v0.1.7

What's Changed

• Support for new list namespaces API via tpuf.namespaces()
• New metadata access functions on namespaces: ns.exists(), ns.dimensions(), ns.approx_count()
• Fix for incorrectly handled positional arguments: ns.upsert(ids, vectors) vs ns.upsert(ids=ids, vectors=vectors)
• Improved request retry logic to use exponential falloff.

Full Changelog: 0.1.6...0.1.7

v0.1.6

What's Changed

• Fixed ns.upsert incompatibility with numpy arrays when not using fast variant.

Full Changelog: 0.1.5...0.1.6

v0.1.5

What's Changed

• Adds String ID support for vectors 🎉

Full Changelog: 0.1.4...0.1.5

v0.1.4

What's Changed

• Fixed compatibility issue with numpy types when using turbopuffer[fast] / orjson
• Added numpy compatibility tests and support passing np.ndarray in more places.

Full Changelog: 0.1.3...0.1.4

v0.1.3

What's Changed

• Add native Parquet / pd.DataFrame support on upsert (See usage in examples/import_parquet.py)
• Add missing VectorResult type to package includes
• Update tests to run automatically in CI, and include flake8 python linting.

Full Changelog: 0.1.2...0.1.3

v0.1.2

Fix bug when calling ns.query() with include_vectors=False hitting an assert.

v0.1.1

Initial public release of the python client.

https://pypi.org/project/turbopuffer/0.1.1/