Struct ExecutionContext 

pub struct ExecutionContext<'a> {
    pub device: &'a Device,
    pub queue: &'a Queue,
    pub encoder: &'a mut CommandEncoder,
    pub registry: &'a ResourceRegistry,
    pub renderer: &'a GpuRenderer,
    pub target_view: &'a TextureView,
    pub depth_view: &'a TextureView,
    pub blur_env_bind_group_a: &'a BindGroup,
    pub blur_env_bind_group_b: &'a BindGroup,
    pub bloom_env_bind_group_a: &'a BindGroup,
    pub bloom_env_bind_group_b: &'a BindGroup,
    pub scale_factor: f32,
}

Context passed to each node during execution.

P1-2 fix: documented the aliasing contract. The struct holds:

• Several &'a shared references to fields of GpuRenderer (device, queue, registry, renderer, target_view, depth_view, bind groups)
• A single &'a mut wgpu::CommandEncoder (the only mutable field)

The renderer field is &'a GpuRenderer (immutable), so nodes cannot call &mut self methods on the renderer during execution. This is intentional: the renderer is being driven by the outer frame loop, and allowing a node to mutate the renderer mid-frame would cause aliasing. The audit flagged the previous implicit split-borrow pattern as a potential safety risk; the fix is to document the contract clearly and ensure no path can construct an ExecutionContext that violates it.

If a future node needs to mutate the renderer (e.g. to record custom draw calls), it should use the encoder field directly (which is &mut) and avoid going through the renderer API.

Fields

device: &'a Device

queue: &'a Queue

encoder: &'a mut CommandEncoder

registry: &'a ResourceRegistry

renderer: &'a GpuRenderer

target_view: &'a TextureView

depth_view: &'a TextureView

blur_env_bind_group_a: &'a BindGroup

blur_env_bind_group_b: &'a BindGroup

bloom_env_bind_group_a: &'a BindGroup

bloom_env_bind_group_b: &'a BindGroup

scale_factor: f32

Implementations

impl<'a> ExecutionContext<'a>

pub fn begin_render_pass( &mut self, desc: &RenderPassDescriptor<'_>, ) -> RenderPass<'_>

pub fn get_or_create_bind_group( &self, key: (ResourceId, u32, bool), layout: &BindGroupLayout, entries: &[BindGroupEntry<'_>], label: Option<&str>, ) -> BindGroup

Get or create a cached bind group for a given resource and mip level. Avoids per-frame GPU allocation when the same bind group is reused across frames.