, MAX_PUSH_BUFFER_SIZE,
 };
@ -208,6 +208,7 @@ where
            ));
        }

+        // Ensure that vertex attributes use location 0 and 1
        let vert_mask = vertex_res.stage_inputs.iter().try_fold(0, |mask, input| {
            Ok::<u32, ErrorCode>(
                mask | 1 << self.vertex.get_decoration(input.id, Decoration::Location)?,
@ -227,9 +228,7 @@ where

        if vertex_res.push_constant_buffers.len() > 1 {
            return Err(ShaderReflectError::VertexSemanticError(
-                SemanticsErrorKind::InvalidUniformBufferCount(
-                    vertex_res.push_constant_buffers.len(),
-                ),
+                SemanticsErrorKind::InvalidPushBufferCount(vertex_res.push_constant_buffers.len()),
            ));
        }

@ -241,7 +240,7 @@ where

        if fragment_res.push_constant_buffers.len() > 1 {
            return Err(ShaderReflectError::FragmentSemanticError(
-                SemanticsErrorKind::InvalidUniformBufferCount(
+                SemanticsErrorKind::InvalidPushBufferCount(
                    fragment_res.push_constant_buffers.len(),
                ),
            ));
@ -332,7 +331,7 @@ where
                                    expected,
                                    received: offset,
                                    ty: offset_type,
-                                    pass: pass_number
+                                    pass: pass_number,
                                });
                            }
                            if meta.size != typeinfo.size {
@ -367,7 +366,7 @@ where
                                    expected,
                                    received: offset,
                                    ty: offset_type,
-                                    pass: pass_number
+                                    pass: pass_number,
                                });
                            }
                            if meta.size != typeinfo.size * typeinfo.columns {
@ -416,7 +415,7 @@ where
                            expected,
                            received: offset,
                            ty: offset_type,
-                            pass: pass_number
+                            pass: pass_number,
                        });
                    }

@ -450,7 +449,7 @@ where
        &mut self,
        vertex_ubo: Option<&Resource>,
        fragment_ubo: Option<&Resource>,
-    ) -> Result<Option<UboReflection>, ShaderReflectError> {
+    ) -> Result<Option<BufferReflection<u32>>, ShaderReflectError> {
        if let Some(vertex_ubo) = vertex_ubo {
            self.vertex
                .set_decoration(vertex_ubo.id, Decoration::Binding, 0)?;
@ -476,7 +475,7 @@ where
                }

                let size = std::cmp::max(vertex_ubo.size, fragment_ubo.size);
-                Ok(Some(UboReflection {
+                Ok(Some(BufferReflection {
                    binding: vertex_ubo.binding,
                    size: align_uniform_size(size),
                    stage_mask: BindingStage::VERTEX | BindingStage::FRAGMENT,
@ -485,7 +484,7 @@ where
            (Some(vertex_ubo), None) => {
                let vertex_ubo =
                    Self::get_ubo_data(&self.vertex, vertex_ubo, SemanticErrorBlame::Vertex)?;
-                Ok(Some(UboReflection {
+                Ok(Some(BufferReflection {
                    binding: vertex_ubo.binding,
                    size: align_uniform_size(vertex_ubo.size),
                    stage_mask: BindingStage::VERTEX,
@ -494,7 +493,7 @@ where
            (None, Some(fragment_ubo)) => {
                let fragment_ubo =
                    Self::get_ubo_data(&self.fragment, fragment_ubo, SemanticErrorBlame::Fragment)?;
-                Ok(Some(UboReflection {
+                Ok(Some(BufferReflection {
                    binding: fragment_ubo.binding,
                    size: align_uniform_size(fragment_ubo.size),
                    stage_mask: BindingStage::FRAGMENT,
@ -570,7 +569,7 @@ where
        &mut self,
        vertex_pcb: Option<&Resource>,
        fragment_pcb: Option<&Resource>,
-    ) -> Result<Option<PushReflection>, ShaderReflectError> {
+    ) -> Result<Option<BufferReflection<Option<u32>>>, ShaderReflectError> {
        if let Some(vertex_pcb) = vertex_pcb {
            self.vertex
                .set_decoration(vertex_pcb.id, Decoration::Binding, 1)?;
@ -594,7 +593,8 @@ where

                let size = std::cmp::max(vertex_size, fragment_size);

-                Ok(Some(PushReflection {
+                Ok(Some(BufferReflection {
+                    binding: None,
                    size: align_uniform_size(size),
                    stage_mask: BindingStage::VERTEX | BindingStage::FRAGMENT,
                }))
@ -602,7 +602,8 @@ where
            (Some(vertex_push), None) => {
                let vertex_size =
                    Self::get_push_size(&self.vertex, vertex_push, SemanticErrorBlame::Vertex)?;
-                Ok(Some(PushReflection {
+                Ok(Some(BufferReflection {
+                    binding: None,
                    size: align_uniform_size(vertex_size),
                    stage_mask: BindingStage::VERTEX,
                }))
@ -613,7 +614,8 @@ where
                    fragment_push,
                    SemanticErrorBlame::Fragment,
                )?;
-                Ok(Some(PushReflection {
+                Ok(Some(BufferReflection {
+                    binding: None,
                    size: align_uniform_size(fragment_size),
                    stage_mask: BindingStage::FRAGMENT,
                }))
@ -922,9 +924,9 @@ mod test {
        let mut opts = CompilerOptions::default();
        opts.version = Version::V4_60;
        opts.enable_420_pack_extension = false;
-        let compiled = reflect.compile(Version::V3_30).unwrap();
-        // eprintln!("{shader_reflection:#?}");
-        eprintln!("{:#}", compiled.fragment)
+        // let compiled: ShaderCompilerOutput<String, CrossWgslContext> = <CrossReflect<glsl::Target> as CompileShader<WGSL>>::compile(reflect, Version::V3_30).unwrap();
+        // // eprintln!("{shader_reflection:#?}");
+        // eprintln!("{}", compiled.fragment)
        // let mut loader = rspirv::dr::Loader::new();
        // rspirv::binary::parse_words(spirv.fragment.as_binary(), &mut loader).unwrap();
        // let module = loader.module();
--- a/librashader-reflect/src/reflect/mod.rs
+++ b/librashader-reflect/src/reflect/mod.rs
@ -12,8 +12,8 @@ pub mod presets;

 mod helper;

-#[cfg(feature = "unstable-naga")]
-mod naga;
+#[cfg(feature = "wgsl")]
+pub mod naga;

 /// A trait for compilation outputs that can provide reflection information.
 pub trait ReflectShader {
--- a/librashader-reflect/src/reflect/naga.rs
+++ b/librashader-reflect/src/reflect/naga.rs
@ -1,126 +1,858 @@
 use crate::error::{SemanticsErrorKind, ShaderReflectError};
-use crate::front::GlslangCompilation;
-use crate::front::NagaCompilation;

-use naga::front::spv::Options;
-use naga::{GlobalVariable, Module, StructMember, Type, TypeInner};
-use std::convert::Infallible;
+use naga::{
+    AddressSpace, Binding, GlobalVariable, Handle, ImageClass, Module, ResourceBinding, Scalar,
+    ScalarKind, TypeInner, VectorSize,
+};
+
+use crate::reflect::helper::{SemanticErrorBlame, TextureData, UboData};
+use crate::reflect::semantics::{
+    BindingMeta, BindingStage, BufferReflection, MemberOffset, ShaderSemantics, TextureBinding,
+    TextureSemanticMap, TextureSemantics, TextureSizeMeta, TypeInfo, UniformMemberBlock,
+    UniqueSemanticMap, UniqueSemantics, ValidateTypeSemantics, VariableMeta, MAX_BINDINGS_COUNT,
+    MAX_PUSH_BUFFER_SIZE,
+};
+use crate::reflect::{align_uniform_size, ReflectShader, ShaderReflection};

 #[derive(Debug)]
-pub struct NagaReflect {
-    vertex: Module,
-    fragment: Module,
+pub(crate) struct NagaReflect {
+    pub(crate) vertex: Module,
+    pub(crate) fragment: Module,
 }

-impl TryFrom<NagaCompilation> for NagaReflect {
-    type Error = ShaderReflectError;
+impl ValidateTypeSemantics<&TypeInner> for UniqueSemantics {
+    fn validate_type(&self, ty: &&TypeInner) -> Option<TypeInfo> {
+        let (TypeInner::Vector { .. } | TypeInner::Scalar { .. } | TypeInner::Matrix { .. }) = *ty
+        else {
+            return None;
+        };

-    fn try_from(value: NagaCompilation) -> Result<Self, Self::Error> {
-        Ok(NagaReflect {
-            vertex: value.vertex,
-            fragment: value.fragment,
-        })
+        match self {
+            UniqueSemantics::MVP => {
+                if matches!(ty, TypeInner::Matrix { columns, rows, scalar: Scalar { width, .. } } if *columns == VectorSize::Quad
+                    && *rows == VectorSize::Quad && *width == 4)
+                {
+                    return Some(TypeInfo {
+                        size: 4,
+                        columns: 4,
+                    });
+                }
+            }
+            UniqueSemantics::FrameCount => {
+                // Uint32 == width 4
+                if matches!(ty, TypeInner::Scalar( Scalar { kind, width }) if *kind == ScalarKind::Uint && *width == 4)
+                {
+                    return Some(TypeInfo {
+                        size: 1,
+                        columns: 1,
+                    });
+                }
+            }
+            UniqueSemantics::FrameDirection => {
+                // Uint32 == width 4
+                if matches!(ty, TypeInner::Scalar( Scalar { kind, width }) if *kind == ScalarKind::Sint && *width == 4)
+                {
+                    return Some(TypeInfo {
+                        size: 1,
+                        columns: 1,
+                    });
+                }
+            }
+            UniqueSemantics::FloatParameter => {
+                // Float32 == width 4
+                if matches!(ty, TypeInner::Scalar( Scalar { kind, width }) if *kind == ScalarKind::Float && *width == 4)
+                {
+                    return Some(TypeInfo {
+                        size: 1,
+                        columns: 1,
+                    });
+                }
+            }
+            _ => {
+                if matches!(ty, TypeInner::Vector { scalar: Scalar { width, kind }, size } if *kind == ScalarKind::Float && *width == 4 && *size == VectorSize::Quad)
+                {
+                    return Some(TypeInfo {
+                        size: 4,
+                        columns: 1,
+                    });
+                }
+            }
+        };
+
+        return None;
    }
 }

-impl TryFrom<&GlslangCompilation> for NagaReflect {
-    type Error = ShaderReflectError;
+impl ValidateTypeSemantics<&TypeInner> for TextureSemantics {
+    fn validate_type(&self, ty: &&TypeInner) -> Option<TypeInfo> {
+        let TypeInner::Vector {
+            scalar: Scalar { width, kind },
+            size,
+        } = ty
+        else {
+            return None;
+        };

-    fn try_from(value: &GlslangCompilation) -> Result<Self, Self::Error> {
-        let ops = Options::default();
-        let vertex =
-            naga::front::spv::Parser::new(value.vertex.clone().into_iter(), &ops).parse()?;
-        let fragment =
-            naga::front::spv::Parser::new(value.fragment.clone().into_iter(), &ops).parse()?;
-        Ok(NagaReflect { vertex, fragment })
-    }
+        if *kind == ScalarKind::Float && *width == 4 && *size == VectorSize::Quad {
+            return Some(TypeInfo {
+                size: 4,
+                columns: 1,
+            });
        }

-struct UboData {
-    // id: u32,
-    // descriptor_set: u32,
-    binding: u32,
-    size: u32,
-}
-
-struct Ubo {
-    members: Vec<StructMember>,
-    span: u32,
-}
-
-impl TryFrom<naga::Type> for Ubo {
-    type Error = Infallible;
-
-    fn try_from(value: Type) -> Result<Self, Infallible> {
-        match value.inner {
-            TypeInner::Struct { members, span } => Ok(Ubo { members, span }),
-            // todo: make this programmer error
-            _ => panic!(),
-        }
+        None
    }
 }

 impl NagaReflect {
-    // pub fn get_ubo_data(arena: Arena, variable: GlobalVariable, blame: SemanticErrorBlame) -> Result<UboData, ShaderReflectError> {
-    //     let binding = match variable.binding {
-    //         Some(ResourceBinding { group: 0, binding }) => binding,
-    //         Some(ResourceBinding { group, .. }) => return Err(blame.error(SemanticsErrorKind::InvalidDescriptorSet(group))),
-    //         None => return Err(blame.error(SemanticsErrorKind::InvalidDescriptorSet(u32::MAX))),
-    //     };
-    //
-    //     if binding >= MAX_BINDINGS_COUNT {
-    //         return Err(blame.error(SemanticsErrorKind::InvalidBinding(binding)));
-    //     }
-    //
-    //     match variable.ty.as {
-    //         Handle { .. } => {}
-    //     }
-    //     Ok(UboData {
-    //         binding,
-    //
-    //     })
-    // }
-    pub fn reflect_ubos(
-        vertex: GlobalVariable,
-        fragment: GlobalVariable,
-    ) -> Result<(), ShaderReflectError> {
-        match (vertex.binding, fragment.binding) {
-            // todo: should emit for both but whatever
-            (None, None) | (Some(_), None) | (None, Some(_)) => {
-                ShaderReflectError::VertexSemanticError(SemanticsErrorKind::InvalidDescriptorSet(
-                    u32::MAX,
-                ))
+    fn reflect_ubos(
+        &mut self,
+        vertex_ubo: Option<Handle<GlobalVariable>>,
+        fragment_ubo: Option<Handle<GlobalVariable>>,
+    ) -> Result<Option<BufferReflection<u32>>, ShaderReflectError> {
+        // todo: merge this with the spirv-cross code
+        match (vertex_ubo, fragment_ubo) {
+            (None, None) => Ok(None),
+            (Some(vertex_ubo), Some(fragment_ubo)) => {
+                let vertex_ubo = Self::get_ubo_data(
+                    &self.vertex,
+                    &self.vertex.global_variables[vertex_ubo],
+                    SemanticErrorBlame::Vertex,
+                )?;
+                let fragment_ubo = Self::get_ubo_data(
+                    &self.fragment,
+                    &self.fragment.global_variables[fragment_ubo],
+                    SemanticErrorBlame::Fragment,
+                )?;
+                if vertex_ubo.binding != fragment_ubo.binding {
+                    return Err(ShaderReflectError::MismatchedUniformBuffer {
+                        vertex: vertex_ubo.binding,
+                        fragment: fragment_ubo.binding,
+                    });
                }
-            (Some(_vert), Some(_frag)) => {
-                todo!();
+
+                let size = std::cmp::max(vertex_ubo.size, fragment_ubo.size);
+                Ok(Some(BufferReflection {
+                    binding: vertex_ubo.binding,
+                    size: align_uniform_size(size),
+                    stage_mask: BindingStage::VERTEX | BindingStage::FRAGMENT,
+                }))
            }
+            (Some(vertex_ubo), None) => {
+                let vertex_ubo = Self::get_ubo_data(
+                    &self.vertex,
+                    &self.vertex.global_variables[vertex_ubo],
+                    SemanticErrorBlame::Vertex,
+                )?;
+                Ok(Some(BufferReflection {
+                    binding: vertex_ubo.binding,
+                    size: align_uniform_size(vertex_ubo.size),
+                    stage_mask: BindingStage::VERTEX,
+                }))
+            }
+            (None, Some(fragment_ubo)) => {
+                let fragment_ubo = Self::get_ubo_data(
+                    &self.fragment,
+                    &self.fragment.global_variables[fragment_ubo],
+                    SemanticErrorBlame::Fragment,
+                )?;
+                Ok(Some(BufferReflection {
+                    binding: fragment_ubo.binding,
+                    size: align_uniform_size(fragment_ubo.size),
+                    stage_mask: BindingStage::FRAGMENT,
+                }))
+            }
+        }
+    }
+
+    fn get_ubo_data(
+        module: &Module,
+        ubo: &GlobalVariable,
+        blame: SemanticErrorBlame,
+    ) -> Result<UboData, ShaderReflectError> {
+        let Some(binding) = &ubo.binding else {
+            return Err(blame.error(SemanticsErrorKind::MissingBinding));
        };

-        todo!();
+        if binding.binding >= MAX_BINDINGS_COUNT {
+            return Err(blame.error(SemanticsErrorKind::InvalidBinding(binding.binding)));
+        }
+
+        if binding.group != 0 {
+            return Err(blame.error(SemanticsErrorKind::InvalidDescriptorSet(binding.group)));
+        }
+
+        let ty = &module.types[ubo.ty];
+        let size = ty.inner.size(module.to_ctx());
+        Ok(UboData {
+            // descriptor_set,
+            // id: ubo.id,
+            binding: binding.binding,
+            size,
+        })
+    }
+
+    fn get_next_binding(&self, bind_group: u32) -> u32 {
+        let mut max_bind = 0;
+        for (_, gv) in self.vertex.global_variables.iter() {
+            let Some(binding) = &gv.binding else {
+                continue;
+            };
+            if binding.group != bind_group {
+                continue;
+            }
+            max_bind = std::cmp::max(max_bind, binding.binding);
+        }
+
+        for (_, gv) in self.fragment.global_variables.iter() {
+            let Some(binding) = &gv.binding else {
+                continue;
+            };
+            if binding.group != bind_group {
+                continue;
+            }
+            max_bind = std::cmp::max(max_bind, binding.binding);
+        }
+
+        max_bind + 1
+    }
+
+    fn get_push_size(
+        module: &Module,
+        push: &GlobalVariable,
+        blame: SemanticErrorBlame,
+    ) -> Result<u32, ShaderReflectError> {
+        let ty = &module.types[push.ty];
+        let size = ty.inner.size(module.to_ctx());
+        if size > MAX_PUSH_BUFFER_SIZE {
+            return Err(blame.error(SemanticsErrorKind::InvalidPushBufferSize(size)));
+        }
+        Ok(size)
+    }
+
+    fn reflect_push_constant_buffer(
+        &mut self,
+        vertex_pcb: Option<Handle<GlobalVariable>>,
+        fragment_pcb: Option<Handle<GlobalVariable>>,
+    ) -> Result<Option<BufferReflection<Option<u32>>>, ShaderReflectError> {
+        let binding = self.get_next_binding(0);
+        // Reassign to UBO later if we want during compilation.
+        if let Some(vertex_pcb) = vertex_pcb {
+            let ubo = &mut self.vertex.global_variables[vertex_pcb];
+            ubo.binding = Some(ResourceBinding { group: 0, binding });
+        }
+
+        if let Some(fragment_pcb) = fragment_pcb {
+            let ubo = &mut self.fragment.global_variables[fragment_pcb];
+            ubo.binding = Some(ResourceBinding { group: 0, binding });
+        };
+
+        match (vertex_pcb, fragment_pcb) {
+            (None, None) => Ok(None),
+            (Some(vertex_push), Some(fragment_push)) => {
+                let vertex_size = Self::get_push_size(
+                    &self.vertex,
+                    &self.vertex.global_variables[vertex_push],
+                    SemanticErrorBlame::Vertex,
+                )?;
+                let fragment_size = Self::get_push_size(
+                    &self.fragment,
+                    &self.fragment.global_variables[fragment_push],
+                    SemanticErrorBlame::Fragment,
+                )?;
+
+                let size = std::cmp::max(vertex_size, fragment_size);
+
+                Ok(Some(BufferReflection {
+                    binding: Some(binding),
+                    size: align_uniform_size(size),
+                    stage_mask: BindingStage::VERTEX | BindingStage::FRAGMENT,
+                }))
+            }
+            (Some(vertex_push), None) => {
+                let vertex_size = Self::get_push_size(
+                    &self.vertex,
+                    &self.vertex.global_variables[vertex_push],
+                    SemanticErrorBlame::Vertex,
+                )?;
+                Ok(Some(BufferReflection {
+                    binding: Some(binding),
+                    size: align_uniform_size(vertex_size),
+                    stage_mask: BindingStage::VERTEX,
+                }))
+            }
+            (None, Some(fragment_push)) => {
+                let fragment_size = Self::get_push_size(
+                    &self.fragment,
+                    &self.fragment.global_variables[fragment_push],
+                    SemanticErrorBlame::Fragment,
+                )?;
+                Ok(Some(BufferReflection {
+                    binding: Some(binding),
+                    size: align_uniform_size(fragment_size),
+                    stage_mask: BindingStage::FRAGMENT,
+                }))
+            }
+        }
+    }
+
+    fn validate(&self) -> Result<(), ShaderReflectError> {
+        // Verify types
+        if self.vertex.global_variables.iter().any(|(_, gv)| {
+            let ty = &self.vertex.types[gv.ty];
+            match ty.inner {
+                TypeInner::Scalar { .. }
+                | TypeInner::Vector { .. }
+                | TypeInner::Matrix { .. }
+                | TypeInner::Struct { .. } => false,
+                _ => true,
+            }
+        }) {
+            return Err(ShaderReflectError::VertexSemanticError(
+                SemanticsErrorKind::InvalidResourceType,
+            ));
+        }
+
+        if self.fragment.global_variables.iter().any(|(_, gv)| {
+            let ty = &self.fragment.types[gv.ty];
+            match ty.inner {
+                TypeInner::Scalar { .. }
+                | TypeInner::Vector { .. }
+                | TypeInner::Matrix { .. }
+                | TypeInner::Struct { .. }
+                | TypeInner::Image { .. }
+                | TypeInner::Sampler { .. } => false,
+                TypeInner::BindingArray { base, .. } => {
+                    let ty = &self.fragment.types[base];
+                    match ty.inner {
+                        TypeInner::Image { class, .. }
+                            if !matches!(class, ImageClass::Storage { .. }) =>
+                        {
+                            false
+                        }
+                        TypeInner::Sampler { .. } => false,
+                        _ => true,
+                    }
+                }
+                _ => true,
+            }
+        }) {
+            return Err(ShaderReflectError::FragmentSemanticError(
+                SemanticsErrorKind::InvalidResourceType,
+            ));
+        }
+
+        // Verify Vertex inputs
+        'vertex: {
+            if self.vertex.entry_points.len() != 1 {
+                return Err(ShaderReflectError::VertexSemanticError(
+                    SemanticsErrorKind::InvalidEntryPointCount(self.vertex.entry_points.len()),
+                ));
+            }
+
+            let vertex_entry_point = &self.vertex.entry_points[0];
+            let vert_inputs = vertex_entry_point.function.arguments.len();
+            if vert_inputs != 2 {
+                return Err(ShaderReflectError::VertexSemanticError(
+                    SemanticsErrorKind::InvalidInputCount(vert_inputs),
+                ));
+            }
+            for input in &vertex_entry_point.function.arguments {
+                let &Some(Binding::Location { location, .. }) = &input.binding else {
+                    return Err(ShaderReflectError::VertexSemanticError(
+                        SemanticsErrorKind::MissingBinding,
+                    ));
+                };
+
+                if location == 0 {
+                    let pos_type = &self.vertex.types[input.ty];
+                    if !matches!(pos_type.inner, TypeInner::Vector { size, ..} if size == VectorSize::Quad)
+                    {
+                        return Err(ShaderReflectError::VertexSemanticError(
+                            SemanticsErrorKind::InvalidLocation(location),
+                        ));
+                    }
+                    break 'vertex;
+                }
+
+                if location == 1 {
+                    let coord_type = &self.vertex.types[input.ty];
+                    if !matches!(coord_type.inner, TypeInner::Vector { size, ..} if size == VectorSize::Bi)
+                    {
+                        return Err(ShaderReflectError::VertexSemanticError(
+                            SemanticsErrorKind::InvalidLocation(location),
+                        ));
+                    }
+                    break 'vertex;
+                }
+
+                return Err(ShaderReflectError::VertexSemanticError(
+                    SemanticsErrorKind::InvalidLocation(location),
+                ));
+            }
+
+            let uniform_buffer_count = self
+                .vertex
+                .global_variables
+                .iter()
+                .filter(|(_, gv)| gv.space == AddressSpace::Uniform)
+                .count();
+
+            if uniform_buffer_count > 1 {
+                return Err(ShaderReflectError::VertexSemanticError(
+                    SemanticsErrorKind::InvalidUniformBufferCount(uniform_buffer_count),
+                ));
+            }
+
+            let push_buffer_count = self
+                .vertex
+                .global_variables
+                .iter()
+                .filter(|(_, gv)| gv.space == AddressSpace::PushConstant)
+                .count();
+
+            if push_buffer_count > 1 {
+                return Err(ShaderReflectError::VertexSemanticError(
+                    SemanticsErrorKind::InvalidPushBufferCount(push_buffer_count),
+                ));
+            }
+        }
+
+        {
+            if self.fragment.entry_points.len() != 1 {
+                return Err(ShaderReflectError::FragmentSemanticError(
+                    SemanticsErrorKind::InvalidEntryPointCount(self.vertex.entry_points.len()),
+                ));
+            }
+
+            let frag_entry_point = &self.fragment.entry_points[0];
+            let Some(frag_output) = &frag_entry_point.function.result else {
+                return Err(ShaderReflectError::FragmentSemanticError(
+                    SemanticsErrorKind::InvalidOutputCount(0),
+                ));
+            };
+
+            let &Some(Binding::Location { location, .. }) = &frag_output.binding else {
+                return Err(ShaderReflectError::VertexSemanticError(
+                    SemanticsErrorKind::MissingBinding,
+                ));
+            };
+
+            if location != 0 {
+                return Err(ShaderReflectError::FragmentSemanticError(
+                    SemanticsErrorKind::InvalidLocation(location),
+                ));
+            }
+
+            let uniform_buffer_count = self
+                .fragment
+                .global_variables
+                .iter()
+                .filter(|(_, gv)| gv.space == AddressSpace::Uniform)
+                .count();
+
+            if uniform_buffer_count > 1 {
+                return Err(ShaderReflectError::FragmentSemanticError(
+                    SemanticsErrorKind::InvalidUniformBufferCount(uniform_buffer_count),
+                ));
+            }
+
+            let push_buffer_count = self
+                .fragment
+                .global_variables
+                .iter()
+                .filter(|(_, gv)| gv.space == AddressSpace::PushConstant)
+                .count();
+
+            if push_buffer_count > 1 {
+                return Err(ShaderReflectError::FragmentSemanticError(
+                    SemanticsErrorKind::InvalidPushBufferCount(push_buffer_count),
+                ));
+            }
+        }
+
+        Ok(())
+    }
+
+    fn reflect_buffer_struct_members(
+        module: &Module,
+        resource: Handle<GlobalVariable>,
+        pass_number: usize,
+        semantics: &ShaderSemantics,
+        meta: &mut BindingMeta,
+        offset_type: UniformMemberBlock,
+        blame: SemanticErrorBlame,
+    ) -> Result<(), ShaderReflectError> {
+        let resource = &module.global_variables[resource];
+        let TypeInner::Struct { members, .. } = &module.types[resource.ty].inner else {
+            return Err(blame.error(SemanticsErrorKind::InvalidResourceType));
+        };
+
+        for member in members {
+            let Some(name) = member.name.clone() else {
+                return Err(blame.error(SemanticsErrorKind::InvalidRange(member.offset)));
+            };
+            let member_type = &module.types[member.ty].inner;
+
+            if let Some(parameter) = semantics.uniform_semantics.get_unique_semantic(&name) {
+                let Some(typeinfo) = parameter.semantics.validate_type(&member_type) else {
+                    return Err(blame.error(SemanticsErrorKind::InvalidTypeForSemantic(name)));
+                };
+
+                match &parameter.semantics {
+                    UniqueSemantics::FloatParameter => {
+                        let offset = member.offset;
+                        if let Some(meta) = meta.parameter_meta.get_mut(&name) {
+                            if let Some(expected) = meta.offset.offset(offset_type)
+                                && expected != offset as usize
+                            {
+                                return Err(ShaderReflectError::MismatchedOffset {
+                                    semantic: name,
+                                    expected,
+                                    received: offset as usize,
+                                    ty: offset_type,
+                                    pass: pass_number,
+                                });
+                            }
+                            if meta.size != typeinfo.size {
+                                return Err(ShaderReflectError::MismatchedSize {
+                                    semantic: name,
+                                    vertex: meta.size,
+                                    fragment: typeinfo.size,
+                                    pass: pass_number,
+                                });
+                            }
+
+                            *meta.offset.offset_mut(offset_type) = Some(offset as usize);
+                        } else {
+                            meta.parameter_meta.insert(
+                                name.clone(),
+                                VariableMeta {
+                                    id: name,
+                                    offset: MemberOffset::new(offset as usize, offset_type),
+                                    size: typeinfo.size,
+                                },
+                            );
+                        }
+                    }
+                    semantics => {
+                        let offset = member.offset;
+                        if let Some(meta) = meta.unique_meta.get_mut(semantics) {
+                            if let Some(expected) = meta.offset.offset(offset_type)
+                                && expected != offset as usize
+                            {
+                                return Err(ShaderReflectError::MismatchedOffset {
+                                    semantic: name,
+                                    expected,
+                                    received: offset as usize,
+                                    ty: offset_type,
+                                    pass: pass_number,
+                                });
+                            }
+                            if meta.size != typeinfo.size * typeinfo.columns {
+                                return Err(ShaderReflectError::MismatchedSize {
+                                    semantic: name,
+                                    vertex: meta.size,
+                                    fragment: typeinfo.size,
+                                    pass: pass_number,
+                                });
+                            }
+
+                            *meta.offset.offset_mut(offset_type) = Some(offset as usize);
+                        } else {
+                            meta.unique_meta.insert(
+                                *semantics,
+                                VariableMeta {
+                                    id: name,
+                                    offset: MemberOffset::new(offset as usize, offset_type),
+                                    size: typeinfo.size * typeinfo.columns,
+                                },
+                            );
+                        }
+                    }
+                }
+            } else if let Some(texture) = semantics.uniform_semantics.get_texture_semantic(&name) {
+                let Some(_typeinfo) = texture.semantics.validate_type(&member_type) else {
+                    return Err(blame.error(SemanticsErrorKind::InvalidTypeForSemantic(name)));
+                };
+
+                if let TextureSemantics::PassOutput = texture.semantics {
+                    if texture.index >= pass_number {
+                        return Err(ShaderReflectError::NonCausalFilterChain {
+                            pass: pass_number,
+                            target: texture.index,
+                        });
+                    }
+                }
+
+                let offset = member.offset;
+                if let Some(meta) = meta.texture_size_meta.get_mut(&texture) {
+                    if let Some(expected) = meta.offset.offset(offset_type)
+                        && expected != offset as usize
+                    {
+                        return Err(ShaderReflectError::MismatchedOffset {
+                            semantic: name,
+                            expected,
+                            received: offset as usize,
+                            ty: offset_type,
+                            pass: pass_number,
+                        });
+                    }
+
+                    meta.stage_mask.insert(match blame {
+                        SemanticErrorBlame::Vertex => BindingStage::VERTEX,
+                        SemanticErrorBlame::Fragment => BindingStage::FRAGMENT,
+                    });
+
+                    *meta.offset.offset_mut(offset_type) = Some(offset as usize);
+                } else {
+                    meta.texture_size_meta.insert(
+                        texture,
+                        TextureSizeMeta {
+                            offset: MemberOffset::new(offset as usize, offset_type),
+                            stage_mask: match blame {
+                                SemanticErrorBlame::Vertex => BindingStage::VERTEX,
+                                SemanticErrorBlame::Fragment => BindingStage::FRAGMENT,
+                            },
+                            id: name,
+                        },
+                    );
+                }
+            } else {
+                return Err(blame.error(SemanticsErrorKind::UnknownSemantics(name)));
+            }
+        }
+        Ok(())
+    }
+
+    fn reflect_texture<'a>(
+        &'a self,
+        texture: &'a GlobalVariable,
+    ) -> Result<TextureData<'a>, ShaderReflectError> {
+        let Some(binding) = &texture.binding else {
+            return Err(ShaderReflectError::FragmentSemanticError(
+                SemanticsErrorKind::MissingBinding,
+            ));
+        };
+
+        let Some(name) = texture.name.as_ref() else {
+            return Err(ShaderReflectError::FragmentSemanticError(
+                SemanticsErrorKind::InvalidBinding(binding.binding),
+            ));
+        };
+
+        if binding.group != 0 {
+            return Err(ShaderReflectError::FragmentSemanticError(
+                SemanticsErrorKind::InvalidDescriptorSet(binding.group),
+            ));
+        }
+        if binding.binding >= MAX_BINDINGS_COUNT {
+            return Err(ShaderReflectError::FragmentSemanticError(
+                SemanticsErrorKind::InvalidBinding(binding.binding),
+            ));
+        }
+
+        Ok(TextureData {
+            // id: texture.id,
+            // descriptor_set,
+            name: &name,
+            binding: binding.binding,
+        })
+    }
+
+    // todo: share this with cross
+    fn reflect_texture_metas(
+        &self,
+        texture: TextureData,
+        pass_number: usize,
+        semantics: &ShaderSemantics,
+        meta: &mut BindingMeta,
+    ) -> Result<(), ShaderReflectError> {
+        let Some(semantic) = semantics
+            .texture_semantics
+            .get_texture_semantic(texture.name)
+        else {
+            return Err(
+                SemanticErrorBlame::Fragment.error(SemanticsErrorKind::UnknownSemantics(
+                    texture.name.to_string(),
+                )),
+            );
+        };
+
+        if semantic.semantics == TextureSemantics::PassOutput && semantic.index >= pass_number {
+            return Err(ShaderReflectError::NonCausalFilterChain {
+                pass: pass_number,
+                target: semantic.index,
+            });
+        }
+
+        meta.texture_meta.insert(
+            semantic,
+            TextureBinding {
+                binding: texture.binding,
+            },
+        );
        Ok(())
    }
 }
+
+impl ReflectShader for NagaReflect {
+    fn reflect(
+        &mut self,
+        pass_number: usize,
+        semantics: &ShaderSemantics,
+    ) -> Result<ShaderReflection, ShaderReflectError> {
+        self.validate()?;
+
+        // Validate verifies that there's only one uniform block.
+        let vertex_ubo = self
+            .vertex
+            .global_variables
+            .iter()
+            .find_map(|(handle, gv)| {
+                if gv.space == AddressSpace::Uniform {
+                    Some(handle)
+                } else {
+                    None
+                }
+            });
+
+        let fragment_ubo = self
+            .fragment
+            .global_variables
+            .iter()
+            .find_map(|(handle, gv)| {
+                if gv.space == AddressSpace::Uniform {
+                    Some(handle)
+                } else {
+                    None
+                }
+            });
+
+        let ubo = self.reflect_ubos(vertex_ubo, fragment_ubo)?;
+
+        let vertex_push = self
+            .vertex
+            .global_variables
+            .iter()
+            .find_map(|(handle, gv)| {
+                if gv.space == AddressSpace::PushConstant {
+                    Some(handle)
+                } else {
+                    None
+                }
+            });
+
+        let fragment_push = self
+            .fragment
+            .global_variables
+            .iter()
+            .find_map(|(handle, gv)| {
+                if gv.space == AddressSpace::PushConstant {
+                    Some(handle)
+                } else {
+                    None
+                }
+            });
+
+        let push_constant = self.reflect_push_constant_buffer(vertex_push, fragment_push)?;
+
+        let mut meta = BindingMeta::default();
+
+        if let Some(ubo) = vertex_ubo {
+            Self::reflect_buffer_struct_members(
+                &self.vertex,
+                ubo,
+                pass_number,
+                semantics,
+                &mut meta,
+                UniformMemberBlock::Ubo,
+                SemanticErrorBlame::Vertex,
+            )?;
+        }
+
+        if let Some(ubo) = fragment_ubo {
+            Self::reflect_buffer_struct_members(
+                &self.fragment,
+                ubo,
+                pass_number,
+                semantics,
+                &mut meta,
+                UniformMemberBlock::Ubo,
+                SemanticErrorBlame::Fragment,
+            )?;
+        }
+
+        if let Some(push) = vertex_push {
+            Self::reflect_buffer_struct_members(
+                &self.vertex,
+                push,
+                pass_number,
+                semantics,
+                &mut meta,
+                UniformMemberBlock::PushConstant,
+                SemanticErrorBlame::Vertex,
+            )?;
+        }
+
+        if let Some(push) = fragment_push {
+            Self::reflect_buffer_struct_members(
+                &self.fragment,
+                push,
+                pass_number,
+                semantics,
+                &mut meta,
+                UniformMemberBlock::PushConstant,
+                SemanticErrorBlame::Fragment,
+            )?;
+        }
+
+        let mut ubo_bindings = 0u16;
+        if vertex_ubo.is_some() || fragment_ubo.is_some() {
+            ubo_bindings = 1 << ubo.as_ref().expect("UBOs should be present").binding;
+        }
+
+        let textures = self.fragment.global_variables.iter().filter(|(_, gv)| {
+            let ty = &self.fragment.types[gv.ty];
+            matches!(ty.inner, TypeInner::Image { .. })
+        });
+
+        for (_, texture) in textures {
+            let texture_data = self.reflect_texture(texture)?;
+            if ubo_bindings & (1 << texture_data.binding) != 0 {
+                return Err(ShaderReflectError::BindingInUse(texture_data.binding));
+            }
+            ubo_bindings |= 1 << texture_data.binding;
+
+            self.reflect_texture_metas(texture_data, pass_number, semantics, &mut meta)?;
+        }
+
+        Ok(ShaderReflection {
+            ubo,
+            push_constant,
+            meta,
+        })
+    }
+}
+
 #[cfg(test)]
 mod test {
-    use librashader_preprocess::ShaderSource;
-    use rspirv::dr::Instruction;
-    use rspirv::spirv::Op;

-    #[test]
-    pub fn test_into() {
-        let result = ShaderSource::load("../test/slang-shaders/crt/shaders/crt-royale/src/crt-royale-scanlines-horizontal-apply-mask.slang").unwrap();
-        let compilation = crate::front::GlslangCompilation::try_from(&result).unwrap();
-
-        let mut loader = rspirv::dr::Loader::new();
-        rspirv::binary::parse_words(compilation.vertex.as_binary(), &mut loader).unwrap();
-        let module = loader.module();
-
-        let outputs: Vec<&Instruction> = module
-            .types_global_values
-            .iter()
-            .filter(|i| i.class.opcode == Op::Variable)
-            .collect();
-
-        println!("{outputs:#?}");
-    }
+    // #[test]
+    // pub fn test_into() {
+    //     let result = ShaderSource::load("../test/slang-shaders/crt/shaders/crt-royale/src/crt-royale-scanlines-horizontal-apply-mask.slang").unwrap();
+    //     let compilation = crate::front::GlslangCompilation::try_from(&result).unwrap();
+    //
+    //     let mut loader = rspirv::dr::Loader::new();
+    //     rspirv::binary::parse_words(compilation.vertex.as_binary(), &mut loader).unwrap();
+    //     let module = loader.module();
+    //
+    //     let outputs: Vec<&Instruction> = module
+    //         .types_global_values
+    //         .iter()
+    //         .filter(|i| i.class.opcode == Op::Variable)
+    //         .collect();
+    //
+    //     println!("{outputs:#?}");
+    // }
 }
--- a/librashader-reflect/src/reflect/semantics.rs
+++ b/librashader-reflect/src/reflect/semantics.rs
@ -169,26 +169,17 @@ bitflags! {
    }
 }

-/// Reflection information for the Uniform Buffer
+/// Reflection information for the Uniform Buffer or Push Constant Block
 #[derive(Debug)]
-pub struct UboReflection {
-    /// The binding point for this UBO.
-    pub binding: u32,
-    /// The size of the UBO buffer. UBO sizes returned by reflection is always aligned to a 16 byte boundary.
+pub struct BufferReflection<T> {
+    /// The binding point for this buffer, if applicable
+    pub binding: T,
+    /// The size of the buffer. Buffer sizes returned by reflection is always aligned to a 16 byte boundary.
    pub size: u32,
    /// The mask indicating for which stages the UBO should be bound.
    pub stage_mask: BindingStage,
 }

-/// Reflection information for the Push Constant Block
-#[derive(Debug)]
-pub struct PushReflection {
-    /// The size of the Push Constant range. The size returned by reflection is always aligned to a 16 byte boundary.
-    pub size: u32,
-    /// The mask indicating for which stages the Push Constant range should be bound.
-    pub stage_mask: BindingStage,
-}
-
 /// The offset of a uniform member.
 ///
 /// A uniform can be bound to both the UBO, or as a Push Constant.
@ -279,9 +270,9 @@ pub struct TextureBinding {
 #[derive(Debug)]
 pub struct ShaderReflection {
    /// Reflection information about the UBO for this shader.
-    pub ubo: Option<UboReflection>,
+    pub ubo: Option<BufferReflection<u32>>,
    /// Reflection information about the Push Constant range for this shader.
-    pub push_constant: Option<PushReflection>,
+    pub push_constant: Option<BufferReflection<Option<u32>>>,
    /// Metadata about the bindings required for this shader.
    pub meta: BindingMeta,
 }
--- a/librashader-runtime-d3d11/Cargo.toml
+++ b/librashader-runtime-d3d11/Cargo.toml
@ -3,7 +3,7 @@ name = "librashader-runtime-d3d11"
 edition = "2021"

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -12,13 +12,13 @@ keywords = ["shader", "retroarch", "SPIR-V"]
 description = "RetroArch shaders for all."

 [dependencies]
-librashader-common = { path = "../librashader-common", features = ["d3d11"], version = "0.2.0-beta.2" }
-librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.2" }
-librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.2" }
-librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.2", features = ["standalone"]  }
-librashader-runtime = { path = "../librashader-runtime", version = "0.2.0-beta.2" }
+librashader-common = { path = "../librashader-common", features = ["d3d11"], version = "0.2.0-beta.5" }
+librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.5" }
+librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.5" }
+librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.5", features = ["standalone"]  }
+librashader-runtime = { path = "../librashader-runtime", version = "0.2.0-beta.5" }
 spirv_cross = { package = "librashader-spirv-cross", version = "0.23" }
-librashader-cache = { path = "../librashader-cache", version = "0.2.0-beta.2", features = ["d3d"] }
+librashader-cache = { path = "../librashader-cache", version = "0.2.0-beta.5", features = ["d3d"] }

 thiserror = "1.0.37"
 rustc-hash = "1.1.0"
--- a/librashader-runtime-d3d11/src/filter_chain.rs
+++ b/librashader-runtime-d3d11/src/filter_chain.rs
@ -265,7 +265,9 @@ impl FilterChainD3D11 {
                disable_cache,
            )?;

-            let ubo_cbuffer = if let Some(ubo) = &reflection.ubo && ubo.size != 0 {
+            let ubo_cbuffer = if let Some(ubo) = &reflection.ubo
+                && ubo.size != 0
+            {
                let buffer = FilterChainD3D11::create_constant_buffer(device, ubo.size)?;
                Some(ConstantBufferBinding {
                    binding: ubo.binding,
@ -277,7 +279,9 @@ impl FilterChainD3D11 {
                None
            };

-            let push_cbuffer = if let Some(push) = &reflection.push_constant && push.size != 0 {
+            let push_cbuffer = if let Some(push) = &reflection.push_constant
+                && push.size != 0
+            {
                let buffer = FilterChainD3D11::create_constant_buffer(device, push.size)?;
                Some(ConstantBufferBinding {
                    binding: if ubo_cbuffer.is_some() { 1 } else { 0 },
--- a/librashader-runtime-d3d11/src/util.rs
+++ b/librashader-runtime-d3d11/src/util.rs
@ -97,7 +97,8 @@ pub fn d3d11_get_closest_format(
    for supported in format_support_list {
        unsafe {
            if let Ok(supported_format) = device.CheckFormatSupport(*supported)
-                && (supported_format & format_support_mask) == format_support_mask {
+                && (supported_format & format_support_mask) == format_support_mask
+            {
                return *supported;
            }
        }
--- a/librashader-runtime-d3d11/tests/triangle.rs
+++ b/librashader-runtime-d3d11/tests/triangle.rs
@ -13,7 +13,7 @@ use librashader_runtime_d3d11::options::FilterChainOptionsD3D11;
 //     "../test/Mega_Bezel_Packs/Duimon-Mega-Bezel/Presets/Advanced/Nintendo_GBA_SP/GBA_SP-[ADV]-[LCD-GRID].slangp";

 const FILTER_PATH: &str =
-    "../test/shaders_slang/bezel/Mega_Bezel/Presets/MBZ__0__SMOOTH-ADV.slangp";
+    "../test/shaders-slang/bezel/Mega_Bezel/Presets/MBZ__0__SMOOTH-ADV.slangp";

 // const FILTER_PATH: &str = "../test/slang-shaders/test/history.slangp";
 // const FILTER_PATH: &str = "../test/slang-shaders/test/feedback.slangp";
--- a/librashader-runtime-d3d12/Cargo.toml
+++ b/librashader-runtime-d3d12/Cargo.toml
@ -3,7 +3,7 @@ name = "librashader-runtime-d3d12"
 edition = "2021"

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -12,12 +12,12 @@ keywords = ["shader", "retroarch", "SPIR-V"]
 description = "RetroArch shaders for all."

 [dependencies]
-librashader-common = { path = "../librashader-common", features = ["d3d12"], version = "0.2.0-beta.2" }
-librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.2" }
-librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.2" }
-librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.2", features = ["dxil", "standalone"]  }
-librashader-runtime = { path = "../librashader-runtime", version = "0.2.0-beta.2" }
-librashader-cache = { path = "../librashader-cache", version = "0.2.0-beta.2", features = ["d3d"] }
+librashader-common = { path = "../librashader-common", features = ["d3d12"], version = "0.2.0-beta.5" }
+librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.5" }
+librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.5" }
+librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.5", features = ["dxil", "standalone"]  }
+librashader-runtime = { path = "../librashader-runtime", version = "0.2.0-beta.5" }
+librashader-cache = { path = "../librashader-cache", version = "0.2.0-beta.5", features = ["d3d"] }

 thiserror = "1.0.37"
 spirv_cross = { package = "librashader-spirv-cross", version = "0.23" }
--- a/librashader-runtime-d3d12/src/descriptor_heap.rs
+++ b/librashader-runtime-d3d12/src/descriptor_heap.rs
@ -15,7 +15,6 @@ use windows::Win32::Graphics::Direct3D12::{
    D3D12_GPU_DESCRIPTOR_HANDLE,
 };

-#[const_trait]
 pub trait D3D12HeapType {
    fn get_desc(size: usize) -> D3D12_DESCRIPTOR_HEAP_DESC;
 }
@ -36,7 +35,7 @@ pub struct ResourceWorkHeap;
 #[derive(Clone)]
 pub struct SamplerWorkHeap;

-impl const D3D12HeapType for SamplerPaletteHeap {
+impl D3D12HeapType for SamplerPaletteHeap {
    // sampler palettes just get set directly
    fn get_desc(size: usize) -> D3D12_DESCRIPTOR_HEAP_DESC {
        D3D12_DESCRIPTOR_HEAP_DESC {
@ -48,7 +47,7 @@ impl const D3D12HeapType for SamplerPaletteHeap {
    }
 }

-impl const D3D12HeapType for CpuStagingHeap {
+impl D3D12HeapType for CpuStagingHeap {
    // Lut texture heaps are CPU only and get bound to the descriptor heap of the shader.
    fn get_desc(size: usize) -> D3D12_DESCRIPTOR_HEAP_DESC {
        D3D12_DESCRIPTOR_HEAP_DESC {
@ -60,7 +59,7 @@ impl const D3D12HeapType for CpuStagingHeap {
    }
 }

-impl const D3D12HeapType for RenderTargetHeap {
+impl D3D12HeapType for RenderTargetHeap {
    // Lut texture heaps are CPU only and get bound to the descriptor heap of the shader.
    fn get_desc(size: usize) -> D3D12_DESCRIPTOR_HEAP_DESC {
        D3D12_DESCRIPTOR_HEAP_DESC {
@ -73,7 +72,7 @@ impl const D3D12HeapType for RenderTargetHeap {
 }

 impl D3D12ShaderVisibleHeapType for ResourceWorkHeap {}
-impl const D3D12HeapType for ResourceWorkHeap {
+impl D3D12HeapType for ResourceWorkHeap {
    // Lut texture heaps are CPU only and get bound to the descriptor heap of the shader.
    fn get_desc(size: usize) -> D3D12_DESCRIPTOR_HEAP_DESC {
        D3D12_DESCRIPTOR_HEAP_DESC {
@ -86,7 +85,7 @@ impl const D3D12HeapType for ResourceWorkHeap {
 }

 impl D3D12ShaderVisibleHeapType for SamplerWorkHeap {}
-impl const D3D12HeapType for SamplerWorkHeap {
+impl D3D12HeapType for SamplerWorkHeap {
    // Lut texture heaps are CPU only and get bound to the descriptor heap of the shader.
    fn get_desc(size: usize) -> D3D12_DESCRIPTOR_HEAP_DESC {
        D3D12_DESCRIPTOR_HEAP_DESC {
--- a/librashader-runtime-d3d12/src/filter_chain.rs
+++ b/librashader-runtime-d3d12/src/filter_chain.rs
@ -477,16 +477,15 @@ impl FilterChainD3D12 {
                    .into();

                    // incredibly cursed.
-                    let (reflection, graphics_pipeline) = if !force_hlsl &&
-                    let Ok(graphics_pipeline) =
-                        D3D12GraphicsPipeline::new_from_dxil(
+                    let (reflection, graphics_pipeline) = if !force_hlsl
+                        && let Ok(graphics_pipeline) = D3D12GraphicsPipeline::new_from_dxil(
                            device,
                            library,
                            validator,
                            &dxil,
                            root_signature,
                            render_format,
-                            disable_cache
+                            disable_cache,
                        ) {
                        (dxil_reflection, graphics_pipeline)
                    } else {
@ -500,7 +499,7 @@ impl FilterChainD3D12 {
                            &hlsl,
                            root_signature,
                            render_format,
-                        disable_cache
+                            disable_cache,
                        )?;
                        (hlsl_reflection, graphics_pipeline)
                    };
--- a/librashader-runtime-d3d12/src/lib.rs
+++ b/librashader-runtime-d3d12/src/lib.rs
@ -1,9 +1,7 @@
 #![cfg(target_os = "windows")]
-#![feature(const_trait_impl)]
+#![deny(unsafe_op_in_unsafe_fn)]
 #![feature(let_chains)]
 #![feature(type_alias_impl_trait)]
-#![feature(int_roundings)]
-#![deny(unsafe_op_in_unsafe_fn)]

 mod buffer;
 mod descriptor_heap;
--- a/librashader-runtime-d3d12/tests/triangle.rs
+++ b/librashader-runtime-d3d12/tests/triangle.rs
@ -7,7 +7,7 @@ fn triangle_d3d12() {
    let sample = hello_triangle::d3d12_hello_triangle::Sample::new(
        // "../test/shaders_slang/crt/crt-lottes.slangp",
        // "../test/basic.slangp",
-        "../test/shaders_slang/handheld/console-border/gbc-lcd-grid-v2.slangp",
+        // "../test/shaders_slang/handheld/console-border/gbc-lcd-grid-v2.slangp",
        // "../test/Mega_Bezel_Packs/Duimon-Mega-Bezel/Presets/Advanced/Nintendo_GBA_SP/GBA_SP-[ADV]-[LCD-GRID]-[Night].slangp",
        // "../test/slang-shaders/test/feedback.slangp",
        // "../test/slang-shaders/test/history.slangp",
--- a/librashader-runtime-gl/Cargo.toml
+++ b/librashader-runtime-gl/Cargo.toml
@ -3,7 +3,7 @@ name = "librashader-runtime-gl"
 edition = "2021"

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -12,12 +12,12 @@ keywords = ["shader", "retroarch", "SPIR-V"]
 description = "RetroArch shaders for all."

 [dependencies]
-librashader-common = { path = "../librashader-common", features = ["opengl"], version = "0.2.0-beta.2" }
-librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.2" }
-librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.2" }
-librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.2", features = ["standalone"]   }
-librashader-runtime = { path = "../librashader-runtime" , version = "0.2.0-beta.2" }
-librashader-cache = { path = "../librashader-cache", version = "0.2.0-beta.2" }
+librashader-common = { path = "../librashader-common", features = ["opengl"], version = "0.2.0-beta.5" }
+librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.5" }
+librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.5" }
+librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.5", features = ["standalone"]   }
+librashader-runtime = { path = "../librashader-runtime" , version = "0.2.0-beta.5" }
+librashader-cache = { path = "../librashader-cache", version = "0.2.0-beta.5" }

 spirv_cross = { package = "librashader-spirv-cross", version = "0.23" }
 rustc-hash = "1.1.0"
@ -26,11 +26,10 @@ bytemuck = "1.12.3"
 thiserror = "1.0.37"
 rayon = "1.6.1"

+sptr = "0.3"
+
 [dev-dependencies]
 glfw = "0.47.0"

-[[test]]
-name = "triangle"
-
 [package.metadata.docs.rs]
 features = ["librashader-cache/docsrs"]
--- a/librashader-runtime-gl/src/binding.rs
+++ b/librashader-runtime-gl/src/binding.rs
@ -68,10 +68,14 @@ where
            && location.bindable()
        {
            if location.is_valid(BindingStage::VERTEX) {
-                unsafe { T::FACTORY(location.vertex, value); }
+                unsafe {
+                    T::FACTORY(location.vertex, value);
+                }
            }
            if location.is_valid(BindingStage::FRAGMENT) {
-                unsafe {  T::FACTORY(location.fragment, value); }
+                unsafe {
+                    T::FACTORY(location.fragment, value);
+                }
            }
            Some(())
        } else {
--- a/librashader-runtime-gl/src/filter_chain/mod.rs
+++ b/librashader-runtime-gl/src/filter_chain/mod.rs
@ -27,12 +27,14 @@ impl FilterChainGL {
        options: Option<&FilterChainOptionsGL>,
    ) -> Result<Self> {
        let result = catch_unwind(|| {
-            if let Some(options) = options && options.use_dsa {
+            if let Some(options) = options
+                && options.use_dsa
+            {
                return Ok(Self {
                    filter: FilterChainDispatch::DirectStateAccess(unsafe {
                        FilterChainImpl::load_from_preset(preset, Some(options))?
-                    })
-                })
+                    }),
+                });
            }
            Ok(Self {
                filter: FilterChainDispatch::Compatibility(unsafe {
--- a/librashader-runtime-gl/src/gl/gl3/draw_quad.rs
+++ b/librashader-runtime-gl/src/gl/gl3/draw_quad.rs
@ -52,7 +52,7 @@ impl DrawQuad for Gl3DrawQuad {
                gl::FLOAT,
                gl::FALSE,
                (4 * std::mem::size_of::<f32>()) as GLsizei,
-                std::ptr::invalid(0),
+                sptr::invalid(0),
            );
            gl::VertexAttribPointer(
                1,
@ -60,7 +60,7 @@ impl DrawQuad for Gl3DrawQuad {
                gl::FLOAT,
                gl::FALSE,
                (4 * std::mem::size_of::<f32>()) as GLsizei,
-                std::ptr::invalid(2 * std::mem::size_of::<f32>()),
+                sptr::invalid(2 * std::mem::size_of::<f32>()),
            );
        }
    }
--- a/librashader-runtime-gl/src/gl/gl3/ubo_ring.rs
+++ b/librashader-runtime-gl/src/gl/gl3/ubo_ring.rs
@ -1,7 +1,7 @@
 use crate::binding::UniformLocation;
 use crate::gl::UboRing;
 use gl::types::{GLsizei, GLsizeiptr, GLuint};
-use librashader_reflect::reflect::semantics::UboReflection;
+use librashader_reflect::reflect::semantics::BufferReflection;
 use librashader_runtime::ringbuffer::InlineRingBuffer;
 use librashader_runtime::ringbuffer::RingBuffer;
 use librashader_runtime::uniforms::UniformStorageAccess;
@ -31,7 +31,7 @@ impl<const SIZE: usize> UboRing<SIZE> for Gl3UboRing<SIZE> {

    fn bind_for_frame(
        &mut self,
-        ubo: &UboReflection,
+        ubo: &BufferReflection<u32>,
        ubo_location: &UniformLocation<GLuint>,
        storage: &impl UniformStorageAccess,
    ) {
--- a/librashader-runtime-gl/src/gl/gl46/ubo_ring.rs
+++ b/librashader-runtime-gl/src/gl/gl46/ubo_ring.rs
@ -1,7 +1,7 @@
 use crate::binding::UniformLocation;
 use crate::gl::UboRing;
 use gl::types::{GLsizei, GLsizeiptr, GLuint};
-use librashader_reflect::reflect::semantics::UboReflection;
+use librashader_reflect::reflect::semantics::BufferReflection;
 use librashader_runtime::ringbuffer::InlineRingBuffer;
 use librashader_runtime::ringbuffer::RingBuffer;
 use librashader_runtime::uniforms::UniformStorageAccess;
@ -30,7 +30,7 @@ impl<const SIZE: usize> UboRing<SIZE> for Gl46UboRing<SIZE> {

    fn bind_for_frame(
        &mut self,
-        ubo: &UboReflection,
+        ubo: &BufferReflection<u32>,
        ubo_location: &UniformLocation<GLuint>,
        storage: &impl UniformStorageAccess,
    ) {
--- a/librashader-runtime-gl/src/gl/mod.rs
+++ b/librashader-runtime-gl/src/gl/mod.rs
@ -13,7 +13,7 @@ use librashader_common::{ImageFormat, Size};
 use librashader_presets::{Scale2D, TextureConfig};
 use librashader_reflect::back::cross::CrossGlslContext;
 use librashader_reflect::back::ShaderCompilerOutput;
-use librashader_reflect::reflect::semantics::{TextureBinding, UboReflection};
+use librashader_reflect::reflect::semantics::{BufferReflection, TextureBinding};
 use librashader_runtime::uniforms::UniformStorageAccess;
 use rustc_hash::FxHashMap;

@ -38,7 +38,7 @@ pub(crate) trait UboRing<const SIZE: usize> {
    fn new(buffer_size: u32) -> Self;
    fn bind_for_frame(
        &mut self,
-        ubo: &UboReflection,
+        ubo: &BufferReflection<u32>,
        ubo_location: &UniformLocation<GLuint>,
        storage: &impl UniformStorageAccess,
    );
--- a/librashader-runtime-gl/src/lib.rs
+++ b/librashader-runtime-gl/src/lib.rs
@ -3,10 +3,8 @@
 //! This crate should not be used directly.
 //! See [`librashader::runtime::gl`](https://docs.rs/librashader/latest/librashader/runtime/gl/index.html) instead.
 #![deny(unsafe_op_in_unsafe_fn)]
-#![feature(strict_provenance)]
 #![feature(let_chains)]
 #![feature(type_alias_impl_trait)]
-#![feature(impl_trait_in_assoc_type)]

 mod binding;
 mod filter_chain;
--- a/librashader-runtime-gl/tests/triangle.rs
+++ b/librashader-runtime-gl/tests/triangle.rs
@ -9,7 +9,7 @@ fn triangle_gl() {

    unsafe {
        let mut filter = FilterChainGL::load_from_path(
-            "../test/shaders_slang/bezel/Mega_Bezel/Presets/MBZ__0__SMOOTH-ADV.slangp",
+            "../test/slang-shaders/bezel/Mega_Bezel/Presets/MBZ__0__SMOOTH-ADV.slangp",
            Some(&FilterChainOptionsGL {
                glsl_version: 0,
                use_dsa: false,
@ -30,8 +30,8 @@ fn triangle_gl46() {
        let mut filter = FilterChainGL::load_from_path(
            // "../test/slang-shaders/vhs/VHSPro.slangp",
            // "../test/slang-shaders/test/history.slangp",
-            // "../test/slang-shaders/bezel/Mega_Bezel/Presets/MBZ__0__SMOOTH-ADV.slangp",
-            "../test/shaders_slang/bezel/Mega_Bezel/Presets/MBZ__0__SMOOTH-ADV.slangp",
+            "../test/slang-shaders/bezel/Mega_Bezel/Presets/MBZ__0__SMOOTH-ADV.slangp",
+            // "../test/shadersslang/bezel/Mega_Bezel/Presets/MBZ__0__SMOOTH-ADV.slangp",
            Some(&FilterChainOptionsGL {
                glsl_version: 0,
                use_dsa: true,
--- a/librashader-runtime-vk/Cargo.toml
+++ b/librashader-runtime-vk/Cargo.toml
@ -3,7 +3,7 @@ name = "librashader-runtime-vk"
 edition = "2021"

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -14,12 +14,12 @@ description = "RetroArch shaders for all."
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html

 [dependencies]
-librashader-common = { path = "../librashader-common", features = ["vulkan"], version = "0.2.0-beta.2" }
-librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.2" }
-librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.2" }
-librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.2", features = []   }
-librashader-runtime = { path = "../librashader-runtime" , version = "0.2.0-beta.2" }
-librashader-cache = { path = "../librashader-cache", version = "0.2.0-beta.2" }
+librashader-common = { path = "../librashader-common", features = ["vulkan"], version = "0.2.0-beta.5" }
+librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.5" }
+librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.5" }
+librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.5", features = []   }
+librashader-runtime = { path = "../librashader-runtime" , version = "0.2.0-beta.5" }
+librashader-cache = { path = "../librashader-cache", version = "0.2.0-beta.5" }

 spirv_cross = { package = "librashader-spirv-cross", version = "0.23" }
 rustc-hash = "1.1.0"
@ -37,9 +37,5 @@ winit = "0.27.5"
 raw-window-handle = "0.5"
 ash-window = "0.12.0"

-
-[[test]]
-name = "triangle"
-
 [package.metadata.docs.rs]
 features = ["librashader-cache/docsrs"]
--- a/librashader-runtime-vk/src/filter_chain.rs
+++ b/librashader-runtime-vk/src/filter_chain.rs
@ -188,7 +188,8 @@ impl FrameResiduals {
        }
        for framebuffer in self.framebuffers.drain(0..) {
            if let Some(framebuffer) = framebuffer
-                && framebuffer != vk::Framebuffer::null() {
+                && framebuffer != vk::Framebuffer::null()
+            {
                unsafe {
                    self.device.destroy_framebuffer(framebuffer, None);
                }
@ -697,8 +698,13 @@ impl FilterChainVulkan {
        // try to hint the optimizer
        assert_eq!(last.len(), 1);
        if let Some(pass) = last.iter_mut().next() {
-            if let Some(format) = pass.graphics_pipeline.render_pass.as_ref().map(|r| r.format)
-                && format != viewport.output.format {
+            if let Some(format) = pass
+                .graphics_pipeline
+                .render_pass
+                .as_ref()
+                .map(|r| r.format)
+                && format != viewport.output.format
+            {
                // need to recompile
                pass.graphics_pipeline.recompile(viewport.output.format)?;
            }
--- a/librashader-runtime-vk/src/filter_pass.rs
+++ b/librashader-runtime-vk/src/filter_pass.rs
@ -124,9 +124,7 @@ impl FilterPass {

        output.output.begin_pass(cmd);

-        let residual = self
-            .graphics_pipeline
-            .begin_rendering(&parent.device, output, cmd)?;
+        let residual = self.graphics_pipeline.begin_rendering(output, cmd)?;

        unsafe {
            parent.device.cmd_bind_pipeline(
@ -178,7 +176,7 @@ impl FilterPass {
                .device
                .cmd_set_viewport(cmd, 0, &[output.output.size.into()]);
            parent.draw_quad.draw_quad(cmd, vbo_type);
-            self.graphics_pipeline.end_rendering(&parent.device, cmd);
+            self.graphics_pipeline.end_rendering(cmd);
        }
        Ok(residual)
    }
--- a/librashader-runtime-vk/src/graphics_pipeline.rs
+++ b/librashader-runtime-vk/src/graphics_pipeline.rs
@ -7,7 +7,7 @@ use crate::render_pass::VulkanRenderPass;
 use ash::vk::PushConstantRange;
 use librashader_cache::cache_pipeline;
 use librashader_reflect::back::ShaderCompilerOutput;
-use librashader_reflect::reflect::semantics::{TextureBinding, UboReflection};
+use librashader_reflect::reflect::semantics::{BufferReflection, TextureBinding};
 use librashader_reflect::reflect::ShaderReflection;
 use librashader_runtime::render_target::RenderTarget;
 use std::ffi::CStr;
@ -30,8 +30,10 @@ impl PipelineDescriptors {
        }
    }

-    pub fn add_ubo_binding(&mut self, ubo_meta: Option<&UboReflection>) {
-        if let Some(ubo_meta) = ubo_meta && !ubo_meta.stage_mask.is_empty() {
+    pub fn add_ubo_binding(&mut self, ubo_meta: Option<&BufferReflection<u32>>) {
+        if let Some(ubo_meta) = ubo_meta
+            && !ubo_meta.stage_mask.is_empty()
+        {
            let ubo_mask = util::binding_stage_to_vulkan_stage(ubo_meta.stage_mask);

            self.layout_bindings.push(vk::DescriptorSetLayoutBinding {
@ -119,18 +121,6 @@ impl PipelineLayoutObjects {

        let pipeline_create_info = pipeline_create_info.push_constant_ranges(push_constant_range);

-        // let pipeline_create_info = if let Some(push_constant) = &reflection.push_constant {
-        //     let stage_mask = util::binding_stage_to_vulkan_stage(push_constant.stage_mask);
-        //     let push_constant_range = vk::PushConstantRange::builder()
-        //         .stage_flags(stage_mask)
-        //         .size(push_constant.size);
-        //     let push_constant_range = [*push_constant_range];
-        //     pipeline_create_info
-        //         .push_constant_ranges(&push_constant_range)
-        // } else {
-        //     pipeline_create_info
-        // };
-
        let layout = unsafe { device.create_pipeline_layout(&pipeline_create_info, None)? };

        let pool_info = vk::DescriptorPoolCreateInfo::builder()
@ -396,14 +386,13 @@ impl VulkanGraphicsPipeline {
    #[inline(always)]
    pub(crate) fn begin_rendering(
        &self,
-        device: &ash::Device,
        output: &RenderTarget<OutputImage>,
        cmd: vk::CommandBuffer,
    ) -> error::Result<Option<vk::Framebuffer>> {
        if let Some(render_pass) = &self.render_pass {
            let attachments = [output.output.image_view];
            let framebuffer = unsafe {
-                device.create_framebuffer(
+                self.device.create_framebuffer(
                    &vk::FramebufferCreateInfo::builder()
                        .render_pass(render_pass.handle)
                        .attachments(&attachments)
@ -430,7 +419,11 @@ impl VulkanGraphicsPipeline {
                    extent: output.output.size.into(),
                });
            unsafe {
-                device.cmd_begin_render_pass(cmd, &render_pass_info, vk::SubpassContents::INLINE);
+                self.device.cmd_begin_render_pass(
+                    cmd,
+                    &render_pass_info,
+                    vk::SubpassContents::INLINE,
+                );
            }
            Ok(Some(framebuffer))
        } else {
@ -450,18 +443,18 @@ impl VulkanGraphicsPipeline {
                .color_attachments(&attachments);

            unsafe {
-                device.cmd_begin_rendering(cmd, &rendering_info);
+                self.device.cmd_begin_rendering(cmd, &rendering_info);
            }
            Ok(None)
        }
    }

-    pub(crate) fn end_rendering(&self, device: &ash::Device, cmd: vk::CommandBuffer) {
+    pub(crate) fn end_rendering(&self, cmd: vk::CommandBuffer) {
        unsafe {
            if self.render_pass.is_none() {
-                device.cmd_end_rendering(cmd);
+                self.device.cmd_end_rendering(cmd);
            } else {
-                device.cmd_end_render_pass(cmd)
+                self.device.cmd_end_render_pass(cmd)
            }
        }
    }
--- a/librashader-runtime-vk/src/lib.rs
+++ b/librashader-runtime-vk/src/lib.rs
@ -5,7 +5,6 @@
 #![deny(unsafe_op_in_unsafe_fn)]
 #![feature(type_alias_impl_trait)]
 #![feature(let_chains)]
-#![feature(strict_provenance)]

 mod draw_quad;
 mod filter_chain;
--- a/librashader-runtime-vk/tests/triangle.rs
+++ b/librashader-runtime-vk/tests/triangle.rs
@ -11,8 +11,8 @@ fn triangle_vk() {

    unsafe {
        let filter = FilterChainVulkan::load_from_path(
-            // "../test/slang-shaders/crt/crt-royale.slangp",
-            "../test/Mega_Bezel_Packs/Duimon-Mega-Bezel/Presets/Advanced/Nintendo_GBA_SP/GBA_SP-[ADV]-[LCD-GRID]-[Night].slangp",
+            "../test/slang-shaders/crt/crt-royale.slangp",
+            // "../test/Mega_Bezel_Packs/Duimon-Mega-Bezel/Presets/Advanced/Nintendo_GBA_SP/GBA_SP-[ADV]-[LCD-GRID]-[Night].slangp",
            &base,
            // "../test/slang-shaders/test/feedback.slancargogp",
            // "../test/basic.slangp",
--- a/librashader-runtime-wgpu/Cargo.toml
+++ b/librashader-runtime-wgpu/Cargo.toml
@ -0,0 +1,38 @@
+[package]
+name = "librashader-runtime-wgpu"
+edition = "2021"
+
+version = "0.2.0-beta.5"
+license = "MPL-2.0 OR GPL-3.0-only"
+authors = ["Ronny Chan <ronny@ronnychan.ca>"]
+repository = "https://github.com/SnowflakePowered/librashader"
+readme = "../README.md"
+categories = ["emulators", "compilers", "graphics"]
+keywords = ["shader", "retroarch", "SPIR-V"]
+description = "RetroArch shaders for all."
+
+# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
+
+[dependencies]
+librashader-common = { path = "../librashader-common", features = ["wgpu"], version = "0.2.0-beta.5" }
+librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.5" }
+librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.5" }
+librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.5", features = ["wgsl"], default-features = false }
+librashader-runtime = { path = "../librashader-runtime" , version = "0.2.0-beta.5" }
+
+wgpu = { version = "0.19.0", features = ["spirv"] }
+rustc-hash = "1.1.0"
+image = "0.24.7"
+thiserror = "1.0.50"
+rayon = "1.8.0"
+bytemuck = { version = "1.14.0", features = ["derive"] }
+array-concat = "0.5.2"
+
+[dev-dependencies]
+config = { version = "0.13.4", features = [] }
+env_logger = "0.10.1"
+raw-window-handle = "0.6.0"
+winit = "0.29.10"
+pollster = "0.3"
+log = "0.4.20"
+
--- a/librashader-runtime-wgpu/shader/blit.wgsl
+++ b/librashader-runtime-wgpu/shader/blit.wgsl
@ -0,0 +1,52 @@
+struct VertexOutput {
+    @builtin(position) position: vec4<f32>,
+    @location(0) tex_coords: vec2<f32>,
+};
+
+// meant to be called with 3 vertex indices: 0, 1, 2
+// draws one large triangle over the clip space like this:
+// (the asterisks represent the clip space bounds)
+//-1,1           1,1
+// ---------------------------------
+// |              *              .
+// |              *           .
+// |              *        .
+// |              *      .
+// |              *    .
+// |              * .
+// |***************
+// |            . 1,-1
+// |          .
+// |       .
+// |     .
+// |   .
+// |.
+@vertex
+fn vs_main(@builtin(vertex_index) vertex_index: u32) -> VertexOutput {
+    var result: VertexOutput;
+    let x = i32(vertex_index) / 2;
+    let y = i32(vertex_index) & 1;
+    let tc = vec2<f32>(
+        f32(x) * 2.0,
+        f32(y) * 2.0
+    );
+    result.position = vec4<f32>(
+        tc.x * 2.0 - 1.0,
+        1.0 - tc.y * 2.0,
+        0.0, 1.0
+    );
+    result.tex_coords = tc;
+    return result;
+}
+
+@group(0)
+@binding(0)
+var r_color: texture_2d<f32>;
+@group(0)
+@binding(1)
+var r_sampler: sampler;
+
+@fragment
+fn fs_main(vertex: VertexOutput) -> @location(0) vec4<f32> {
+    return textureSample(r_color, r_sampler, vertex.tex_coords);
+}
--- a/librashader-runtime-wgpu/shader/triangle.wgsl
+++ b/librashader-runtime-wgpu/shader/triangle.wgsl
@ -0,0 +1,24 @@
+struct VertexInput {
+    @location(0) position: vec3<f32>,
+    @location(1) color: vec3<f32>,
+}
+
+
+struct VertexOutput {
+    @builtin(position) clip_position: vec4<f32>,
+    @location(0) color: vec3<f32>,
+};
+
+
+@vertex
+fn vs_main(model: VertexInput) -> VertexOutput {
+    var out: VertexOutput;
+    out.color = model.color;
+    out.clip_position = vec4(model.position, 1.0);
+    return out;
+}
+
+@fragment
+fn fs_main(in: VertexOutput) -> @location(0) vec4<f32> {
+    return vec4(in.color, 1.0);
+}
--- a/librashader-runtime-wgpu/src/buffer.rs
+++ b/librashader-runtime-wgpu/src/buffer.rs
@ -0,0 +1,51 @@
+use std::ops::{Deref, DerefMut};
+use std::sync::Arc;
+
+pub struct WgpuStagedBuffer {
+    buffer: wgpu::Buffer,
+    shadow: Box<[u8]>,
+}
+
+impl WgpuStagedBuffer {
+    pub fn new(
+        device: &Arc<wgpu::Device>,
+        usage: wgpu::BufferUsages,
+        size: wgpu::BufferAddress,
+        label: wgpu::Label<'static>,
+    ) -> WgpuStagedBuffer {
+        let buffer = device.create_buffer(&wgpu::BufferDescriptor {
+            label,
+            size,
+            usage: usage | wgpu::BufferUsages::COPY_DST,
+            mapped_at_creation: false,
+        });
+
+        WgpuStagedBuffer {
+            buffer,
+            shadow: vec![0u8; size as usize].into_boxed_slice(),
+        }
+    }
+
+    pub fn buffer(&self) -> &wgpu::Buffer {
+        &self.buffer
+    }
+
+    /// Write the contents of the backing buffer to the device buffer.
+    pub fn flush(&self, queue: &wgpu::Queue) {
+        queue.write_buffer(&self.buffer, 0, &self.shadow);
+    }
+}
+
+impl Deref for WgpuStagedBuffer {
+    type Target = [u8];
+
+    fn deref(&self) -> &Self::Target {
+        self.shadow.deref()
+    }
+}
+
+impl DerefMut for WgpuStagedBuffer {
+    fn deref_mut(&mut self) -> &mut Self::Target {
+        self.shadow.deref_mut()
+    }
+}
--- a/librashader-runtime-wgpu/src/draw_quad.rs
+++ b/librashader-runtime-wgpu/src/draw_quad.rs
@ -0,0 +1,51 @@
+use array_concat::concat_arrays;
+use librashader_runtime::quad::QuadType;
+use wgpu::util::{BufferInitDescriptor, DeviceExt};
+use wgpu::{Buffer, Device, RenderPass};
+
+#[rustfmt::skip]
+const VBO_OFFSCREEN: [f32; 16] = [
+    // Offscreen
+    -1.0f32, -1.0, 0.0, 0.0,
+    -1.0, 1.0, 0.0, 1.0,
+    1.0, -1.0, 1.0, 0.0,
+    1.0, 1.0, 1.0, 1.0,
+];
+
+#[rustfmt::skip]
+const VBO_DEFAULT_FINAL: [f32; 16] = [
+    // Final
+    0.0f32, 0.0, 0.0, 0.0,
+    0.0, 1.0, 0.0, 1.0,
+    1.0, 0.0, 1.0, 0.0,
+    1.0, 1.0, 1.0, 1.0,
+];
+
+const VBO_DATA: [f32; 32] = concat_arrays!(VBO_OFFSCREEN, VBO_DEFAULT_FINAL);
+
+pub struct DrawQuad {
+    buffer: Buffer,
+}
+
+impl DrawQuad {
+    pub fn new(device: &Device) -> DrawQuad {
+        let buffer = device.create_buffer_init(&BufferInitDescriptor {
+            label: Some("librashader vbo"),
+            contents: bytemuck::cast_slice(&VBO_DATA),
+            usage: wgpu::BufferUsages::VERTEX,
+        });
+
+        DrawQuad { buffer }
+    }
+
+    pub fn draw_quad<'a, 'b: 'a>(&'b self, cmd: &mut RenderPass<'a>, vbo: QuadType) {
+        cmd.set_vertex_buffer(0, self.buffer.slice(0..));
+
+        let offset = match vbo {
+            QuadType::Offscreen => 0..4,
+            QuadType::Final => 4..8,
+        };
+
+        cmd.draw(offset, 0..1)
+    }
+}
--- a/librashader-runtime-wgpu/src/error.rs
+++ b/librashader-runtime-wgpu/src/error.rs
@ -0,0 +1,24 @@
+//! wgpu shader runtime errors.
+use librashader_preprocess::PreprocessError;
+use librashader_presets::ParsePresetError;
+use librashader_reflect::error::{ShaderCompileError, ShaderReflectError};
+use librashader_runtime::image::ImageError;
+use thiserror::Error;
+
+/// Cumulative error type for wgpu filter chains.
+#[derive(Error, Debug)]
+pub enum FilterChainError {
+    #[error("shader preset parse error")]
+    ShaderPresetError(#[from] ParsePresetError),
+    #[error("shader preprocess error")]
+    ShaderPreprocessError(#[from] PreprocessError),
+    #[error("shader compile error")]
+    ShaderCompileError(#[from] ShaderCompileError),
+    #[error("shader reflect error")]
+    ShaderReflectError(#[from] ShaderReflectError),
+    #[error("lut loading error")]
+    LutLoadError(#[from] ImageError),
+}
+
+/// Result type for wgpu filter chains.
+pub type Result<T> = std::result::Result<T, FilterChainError>;
--- a/librashader-runtime-wgpu/src/filter_chain.rs
+++ b/librashader-runtime-wgpu/src/filter_chain.rs
@ -0,0 +1,465 @@
+use librashader_presets::{ShaderPassConfig, ShaderPreset, TextureConfig};
+use librashader_reflect::back::targets::WGSL;
+use librashader_reflect::back::{CompileReflectShader, CompileShader};
+use librashader_reflect::front::GlslangCompilation;
+use librashader_reflect::reflect::presets::{CompilePresetTarget, ShaderPassArtifact};
+use librashader_reflect::reflect::semantics::ShaderSemantics;
+use librashader_reflect::reflect::ReflectShader;
+use librashader_runtime::binding::BindingUtil;
+use librashader_runtime::image::{Image, ImageError, UVDirection};
+use librashader_runtime::quad::QuadType;
+use librashader_runtime::uniforms::UniformStorage;
+use rustc_hash::FxHashMap;
+use std::collections::VecDeque;
+use std::path::Path;
+
+use std::sync::Arc;
+
+use crate::buffer::WgpuStagedBuffer;
+use crate::draw_quad::DrawQuad;
+use librashader_common::{FilterMode, ImageFormat, Size, Viewport, WrapMode};
+use librashader_reflect::back::wgsl::WgslCompileOptions;
+use librashader_runtime::framebuffer::FramebufferInit;
+use librashader_runtime::render_target::RenderTarget;
+use librashader_runtime::scaling::ScaleFramebuffer;
+use rayon::prelude::*;
+use wgpu::{Device, TextureFormat};
+
+use crate::error;
+use crate::error::FilterChainError;
+use crate::filter_pass::FilterPass;
+use crate::framebuffer::WgpuOutputView;
+use crate::graphics_pipeline::WgpuGraphicsPipeline;
+use crate::luts::LutTexture;
+use crate::mipmap::MipmapGen;
+use crate::options::{FilterChainOptionsWgpu, FrameOptionsWgpu};
+use crate::samplers::SamplerSet;
+use crate::texture::{InputImage, OwnedImage};
+
+type ShaderPassMeta =
+    ShaderPassArtifact<impl CompileReflectShader<WGSL, GlslangCompilation> + Send>;
+fn compile_passes(
+    shaders: Vec<ShaderPassConfig>,
+    textures: &[TextureConfig],
+) -> Result<(Vec<ShaderPassMeta>, ShaderSemantics), FilterChainError> {
+    let (passes, semantics) =
+        WGSL::compile_preset_passes::<GlslangCompilation, FilterChainError>(shaders, &textures)?;
+    Ok((passes, semantics))
+}
+
+/// A wgpu filter chain.
+pub struct FilterChainWgpu {
+    pub(crate) common: FilterCommon,
+    passes: Box<[FilterPass]>,
+    output_framebuffers: Box<[OwnedImage]>,
+    feedback_framebuffers: Box<[OwnedImage]>,
+    history_framebuffers: VecDeque<OwnedImage>,
+    disable_mipmaps: bool,
+    mipmapper: MipmapGen,
+}
+
+pub struct FilterMutable {
+    pub passes_enabled: usize,
+    pub(crate) parameters: FxHashMap<String, f32>,
+}
+
+pub(crate) struct FilterCommon {
+    pub output_textures: Box<[Option<InputImage>]>,
+    pub feedback_textures: Box<[Option<InputImage>]>,
+    pub history_textures: Box<[Option<InputImage>]>,
+    pub luts: FxHashMap<usize, LutTexture>,
+    pub samplers: SamplerSet,
+    pub config: FilterMutable,
+    pub internal_frame_count: i32,
+    pub(crate) draw_quad: DrawQuad,
+    device: Arc<Device>,
+    pub(crate) queue: Arc<wgpu::Queue>,
+}
+
+impl FilterChainWgpu {
+    /// Load the shader preset at the given path into a filter chain.
+    pub fn load_from_path(
+        path: impl AsRef<Path>,
+        device: Arc<Device>,
+        queue: Arc<wgpu::Queue>,
+        options: Option<&FilterChainOptionsWgpu>,
+    ) -> error::Result<FilterChainWgpu> {
+        // load passes from preset
+        let preset = ShaderPreset::try_parse(path)?;
+
+        Self::load_from_preset(preset, device, queue, options)
+    }
+
+    /// Load a filter chain from a pre-parsed `ShaderPreset`.
+    pub fn load_from_preset(
+        preset: ShaderPreset,
+        device: Arc<Device>,
+        queue: Arc<wgpu::Queue>,
+        options: Option<&FilterChainOptionsWgpu>,
+    ) -> error::Result<FilterChainWgpu> {
+        let mut cmd = device.create_command_encoder(&wgpu::CommandEncoderDescriptor {
+            label: Some("librashader load cmd"),
+        });
+        let filter_chain = Self::load_from_preset_deferred(
+            preset,
+            Arc::clone(&device),
+            Arc::clone(&queue),
+            &mut cmd,
+            options,
+        )?;
+
+        let cmd = cmd.finish();
+
+        // Wait for device
+        let index = queue.submit([cmd]);
+        device.poll(wgpu::Maintain::WaitForSubmissionIndex(index));
+
+        Ok(filter_chain)
+    }
+
+    /// Load a filter chain from a pre-parsed `ShaderPreset`, deferring and GPU-side initialization
+    /// to the caller. This function therefore requires no external synchronization of the device queue.
+    ///
+    /// ## Safety
+    /// The provided command buffer must be ready for recording and contain no prior commands.
+    /// The caller is responsible for ending the command buffer and immediately submitting it to a
+    /// graphics queue. The command buffer must be completely executed before calling [`frame`](Self::frame).
+    pub fn load_from_preset_deferred(
+        preset: ShaderPreset,
+        device: Arc<Device>,
+        queue: Arc<wgpu::Queue>,
+        cmd: &mut wgpu::CommandEncoder,
+        options: Option<&FilterChainOptionsWgpu>,
+    ) -> error::Result<FilterChainWgpu> {
+        let (passes, semantics) = compile_passes(preset.shaders, &preset.textures)?;
+
+        // // initialize passes
+        let filters = Self::init_passes(Arc::clone(&device), passes, &semantics)?;
+
+        let samplers = SamplerSet::new(&device);
+        let mut mipmapper = MipmapGen::new(Arc::clone(&device));
+        let luts = FilterChainWgpu::load_luts(
+            &device,
+            &queue,
+            cmd,
+            &mut mipmapper,
+            &samplers,
+            &preset.textures,
+        )?;
+        //
+        let framebuffer_gen = || {
+            Ok::<_, error::FilterChainError>(OwnedImage::new(
+                Arc::clone(&device),
+                Size::new(1, 1),
+                1,
+                ImageFormat::R8G8B8A8Unorm,
+            ))
+        };
+        let input_gen = || None;
+        let framebuffer_init = FramebufferInit::new(
+            filters.iter().map(|f| &f.reflection.meta),
+            &framebuffer_gen,
+            &input_gen,
+        );
+
+        //
+        // // initialize output framebuffers
+        let (output_framebuffers, output_textures) = framebuffer_init.init_output_framebuffers()?;
+        //
+        // initialize feedback framebuffers
+        let (feedback_framebuffers, feedback_textures) =
+            framebuffer_init.init_output_framebuffers()?;
+        //
+        // initialize history
+        let (history_framebuffers, history_textures) = framebuffer_init.init_history()?;
+
+        let draw_quad = DrawQuad::new(&device);
+
+        Ok(FilterChainWgpu {
+            common: FilterCommon {
+                luts,
+                samplers,
+                config: FilterMutable {
+                    passes_enabled: preset.shader_count as usize,
+                    parameters: preset
+                        .parameters
+                        .into_iter()
+                        .map(|param| (param.name, param.value))
+                        .collect(),
+                },
+                draw_quad,
+                device,
+                queue,
+                output_textures,
+                feedback_textures,
+                history_textures,
+                internal_frame_count: 0,
+            },
+            passes: filters,
+            output_framebuffers,
+            feedback_framebuffers,
+            history_framebuffers,
+            disable_mipmaps: options.map(|f| f.force_no_mipmaps).unwrap_or(false),
+            mipmapper,
+        })
+    }
+
+    fn load_luts(
+        device: &wgpu::Device,
+        queue: &wgpu::Queue,
+        cmd: &mut wgpu::CommandEncoder,
+        mipmapper: &mut MipmapGen,
+        sampler_set: &SamplerSet,
+        textures: &[TextureConfig],
+    ) -> error::Result<FxHashMap<usize, LutTexture>> {
+        let mut luts = FxHashMap::default();
+        let images = textures
+            .par_iter()
+            .map(|texture| Image::load(&texture.path, UVDirection::TopLeft))
+            .collect::<Result<Vec<Image>, ImageError>>()?;
+        for (index, (texture, image)) in textures.iter().zip(images).enumerate() {
+            let texture =
+                LutTexture::new(device, queue, cmd, image, texture, mipmapper, sampler_set);
+            luts.insert(index, texture);
+        }
+        Ok(luts)
+    }
+
+    fn push_history(&mut self, input: &wgpu::Texture, cmd: &mut wgpu::CommandEncoder) {
+        if let Some(mut back) = self.history_framebuffers.pop_back() {
+            if back.image.size() != input.size() || input.format() != back.image.format() {
+                // old back will get dropped.. do we need to defer?
+                let _old_back = std::mem::replace(
+                    &mut back,
+                    OwnedImage::new(
+                        Arc::clone(&self.common.device),
+                        input.size().into(),
+                        1,
+                        input.format().into(),
+                    ),
+                );
+            }
+
+            back.copy_from(cmd, input);
+
+            self.history_framebuffers.push_front(back)
+        }
+    }
+
+    fn init_passes(
+        device: Arc<Device>,
+        passes: Vec<ShaderPassMeta>,
+        semantics: &ShaderSemantics,
+    ) -> error::Result<Box<[FilterPass]>> {
+        let filters: Vec<error::Result<FilterPass>> = passes
+            .into_par_iter()
+            .enumerate()
+            .map(|(index, (config, source, mut reflect))| {
+                let reflection = reflect.reflect(index, semantics)?;
+                let wgsl = reflect.compile(WgslCompileOptions {
+                    write_pcb_as_ubo: true,
+                    sampler_bind_group: 1,
+                })?;
+
+                let ubo_size = reflection.ubo.as_ref().map_or(0, |ubo| ubo.size as usize);
+                let push_size = reflection
+                    .push_constant
+                    .as_ref()
+                    .map_or(0, |push| push.size as wgpu::BufferAddress);
+
+                let uniform_storage = UniformStorage::new_with_storage(
+                    WgpuStagedBuffer::new(
+                        &device,
+                        wgpu::BufferUsages::UNIFORM,
+                        ubo_size as wgpu::BufferAddress,
+                        Some("ubo"),
+                    ),
+                    WgpuStagedBuffer::new(
+                        &device,
+                        wgpu::BufferUsages::UNIFORM,
+                        push_size as wgpu::BufferAddress,
+                        Some("push"),
+                    ),
+                );
+
+                let uniform_bindings = reflection.meta.create_binding_map(|param| param.offset());
+
+                let render_pass_format: Option<TextureFormat> =
+                    if let Some(format) = config.get_format_override() {
+                        format.into()
+                    } else {
+                        source.format.into()
+                    };
+
+                let graphics_pipeline = WgpuGraphicsPipeline::new(
+                    Arc::clone(&device),
+                    &wgsl,
+                    &reflection,
+                    render_pass_format.unwrap_or(TextureFormat::Rgba8Unorm),
+                );
+
+                Ok(FilterPass {
+                    device: Arc::clone(&device),
+                    reflection,
+                    uniform_storage,
+                    uniform_bindings,
+                    source,
+                    config,
+                    graphics_pipeline,
+                })
+            })
+            .collect();
+        //
+        let filters: error::Result<Vec<FilterPass>> = filters.into_iter().collect();
+        let filters = filters?;
+        Ok(filters.into_boxed_slice())
+    }
+
+    /// Records shader rendering commands to the provided command encoder.
+    pub fn frame<'a>(
+        &mut self,
+        input: Arc<wgpu::Texture>,
+        viewport: &Viewport<WgpuOutputView<'a>>,
+        cmd: &mut wgpu::CommandEncoder,
+        frame_count: usize,
+        options: Option<&FrameOptionsWgpu>,
+    ) -> error::Result<()> {
+        let max = std::cmp::min(self.passes.len(), self.common.config.passes_enabled);
+        let passes = &mut self.passes[0..max];
+
+        if let Some(options) = &options {
+            if options.clear_history {
+                for history in &mut self.history_framebuffers {
+                    history.clear(cmd);
+                }
+            }
+        }
+
+        if passes.is_empty() {
+            return Ok(());
+        }
+
+        let original_image_view = input.create_view(&wgpu::TextureViewDescriptor::default());
+
+        let filter = passes[0].config.filter;
+        let wrap_mode = passes[0].config.wrap_mode;
+
+        // update history
+        for (texture, image) in self
+            .common
+            .history_textures
+            .iter_mut()
+            .zip(self.history_framebuffers.iter())
+        {
+            *texture = Some(image.as_input(filter, wrap_mode));
+        }
+
+        let original = InputImage {
+            image: Arc::clone(&input),
+            view: Arc::new(original_image_view),
+            wrap_mode,
+            filter_mode: filter,
+            mip_filter: filter,
+        };
+
+        let mut source = original.clone();
+
+        // swap output and feedback **before** recording command buffers
+        std::mem::swap(
+            &mut self.output_framebuffers,
+            &mut self.feedback_framebuffers,
+        );
+
+        // rescale render buffers to ensure all bindings are valid.
+        OwnedImage::scale_framebuffers_with_context(
+            source.image.size().into(),
+            viewport.output.size,
+            &mut self.output_framebuffers,
+            &mut self.feedback_framebuffers,
+            passes,
+            &(),
+            Some(&mut |index: usize,
+                       pass: &FilterPass,
+                       output: &OwnedImage,
+                       feedback: &OwnedImage| {
+                // refresh inputs
+                self.common.feedback_textures[index] =
+                    Some(feedback.as_input(pass.config.filter, pass.config.wrap_mode));
+                self.common.output_textures[index] =
+                    Some(output.as_input(pass.config.filter, pass.config.wrap_mode));
+                Ok(())
+            }),
+        )?;
+
+        let passes_len = passes.len();
+        let (pass, last) = passes.split_at_mut(passes_len - 1);
+
+        let frame_direction = options.map_or(1, |f| f.frame_direction);
+
+        for (index, pass) in pass.iter_mut().enumerate() {
+            let target = &self.output_framebuffers[index];
+            source.filter_mode = pass.config.filter;
+            source.wrap_mode = pass.config.wrap_mode;
+            source.mip_filter = pass.config.filter;
+
+            let output_image = WgpuOutputView::from(target);
+            let out = RenderTarget::identity(&output_image);
+
+            pass.draw(
+                cmd,
+                index,
+                &self.common,
+                pass.config.get_frame_count(frame_count),
+                frame_direction,
+                viewport,
+                &original,
+                &source,
+                &out,
+                QuadType::Offscreen,
+            )?;
+
+            if target.max_miplevels > 1 && !self.disable_mipmaps {
+                let sampler = self.common.samplers.get(
+                    WrapMode::ClampToEdge,
+                    FilterMode::Linear,
+                    FilterMode::Nearest,
+                );
+
+                target.generate_mipmaps(cmd, &mut self.mipmapper, &sampler);
+            }
+
+            source = self.common.output_textures[index].clone().unwrap();
+        }
+
+        // try to hint the optimizer
+        assert_eq!(last.len(), 1);
+
+        if let Some(pass) = last.iter_mut().next() {
+            if pass.graphics_pipeline.format != viewport.output.format {
+                // need to recompile
+                pass.graphics_pipeline.recompile(viewport.output.format);
+            }
+            source.filter_mode = pass.config.filter;
+            source.wrap_mode = pass.config.wrap_mode;
+            source.mip_filter = pass.config.filter;
+            let output_image = &viewport.output;
+            let out = RenderTarget::viewport_with_output(output_image, viewport);
+            pass.draw(
+                cmd,
+                passes_len - 1,
+                &self.common,
+                pass.config.get_frame_count(frame_count),
+                frame_direction,
+                viewport,
+                &original,
+                &source,
+                &out,
+                QuadType::Final,
+            )?;
+        }
+
+        self.push_history(&input, cmd);
+        self.common.internal_frame_count = self.common.internal_frame_count.wrapping_add(1);
+        Ok(())
+    }
+}
--- a/librashader-runtime-wgpu/src/filter_pass.rs
+++ b/librashader-runtime-wgpu/src/filter_pass.rs
@ -0,0 +1,247 @@
+use crate::buffer::WgpuStagedBuffer;
+use crate::error;
+use crate::filter_chain::FilterCommon;
+use crate::framebuffer::WgpuOutputView;
+use crate::graphics_pipeline::WgpuGraphicsPipeline;
+use crate::samplers::SamplerSet;
+use crate::texture::InputImage;
+use librashader_common::{ImageFormat, Size, Viewport};
+use librashader_preprocess::ShaderSource;
+use librashader_presets::ShaderPassConfig;
+use librashader_reflect::reflect::semantics::{
+    BindingStage, MemberOffset, TextureBinding, UniformBinding,
+};
+use librashader_reflect::reflect::ShaderReflection;
+use librashader_runtime::binding::{BindSemantics, TextureInput};
+use librashader_runtime::filter_pass::FilterPassMeta;
+use librashader_runtime::quad::QuadType;
+use librashader_runtime::render_target::RenderTarget;
+use librashader_runtime::uniforms::{NoUniformBinder, UniformStorage, UniformStorageAccess};
+use rustc_hash::FxHashMap;
+use std::sync::Arc;
+use wgpu::{BindGroupDescriptor, BindGroupEntry, BindingResource, BufferBinding, ShaderStages};
+
+pub struct FilterPass {
+    pub device: Arc<wgpu::Device>,
+    pub reflection: ShaderReflection,
+    pub(crate) uniform_storage:
+        UniformStorage<NoUniformBinder, Option<()>, WgpuStagedBuffer, WgpuStagedBuffer>,
+    pub uniform_bindings: FxHashMap<UniformBinding, MemberOffset>,
+    pub source: ShaderSource,
+    pub config: ShaderPassConfig,
+    pub graphics_pipeline: WgpuGraphicsPipeline,
+}
+
+impl TextureInput for InputImage {
+    fn size(&self) -> Size<u32> {
+        self.image.size().into()
+    }
+}
+
+pub struct WgpuArcBinding<T> {
+    binding: u32,
+    resource: Arc<T>,
+}
+
+impl BindSemantics<NoUniformBinder, Option<()>, WgpuStagedBuffer, WgpuStagedBuffer> for FilterPass {
+    type InputTexture = InputImage;
+    type SamplerSet = SamplerSet;
+    type DescriptorSet<'a> = (
+        &'a mut FxHashMap<u32, WgpuArcBinding<wgpu::TextureView>>,
+        &'a mut FxHashMap<u32, WgpuArcBinding<wgpu::Sampler>>,
+    );
+    type DeviceContext = Arc<wgpu::Device>;
+    type UniformOffset = MemberOffset;
+
+    #[inline(always)]
+    fn bind_texture<'a>(
+        descriptors: &mut Self::DescriptorSet<'a>,
+        samplers: &Self::SamplerSet,
+        binding: &TextureBinding,
+        texture: &Self::InputTexture,
+        _device: &Self::DeviceContext,
+    ) {
+        let sampler = samplers.get(texture.wrap_mode, texture.filter_mode, texture.mip_filter);
+
+        let (texture_binding, sampler_binding) = descriptors;
+        texture_binding.insert(
+            binding.binding,
+            WgpuArcBinding {
+                binding: binding.binding,
+                resource: Arc::clone(&texture.view),
+            },
+        );
+
+        sampler_binding.insert(
+            binding.binding,
+            WgpuArcBinding {
+                binding: binding.binding,
+                resource: sampler,
+            },
+        );
+    }
+}
+
+impl FilterPass {
+    pub(crate) fn draw(
+        &mut self,
+        cmd: &mut wgpu::CommandEncoder,
+        pass_index: usize,
+        parent: &FilterCommon,
+        frame_count: u32,
+        frame_direction: i32,
+        viewport: &Viewport<WgpuOutputView>,
+        original: &InputImage,
+        source: &InputImage,
+        output: &RenderTarget<WgpuOutputView>,
+        vbo_type: QuadType,
+    ) -> error::Result<()> {
+        let mut main_heap = FxHashMap::default();
+        let mut sampler_heap = FxHashMap::default();
+
+        self.build_semantics(
+            pass_index,
+            parent,
+            output.mvp,
+            frame_count,
+            frame_direction,
+            output.output.size,
+            viewport.output.size,
+            original,
+            source,
+            &mut main_heap,
+            &mut sampler_heap,
+        );
+
+        let mut main_heap_array = Vec::with_capacity(main_heap.len() + 1);
+        let mut sampler_heap_array = Vec::with_capacity(sampler_heap.len() + 1);
+
+        for binding in main_heap.values() {
+            main_heap_array.push(BindGroupEntry {
+                binding: binding.binding,
+                resource: BindingResource::TextureView(&binding.resource),
+            })
+        }
+
+        for binding in sampler_heap.values() {
+            sampler_heap_array.push(BindGroupEntry {
+                binding: binding.binding,
+                resource: BindingResource::Sampler(&binding.resource),
+            })
+        }
+
+        if let Some(ubo) = &self.reflection.ubo {
+            main_heap_array.push(BindGroupEntry {
+                binding: ubo.binding,
+                resource: BindingResource::Buffer(BufferBinding {
+                    buffer: self.uniform_storage.inner_ubo().buffer(),
+                    offset: 0,
+                    size: None,
+                }),
+            });
+        }
+
+        let mut has_pcb_buffer = false;
+        if let Some(pcb) = &self.reflection.push_constant {
+            if let Some(binding) = pcb.binding {
+                main_heap_array.push(BindGroupEntry {
+                    binding,
+                    resource: BindingResource::Buffer(BufferBinding {
+                        buffer: self.uniform_storage.inner_push().buffer(),
+                        offset: 0,
+                        size: None,
+                    }),
+                });
+                has_pcb_buffer = true;
+            }
+        }
+
+        let main_bind_group = self.device.create_bind_group(&BindGroupDescriptor {
+            label: Some("main bind group"),
+            layout: &self.graphics_pipeline.layout.main_bind_group_layout,
+            entries: &main_heap_array,
+        });
+
+        let sampler_bind_group = self.device.create_bind_group(&BindGroupDescriptor {
+            label: Some("sampler bind group"),
+            layout: &self.graphics_pipeline.layout.sampler_bind_group_layout,
+            entries: &sampler_heap_array,
+        });
+
+        let mut render_pass = self.graphics_pipeline.begin_rendering(output, cmd);
+
+        render_pass.set_bind_group(0, &main_bind_group, &[]);
+
+        render_pass.set_bind_group(1, &sampler_bind_group, &[]);
+
+        if let Some(push) = &self.reflection.push_constant
+            && !has_pcb_buffer
+        {
+            let mut stage_mask = ShaderStages::empty();
+            if push.stage_mask.contains(BindingStage::FRAGMENT) {
+                stage_mask |= ShaderStages::FRAGMENT;
+            }
+            if push.stage_mask.contains(BindingStage::VERTEX) {
+                stage_mask |= ShaderStages::VERTEX;
+            }
+            render_pass.set_push_constants(stage_mask, 0, self.uniform_storage.push_slice())
+        }
+
+        parent.draw_quad.draw_quad(&mut render_pass, vbo_type);
+
+        Ok(())
+    }
+
+    fn build_semantics<'a>(
+        &mut self,
+        pass_index: usize,
+        parent: &FilterCommon,
+        mvp: &[f32; 16],
+        frame_count: u32,
+        frame_direction: i32,
+        fb_size: Size<u32>,
+        viewport_size: Size<u32>,
+        original: &InputImage,
+        source: &InputImage,
+        main_heap: &'a mut FxHashMap<u32, WgpuArcBinding<wgpu::TextureView>>,
+        sampler_heap: &'a mut FxHashMap<u32, WgpuArcBinding<wgpu::Sampler>>,
+    ) {
+        Self::bind_semantics(
+            &self.device,
+            &parent.samplers,
+            &mut self.uniform_storage,
+            &mut (main_heap, sampler_heap),
+            mvp,
+            frame_count,
+            frame_direction,
+            fb_size,
+            viewport_size,
+            original,
+            source,
+            &self.uniform_bindings,
+            &self.reflection.meta.texture_meta,
+            parent.output_textures[0..pass_index]
+                .iter()
+                .map(|o| o.as_ref()),
+            parent.feedback_textures.iter().map(|o| o.as_ref()),
+            parent.history_textures.iter().map(|o| o.as_ref()),
+            parent.luts.iter().map(|(u, i)| (*u, i.as_ref())),
+            &self.source.parameters,
+            &parent.config.parameters,
+        );
+
+        // flush to buffers
+        self.uniform_storage.inner_ubo().flush(&parent.queue);
+        self.uniform_storage.inner_push().flush(&parent.queue);
+    }
+}
+
+impl FilterPassMeta for FilterPass {
+    fn framebuffer_format(&self) -> ImageFormat {
+        self.source.format
+    }
+
+    fn config(&self) -> &ShaderPassConfig {
+        &self.config
+    }
+}
--- a/librashader-runtime-wgpu/src/framebuffer.rs
+++ b/librashader-runtime-wgpu/src/framebuffer.rs
@ -0,0 +1,47 @@
+use crate::handle::Handle;
+use crate::texture::OwnedImage;
+use librashader_common::Size;
+use wgpu::TextureViewDescriptor;
+
+/// A wgpu `TextureView` with size and texture information to output.
+pub struct WgpuOutputView<'a> {
+    pub(crate) size: Size<u32>,
+    pub(crate) view: Handle<'a, wgpu::TextureView>,
+    pub(crate) format: wgpu::TextureFormat,
+}
+
+impl<'a> WgpuOutputView<'a> {
+    /// Create an output view from an existing texture view, size, and format.
+    pub fn new_from_raw(
+        view: &'a wgpu::TextureView,
+        size: Size<u32>,
+        format: wgpu::TextureFormat,
+    ) -> Self {
+        Self {
+            size,
+            view: Handle::Borrowed(&view),
+            format,
+        }
+    }
+}
+
+#[doc(hidden)]
+impl<'a> From<&'a OwnedImage> for WgpuOutputView<'a> {
+    fn from(image: &'a OwnedImage) -> Self {
+        Self {
+            size: image.size,
+            view: Handle::Borrowed(&image.view),
+            format: image.image.format(),
+        }
+    }
+}
+
+impl From<&wgpu::Texture> for WgpuOutputView<'static> {
+    fn from(image: &wgpu::Texture) -> Self {
+        Self {
+            size: image.size().into(),
+            view: Handle::Owned(image.create_view(&TextureViewDescriptor::default())),
+            format: image.format(),
+        }
+    }
+}
--- a/librashader-runtime-wgpu/src/graphics_pipeline.rs
+++ b/librashader-runtime-wgpu/src/graphics_pipeline.rs
@ -0,0 +1,265 @@
+use crate::framebuffer::WgpuOutputView;
+use crate::util;
+use librashader_reflect::back::wgsl::NagaWgslContext;
+use librashader_reflect::back::ShaderCompilerOutput;
+use librashader_reflect::reflect::ShaderReflection;
+use librashader_runtime::render_target::RenderTarget;
+use std::borrow::Cow;
+use std::sync::Arc;
+use wgpu::{
+    BindGroupLayout, BindGroupLayoutDescriptor, BindGroupLayoutEntry, BindingType,
+    BufferBindingType, BufferSize, CommandEncoder, Operations, PipelineLayout, PushConstantRange,
+    RenderPass, RenderPassColorAttachment, RenderPassDescriptor, SamplerBindingType, ShaderModule,
+    ShaderSource, ShaderStages, TextureFormat, TextureSampleType, TextureViewDimension,
+    VertexBufferLayout,
+};
+
+pub struct WgpuGraphicsPipeline {
+    pub layout: PipelineLayoutObjects,
+    render_pipeline: wgpu::RenderPipeline,
+    pub format: wgpu::TextureFormat,
+}
+
+pub struct PipelineLayoutObjects {
+    layout: PipelineLayout,
+    pub main_bind_group_layout: BindGroupLayout,
+    pub sampler_bind_group_layout: BindGroupLayout,
+    fragment_entry_name: String,
+    vertex_entry_name: String,
+    vertex: ShaderModule,
+    fragment: ShaderModule,
+    device: Arc<wgpu::Device>,
+}
+
+impl PipelineLayoutObjects {
+    pub fn new(
+        reflection: &ShaderReflection,
+        shader_assembly: &ShaderCompilerOutput<String, NagaWgslContext>,
+        device: Arc<wgpu::Device>,
+    ) -> Self {
+        let vertex = device.create_shader_module(wgpu::ShaderModuleDescriptor {
+            label: Some("vertex"),
+            source: ShaderSource::Wgsl(Cow::from(&shader_assembly.vertex)),
+        });
+
+        let fragment = device.create_shader_module(wgpu::ShaderModuleDescriptor {
+            label: Some("fragment"),
+            source: ShaderSource::Wgsl(Cow::from(&shader_assembly.fragment)),
+        });
+
+        let mut main_bindings = Vec::new();
+        let mut sampler_bindings = Vec::new();
+
+        let mut push_constant_range = Vec::new();
+
+        if let Some(push_meta) = reflection.push_constant.as_ref()
+            && !push_meta.stage_mask.is_empty()
+        {
+            let push_mask = util::binding_stage_to_wgpu_stage(push_meta.stage_mask);
+
+            if let Some(binding) = push_meta.binding {
+                main_bindings.push(BindGroupLayoutEntry {
+                    binding,
+                    visibility: push_mask,
+                    ty: BindingType::Buffer {
+                        ty: BufferBindingType::Uniform,
+                        has_dynamic_offset: false,
+                        min_binding_size: BufferSize::new(push_meta.size as u64),
+                    },
+                    count: None,
+                });
+            } else {
+                push_constant_range.push(PushConstantRange {
+                    stages: push_mask,
+                    range: 0..push_meta.size,
+                })
+            }
+        }
+
+        if let Some(ubo_meta) = reflection.ubo.as_ref()
+            && !ubo_meta.stage_mask.is_empty()
+        {
+            let ubo_mask = util::binding_stage_to_wgpu_stage(ubo_meta.stage_mask);
+            main_bindings.push(BindGroupLayoutEntry {
+                binding: ubo_meta.binding,
+                visibility: ubo_mask,
+                ty: BindingType::Buffer {
+                    ty: BufferBindingType::Uniform,
+                    has_dynamic_offset: false,
+                    min_binding_size: BufferSize::new(ubo_meta.size as u64),
+                },
+                count: None,
+            });
+        }
+
+        for texture in reflection.meta.texture_meta.values() {
+            main_bindings.push(BindGroupLayoutEntry {
+                binding: texture.binding,
+                visibility: ShaderStages::FRAGMENT,
+                ty: BindingType::Texture {
+                    sample_type: TextureSampleType::Float { filterable: true },
+                    view_dimension: TextureViewDimension::D2,
+                    multisampled: false,
+                },
+                count: None,
+            });
+
+            sampler_bindings.push(BindGroupLayoutEntry {
+                binding: texture.binding,
+                visibility: ShaderStages::FRAGMENT,
+                ty: BindingType::Sampler(SamplerBindingType::Filtering),
+                count: None,
+            })
+        }
+        let main_bind_group = device.create_bind_group_layout(&BindGroupLayoutDescriptor {
+            label: Some("bind group 0"),
+            entries: &main_bindings,
+        });
+
+        let sampler_bind_group = device.create_bind_group_layout(&BindGroupLayoutDescriptor {
+            label: Some("bind group 1"),
+            entries: &sampler_bindings,
+        });
+
+        let bind_group_layout_refs = [&main_bind_group, &sampler_bind_group];
+
+        let layout = device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
+            label: Some("shader pipeline layout"),
+            bind_group_layouts: &bind_group_layout_refs,
+            push_constant_ranges: &push_constant_range.as_ref(),
+        });
+
+        Self {
+            layout,
+            main_bind_group_layout: main_bind_group,
+            sampler_bind_group_layout: sampler_bind_group,
+            fragment_entry_name: shader_assembly.context.fragment.entry_points[0]
+                .name
+                .clone(),
+            vertex_entry_name: shader_assembly.context.vertex.entry_points[0].name.clone(),
+            vertex,
+            fragment,
+            device,
+        }
+    }
+
+    pub fn create_pipeline(&self, framebuffer_format: TextureFormat) -> wgpu::RenderPipeline {
+        self.device
+            .create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+                label: Some("Render Pipeline"),
+                layout: Some(&self.layout),
+                vertex: wgpu::VertexState {
+                    module: &self.vertex,
+                    entry_point: &self.vertex_entry_name,
+                    buffers: &[VertexBufferLayout {
+                        array_stride: 4 * std::mem::size_of::<f32>() as wgpu::BufferAddress,
+                        step_mode: wgpu::VertexStepMode::Vertex,
+                        attributes: &[
+                            wgpu::VertexAttribute {
+                                format: wgpu::VertexFormat::Float32x2,
+                                offset: 0,
+                                shader_location: 0,
+                            },
+                            wgpu::VertexAttribute {
+                                format: wgpu::VertexFormat::Float32x2,
+                                offset: (2 * std::mem::size_of::<f32>()) as wgpu::BufferAddress,
+                                shader_location: 1,
+                            },
+                        ],
+                    }],
+                },
+                fragment: Some(wgpu::FragmentState {
+                    module: &self.fragment,
+                    entry_point: &self.fragment_entry_name,
+                    targets: &[Some(wgpu::ColorTargetState {
+                        format: framebuffer_format,
+                        blend: Some(wgpu::BlendState::REPLACE),
+                        write_mask: wgpu::ColorWrites::ALL,
+                    })],
+                }),
+                primitive: wgpu::PrimitiveState {
+                    topology: wgpu::PrimitiveTopology::TriangleStrip,
+                    strip_index_format: None,
+                    front_face: wgpu::FrontFace::Ccw,
+                    cull_mode: None,
+                    unclipped_depth: false,
+                    polygon_mode: wgpu::PolygonMode::Fill,
+                    conservative: false,
+                },
+                depth_stencil: None,
+                multisample: wgpu::MultisampleState {
+                    count: 1,
+                    mask: !0,
+                    alpha_to_coverage_enabled: false,
+                },
+                multiview: None,
+            })
+    }
+}
+
+impl WgpuGraphicsPipeline {
+    pub fn new(
+        device: Arc<wgpu::Device>,
+        shader_assembly: &ShaderCompilerOutput<String, NagaWgslContext>,
+        reflection: &ShaderReflection,
+        render_pass_format: TextureFormat,
+    ) -> Self {
+        let layout = PipelineLayoutObjects::new(reflection, shader_assembly, device);
+        let render_pipeline = layout.create_pipeline(render_pass_format);
+        Self {
+            layout,
+            render_pipeline,
+            format: render_pass_format,
+        }
+    }
+
+    pub fn recompile(&mut self, format: TextureFormat) {
+        let render_pipeline = self.layout.create_pipeline(format);
+        self.render_pipeline = render_pipeline;
+    }
+
+    pub(crate) fn begin_rendering<'pass>(
+        &'pass self,
+        output: &RenderTarget<'pass, WgpuOutputView>,
+        cmd: &'pass mut CommandEncoder,
+    ) -> RenderPass<'pass> {
+        let mut render_pass = cmd.begin_render_pass(&RenderPassDescriptor {
+            label: Some("librashader"),
+            color_attachments: &[Some(RenderPassColorAttachment {
+                view: &output.output.view,
+                resolve_target: None,
+                ops: Operations {
+                    load: wgpu::LoadOp::Clear(wgpu::Color {
+                        r: 0.0,
+                        g: 0.0,
+                        b: 0.0,
+                        a: 0.0,
+                    }),
+                    store: wgpu::StoreOp::Store,
+                },
+            })],
+            depth_stencil_attachment: None,
+            timestamp_writes: None,
+            occlusion_query_set: None,
+        });
+
+        render_pass.set_scissor_rect(
+            output.x as u32,
+            output.y as u32,
+            output.output.size.width,
+            output.output.size.height,
+        );
+
+        render_pass.set_viewport(
+            output.x,
+            output.y,
+            output.output.size.width as f32,
+            output.output.size.height as f32,
+            1.0,
+            1.0,
+        );
+
+        render_pass.set_pipeline(&self.render_pipeline);
+        render_pass
+    }
+}
--- a/librashader-runtime-wgpu/src/handle.rs
+++ b/librashader-runtime-wgpu/src/handle.rs
@ -0,0 +1,17 @@
+use std::ops::Deref;
+
+pub enum Handle<'a, T> {
+    Borrowed(&'a T),
+    Owned(T),
+}
+
+impl<T> Deref for Handle<'_, T> {
+    type Target = T;
+
+    fn deref(&self) -> &Self::Target {
+        match self {
+            Handle::Borrowed(r) => &r,
+            Handle::Owned(r) => &r,
+        }
+    }
+}
--- a/librashader-runtime-wgpu/src/lib.rs
+++ b/librashader-runtime-wgpu/src/lib.rs
@ -0,0 +1,28 @@
+//! librashader WGPU runtime
+//!
+//! This crate should not be used directly.
+//! See [`librashader::runtime::wgpu`](https://docs.rs/librashader/latest/librashader/runtime/wgpu/index.html) instead.
+#![deny(unsafe_op_in_unsafe_fn)]
+#![feature(type_alias_impl_trait)]
+#![feature(let_chains)]
+#![feature(strict_provenance)]
+
+mod buffer;
+mod draw_quad;
+mod filter_chain;
+mod filter_pass;
+mod framebuffer;
+mod graphics_pipeline;
+mod handle;
+mod luts;
+mod mipmap;
+mod samplers;
+mod texture;
+mod util;
+
+pub use filter_chain::FilterChainWgpu;
+pub use framebuffer::WgpuOutputView;
+
+
+pub mod error;
+pub mod options;
--- a/librashader-runtime-wgpu/src/luts.rs
+++ b/librashader-runtime-wgpu/src/luts.rs
@ -0,0 +1,87 @@
+use crate::mipmap::MipmapGen;
+use crate::samplers::SamplerSet;
+use crate::texture::InputImage;
+use librashader_common::{Size, WrapMode};
+use librashader_presets::TextureConfig;
+use librashader_runtime::image::Image;
+use librashader_runtime::scaling::MipmapSize;
+use std::sync::Arc;
+use wgpu::TextureDescriptor;
+
+pub(crate) struct LutTexture(InputImage);
+impl AsRef<InputImage> for LutTexture {
+    fn as_ref(&self) -> &InputImage {
+        &self.0
+    }
+}
+
+impl LutTexture {
+    pub fn new(
+        device: &wgpu::Device,
+        queue: &wgpu::Queue,
+        cmd: &mut wgpu::CommandEncoder,
+        image: Image,
+        config: &TextureConfig,
+        mipmapper: &mut MipmapGen,
+        sampler_set: &SamplerSet,
+    ) -> LutTexture {
+        let texture = device.create_texture(&TextureDescriptor {
+            label: Some(&config.name),
+            size: image.size.into(),
+            mip_level_count: if config.mipmap {
+                image.size.calculate_miplevels()
+            } else {
+                1
+            },
+            sample_count: 1,
+            dimension: wgpu::TextureDimension::D2,
+            format: wgpu::TextureFormat::Rgba8Unorm,
+            usage: wgpu::TextureUsages::TEXTURE_BINDING
+                | wgpu::TextureUsages::COPY_DST
+                // need render attachment for mipmaps...
+                | wgpu::TextureUsages::RENDER_ATTACHMENT,
+            view_formats: &[wgpu::TextureFormat::Rgba8Unorm],
+        });
+
+        queue.write_texture(
+            wgpu::ImageCopyTexture {
+                texture: &texture,
+                mip_level: 0,
+                origin: wgpu::Origin3d::ZERO,
+                aspect: wgpu::TextureAspect::All,
+            },
+            &image.bytes,
+            wgpu::ImageDataLayout {
+                offset: 0,
+                bytes_per_row: Some(4 * image.size.width),
+                rows_per_image: None,
+            },
+            image.size.into(),
+        );
+
+        if config.mipmap {
+            mipmapper.generate_mipmaps(
+                cmd,
+                &texture,
+                &*sampler_set.get(
+                    WrapMode::ClampToEdge,
+                    config.filter_mode,
+                    config.filter_mode,
+                ),
+                Size::<u32>::from(texture.size()).calculate_miplevels(),
+            );
+        }
+
+        let view = texture.create_view(&wgpu::TextureViewDescriptor::default());
+
+        let image = InputImage {
+            image: Arc::new(texture),
+            view: Arc::new(view),
+            wrap_mode: config.wrap_mode,
+            filter_mode: config.filter_mode,
+            mip_filter: config.filter_mode,
+        };
+
+        Self(image)
+    }
+}
--- a/librashader-runtime-wgpu/src/mipmap.rs
+++ b/librashader-runtime-wgpu/src/mipmap.rs
@ -0,0 +1,119 @@
+use rustc_hash::FxHashMap;
+use std::borrow::Cow;
+use std::sync::Arc;
+
+pub struct MipmapGen {
+    device: Arc<wgpu::Device>,
+    shader: wgpu::ShaderModule,
+    pipeline_cache: FxHashMap<wgpu::TextureFormat, wgpu::RenderPipeline>,
+}
+
+impl MipmapGen {
+    fn create_pipeline(
+        device: &wgpu::Device,
+        shader: &wgpu::ShaderModule,
+        format: wgpu::TextureFormat,
+    ) -> wgpu::RenderPipeline {
+        let pipeline = device.create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+            label: Some("blit"),
+            layout: None,
+            vertex: wgpu::VertexState {
+                module: &shader,
+                entry_point: "vs_main",
+                buffers: &[],
+            },
+            fragment: Some(wgpu::FragmentState {
+                module: &shader,
+                entry_point: "fs_main",
+                targets: &[Some(format.into())],
+            }),
+            primitive: wgpu::PrimitiveState {
+                topology: wgpu::PrimitiveTopology::TriangleList,
+                ..Default::default()
+            },
+            depth_stencil: None,
+            multisample: wgpu::MultisampleState::default(),
+            multiview: None,
+        });
+
+        pipeline
+    }
+    pub fn new(device: Arc<wgpu::Device>) -> Self {
+        let shader = device.create_shader_module(wgpu::ShaderModuleDescriptor {
+            label: None,
+            source: wgpu::ShaderSource::Wgsl(Cow::Borrowed(include_str!("../shader/blit.wgsl"))),
+        });
+
+        Self {
+            device,
+            shader,
+            pipeline_cache: Default::default(),
+        }
+    }
+
+    pub fn generate_mipmaps(
+        &mut self,
+        cmd: &mut wgpu::CommandEncoder,
+        texture: &wgpu::Texture,
+        sampler: &wgpu::Sampler,
+        miplevels: u32,
+    ) {
+        let format = texture.format();
+        let pipeline = &*self
+            .pipeline_cache
+            .entry(format)
+            .or_insert_with(|| Self::create_pipeline(&self.device, &self.shader, format));
+
+        let views = (0..miplevels)
+            .map(|mip| {
+                texture.create_view(&wgpu::TextureViewDescriptor {
+                    label: Some("mip"),
+                    format: None,
+                    dimension: None,
+                    aspect: wgpu::TextureAspect::All,
+                    base_mip_level: mip,
+                    mip_level_count: Some(1),
+                    base_array_layer: 0,
+                    array_layer_count: None,
+                })
+            })
+            .collect::<Vec<_>>();
+
+        for target_mip in 1..miplevels as usize {
+            let bind_group_layout = pipeline.get_bind_group_layout(0);
+            let bind_group = self.device.create_bind_group(&wgpu::BindGroupDescriptor {
+                layout: &bind_group_layout,
+                entries: &[
+                    wgpu::BindGroupEntry {
+                        binding: 0,
+                        resource: wgpu::BindingResource::TextureView(&views[target_mip - 1]),
+                    },
+                    wgpu::BindGroupEntry {
+                        binding: 1,
+                        resource: wgpu::BindingResource::Sampler(&sampler),
+                    },
+                ],
+                label: None,
+            });
+
+            let mut pass = cmd.begin_render_pass(&wgpu::RenderPassDescriptor {
+                label: None,
+                color_attachments: &[Some(wgpu::RenderPassColorAttachment {
+                    view: &views[target_mip],
+                    resolve_target: None,
+                    ops: wgpu::Operations {
+                        load: wgpu::LoadOp::Clear(wgpu::Color::BLACK),
+                        store: wgpu::StoreOp::Store,
+                    },
+                })],
+                depth_stencil_attachment: None,
+                timestamp_writes: None,
+                occlusion_query_set: None,
+            });
+
+            pass.set_pipeline(&pipeline);
+            pass.set_bind_group(0, &bind_group, &[]);
+            pass.draw(0..3, 0..1);
+        }
+    }
+}
--- a/librashader-runtime-wgpu/src/options.rs
+++ b/librashader-runtime-wgpu/src/options.rs
@ -0,0 +1,20 @@
+//! wgpu shader runtime options.
+
+/// Options for each wgpu shader frame.
+#[repr(C)]
+#[derive(Default, Debug, Clone)]
+pub struct FrameOptionsWgpu {
+    /// Whether or not to clear the history buffers.
+    pub clear_history: bool,
+    /// The direction of rendering.
+    /// -1 indicates that the frames are played in reverse order.
+    pub frame_direction: i32,
+}
+
+/// Options for filter chain creation.
+#[repr(C)]
+#[derive(Default, Debug, Clone)]
+pub struct FilterChainOptionsWgpu {
+    /// Whether or not to explicitly disable mipmap generation regardless of shader preset settings.
+    pub force_no_mipmaps: bool,
+}
--- a/librashader-runtime-wgpu/src/samplers.rs
+++ b/librashader-runtime-wgpu/src/samplers.rs
@ -0,0 +1,63 @@
+use librashader_common::{FilterMode, WrapMode};
+use rustc_hash::FxHashMap;
+use std::sync::Arc;
+use wgpu::{Sampler, SamplerBorderColor, SamplerDescriptor};
+
+pub struct SamplerSet {
+    // todo: may need to deal with differences in mip filter.
+    samplers: FxHashMap<(WrapMode, FilterMode, FilterMode), Arc<Sampler>>,
+}
+
+impl SamplerSet {
+    #[inline(always)]
+    pub fn get(&self, wrap: WrapMode, filter: FilterMode, mipmap: FilterMode) -> Arc<Sampler> {
+        // eprintln!("{wrap}, {filter}, {mip}");
+        // SAFETY: the sampler set is complete for the matrix
+        // wrap x filter x mipmap
+        unsafe {
+            Arc::clone(
+                &self
+                    .samplers
+                    .get(&(wrap, filter, mipmap))
+                    .unwrap_unchecked(),
+            )
+        }
+    }
+
+    pub fn new(device: &wgpu::Device) -> SamplerSet {
+        let mut samplers = FxHashMap::default();
+        let wrap_modes = &[
+            WrapMode::ClampToBorder,
+            WrapMode::ClampToEdge,
+            WrapMode::Repeat,
+            WrapMode::MirroredRepeat,
+        ];
+        for wrap_mode in wrap_modes {
+            for filter_mode in &[FilterMode::Linear, FilterMode::Nearest] {
+                for mipmap_filter in &[FilterMode::Linear, FilterMode::Nearest] {
+                    samplers.insert(
+                        (*wrap_mode, *filter_mode, *mipmap_filter),
+                        Arc::new(device.create_sampler(&SamplerDescriptor {
+                            label: None,
+                            address_mode_u: (*wrap_mode).into(),
+                            address_mode_v: (*wrap_mode).into(),
+                            address_mode_w: (*wrap_mode).into(),
+                            mag_filter: (*filter_mode).into(),
+                            min_filter: (*filter_mode).into(),
+                            mipmap_filter: (*mipmap_filter).into(),
+                            lod_min_clamp: 0.0,
+                            lod_max_clamp: 1000.0,
+                            compare: None,
+                            anisotropy_clamp: 1,
+                            border_color: Some(SamplerBorderColor::TransparentBlack),
+                        })),
+                    );
+                }
+            }
+        }
+
+        // assert all samplers were created.
+        assert_eq!(samplers.len(), wrap_modes.len() * 2 * 2);
+        SamplerSet { samplers }
+    }
+}
--- a/librashader-runtime-wgpu/src/texture.rs
+++ b/librashader-runtime-wgpu/src/texture.rs
@ -0,0 +1,152 @@
+use crate::error::FilterChainError;
+use crate::mipmap::MipmapGen;
+use librashader_common::{FilterMode, ImageFormat, Size, WrapMode};
+use librashader_presets::Scale2D;
+use librashader_runtime::scaling::{MipmapSize, ScaleFramebuffer, ViewportSize};
+use std::sync::Arc;
+
+pub struct OwnedImage {
+    device: Arc<wgpu::Device>,
+    pub image: Arc<wgpu::Texture>,
+    pub view: Arc<wgpu::TextureView>,
+    pub max_miplevels: u32,
+    pub levels: u32,
+    pub size: Size<u32>,
+}
+
+#[derive(Clone)]
+pub struct InputImage {
+    /// A handle to the `VkImage`.
+    pub image: Arc<wgpu::Texture>,
+    pub view: Arc<wgpu::TextureView>,
+    pub wrap_mode: WrapMode,
+    pub filter_mode: FilterMode,
+    pub mip_filter: FilterMode,
+}
+
+impl AsRef<InputImage> for InputImage {
+    fn as_ref(&self) -> &InputImage {
+        &self
+    }
+}
+
+impl OwnedImage {
+    pub fn new(
+        device: Arc<wgpu::Device>,
+        size: Size<u32>,
+        max_miplevels: u32,
+        format: ImageFormat,
+    ) -> Self {
+        let format: Option<wgpu::TextureFormat> = format.into();
+        let format = format.unwrap_or(wgpu::TextureFormat::Rgba8Unorm);
+
+        let texture = device.create_texture(&wgpu::TextureDescriptor {
+            label: None,
+            size: size.into(),
+            mip_level_count: std::cmp::min(max_miplevels, size.calculate_miplevels()),
+            sample_count: 1,
+            dimension: wgpu::TextureDimension::D2,
+            format,
+            usage: wgpu::TextureUsages::TEXTURE_BINDING
+                | wgpu::TextureUsages::RENDER_ATTACHMENT
+                | wgpu::TextureUsages::COPY_DST
+                | wgpu::TextureUsages::COPY_SRC,
+            view_formats: &[format.into()],
+        });
+
+        let view = texture.create_view(&wgpu::TextureViewDescriptor {
+            label: None,
+            format: Some(format),
+            dimension: Some(wgpu::TextureViewDimension::D2),
+            aspect: wgpu::TextureAspect::All,
+            base_mip_level: 0,
+            mip_level_count: None,
+            base_array_layer: 0,
+            array_layer_count: None,
+        });
+
+        Self {
+            device,
+            image: Arc::new(texture),
+            view: Arc::new(view),
+            max_miplevels,
+            levels: std::cmp::min(max_miplevels, size.calculate_miplevels()),
+            size,
+        }
+    }
+
+    pub fn scale(
+        &mut self,
+        scaling: Scale2D,
+        format: ImageFormat,
+        viewport_size: &Size<u32>,
+        source_size: &Size<u32>,
+        mipmap: bool,
+    ) -> Size<u32> {
+        let size = source_size.scale_viewport(scaling, *viewport_size);
+        let format: Option<wgpu::TextureFormat> = format.into();
+        let format = format.unwrap_or(wgpu::TextureFormat::Rgba8Unorm);
+
+        if self.size != size
+            || (mipmap && self.max_miplevels == 1)
+            || (!mipmap && self.max_miplevels != 1)
+            || format != self.image.format()
+        {
+            let mut new = OwnedImage::new(
+                Arc::clone(&self.device),
+                size,
+                self.max_miplevels,
+                format.into(),
+            );
+            std::mem::swap(self, &mut new);
+        }
+        size
+    }
+
+    pub(crate) fn as_input(&self, filter: FilterMode, wrap_mode: WrapMode) -> InputImage {
+        InputImage {
+            image: Arc::clone(&self.image),
+            view: Arc::clone(&self.view),
+            wrap_mode,
+            filter_mode: filter,
+            mip_filter: filter,
+        }
+    }
+
+    pub fn copy_from(&self, cmd: &mut wgpu::CommandEncoder, source: &wgpu::Texture) {
+        cmd.copy_texture_to_texture(
+            source.as_image_copy(),
+            self.image.as_image_copy(),
+            source.size(),
+        )
+    }
+
+    pub fn clear(&self, cmd: &mut wgpu::CommandEncoder) {
+        cmd.clear_texture(&self.image, &wgpu::ImageSubresourceRange::default());
+    }
+    pub fn generate_mipmaps(
+        &self,
+        cmd: &mut wgpu::CommandEncoder,
+        mipmapper: &mut MipmapGen,
+        sampler: &wgpu::Sampler,
+    ) {
+        mipmapper.generate_mipmaps(cmd, &self.image, sampler, self.max_miplevels);
+    }
+}
+
+impl ScaleFramebuffer for OwnedImage {
+    type Error = FilterChainError;
+    type Context = ();
+
+    fn scale(
+        &mut self,
+        scaling: Scale2D,
+        format: ImageFormat,
+        viewport_size: &Size<u32>,
+        source_size: &Size<u32>,
+        should_mipmap: bool,
+        _context: &Self::Context,
+    ) -> Result<Size<u32>, Self::Error> {
+        Ok(self.scale(scaling, format, viewport_size, source_size, should_mipmap))
+    }
+}
--- a/librashader-runtime-wgpu/src/util.rs
+++ b/librashader-runtime-wgpu/src/util.rs
@ -0,0 +1,15 @@
+use librashader_reflect::reflect::semantics::BindingStage;
+use wgpu::ShaderStages;
+
+pub fn binding_stage_to_wgpu_stage(stage_mask: BindingStage) -> ShaderStages {
+    let mut mask = ShaderStages::empty();
+    if stage_mask.contains(BindingStage::VERTEX) {
+        mask |= ShaderStages::VERTEX;
+    }
+
+    if stage_mask.contains(BindingStage::FRAGMENT) {
+        mask |= ShaderStages::FRAGMENT;
+    }
+
+    mask
+}
--- a/librashader-runtime-wgpu/tests/hello_triangle.rs
+++ b/librashader-runtime-wgpu/tests/hello_triangle.rs
@ -0,0 +1,357 @@
+use std::sync::Arc;
+use wgpu::Maintain;
+use winit::{
+    event::*,
+    event_loop::{ControlFlow, EventLoop},
+    window::{Window, WindowBuilder},
+};
+
+use librashader_common::Viewport;
+use librashader_presets::ShaderPreset;
+use librashader_runtime_wgpu::FilterChainWgpu;
+use wgpu::util::DeviceExt;
+use winit::event_loop::EventLoopBuilder;
+use winit::keyboard::{Key, KeyCode, PhysicalKey};
+use winit::platform::windows::EventLoopBuilderExtWindows;
+
+#[cfg(target_arch = "wasm32")]
+use wasm_bindgen::prelude::*;
+
+#[repr(C)]
+#[derive(Clone, Copy, Debug, bytemuck::Pod, bytemuck::Zeroable)]
+struct Vertex {
+    position: [f32; 3],
+    color: [f32; 3],
+}
+impl Vertex {
+    fn desc<'a>() -> wgpu::VertexBufferLayout<'a> {
+        wgpu::VertexBufferLayout {
+            array_stride: std::mem::size_of::<Vertex>() as wgpu::BufferAddress,
+            step_mode: wgpu::VertexStepMode::Vertex,
+            attributes: &[
+                wgpu::VertexAttribute {
+                    offset: 0,
+                    shader_location: 0,
+                    format: wgpu::VertexFormat::Float32x3,
+                },
+                wgpu::VertexAttribute {
+                    offset: std::mem::size_of::<[f32; 3]>() as wgpu::BufferAddress,
+                    shader_location: 1,
+                    format: wgpu::VertexFormat::Float32x3,
+                },
+            ],
+        }
+    }
+}
+
+const VERTICES: &[Vertex] = &[
+    Vertex {
+        // top
+        position: [0.0, 0.5, 0.0],
+        color: [1.0, 0.0, 0.0],
+    },
+    Vertex {
+        // bottom left
+        position: [-0.5, -0.5, 0.0],
+        color: [0.0, 1.0, 0.0],
+    },
+    Vertex {
+        // bottom right
+        position: [0.5, -0.5, 0.0],
+        color: [0.0, 0.0, 1.0],
+    },
+];
+
+struct State<'a> {
+    surface: wgpu::Surface<'a>,
+    device: Arc<wgpu::Device>,
+    queue: Arc<wgpu::Queue>,
+    config: wgpu::SurfaceConfiguration,
+    size: winit::dpi::PhysicalSize<u32>,
+    clear_color: wgpu::Color,
+
+    render_pipeline: wgpu::RenderPipeline,
+
+    vertex_buffer: wgpu::Buffer,
+    num_vertices: u32,
+    chain: FilterChainWgpu,
+    frame_count: usize,
+}
+impl<'a> State<'a> {
+    async fn new(window: &'a Window) -> Self {
+        let size = window.inner_size();
+
+        let instance = wgpu::Instance::default();
+        let surface = instance.create_surface(window).unwrap();
+        // NOTE: could be none, see: https://sotrh.github.io/learn-wgpu/beginner/tutorial2-surface/#state-new
+        let adapter = instance
+            .request_adapter(&wgpu::RequestAdapterOptions {
+                power_preference: wgpu::PowerPreference::default(),
+                compatible_surface: Some(&surface),
+                force_fallback_adapter: false,
+            })
+            .await
+            .unwrap();
+
+        let (device, queue) = adapter
+            .request_device(
+                &wgpu::DeviceDescriptor {
+                    required_features: wgpu::Features::ADDRESS_MODE_CLAMP_TO_BORDER,
+                    required_limits: wgpu::Limits::default(),
+                    label: None,
+                },
+                None,
+            )
+            .await
+            .unwrap();
+        let swapchain_capabilities = surface.get_capabilities(&adapter);
+        let swapchain_format = swapchain_capabilities.formats[0];
+
+        let mut config = wgpu::SurfaceConfiguration {
+            usage: wgpu::TextureUsages::RENDER_ATTACHMENT | wgpu::TextureUsages::COPY_DST,
+            format: swapchain_format,
+            width: size.width,
+            height: size.height,
+            present_mode: wgpu::PresentMode::Fifo,
+            desired_maximum_frame_latency: 2,
+            alpha_mode: swapchain_capabilities.alpha_modes[0],
+            view_formats: vec![],
+        };
+
+        let device = Arc::new(device);
+        let queue = Arc::new(queue);
+
+        let preset =
+            ShaderPreset::try_parse("../test/shaders_slang/crt/crt-royale.slangp").unwrap();
+
+        let chain = FilterChainWgpu::load_from_preset(
+            preset,
+            Arc::clone(&device),
+            Arc::clone(&queue),
+            None,
+        )
+        .unwrap();
+
+        let shader = device.create_shader_module(wgpu::ShaderModuleDescriptor {
+            label: Some("Shader"),
+            source: wgpu::ShaderSource::Wgsl(include_str!("../shader/triangle.wgsl").into()),
+        });
+        let render_pipeline_layout =
+            device.create_pipeline_layout(&wgpu::PipelineLayoutDescriptor {
+                label: Some("Render Pipeline Layout"),
+                bind_group_layouts: &[],
+                push_constant_ranges: &[],
+            });
+        let render_pipeline = device.create_render_pipeline(&wgpu::RenderPipelineDescriptor {
+            label: Some("Render Pipeline"),
+            layout: Some(&render_pipeline_layout),
+            vertex: wgpu::VertexState {
+                module: &shader,
+                entry_point: "vs_main",
+                buffers: &[Vertex::desc()],
+            },
+            fragment: Some(wgpu::FragmentState {
+                module: &shader,
+                entry_point: "fs_main",
+                targets: &[Some(wgpu::ColorTargetState {
+                    format: config.format,
+                    blend: Some(wgpu::BlendState::REPLACE),
+                    write_mask: wgpu::ColorWrites::ALL,
+                })],
+            }),
+            primitive: wgpu::PrimitiveState {
+                topology: wgpu::PrimitiveTopology::TriangleList,
+                strip_index_format: None,
+                front_face: wgpu::FrontFace::Ccw,
+                cull_mode: Some(wgpu::Face::Back),
+                polygon_mode: wgpu::PolygonMode::Fill,
+                unclipped_depth: false,
+                conservative: false,
+            },
+            depth_stencil: None,
+            multisample: wgpu::MultisampleState {
+                count: 1,
+                mask: !0,
+                alpha_to_coverage_enabled: false,
+            },
+            multiview: None,
+        });
+
+        let vertex_buffer = device.create_buffer_init(&wgpu::util::BufferInitDescriptor {
+            label: Some("triangle vertices"),
+            contents: bytemuck::cast_slice(VERTICES),
+            usage: wgpu::BufferUsages::VERTEX,
+        });
+
+        let clear_color = wgpu::Color {
+            r: 0.1,
+            g: 0.2,
+            b: 0.3,
+            a: 1.0,
+        };
+        let num_vertices = VERTICES.len() as u32;
+        Self {
+            surface,
+            device,
+            queue,
+            config,
+            size,
+            clear_color,
+            render_pipeline,
+            vertex_buffer,
+            num_vertices,
+            chain,
+            frame_count: 0,
+        }
+    }
+    fn resize(&mut self, new_size: winit::dpi::PhysicalSize<u32>) {
+        if new_size.width > 0 && new_size.height > 0 {
+            self.size = new_size;
+            self.config.width = new_size.width;
+            self.config.height = new_size.height;
+            self.surface.configure(&self.device, &self.config);
+        }
+    }
+    fn input(&mut self, event: &WindowEvent) -> bool {
+        false
+    }
+    fn update(&mut self) {}
+    fn render(&mut self) -> Result<(), wgpu::SurfaceError> {
+        let output = self.surface.get_current_texture()?;
+
+        let render_output = Arc::new(self.device.create_texture(&wgpu::TextureDescriptor {
+            label: Some("rendertexture"),
+            size: output.texture.size(),
+            mip_level_count: output.texture.mip_level_count(),
+            sample_count: output.texture.sample_count(),
+            dimension: output.texture.dimension(),
+            format: output.texture.format(),
+            usage: wgpu::TextureUsages::TEXTURE_BINDING
+                | wgpu::TextureUsages::RENDER_ATTACHMENT
+                | wgpu::TextureUsages::COPY_DST
+                | wgpu::TextureUsages::COPY_SRC,
+            view_formats: &[output.texture.format()],
+        }));
+
+        let filter_output = Arc::new(self.device.create_texture(&wgpu::TextureDescriptor {
+            label: Some("filteroutput"),
+            size: output.texture.size(),
+            mip_level_count: output.texture.mip_level_count(),
+            sample_count: output.texture.sample_count(),
+            dimension: output.texture.dimension(),
+            format: output.texture.format(),
+            usage: wgpu::TextureUsages::TEXTURE_BINDING
+                | wgpu::TextureUsages::RENDER_ATTACHMENT
+                | wgpu::TextureUsages::COPY_DST
+                | wgpu::TextureUsages::COPY_SRC,
+            view_formats: &[output.texture.format()],
+        }));
+
+        let view = render_output.create_view(&wgpu::TextureViewDescriptor::default());
+
+        let filter_view = filter_output.create_view(&wgpu::TextureViewDescriptor::default());
+
+        let mut encoder = self
+            .device
+            .create_command_encoder(&wgpu::CommandEncoderDescriptor {
+                label: Some("Render Encoder"),
+            });
+
+        {
+            let mut render_pass = encoder.begin_render_pass(&wgpu::RenderPassDescriptor {
+                label: Some("Render Pass"),
+                color_attachments: &[Some(wgpu::RenderPassColorAttachment {
+                    view: &view,
+                    resolve_target: None,
+                    ops: wgpu::Operations {
+                        load: wgpu::LoadOp::Clear(self.clear_color),
+                        store: wgpu::StoreOp::Store,
+                    },
+                })],
+                depth_stencil_attachment: None,
+                timestamp_writes: None,
+                occlusion_query_set: None,
+            });
+            render_pass.set_pipeline(&self.render_pipeline);
+            render_pass.set_vertex_buffer(0, self.vertex_buffer.slice(..));
+            render_pass.draw(0..self.num_vertices, 0..1);
+        }
+
+        self.chain
+            .frame(
+                Arc::clone(&render_output),
+                &Viewport {
+                    x: 0.0,
+                    y: 0.0,
+                    mvp: None,
+                    output: librashader_runtime_wgpu::WgpuOutputView::new_from_raw(
+                        &filter_view,
+                        filter_output.size().into(),
+                        filter_output.format(),
+                    ),
+                },
+                &mut encoder,
+                self.frame_count,
+                None,
+            )
+            .expect("failed to draw frame");
+
+        encoder.copy_texture_to_texture(
+            filter_output.as_image_copy(),
+            output.texture.as_image_copy(),
+            output.texture.size(),
+        );
+
+        self.queue.submit(std::iter::once(encoder.finish()));
+        output.present();
+
+        self.frame_count += 1;
+        Ok(())
+    }
+}
+
+pub fn run() {
+    env_logger::init();
+
+    let event_loop = EventLoopBuilder::new()
+        .with_any_thread(true)
+        .with_dpi_aware(true)
+        .build()
+        .unwrap();
+    let window = WindowBuilder::new().build(&event_loop).unwrap();
+
+    pollster::block_on(async {
+        let mut state = State::new(&window).await;
+        event_loop
+            .run(|event, target| {
+                match event {
+                    Event::WindowEvent {
+                        window_id: _,
+                        event,
+                    } => match event {
+                        WindowEvent::Resized(new_size) => {
+                            state.resize(new_size);
+                            // On macos the window needs to be redrawn manually after resizing
+                            window.request_redraw();
+                        }
+                        WindowEvent::RedrawRequested => {
+                            state.update();
+                            match state.render() {
+                                Ok(_) => {}
+                                Err(wgpu::SurfaceError::Lost | wgpu::SurfaceError::Outdated) => {
+                                    state.resize(state.size)
+                                }
+                                Err(wgpu::SurfaceError::OutOfMemory) => target.exit(),
+                                Err(wgpu::SurfaceError::Timeout) => log::warn!("Surface timeout"),
+                            }
+                        }
+                        WindowEvent::CloseRequested => target.exit(),
+                        _ => {}
+                    },
+                    Event::AboutToWait => window.request_redraw(),
+                    _ => {}
+                }
+            })
+            .unwrap();
+    });
+}
--- a/librashader-runtime-wgpu/tests/triangle.rs
+++ b/librashader-runtime-wgpu/tests/triangle.rs
@ -0,0 +1,6 @@
+mod hello_triangle;
+
+#[test]
+fn triangle_wgpu() {
+    hello_triangle::run()
+}
--- a/librashader-runtime/Cargo.toml
+++ b/librashader-runtime/Cargo.toml
@ -3,7 +3,7 @@ name = "librashader-runtime"
 edition = "2021"

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -12,10 +12,10 @@ keywords = ["shader", "retroarch", "SPIR-V"]
 description = "RetroArch shaders for all."

 [dependencies]
-librashader-common = { path = "../librashader-common", version = "0.2.0-beta.2" }
-librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.2" }
-librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.2" }
-librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.2" }
+librashader-common = { path = "../librashader-common", version = "0.2.0-beta.5" }
+librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.5" }
+librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.5" }
+librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.5" }
 bytemuck = "1.12.3"
 rustc-hash = "1.1.0"
 num-traits = "0.2.15"
--- a/librashader-runtime/src/array_chunks_mut.rs
+++ b/librashader-runtime/src/array_chunks_mut.rs
@ -0,0 +1,87 @@
+/// An iterator over a slice in (non-overlapping) mutable chunks (`N` elements
+/// at a time), starting at the beginning of the slice.
+///
+/// When the slice len is not evenly divided by the chunk size, the last
+/// up to `N-1` elements will be omitted but can be retrieved from
+/// the [`into_remainder`] function from the iterator.
+///
+/// This struct is created by the [`array_chunks_mut`] method on [slices].
+///
+///
+/// [`array_chunks_mut`]: slice::array_chunks_mut
+/// [`into_remainder`]: ../../std/slice/struct.ArrayChunksMut.html#method.into_remainder
+/// [slices]: slice
+#[derive(Debug)]
+#[must_use = "iterators are lazy and do nothing unless consumed"]
+pub struct ArrayChunksMut<'a, T: 'a, const N: usize> {
+    iter: core::slice::IterMut<'a, [T; N]>,
+}
+
+impl<'a, T, const N: usize> ArrayChunksMut<'a, T, N> {
+    #[inline]
+    pub(super) fn new(slice: &'a mut [T]) -> Self {
+        let (array_slice, _rem) = as_chunks_mut(slice);
+        Self {
+            iter: array_slice.iter_mut(),
+        }
+    }
+}
+
+impl<'a, T, const N: usize> Iterator for ArrayChunksMut<'a, T, N> {
+    type Item = &'a mut [T; N];
+
+    #[inline]
+    fn next(&mut self) -> Option<&'a mut [T; N]> {
+        self.iter.next()
+    }
+
+    #[inline]
+    fn size_hint(&self) -> (usize, Option<usize>) {
+        self.iter.size_hint()
+    }
+
+    #[inline]
+    fn count(self) -> usize {
+        self.iter.count()
+    }
+
+    #[inline]
+    fn nth(&mut self, n: usize) -> Option<Self::Item> {
+        self.iter.nth(n)
+    }
+
+    #[inline]
+    fn last(self) -> Option<Self::Item> {
+        self.iter.last()
+    }
+}
+
+/// Splits the slice into a slice of `N`-element arrays,
+/// starting at the beginning of the slice,
+/// and a remainder slice with length strictly less than `N`.
+///
+/// # Panics
+///
+/// Panics if `N` is 0. This check will most probably get changed to a compile time
+/// error before this method gets stabilized.
+///
+#[inline]
+#[must_use]
+fn as_chunks_mut<T, const N: usize>(slice: &mut [T]) -> (&mut [[T; N]], &mut [T]) {
+    unsafe fn as_chunks_unchecked_mut<T, const N: usize>(slice: &mut [T]) -> &mut [[T; N]] {
+        // SAFETY: Caller must guarantee that `N` is nonzero and exactly divides the slice length
+        let new_len = slice.len() / N;
+
+        // SAFETY: We cast a slice of `new_len * N` elements into
+        // a slice of `new_len` many `N` elements chunks.
+        unsafe { core::slice::from_raw_parts_mut(slice.as_mut_ptr().cast(), new_len) }
+    }
+
+    assert!(N != 0, "chunk size must be non-zero");
+    let len = slice.len() / N;
+    let (multiple_of_n, remainder) = slice.split_at_mut(len * N);
+    // SAFETY: We already panicked for zero, and ensured by construction
+    // that the length of the subslice is a multiple of N.
+    let array_slice = unsafe { as_chunks_unchecked_mut(multiple_of_n) };
+    (array_slice, remainder)
+}
--- a/librashader-runtime/src/image.rs
+++ b/librashader-runtime/src/image.rs
@ -2,6 +2,7 @@ pub use image::ImageError;
 use librashader_common::Size;
 use std::marker::PhantomData;

+use crate::array_chunks_mut::ArrayChunksMut;
 use std::path::Path;

 /// An uncompressed raw image ready to upload to GPU buffers.
@ -37,7 +38,8 @@ impl PixelFormat for RGBA8 {

 impl PixelFormat for BGRA8 {
    fn convert(pixels: &mut Vec<u8>) {
-        for [r, _g, b, _a] in pixels.array_chunks_mut::<4>() {
+        assert!(pixels.len() % 4 == 0);
+        for [r, _g, b, _a] in ArrayChunksMut::new(pixels) {
            std::mem::swap(b, r)
        }
    }
--- a/librashader-runtime/src/lib.rs
+++ b/librashader-runtime/src/lib.rs
@ -6,7 +6,6 @@
 //!
 //! If you are _writing_ a librashader runtime implementation, using these traits and helpers will
 //! help in maintaining consistent behaviour in binding semantics and image handling.
-#![feature(array_chunks)]

 /// Scaling helpers.
 pub mod scaling;
@ -37,3 +36,6 @@ pub mod render_target;

 /// Helpers for handling framebuffers.
 pub mod framebuffer;
+
+/// array_chunks_mut polyfill
+mod array_chunks_mut;
--- a/librashader.copr.spec
+++ b/librashader.copr.spec
@ -0,0 +1,51 @@
+%global commit 1dca2a97d03fc6aa531a03ba7aaa9ca3dbcb5a61
+%global shortcommit %(c=%{commit}; echo ${c:0:7})
+
+Name:     librashader
+%define lname librashader0
+%define profile optimized
+Version:    {{{ git_dir_version }}}
+Release:  %autorelease
+Summary:  RetroArch shaders for all
+License:  MPL-2.0
+URL:      https://github.com/SnowflakePowered/%{name}
+%undefine _disable_source_fetch
+Source:   {{{ git_dir_pack }}}
+BuildRequires: pkgconfig(vulkan)
+BuildRequires: pkgconfig(shaderc)
+BuildRequires: cmake
+BuildRequires: gcc
+BuildRequires: git
+BuildRequires: g++
+BuildRequires: ninja-build
+BuildRequires: patchelf
+BuildRequires: rustc
+BuildRequires: cargo
+Requires: vulkan
+
+%description
+RetroArch shader runtime
+
+Summary:        RetroArch shader runtime
+Provides:       librashader
+
+%prep
+{{{ git_dir_setup_macro }}}
+
+%build
+# need to use stable compiler, but enable nightly features
+RUSTC_BOOTSTRAP=1 cargo run -p librashader-build-script -- --profile %{profile}
+
+%install
+mkdir -p %{buildroot}/%{_libdir}
+mkdir -p %{buildroot}/%{_includedir}/librashader
+patchelf --set-soname librashader.so.1 target/%{profile}/librashader.so
+install -m 0755 target/%{profile}/librashader.so %{buildroot}%{_libdir}/librashader.so
+cp target/%{profile}/librashader.h %{buildroot}%{_includedir}/librashader/librashader.h 
+cp include/librashader_ld.h %{buildroot}%{_includedir}/librashader/librashader_ld.h 
+
+
+%files 
+%{_libdir}/librashader.so
+%{_libdir}/librashader.so.1
+%{_includedir}/librashader/
--- a/librashader/Cargo.toml
+++ b/librashader/Cargo.toml
@ -4,7 +4,7 @@ edition = "2021"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html

 license = "MPL-2.0 OR GPL-3.0-only"
-version = "0.2.0-beta.2"
+version = "0.2.0-beta.5"
 authors = ["Ronny Chan <ronny@ronnychan.ca>"]
 repository = "https://github.com/SnowflakePowered/librashader"
 readme = "../README.md"
@ -13,19 +13,21 @@ keywords = ["shader", "retroarch", "SPIR-V"]
 description = "RetroArch shaders for all."

 [dependencies]
-librashader-common = { path = "../librashader-common", version = "0.2.0-beta.2" }
-librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.2" }
-librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.2" }
-librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.2", features = ["standalone"] }
-librashader-runtime  = { path = "../librashader-runtime", version = "0.2.0-beta.2" }
-librashader-runtime-d3d11  = { path = "../librashader-runtime-d3d11", version = "0.2.0-beta.2", optional = true }
-librashader-runtime-d3d12  = { path = "../librashader-runtime-d3d12", version = "0.2.0-beta.2", optional = true }
-librashader-runtime-gl = { path = "../librashader-runtime-gl", version = "0.2.0-beta.2", optional = true }
-librashader-runtime-vk = { path = "../librashader-runtime-vk", version = "0.2.0-beta.2", optional = true }
+librashader-common = { path = "../librashader-common", version = "0.2.0-beta.5" }
+librashader-presets = { path = "../librashader-presets", version = "0.2.0-beta.5" }
+librashader-preprocess = { path = "../librashader-preprocess", version = "0.2.0-beta.5" }
+librashader-reflect = { path = "../librashader-reflect", version = "0.2.0-beta.5", features = ["standalone"] }
+librashader-runtime  = { path = "../librashader-runtime", version = "0.2.0-beta.5" }
+librashader-runtime-d3d11  = { path = "../librashader-runtime-d3d11", version = "0.2.0-beta.5", optional = true }
+librashader-runtime-d3d12  = { path = "../librashader-runtime-d3d12", version = "0.2.0-beta.5", optional = true }
+librashader-runtime-gl = { path = "../librashader-runtime-gl", version = "0.2.0-beta.5", optional = true }
+librashader-runtime-vk = { path = "../librashader-runtime-vk", version = "0.2.0-beta.5", optional = true }
+librashader-runtime-wgpu = { path = "../librashader-runtime-wgpu", version = "0.2.0-beta.5", optional = true }

-librashader-cache = { path = "../librashader-cache", version = "0.2.0-beta.2" }
+librashader-cache = { path = "../librashader-cache", version = "0.2.0-beta.5" }

 ash = { version = "0.37", optional = true }
+wgpu = { version = "0.19.1", optional = true }

 [target.'cfg(windows)'.dependencies.windows]
 version = "0.48.0"
@ -39,17 +41,20 @@ preprocess = []
 presets = []

 # runtimes
+
 runtime-gl = [ "runtime", "reflect-cross", "librashader-common/opengl", "librashader-runtime-gl" ]
 runtime-d3d11 = [ "runtime", "reflect-cross","librashader-common/d3d11", "librashader-runtime-d3d11", "windows/Win32_Graphics_Direct3D11" ]
 runtime-d3d12 = [ "runtime", "reflect-cross", "reflect-dxil", "librashader-common/d3d12", "librashader-runtime-d3d12", "windows/Win32_Graphics_Direct3D12" ]
 runtime-vk = ["runtime", "reflect-cross", "librashader-common/vulkan", "librashader-runtime-vk", "ash" ]
+runtime-wgpu = [ "runtime", "reflect-naga", "librashader-common/wgpu", "librashader-runtime-wgpu", "wgpu" ]

 # reflection
 reflect-cross = ["reflect", "librashader-reflect/cross"]
 reflect-dxil = ["reflect", "librashader-reflect/dxil"]
+reflect-naga = ["reflect", "librashader-reflect/naga"]

-runtime-all = ["runtime-gl", "runtime-d3d11", "runtime-d3d12", "runtime-vk"]
-reflect-all = ["reflect-cross", "reflect-dxil"]
+runtime-all = ["runtime-gl", "runtime-d3d11", "runtime-d3d12", "runtime-vk", "runtime-wgpu"]
+reflect-all = ["reflect-cross", "reflect-dxil", "reflect-naga"]

 # enable all features by default
 default = [ "full" ]
--- a/librashader/src/lib.rs
+++ b/librashader/src/lib.rs
@ -15,7 +15,7 @@
 //! called with appropriate input and output parameters to draw a frame with the shader effect applied.
 //!
 //! ## Runtimes
-//! Currently available runtimes are Vulkan, OpenGL 3.3+ and 4.6 (with DSA), Direct3D 11, and Direct3D 12.
+//! Currently available runtimes are wgpu, Vulkan, OpenGL 3.3+ and 4.6 (with DSA), Direct3D 11, and Direct3D 12.
 //!
 //! The Vulkan runtime requires [`VK_KHR_dynamic_rendering`](https://registry.khronos.org/vulkan/specs/1.3-extensions/man/html/VK_KHR_dynamic_rendering.html)
 //! by default, unless [`FilterChainOptions::use_render_pass`](crate::runtime::vk::FilterChainOptions) is explicitly set. Note that dynamic rendering
@ -24,6 +24,8 @@
 //! The Direct3D 12 runtime requires support for [render passes](https://learn.microsoft.com/en-us/windows/win32/direct3d12/direct3d-12-render-passes), which
 //! have been available since Windows 10, version 1809.
 //!
+//! wgpu support is not available in the librashader C API.
+//!
 //! | **API**     | **Status** | **`librashader` feature** |
 //! |-------------|------------|---------------------------|
 //! | OpenGL 3.3+ | ✔         | `gl`                      |
@ -31,8 +33,9 @@
 //! | Vulkan      | ✔         | `vk`                     |
 //! | Direct3D 11  | ✔        | `d3d11`                 |
 //! | Direct3D 12  | ✔        | `d3d12`                 |
+//! | wgpu        | ✔        |  `wgpu`                 |
 //! | Metal       | ❌        |                         |
-//! | WebGPU      | ❌        |                        |
+//!
 //! ## C API
 //! For documentation on the librashader C API, see [librashader-capi](https://docs.rs/librashader-capi/latest/librashader_capi/),
 //! or [`librashader.h`](https://github.com/SnowflakePowered/librashader/blob/master/include/librashader.h).
@ -135,6 +138,7 @@ pub mod reflect {
        pub use librashader_reflect::back::targets::GLSL;
        pub use librashader_reflect::back::targets::HLSL;
        pub use librashader_reflect::back::targets::SPIRV;
+        pub use librashader_reflect::back::targets::WGSL;
    }

    pub use librashader_reflect::error::*;
@ -146,12 +150,12 @@ pub mod reflect {
        FromCompilation, ShaderCompilerOutput,
    };

+    pub use librashader_reflect::front::GlslangCompilation;
+
    /// Reflection via SPIRV-Cross.
    #[cfg(feature = "reflect-cross")]
    #[doc(cfg(feature = "reflect-cross"))]
    pub mod cross {
-        pub use librashader_reflect::front::GlslangCompilation;
-
        /// The version of GLSL to target.
        ///
        pub use librashader_reflect::back::cross::GlslVersion;
@ -180,6 +184,14 @@ pub mod reflect {
        pub use librashader_reflect::back::dxil::DxilObject;
    }

+    /// Reflection via Naga
+    #[cfg(feature = "reflect-naga")]
+    #[doc(cfg(feature = "reflect-naga"))]
+    pub mod naga {
+        pub use librashader_reflect::back::wgsl::NagaWgslContext;
+        pub use librashader_reflect::back::wgsl::WgslCompileOptions;
+    }
+
    pub use librashader_reflect::reflect::semantics::BindingMeta;

    pub use librashader_reflect::reflect::presets::{CompilePresetTarget, ShaderPassArtifact};
@ -292,6 +304,19 @@ pub mod runtime {
            pub use librashader_runtime_vk::*;
        }
    }
+
+    #[cfg(feature = "runtime-wgpu")]
+    #[doc(cfg(feature = "runtime-wgpu"))]
+    /// Shader runtime for wgpu
+    pub mod wgpu {
+        pub use librashader_runtime_wgpu::{
+            error,
+            options::{
+                FilterChainOptionsWgpu as FilterChainOptions, FrameOptionsWgpu as FrameOptions,
+            },
+            FilterChainWgpu as FilterChain, WgpuOutputView,
+        };
+    }
 }

 pub use librashader_common::{FilterMode, ImageFormat, WrapMode};
--- a/pkg/librashader.pc
+++ b/pkg/librashader.pc
@ -0,0 +1,10 @@
+prefix=/usr/local
+exec_prefix=${prefix}
+libdir=${exec_prefix}/lib
+includedir=${exec_prefix}/include
+ 
+Name: librashader
+Description: RetroArch shaders for all
+Version: 0.2.0
+Libs: -L${libdir} -lvulkan -lrashader
+Cflags: -I${includedir}
--- a/pkg/librashader.spec
+++ b/pkg/librashader.spec
@ -0,0 +1,49 @@
+%global commit 1dca2a97d03fc6aa531a03ba7aaa9ca3dbcb5a61
+%global shortcommit %(c=%{commit}; echo ${c:0:7})
+
+Name:     librashader
+%define lname librashader0
+%define profile optimized
+Version:  0.2.0~beta.2
+Release:  %autorelease
+Summary:  RetroArch shaders for all
+License:  MPL-2.0
+URL:      https://github.com/SnowflakePowered/%{name}
+%undefine _disable_source_fetch
+Source:  https://github.com/SnowflakePowered/%{name}/archive/%{commit}/%{name}-%{shortcommit}.tar.gz
+BuildRequires: pkgconfig(vulkan)
+BuildRequires: pkgconfig(shaderc)
+BuildRequires: cmake
+BuildRequires: gcc
+BuildRequires: g++
+BuildRequires: ninja-build
+BuildRequires: patchelf
+BuildRequires: rustc
+BuildRequires: cargo
+Requires: vulkan
+
+%description
+RetroArch shader runtime
+
+Summary:        RetroArch shader runtime
+Provides:       librashader
+
+%prep
+%autosetup -n %{name}-%{commit}
+
+%build
+# need to use stable compiler, but enable nightly features
+RUSTC_BOOTSTRAP=1 cargo run -p librashader-build-script -- --profile %{profile}
+
+%install
+mkdir -p %{buildroot}/%{_libdir}
+mkdir -p %{buildroot}/%{_includedir}/librashader
+patchelf --set-soname librashader.so.1 target/%{profile}/librashader.so
+install -m 0755 target/%{profile}/librashader.so %{buildroot}%{_libdir}/librashader.so
+cp target/%{profile}/librashader.h %{buildroot}%{_includedir}/librashader/librashader.h 
+cp include/librashader_ld.h %{buildroot}%{_includedir}/librashader/librashader_ld.h 
+
+
+%files 
+%{_libdir}/librashader.so
+%{_includedir}/librashader/
--- a/pkg/obs/PKGBUILD
+++ b/pkg/obs/PKGBUILD
@ -0,0 +1,33 @@
+pkgname=librashader
+pkgver=0.2.0~beta.2
+pkgrel=0
+pkgdesc="RetroArch shader runtime"
+arch=('x86_64' 'aarch64')
+url="https://github.com/SnowflakePowered/librashader"
+license=('MPL-2.0')
+groups=('')
+depends=('vulkan-icd-loader' 'shaderc' 'cmake' 'gcc' 'rust' 'patchelf' 'ninja')
+provides=("$pkgname=$pkgver" 'librashader.so')
+backup=('')
+source=("$pkgname-$pkgver.tar.xz" 'vendor.tar.xz' 'cargo_config')
+cksums=("SKIP" "SKIP" "SKIP") 
+profile="optimized"
+
+build() {
+  cd $pkgname-$pkgver
+  mkdir .cargo               
+  cp "$srcdir/cargo_config" .cargo/config
+  cp -r "$srcdir/vendor" "vendor"
+  cp "$srcdir/Cargo.lock" "Cargo.lock"
+  RUSTC_BOOTSTRAP=1 cargo run -p librashader-build-script -- --profile ${profile}
+
+}
+
+package() {
+  mkdir -p $pkgdir/usr/lib
+  mkdir -p $pkgdir/usr/include/librashader
+  patchelf --set-soname librashader.so.1 $srcdir/$pkgname-$pkgver/target/$profile/librashader.so
+  install -m 0755 $srcdir/$pkgname-$pkgver/target/${profile}/librashader.so $pkgdir/usr/lib/librashader.so
+  cp $srcdir/$pkgname-$pkgver/target/${profile}/librashader.h $pkgdir/usr/include/librashader/librashader.h
+  cp $srcdir/$pkgname-$pkgver/include/librashader_ld.h $pkgdir/usr/include/librashader/librashader_ld.h
+}
--- a/pkg/obs/_service
+++ b/pkg/obs/_service
@ -0,0 +1,17 @@
+<services>
+    <service name="obs_scm" mode="manual">
+        <param name="scm">git</param>
+        <param name="url">https://github.com/SnowflakePowered/librashader</param>
+        <param name="revision">master</param>
+    </service>
+    <service name="cargo_vendor" mode="manual">
+        <param name="srcdir">librashader</param>
+        <param name="compression">xz</param>
+    </service>
+    <service name="tar" mode="buildtime"/>
+    <service name="recompress" mode="buildtime">
+        <param name="file">*.tar</param>
+        <param name="compression">xz</param>
+    </service>
+    <service name="set_version" mode="buildtime"/>
+</services>
--- a/pkg/obs/librashader.spec
+++ b/pkg/obs/librashader.spec
@ -0,0 +1,51 @@
+Name:     librashader
+%define lname librashader0
+%define profile optimized
+Summary:  RetroArch shaders for all
+License:  MPL-2.0
+Version: 0.2.0~beta.2
+Release: 0
+URL:      https://github.com/SnowflakePowered/%{name}
+Source0:  librashader-%{version}.tar.xz
+Source1:  vendor.tar.xz
+Source2:  cargo_config
+BuildRequires: pkgconfig(vulkan)
+BuildRequires: pkgconfig(shaderc)
+BuildRequires: ninja-build
+BuildRequires: patchelf
+BuildRequires: cmake
+BuildRequires: gcc
+BuildRequires: gcc-c++
+BuildRequires: cargo
+BuildRequires: rust
+Requires: vulkan
+
+%description
+RetroArch shader runtime
+
+Summary:        RetroArch shader runtime
+Provides:       librashader
+
+%prep
+%setup -qa1 
+mkdir .cargo                # cargo automatically uses this dir
+cp %{SOURCE2} .cargo/config # and automatically uses this config
+
+%build
+RUSTC_BOOTSTRAP=1 cargo run -p librashader-build-script -- --profile %{profile}
+
+%install
+mkdir -p %{buildroot}/%{_libdir}
+mkdir -p %{buildroot}/%{_includedir}/librashader
+patchelf --set-soname librashader.so.1 target/%{profile}/librashader.so
+install -m 0755 target/%{profile}/librashader.so %{buildroot}%{_libdir}/librashader.so
+cp target/%{profile}/librashader.h %{buildroot}%{_includedir}/librashader/librashader.h 
+cp include/librashader_ld.h %{buildroot}%{_includedir}/librashader/librashader_ld.h 
+
+
+%files 
+%{_libdir}/librashader.so
+%if !0%{?suse_version}
+%{_libdir}/librashader.so.1
+%endif
+%{_includedir}/librashader/
--- a/test/extreme_basic.glsl
+++ b/test/extreme_basic.glsl
@ -0,0 +1,24 @@
+#version 450
+layout(set = 0, binding = 0, std140) uniform UBO
+{
+    mat4 MVP;
+};
+
+#pragma stage vertex
+layout(location = 0) in vec4 Position;
+layout(location = 1) in vec2 TexCoord;
+layout(location = 0) out vec2 vTexCoord;
+void main()
+{
+    gl_Position = MVP * Position;
+    vTexCoord = TexCoord;
+}
+
+#pragma stage fragment
+
+layout(location = 0) out vec4 color;
+layout(binding = 1) uniform sampler2D tex;
+
+void main() {
+    color = texture(tex, vec2(0.0));
+}
--- a/test/extreme_basic.spv
+++ b/test/extreme_basic.spv
--- a/test/extreme_basic_split.frag.glsl
+++ b/test/extreme_basic_split.frag.glsl
@ -0,0 +1,9 @@
+#version 450
+
+layout(location = 0) out vec4 color;
+layout(set = 0, binding = 1) uniform texture2D tex;
+layout(set = 1, binding = 1) uniform sampler tex_sampler;
+
+void main() {
+    color = texture(sampler2D(tex, tex_sampler), vec2(0.0));
+}
--- a/test/extreme_basic_split.spirt
+++ b/test/extreme_basic_split.spirt
@ -0,0 +1,47 @@
+module.dialect = spv.Module(version: 1.0, spv.Capability.{Shader, Sampled1D}, spv.MemoryModel.GLSL450)
+
+module.debug_info = spv.Module.DebugInfo(generator: spv.Tool(id: 0, version: 28))
+
+type T0 = spv.OpTypeImage(SampledType: f32, spv.Dim.2D, Depth: 0, Arrayed: 0, MS: 0, Sampled: 1, spv.ImageFormat.Unknown)
+
+#[spv.Decoration.Location(Location: 0)]
+global_var GV0(spv.StorageClass.Output): f32×4
+
+#[spv.Decoration.Binding(BindingPoint: 1)]
+#[spv.Decoration.DescriptorSet(DescriptorSet: 0)]
+global_var GV1(spv.StorageClass.UniformConstant): T0
+
+#[spv.Decoration.Binding(BindingPoint: 1)]
+#[spv.Decoration.DescriptorSet(DescriptorSet: 1)]
+global_var GV2(spv.StorageClass.UniformConstant): spv.OpTypeSampler
+
+global_var GV3(spv.StorageClass.Private, init: f32×4(0.0, 0.0, 0.0, 0.0)): f32×4
+
+func F0() {
+    v0 = spv.OpLoad(Pointer: &GV1): T0
+    v1 = spv.OpLoad(Pointer: &GV2): spv.OpTypeSampler
+  branch L0
+
+  label L0:
+    v2 = spv.OpSampledImage(Image: v0, Sampler: v1): spv.OpTypeSampledImage(ImageType: T0)
+    v3 = spv.OpImageSampleImplicitLod(SampledImage: v2, Coordinate: f32×2(0.0, 0.0)): f32×4
+    spv.OpStore(Pointer: &GV3, Object: v3)
+  return
+}
+
+#[spv.ExecutionMode.OriginUpperLeft]
+func F1() {
+    _ = spv.OpLoad(Pointer: &GV1): T0
+    _ = spv.OpLoad(Pointer: &GV2): spv.OpTypeSampler
+  branch L0
+
+  label L0:
+    call F0()
+    v0 = spv.OpLoad(Pointer: &GV3): f32×4
+    spv.OpStore(Pointer: &GV0, Object: v0)
+  return
+}
+
+export {
+  spv.OpEntryPoint(spv.ExecutionModel.Fragment, Name: "main"): F1,
+}
--- a/Show more
+++ b/Show more
Author	SHA1	Message	Date
Alex Janka	446b052657	fix double free	2024-02-07 15:52:36 +11:00
Alex Janka	9d926b0051	shaders can be either a path or a string	2024-02-07 15:52:36 +11:00
chyyran	fb2bcc5d52	chore: Release	2024-02-06 21:41:00 -05:00
chyyran	8015a2a796	fmt: cleanup repo	2024-02-06 21:38:17 -05:00
chyyran	5c8428eac8	lib: add wgpu dependency in runtime-wgpu	2024-02-06 21:38:17 -05:00
chyyran	8fb2179ae8	lib: add wgpu dependency in runtime-wgpu	2024-02-06 21:31:00 -05:00
chyyran	f6268a621c	chore: Release	2024-02-06 19:20:37 -05:00
chyyran	bbfd5153da	rt(wgpu): rename OutputView to WgpuOutputView	2024-02-06 19:04:28 -05:00
chyyran	37397ff216	lib: add wgpu to librashader library	2024-02-06 19:04:28 -05:00
chyyran	754e8da620	runtime: remove unused AsDerefable trait	2024-02-06 19:04:28 -05:00
chyyran	6c50880600	reflect: remove spirt test outputs	2024-02-06 19:04:28 -05:00
chyyran	acc9bfeb53	rt(wgpu): add blocking submission API	2024-02-06 19:04:28 -05:00
chyyran	54e86e7b06	fmt: clean up unused imports	2024-02-06 19:04:28 -05:00
chyyran	2450217c29	fmt: cargo fmt	2024-02-06 19:04:28 -05:00
chyyran	962a81c2e3	rt(wgsl): mipmaps	2024-02-06 19:04:28 -05:00
chyyran	11ab4b7c9a	build: tag RPITIT cause ubuntu 23.10 uses an older compiler	2024-02-06 19:04:28 -05:00
chyyran	121dbc4ed6	reflect(wgsl): properly adjust coordinate space to WGSL-expected	2024-02-06 19:04:28 -05:00
chyyran	31891e414f	rt(wgpu): fix enough stuff to get it to draw a frame	2024-02-06 19:04:28 -05:00
chyyran	e39834547c	rt(wgpu): take an arc of the source texture	2024-02-06 19:04:28 -05:00
chyyran	d5aa6b2e4a	rt(wgpu): sketch out skeleton for filter chain logic	2024-02-06 19:04:28 -05:00
chyyran	cc26be486b	rt(wgpu): fix compiler errors for filer pass	2024-02-06 19:04:28 -05:00
chyyran	10358b4966	rt(wgpu): wip filter chain logic	2024-02-06 19:04:28 -05:00
chyyran	32148cdff4	rt(wgpu): filter pass logic	2024-02-06 19:04:28 -05:00
chyyran	555ff6f9fc	rt(wgpu): update to wgpu 0.19	2024-02-06 19:04:28 -05:00
chyyran	34f224cc5d	build: try to fix build for arm	2024-02-06 19:04:28 -05:00
chyyran	7586ed4633	build: don't specify python patch # Conflicts: # .idea/workspace.xml	2024-02-06 19:04:28 -05:00
chyyran	2b995539f2	rt(wgpu): add structure to wgpu backend	2024-02-06 19:04:28 -05:00
chyyran	8cfe1e9da2	rt(wgpu): create pipeline	2024-02-06 19:04:28 -05:00
chyyran	f9df72a02d	rt(wgpu): sampler set	2024-02-06 19:04:28 -05:00
chyyran	4e052159e7	rt(wgpu): create pipeline bind group layouts	2024-02-06 19:04:28 -05:00
chyyran	171c842c97	reflect(wgsl): implement WGSL reflection	2024-02-06 19:04:28 -05:00
chyyran	4dfcdf2725	reflect(wgsl): wgsl compile backend	2024-02-06 19:04:28 -05:00
chyyran	1a16c4fadf	rt(wgpu): load shaders	2024-02-06 19:04:28 -05:00
chyyran	c05d8ff06a	rt(wgpu): basic triangle example	2024-02-06 19:04:28 -05:00
Ronny Chan	ec98494202	build(obs): remove old obscpio before pushing	2024-02-04 18:06:39 -05:00
Ronny Chan	ef4e2353ff	build(obs): remove old oscpio	2024-02-04 18:05:58 -05:00
chyyran	a5e978f158	doc: add download link	2024-02-04 13:16:48 -05:00
chyyran	97ff76276f	doc: add obs badge	2024-02-04 13:16:48 -05:00
chyyran	797625903a	build(obs): add openbuildservice	2024-02-04 13:16:48 -05:00
Ronny Chan	81ba694ba4	build(copr): require git	2024-02-04 13:16:48 -05:00
chyyran	18ff9cf05a	build(copr): add copr	2024-02-04 13:16:48 -05:00
chyyran	ae2a427b5e	pkg(rpm): add RPM spec file	2024-02-04 13:16:48 -05:00
chyyran	552be8c34e	rt(gl): remove unnecessary unstable features polyfill strict_provenance with sptr	2024-02-03 22:06:55 -05:00
chyyran	7c0190004f	rt(d3d12): remove unnecessary unstable features div_ceil has been stabilized, and we don't really make use of const trait impl	2024-02-03 22:06:55 -05:00
chyyran	a6c91a07df	build: use stable polyfills for extract_if and array_chunks_mut	2024-02-03 22:06:55 -05:00
chyyran	d700234c3c	presets: polyfill extract_if	2024-02-03 22:06:55 -05:00
chyyran	96f937586c	build: clean unused features	2024-02-03 03:01:16 -05:00
chyyran	9c5a8f4042	gh: fix python requirement	2024-02-03 02:46:18 -05:00
chyyran	d5bf7e312c	build: update lockfile	2024-02-03 02:42:52 -05:00
chyyran	80325fda9e	fmt: cargo fmt	2024-02-03 02:42:52 -05:00
chyyran	60fac06332	build: fix rename when pdb is missing	2024-02-03 02:42:52 -05:00
chyyran	92e8a05f8a	build: fix cbindgen version	2024-02-03 02:42:52 -05:00
chyyran	617bfdd93e	ide: fix rustrover iml	2024-02-03 02:42:52 -05:00