Struct SurfaceFactory 

pub struct SurfaceFactory { /* private fields */ }

A cheap, cloneable handle to a wgpu::Instance used to create a surface on a different thread than the one that owns the renderer.

Why this exists

wgpu surface creation reads the platform window handle, which several windowing backends (notably winit on macOS/AppKit) only make available on the main/UI thread. A render-thread split therefore cannot create the surface where the renderer lives; instead the UI thread creates a DetachedSurface via this factory and hands it across to the render thread, which configures it there. A wgpu::Instance is Send + Sync + Clone (Arc-backed) and a Surface it produces stays compatible with any adapter/device the cloned instance requests, so the two threads share one instance with no unsafe.

The mobile shells do not need this: they receive a platform-created surface pointer and go through create_android_surface / create_metal_surface instead.

Implementations

impl SurfaceFactory

pub fn new(instance: &Instance) -> SurfaceFactory

Clones instance into a factory that can be moved to the UI thread.

pub fn create_detached_surface( &self, target: impl Into<SurfaceTarget<'static>>, ) -> Result<DetachedSurface, Error>

Create a DetachedSurface from a window handle on the calling thread — call this on the thread the windowing backend requires (the main/UI thread for winit). The returned surface is Send and may then be moved to the render thread and configured there.

This performs only the window-handle-dependent step (surface creation); the device and the swapchain configuration are both built later, on the configuring thread, so nothing here touches a GPU device.

Trait Implementations

impl Clone for SurfaceFactory

fn clone(&self) -> SurfaceFactory

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.