Currently your project says that it's compatible with "numpy>=1.25.0". However, ingesting vectors / creating an index fails on NumPy 2.4 and newer with: "AttributeError: module 'numpy' has no attribute 'in1d'".
This is because Numpy deprecated np.in1d in 2.0 and removed it in 2.4.0:
numpy/numpy#29978
https://github.com/numpy/numpy/releases/tag/v2.4.0
tiledb/vector_search/ingestion.py, inside ingest_flat() still calls the old `np.in1d'
np.isin is a direct replacement and takes the same arguments (including kind="sort")
Temporary workaround (for others hitting this before a release)
Run once at startup, before ingesting:
import numpy as np
if not hasattr(np, "in1d"):
np.in1d = np.isin
Currently your project says that it's compatible with "numpy>=1.25.0". However, ingesting vectors / creating an index fails on NumPy 2.4 and newer with: "AttributeError: module 'numpy' has no attribute 'in1d'".
This is because Numpy deprecated np.in1d in 2.0 and removed it in 2.4.0:
numpy/numpy#29978
https://github.com/numpy/numpy/releases/tag/v2.4.0
tiledb/vector_search/ingestion.py, insideingest_flat()still calls the old `np.in1d'np.isinis a direct replacement and takes the same arguments (includingkind="sort")Temporary workaround (for others hitting this before a release)
Run once at startup, before ingesting: