G10: Use the shared wgpu fork for OCLP Macs

Cargo.toml

@@ -141,6 +141,8 @@ lto = "thin"
 codegen-units = 1
 
 [patch.crates-io]
+# Shared OCLP Metal compatibility backport; see docs/oclp-wgpu-fork.md.
+wgpu = { git = "https://github.com/ITJamie/wgpu-fork", rev = "21c08e23135ba27ed1c6b68177ecc20c0d076dd9" }
 hayro-syntax = { git = "https://github.com/storytold/craft-libs", rev = "a0916808f7d78bed8c7b90cdf937176015454a7f" }
 hayro-interpret = { git = "https://github.com/storytold/craft-libs", rev = "a0916808f7d78bed8c7b90cdf937176015454a7f" }
 hayro-jbig2 = { git = "https://github.com/storytold/craft-libs", rev = "a0916808f7d78bed8c7b90cdf937176015454a7f" }

ROADMAP.md

@@ -1,5 +1,7 @@
 # VectorCraft Roadmap
 
+> **Last updated:** 2026-10-26 · **Change:** minor (shared OCLP wgpu compatibility proposal; scores and estimates unchanged).
+
 - PDF.R1: editable PDF import uses the shared reader, scans page content in windows, and rejects known incomplete input before returning a document.
 - PDF.W1: PDF export and print composition use the shared optional writer for configuration, metadata, font preparation and guarded finalization; drawing and output policy remain app-owned.
 
@@ -54,7 +56,7 @@ everything is measured from our code against Illustrator's public documentation.
 | Features (depth) | 75% | 220–360 | [target-app-parity.md](docs/target-app-parity.md) |
 | UI/UX fidelity | 55% | 70–110 | [ui-parity.md](docs/ui-parity.md) |
 | File formats | 75% | 15–30 | [file-format-parity.md](docs/file-format-parity.md) |
-| Hardware (GPU, pen, touch, displays) | 35% | 25–45 | [hardware-parity.md](docs/hardware-parity.md) |
+| Hardware (GPU, pen, touch, displays) | 35% | 25–45 | [hardware-parity.md](docs/hardware-parity.md); [OCLP proposal](docs/oclp-wgpu-fork.md) |
 | Localization (12 key languages) | 74% (measured) | 50–85 | [localization-parity.md](docs/localization-parity.md) |
 | Performance | ~60% | 15–25 | [gaps.md › G15](docs/gaps.md#g15-performance-budgets-and-a-gpu-renderer) |
 | Stability | ~60% | 30–50 | [gaps.md › G10](docs/gaps.md#g10-launch-and-stability-on-real-machines) |
@@ -198,6 +200,7 @@ The full pre-standard record of what landed is kept in [docs/roadmap.md](docs/ro
 
 | Date | Change | Summary |
 |---|---|---|
+| 2026-10-26 | minor | Propose the shared OCLP wgpu backport with internal validation preserved; qualification and known compiler-probe delay documented; scores and estimates unchanged. |
 | 2026-10-23 | minor | Progress log row for Pathfinder › Trap; Live vector effects hours 6–10 → 4–8; headline numbers unchanged. |
 | 2026-10-23 | minor | Progress log row for OpenType SVG colour fonts; headline numbers unchanged. |
 | 2026-10-23 | minor | Progress log row for COLR colour fonts; headline numbers unchanged. |

apps/vectorcraft/src/gpu.rs

@@ -59,6 +59,11 @@ use std::time::{Duration, Instant};
 use eframe::egui_wgpu::NativeAdapterSelectorMethod;
 use eframe::wgpu::{self, Backend, Backends, DeviceType, PowerPreference};
 
+pub fn configure_instance(descriptor: &mut wgpu::InstanceDescriptor, backends: Backends, compiler: wgpu::Dx12Compiler) {
+    descriptor.backends = backends;
+    descriptor.backend_options.dx12.shader_compiler = compiler;
+}
+
 /// The adapters a restart leaves out, those the window failed on, as comma-separated [`key`]s
 /// (`Vulkan:1002:164e`, `Metal:Intel Iris Pro Graphics`). Set by the app itself when it starts
 /// again.
@@ -881,6 +886,23 @@ fn restart(skip: &str) -> std::io::Result<()> {
 mod tests {
     use super::*;
 
+    #[test]
+    fn instance_configuration_preserves_validation_flags() {
+        use eframe::wgpu::{Dx12Compiler, InstanceDescriptor, InstanceFlags, MetalShadingLanguageVersion as Msl};
+        for flags in [InstanceFlags::VALIDATION_INDIRECT_CALL, InstanceFlags::empty(), InstanceFlags::default()] {
+            let mut descriptor = InstanceDescriptor::new_without_display_handle();
+            descriptor.flags = flags;
+            descriptor.backend_options.metal.msl_version = Some(Msl::Version3_1);
+            descriptor.backend_options.metal.max_msl_version = Some(Msl::Version3_0);
+            super::configure_instance(&mut descriptor, super::Backends::METAL, Dx12Compiler::Fxc);
+            assert_eq!(descriptor.flags, flags);
+            assert_eq!(descriptor.backend_options.metal.msl_version, Some(Msl::Version3_1));
+            assert_eq!(descriptor.backend_options.metal.max_msl_version, Some(Msl::Version3_0));
+            assert_eq!(descriptor.backends, super::Backends::METAL);
+            assert!(matches!(descriptor.backend_options.dx12.shader_compiler, Dx12Compiler::Fxc));
+        }
+    }
+
     fn gpu(name: &str, backend: Backend, device_type: DeviceType) -> Candidate {
         Candidate { key: format!("{backend:?}:{name}"), name: name.into(), backend, device_type, pci: (0, 0), presents: true }
     }

apps/vectorcraft/src/main.rs

@@ -1000,18 +1000,10 @@ fn main() -> std::process::ExitCode {
         // Only adapters that can show the window, in the order `gpu` gives (#306, #502).
         if let eframe::egui_wgpu::WgpuSetup::CreateNew(create) = &mut options.wgpu_options.wgpu_setup {
             create.native_adapter_selector = Some(gpu::selector(power, displays, compositor, left_out.adapters.clone(), startup.clone()));
-            // Nothing draws or dispatches indirectly, so wgpu's check of indirect arguments only
-            // costs a compute shader at start-up, one some drivers can't compile (OCLP-patched
-            // Metal on an Iris Pro, #651). `WGPU_VALIDATION_INDIRECT_CALL=1` turns it back on.
-            create.instance_descriptor.flags =
-                (eframe::wgpu::InstanceFlags::from_build_config() - eframe::wgpu::InstanceFlags::VALIDATION_INDIRECT_CALL).with_env();
-            create.instance_descriptor.backends =
-                left_out.backends(gpu::backends(create.instance_descriptor.backends, std::env::var_os("WGPU_BACKEND").is_some()));
-            // DX12 shaders compile with FXC, part of Windows: wgpu's default loads whichever
-            // dxcompiler.dll the search path finds, and one without its dxil.dll fails every shader,
-            // losing the device before the first frame (#1194). `WGPU_DX12_COMPILER` picks another.
-            create.instance_descriptor.backend_options.dx12.shader_compiler =
-                eframe::wgpu::Dx12Compiler::from_env().unwrap_or(eframe::wgpu::Dx12Compiler::Fxc);
+            let backends = left_out.backends(gpu::backends(create.instance_descriptor.backends, std::env::var_os("WGPU_BACKEND").is_some()));
+            // Keep FXC as the default on Windows (#1194), unless explicitly overridden.
+            let compiler = eframe::wgpu::Dx12Compiler::from_env().unwrap_or(eframe::wgpu::Dx12Compiler::Fxc);
+            gpu::configure_instance(&mut create.instance_descriptor, backends, compiler);
             // The window's surface may be as large as the adapter allows, not eframe's fixed
             // 8192 px: a window stretched over several monitors, or grown by a display scale
             // change, otherwise fails to configure its surface and the app closes (#1459).

contributors/people.toml

@@ -169,3 +169,6 @@ sync_github_name = false
 real_name = "Andrés Ortega Montoya"
 display_name = "Andrés Ortega Montoya"
 sync_github_name = false
+
+[people.ITJamie]
+sync_github_name = true

docs/development.md

@@ -74,7 +74,7 @@ The canvas is rasterized on the CPU (`vectorcraft-render`, vello_cpu); the GPU (
   - `WGPU_POWER_PREF`: `low`, `high` or `none` (the system's order).
   - `WGPU_ADAPTER_NAME`: part of an adapter's name, any case (`WGPU_ADAPTER_NAME=nvidia`, `=radv`, `=intel`); the names are in the log.
   - `WGPU_BACKEND`: the backends to use, such as `vulkan`, `dx12`, `metal` or `gl`.
-  - `WGPU_VALIDATION_INDIRECT_CALL=1`: turns wgpu's check of indirect draw arguments back on. VectorCraft draws nothing indirectly and turns it off, since the compute shader it needs fails to compile on some drivers (#651).
+  - `WGPU_VALIDATION_INDIRECT_CALL`: explicitly enables (`1`) or disables (`0`) wgpu's internal indirect-call validation. The shared [OCLP wgpu proposal](oclp-wgpu-fork.md) preserves the instance's configured flags instead of disabling this check at startup.
   - On Linux with Mesa, `MESA_VK_DEVICE_SELECT=10de:2f04!` (vendor:device as `vulkaninfo --summary` or `lspci -nn` show it) leaves only that GPU to Vulkan, and `DRI_PRIME=1` picks the other GPU.
 - Help › About and the control channel's `ui.inspect` (`graphicsAdapter`) show the adapter in use, and the app logs it at startup.
 

docs/hardware-parity.md

@@ -1,5 +1,7 @@
 # Hardware parity
 
+> **Last updated:** 2026-10-26 · **Change:** minor (shared OCLP wgpu compatibility proposal; scores and estimates unchanged).
+
 > **Last reviewed:** 2026-10-10 · **Last updated:** 2026-10-24 · **Change:** minor (pen pressure and the pen's eraser end on macOS, read from AppKit's tablet events, #852, not yet tried on a Mac; previously: patch: web pen buttons/modifiers preserved, touch-only input left to eframe, session-only palm rejection; tool cursors painted into the window on Wayland, #1652, not run on Linux; borderless title-bar window on Windows, #1758, awaiting confirmation; a window surface larger than 8192 px keeps working, #1459 in part; Linux Display Server preference opens the window through XWayland, #1151) · **Target:** Adobe Illustrator 2026 (30.x)
 
 What hardware Illustrator uses, per platform, against VectorCraft. Illustrator's side is from its public
@@ -9,6 +11,8 @@ documentation (GPU Performance, touch workspace, pen pressure pages); ours from
 
 ## Summary
 
+The [OCLP wgpu proposal](oclp-wgpu-fork.md) records the pinned compatibility fork, hardware evidence and remaining compiler-probe delay. It does not change hardware parity scores.
+
 **~35% (estimated), 25–45 h.** VectorCraft rasterizes the canvas on the CPU (multithreaded `vello_cpu`) and uses
 the GPU only to composite, so any GPU that can show a window works, at the cost of GPU-accelerated preview
 on huge documents. VectorCraft reads pen pressure on Windows, on the web, on macOS, and on Linux under X11 and XWayland, and it
@@ -36,6 +40,7 @@ reads the pen's eraser end on Windows, on macOS and on Linux under X11 and XWayl
 
 | Date | Change | Summary |
 |---|---|---|
+| 2026-10-26 | minor | Propose the shared OCLP wgpu backport with internal validation preserved; qualification and known compiler-probe delay documented; scores and estimates unchanged. |
 | 2026-10-24 | minor | macOS: an AppKit local event monitor reads each mouse and tablet event's pen pressure and the pen's end before the window gets it and writes them into the canvas's pen feed (#852): each pointer move of a stroke keeps its own pressure, mouse event coalescing is off, and the eraser end selects the Eraser tool. The mapping is unit-tested on every platform and the macOS code type-checks for `aarch64-apple-darwin`; not yet tried on a Mac with a tablet, so the score is unchanged (estimated) |
 | 2026-10-15 | patch | Web review: preserve pen buttons and modifiers; leave fingers to eframe before the first pen contact; remove persistent pen policy. Canvas regression tests cover intermediate moves and long-touch cancellation on native and web input. Physical-device QA pending. |
 | 2026-10-14 | minor | Tool cursors are painted into the window on Wayland (#1652; not run on Linux, #1645 stays open until confirmed); On Windows the custom title bar's window stays truly borderless, so the pointer lines up with what is drawn on systems where the hidden caption offset it (#1791, #1758; awaiting the reporter's confirmation); neither is confirmed on the hardware that reported it, so nothing is marked done |