Module shape 

Symbolic shape inference over the shared runtime IR (ONNX_RS §9).

This module is a thin standard-library wrapper around onnx_runtime_shape_inference. It does not duplicate operator rules: infer_shapes uses the crate’s built-in InferenceRegistry, while infer_shapes_with_registry accepts a caller-supplied registry containing custom rules.

Structs

DimExpr

A canonical integer polynomial over symbolic dimensions.

InferenceContext

The context passed to every op inference rule.

InferenceRegistry

A registry mapping (domain, op_type, opset) to an InferenceFn.

NodeIo

The resolved inference state of a single input or output slot: an optional type and an optional ShapeData side-value.

ShapeData

The known element values of a rank-0 or rank-1 integer tensor flowing through a shape-computation subgraph.

ShapeInferenceResult

Summary of a whole-model shape-inference run.

TypeInfo

The inferred type of a value: element dtype plus a symbolic shape.

Enums

MergePolicy

How to reconcile an inferred shape with a value’s declared shape.

ShapeError

An error produced while inferring shapes.

Functions

infer_shapes

Infer value types and shapes with the built-in ONNX operator registry.

infer_shapes_with_registry

Infer value types and shapes with a caller-supplied registry.

register_shape_inference

Register an opset-aware shape-inference function on registry.

Type Aliases

InferenceFn

An operator inference rule: reads inputs from the InferenceContext and sets its outputs’ types (and, where applicable, shape-data).

TypedShape

An inferred shape: an ordered list of symbolic dimension expressions. The rank is always known (unknown-rank tensors are represented by the absence of a TypeInfo, never by a TypedShape).