Struct GpuCanvasContext 

pub struct GpuCanvasContext<'a, 'b> { /* private fields */ }

A slice of Window — the only place raw GPU import is reachable.

Handed to a gpu_canvas callback. A plain &mut Window or an App cannot composite a foreign surface; a canvas narrows the window to exactly the operations that can. Bounded by the canvas, so paint_surface/paint_texture composite into its bounds.

Implementations

impl GpuCanvasContext<'_, '_>

pub fn paint_surface(&mut self, source: impl Into<SurfaceSource>)

Composite source (a dma-buf, a Direct3D texture/view, or a CoreVideo buffer) into the canvas’s bounds for this frame.

pub fn paint_texture( &mut self, handle: ImportedTextureHandle, corner_radii: Corners<Pixels>, opacity: f32, flip_v: bool, )

Composite a texture produced on the window renderer’s own device into the canvas’s bounds.

pub fn bounds(&self) -> Bounds<Pixels>

The canvas’s bounds.

pub fn device<R: GpuRenderer>(&mut self) -> R::Device

The window renderer’s device, downcast to R.

This is the same-device producer’s door: it hands back an owned handle (ID3D11Device, a metal::Device, a wgpu device and queue) that the caller may keep, because a borrow of the renderer’s device cannot outlive the call. Naming R is the assertion that the window draws through that backend — a #[cfg]-selected producer already knows — so a mismatch panics; use try_device to handle it instead.

pub fn try_device<R: GpuRenderer>(&mut self) -> Option<R::Device>

The window renderer’s device, downcast to R, or None when the renderer is not R or has no device to lend right now.

The fallible form of device, for a producer that degrades — paints a dma-buf instead of a same-device texture, or skips the frame — rather than asserting the backend it is running under.

pub fn cx(&mut self) -> &mut App

The application, mirroring the cx a canvas() callback receives.