Crate edgefirst_tensor_abi 

Layout-only C ABI contract for the EdgeFirst tensor surface.

Plain #[repr(C)] values and the integer vocabularies, declared exactly once. Every -capi crate imports these instead of carrying an “identical copy” — r2’s duplication was safe but unverified-by-construction; a single declaration makes divergence impossible instead of merely absent.

This crate has no dependencies and must never gain one. The vocabulary values are asserted against edgefirst-tensor’s authority from the crates that depend on both (see edgefirst-tensor-capi/src/codes.rs) — this crate is emitted surface, never a second authority.

Structs

EfClientState

Client-owned state a library entry point calls back into — frozen by value, forever, and reused verbatim by every domain that needs one (PBO today; CUDA-GL next).

EfD3d11Layout

The D3D11 texture behind a Windows texture tensor.

EfImageDescView

Flattened, #[repr(C)] view of an image-request descriptor’s fields.

EfQuantizationInfo

Presence/shape summary of a tensor’s quantization metadata.

EfTensorPlane

One plane’s location, mirroring TensorPlane on the wire.

EfTensorView

A mapped CPU window over a tensor’s bytes.

EfViewOrigin

Parent-region snapshot for a tensor that is a view/batch sub-region.

Enums

EfCompression

Vendor tile-compression scheme actually in force for a tensor’s allocation.

EfCompressionRequest

Compression request, as ef_tensor_image_desc_set_compression takes it and EfImageDescView::compression reports it.

EfContiguity

Whether a tensor’s memory is known to be physically contiguous (ef_tensor_contiguity).

EfCpuAccess

CPU access: a tensor’s declaration, or the direction of a map window.

EfDtype

Element type of a tensor’s addressing grid.

EfErrorClass

Which kind of failure the calling thread’s last failing ef_tensor_* call was.

EfStorageKind

Backing store for a tensor.

Type Aliases

EfClientRefFn

A retain or release call on a client-owned callback context.

EfCudaMapFn

Map a CUDA graphics resource to a device pointer. 0 on success.

EfCudaMapFnNullable

Map a CUDA graphics resource to a device pointer. 0 on success.

EfCudaUnmapFn

Unmap a resource previously mapped by an EfCudaMapFn.

EfCudaUnmapFnNullable

Unmap a resource previously mapped by an ef_cuda_map_fn.

EfCudaUnregisterFn

Unregister a CUDA graphics resource.

EfCudaUnregisterFnNullable

Unregister a CUDA graphics resource.

EfPboMapFn

Map a PBO for CPU access. 0 on success; -1 when the GL context is gone; any other non-zero value a generic failure.

EfPboMapFnNullable

Map a PBO for CPU access. 0 on success; -1 when the GL context is gone; any other non-zero value a generic failure.

EfPboUnmapFn

Unmap a PBO previously mapped by an EfPboMapFn. Same return codes.

EfPboUnmapFnNullable

Unmap a PBO previously mapped by an ef_pbo_map_fn. Same return codes.