damascene-vulkano

Native Vulkan backend for Damascene using vulkano.

Most applications should use damascene-core plus damascene-winit-wgpu. Use this crate directly when validating backend parity or writing a custom Vulkan host.

The public entry point mirrors damascene-wgpu::Runner where the GPU API allows it. A host owns the window, device, queue, swapchain, and event loop; the runner owns Damascene interaction state, layout/draw-op preparation, Vulkan pipelines, text atlas images, and icon rendering.

Swapchain requirements

Create the swapchain with

image_usage: COLOR_ATTACHMENT | TRANSFER_SRC | TRANSFER_DST

TRANSFER_SRC lets Runner::render copy the mid-frame surface into its backdrop snapshot for backdrop-sampling shaders (modal scrim blur, …). TRANSFER_DST is required under MSAA: vulkano's render-pass macro gives the resolve attachment a TransferDstOptimal layout, which needs the flag on the underlying image. The runner validates this on the first frame and fails with an actionable message; most surfaces advertise both flags, so the cost of always setting them is nil. See damascene-vulkano-demo for a complete bring-up.

WGSL remains the shader source language. This backend uses naga to compile WGSL to SPIR-V when building pipelines.