Module intrinsics

Expand description

High-performance intrinsic functions for Shape

Intrinsics are Rust-implemented functions that provide performance-critical operations while keeping domain logic in Shape stdlib.

These functions are prefixed with __intrinsic_ and should not be called directly by users - they are wrapped by Shape stdlib functions.

Modules§

array_transforms: Array-transform intrinsics — partial migration to typed marshal layer.
convolution: Convolution intrinsics — full migration to typed marshal layer.
distributions: Statistical distribution intrinsics — full migration to typed marshal layer.
fft: FFT (Fast Fourier Transform) intrinsics — full migration to typed marshal layer.
math: Math intrinsics — partial migration to typed marshal layer.
matrix: Matrix intrinsics — full migration to typed marshal layer.
matrix_kernels: SIMD-accelerated matrix kernels operating on MatrixData.
random: Random number generation intrinsics — full migration to typed marshal layer.
recurrence: Recurrence intrinsics — partial migration to typed marshal layer (cross-crate-dual-consumer pattern; see docs/defections.md 2026-05-07 intrinsics-typed-CC entry’s N1 sub-decision).
rolling: Rolling window intrinsics — partial migration to typed marshal layer.
statistical: Statistical intrinsics — full migration to typed marshal layer.
stochastic: Stochastic process intrinsics — full migration to typed marshal layer.
vector: Vector intrinsics — element-wise SIMD operations on f64 / i64 typed arrays.

extract_f64: Extract a f64 from an intrinsic argument. Phase 1.B reads the slot’s 8 bytes as f64 directly — variadic intrinsic callers carry the kind contract per registration.
extract_f64_array: Extract a Vec<f64> from an intrinsic array argument.
extract_str: Extract a string reference from an intrinsic argument. Phase 1.B reads the slot bits as Arc<String>::into_raw-shaped per registered string param; returns the borrowed string.
extract_usize: Extract a usize from an intrinsic argument (window size / period).
f64_vec_to_float_array: Build a KindedSlot typed FloatArray from a Vec<f64>. See f64_vec_to_nb_array — Phase 2c rebuild deferral.
f64_vec_to_nb_array: Build a KindedSlot array from a Vec<f64>. Phase 2c lands the proper HeapValue::TypedArray(TypedArrayData::F64) constructor.
i64_vec_to_nb_int_array: Build a KindedSlot typed IntArray from a Vec<i64>. See above.
option_i64_vec_to_nb: Build a KindedSlot IntArray with validity bitmap from Vec<Option<i64>>. Phase 2c rebuild deferral.
try_extract_i64_slice: Try to read an i64 slice directly from a KindedSlot IntArray. Phase 1.B: deferred — returns None.