Struct WaiterThread 

pub struct WaiterThread { /* private fields */ }

Permanently-attached waiter thread + request channel.

Cheap to construct (one mpsc::channel + one thread::spawn). Joined cleanly on Drop: the channel sender is dropped, the thread’s recv() returns Err, the loop exits, and the join handle is consumed.

Every request that is in flight or still queued when Drop runs is resolved with Error::Cancelled before the thread exits. BackendWaitFuture borrows nothing from this type, so a caller may hold one across the drop; without that resolution its await would pend forever.

Per-instance ownership is supported, but the usual shape is a per-backend-kind static singleton in each backend’s waiter module. See module-level docs.

Implementations

impl WaiterThread

pub fn new(name: &str) -> Self

Spawn a new waiter thread. name is appended to "gpu-wait-" for the thread name visible to debuggers and ps -L / Process Explorer. Pick a short, stable backend-identifying string ("vulkan", "wgpu", "d3d12", "metal", "opencl", …) — Linux truncates thread names to 15 bytes.

pub fn enqueue( &self, slice_fn: SliceFn, deadline: Option<Instant>, ) -> BackendWaitFuture

Enqueue a wait request. Returns a BackendWaitFuture the executor can .await; Drop of the future signals cancellation to the waiter thread.

deadline = None means “wait forever” — the slice loop never transitions to Err(Timeout) and only exits on Signaled / Failed / cancelled.

Trait Implementations

impl Debug for WaiterThread

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Drop for WaiterThread

fn drop(&mut self)

Executes the destructor for this type.

fn pin_drop(self: Pin<&mut Self>)

🔬This is a nightly-only experimental API. (pin_ergonomics)

Execute the destructor for this type, but different to Drop::drop, it requires self to be pinned.