Ejemplo
Prevent silent numeric narrowing with NumPy 2.5.2 astype casting rules and distinguish copy=False from a no-allocation guarantee.
sha256:37b97780098278fbc636af0c6a293174b30101167e1aac82a81b85a9b5345f3c
PUBLISHED
L3_CONTRACT_PASS
MIT-0
Caso
- Objetivo
- Prevent silent numeric narrowing with NumPy 2.5.2 astype casting rules and distinguish copy=False from a no-allocation guarantee. HOW
- Paquetes
- numpy 2.5.2
- Entorno
- python 3.12
- Creado
- 2026-08-17T02:13:42Z
Lo que suele suponerse
casting='same_kind' prevents integer overflow during narrowing, while astype(copy=False) refuses any conversion that must allocate a new array.
El autor de la muestra anotó aquí lo que un desarrollador o un modelo esperaría. El contrato de abajo es lo que realmente se ejecutó.
Contrato
- astype(casting='same_kind') accepts int64-to-int8 narrowing and wraps out-of-range values instead of checking them.
- astype(casting='safe') rejects int64-to-int8 narrowing even when every runtime value fits in int8.
- astype(casting='same_value') accepts narrowing only when every value is preserved and raises ValueError for integer overflow or fractional loss.
- astype(copy=False) returns the input when dtype requirements already match, but still allocates when a dtype conversion is required.
Archivos
- NOTES.md
- csx.json
- requirements.in
- requirements.txt
- test/contract.py
Descargar el artefacto verificado (tar.gz): los bytes exactos con los que se ejecutó el contrato
Seeder de origen
Recibos de verificación
- python 3.12 · CONTAINER_RUN · compile:SKIPPED · contract:PASS · load:PASS · resolve:PASS · python@1 · 2026-08-17 · ed25519:d91480838ac982c9