feat(python): add LanceTorchDataset / LanceIterableTorchDataset wrappers

Provides first-class PyTorch `Dataset`/`IterableDataset` wrappers around a
LanceDB table or permutation. The wrapper:

* Captures only the URI / table name / connect kwargs needed to re-open
  the table — no Rust handles in pickle output. Works out of the box with
  `DataLoader(num_workers > 0)`, which would otherwise crash a
  hand-rolled subclass.
* Implements both `__getitem__` and PyTorch's `__getitems__` dunder so
  the underlying batched `Permutation.fetch` is used when DataLoader
  fetches a batch of indices.
* Forwards column selection / format / transform / batch_size to the
  underlying Permutation, so users do not have to hand-roll the
  `_ensure_open` boilerplate from the issue.

Builds on the public `Permutation.fetch` API (#3243).

Closes lancedb/lancedb#3242

.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.27.0-beta.5"
+current_version = "0.28.0-beta.10"
 parse = """(?x)
     (?P<major>0|[1-9]\\d*)\\.
     (?P<minor>0|[1-9]\\d*)\\.

.github/ISSUE_TEMPLATE/documentation.yml

@@ -18,6 +18,6 @@ body:
       label: Link
       description: >
         Provide a link to the existing documentation, if applicable.
-      placeholder: ex. https://lancedb.com/docs/tables/...
+      placeholder: ex. https://docs.lancedb.com/tables/...
     validations:
       required: false

.github/dependabot.yml

@@ -0,0 +1,18 @@
+version: 2
+
+# Scope: the root Cargo workspace, which produces the Rust binaries we
+# ship to users (the Node.js and Python native extensions). The
+# `rust/lancedb` library crate shares the same lockfile; its consumers
+# pick their own dependency versions, but bumping transitive deps here
+# keeps the binaries we ship current.
+updates:
+  - package-ecosystem: cargo
+    directory: /
+    schedule:
+      interval: weekly
+    open-pull-requests-limit: 10
+    groups:
+      rust-minor-patch:
+        update-types:
+          - minor
+          - patch

.github/workflows/build_linux_wheel/action.yml

@@ -23,8 +23,10 @@ runs:
   steps:
     - name: CONFIRM ARM BUILD
       shell: bash
+      env:
+        ARM_BUILD: ${{ inputs.arm-build }}
       run: |
-        echo "ARM BUILD: ${{ inputs.arm-build }}"
+        echo "ARM BUILD: $ARM_BUILD"
     - name: Build x86_64 Manylinux wheel
       if: ${{ inputs.arm-build == 'false' }}
       uses: PyO3/maturin-action@v1

.github/workflows/dev.yml

@@ -8,6 +8,9 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
   cancel-in-progress: true
 
+permissions:
+  contents: read
+
 jobs:
   labeler:
     permissions:

.github/workflows/java-publish.yml

@@ -19,6 +19,9 @@ on:
     paths:
       - .github/workflows/java-publish.yml
 
+permissions:
+  contents: read
+
 jobs:
   publish:
     name: Build and Publish

.github/workflows/java.yml

@@ -24,6 +24,9 @@ on:
       - java/**
       - .github/workflows/java.yml
 
+permissions:
+  contents: read
+
 jobs:
   build-java:
     runs-on: ubuntu-24.04

.github/workflows/license-header-check.yml

@@ -10,6 +10,10 @@ on:
       - nodejs/**
       - java/**
       - .github/workflows/license-header-check.yml
+
+permissions:
+  contents: read
+
 jobs:
   check-licenses:
     runs-on: ubuntu-latest

.github/workflows/nodejs.yml

@@ -7,12 +7,17 @@ on:
   pull_request:
     paths:
       - Cargo.toml
+      - Cargo.lock
+      - rust-toolchain.toml
       - nodejs/**
       - rust/**
       - docs/src/js/**
       - .github/workflows/nodejs.yml
       - docker-compose.yml
 
+permissions:
+  contents: read
+
 concurrency:
   group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
   cancel-in-progress: true

.github/workflows/npm-publish.yml

@@ -19,6 +19,7 @@ on:
     paths:
       - .github/workflows/npm-publish.yml
       - Cargo.toml # Change in dependency frequently breaks builds
+      - Cargo.lock
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
@@ -124,7 +125,12 @@ jobs:
             pre_build: |-
               set -e &&
               apt-get update &&
-              apt-get install -y protobuf-compiler pkg-config
+              apt-get install -y protobuf-compiler pkg-config &&
+              # The base image (manylinux2014-cross) sets TARGET_CC to the old
+              # GCC 4.8 cross-compiler. aws-lc-sys checks TARGET_CC before CC,
+              # so it picks up GCC even though the napi-rs image sets CC=clang.
+              # Override to use the image's clang-18 which supports -fuse-ld=lld.
+              export TARGET_CC=clang TARGET_CXX=clang++
           - target: x86_64-unknown-linux-musl
             # This one seems to need some extra memory
             host: ubuntu-2404-8x-x64
@@ -144,9 +150,10 @@ jobs:
               set -e &&
               apt-get update &&
               apt-get install -y protobuf-compiler pkg-config &&
-              # https://github.com/aws/aws-lc-rs/issues/737#issuecomment-2725918627
-              ln -s /usr/aarch64-unknown-linux-gnu/lib/gcc/aarch64-unknown-linux-gnu/4.8.5/crtbeginS.o /usr/aarch64-unknown-linux-gnu/aarch64-unknown-linux-gnu/sysroot/usr/lib/crtbeginS.o &&
-              ln -s /usr/aarch64-unknown-linux-gnu/lib/gcc /usr/aarch64-unknown-linux-gnu/aarch64-unknown-linux-gnu/sysroot/usr/lib/gcc &&
+              export TARGET_CC=clang TARGET_CXX=clang++ &&
+              # The manylinux2014 sysroot has glibc 2.17 headers which lack
+              # AT_HWCAP2 (added in Linux 3.17). Define it for aws-lc-sys.
+              export CFLAGS="$CFLAGS -DAT_HWCAP2=26" &&
               rustup target add aarch64-unknown-linux-gnu
           - target: aarch64-unknown-linux-musl
             host: ubuntu-2404-8x-x64
@@ -266,7 +273,7 @@ jobs:
           - target: x86_64-unknown-linux-gnu
             host: ubuntu-latest
           - target: aarch64-unknown-linux-gnu
-            host: buildjet-16vcpu-ubuntu-2204-arm
+            host: ubuntu-2404-8x-arm64
         node:
           - '20'
     runs-on: ${{ matrix.settings.host }}

.github/workflows/pypi-publish.yml

@@ -9,14 +9,21 @@ on:
     paths:
       - .github/workflows/pypi-publish.yml
       - Cargo.toml # Change in dependency frequently breaks builds
+      - Cargo.lock
 
 env:
   PIP_EXTRA_INDEX_URL: "https://pypi.fury.io/lance-format/ https://pypi.fury.io/lancedb/"
 
+permissions:
+  contents: read
+
 jobs:
   linux:
     name: Python ${{ matrix.config.platform }} manylinux${{ matrix.config.manylinux }}
     timeout-minutes: 60
+    permissions:
+      id-token: write
+      contents: read
     strategy:
       matrix:
         config:
@@ -56,10 +63,12 @@ jobs:
       - uses: ./.github/workflows/upload_wheel
         if: startsWith(github.ref, 'refs/tags/python-v')
         with:
-          pypi_token: ${{ secrets.LANCEDB_PYPI_API_TOKEN }}
           fury_token: ${{ secrets.FURY_TOKEN }}
   mac:
     timeout-minutes: 90
+    permissions:
+      id-token: write
+      contents: read
     runs-on: ${{ matrix.config.runner }}
     strategy:
       matrix:
@@ -84,10 +93,12 @@ jobs:
       - uses: ./.github/workflows/upload_wheel
         if: startsWith(github.ref, 'refs/tags/python-v')
         with:
-          pypi_token: ${{ secrets.LANCEDB_PYPI_API_TOKEN }}
           fury_token: ${{ secrets.FURY_TOKEN }}
   windows:
     timeout-minutes: 60
+    permissions:
+      id-token: write
+      contents: read
     runs-on: windows-latest
     steps:
       - uses: actions/checkout@v4
@@ -106,7 +117,6 @@ jobs:
       - uses: ./.github/workflows/upload_wheel
         if: startsWith(github.ref, 'refs/tags/python-v')
         with:
-          pypi_token: ${{ secrets.LANCEDB_PYPI_API_TOKEN }}
           fury_token: ${{ secrets.FURY_TOKEN }}
   gh-release:
     if: startsWith(github.ref, 'refs/tags/python-v')

.github/workflows/python.yml

@@ -7,6 +7,8 @@ on:
   pull_request:
     paths:
       - Cargo.toml
+      - Cargo.lock
+      - rust-toolchain.toml
       - python/**
       - rust/**
       - .github/workflows/python.yml
@@ -15,6 +17,9 @@ on:
       - .github/workflows/build_windows_wheel/**
       - .github/workflows/run_tests/**
 
+permissions:
+  contents: read
+
 concurrency:
   group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
   cancel-in-progress: true
@@ -106,7 +111,6 @@ jobs:
       - name: Install
         run: |
           pip install --extra-index-url https://pypi.fury.io/lance-format/ --extra-index-url https://pypi.fury.io/lancedb/ -e .[tests,dev,embeddings]
-          pip install tantivy
           pip install mlx
       - name: Doctest
         run: pytest --doctest-modules python/lancedb
@@ -225,6 +229,5 @@ jobs:
           pip install "pydantic<2"
           pip install pyarrow==16
           pip install --extra-index-url https://pypi.fury.io/lance-format/ --extra-index-url https://pypi.fury.io/lancedb/ -e .[tests]
-          pip install tantivy
       - name: Run tests
         run: pytest -m "not slow and not s3_test" -x -v --durations=30 python/tests

.github/workflows/rust.yml

@@ -7,9 +7,17 @@ on:
   pull_request:
     paths:
       - Cargo.toml
+      - Cargo.lock
+      - rust-toolchain.toml
+      - deny.toml
       - rust/**
+      - nodejs/Cargo.toml
+      - python/Cargo.toml
       - .github/workflows/rust.yml
 
+permissions:
+  contents: read
+
 concurrency:
   group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }}
   cancel-in-progress: true
@@ -51,6 +59,17 @@ jobs:
       - name: Run clippy (without remote feature)
         run: cargo clippy --profile ci --workspace --tests -- -D warnings
 
+  deny:
+    # Supply-chain checks: advisories, licenses, banned crates, and source
+    # restrictions. Configuration lives in `deny.toml` at the workspace root.
+    timeout-minutes: 10
+    runs-on: ubuntu-24.04
+    steps:
+      - uses: actions/checkout@v4
+      - uses: EmbarkStudios/cargo-deny-action@v2
+        with:
+          command: check advisories bans licenses sources
+
   build-no-lock:
     runs-on: ubuntu-24.04
     timeout-minutes: 30
@@ -206,14 +225,14 @@ jobs:
       - name: Downgrade  dependencies
         # These packages have newer requirements for MSRV
         run: |
-          cargo update -p aws-sdk-bedrockruntime --precise 1.64.0
-          cargo update -p aws-sdk-dynamodb --precise 1.55.0
-          cargo update -p aws-config --precise 1.5.10
-          cargo update -p aws-sdk-kms --precise 1.51.0
-          cargo update -p aws-sdk-s3 --precise 1.65.0
-          cargo update -p aws-sdk-sso --precise 1.50.0
-          cargo update -p aws-sdk-ssooidc --precise 1.51.0
-          cargo update -p aws-sdk-sts --precise 1.51.0
+          cargo update -p aws-sdk-bedrockruntime --precise 1.77.0
+          cargo update -p aws-sdk-dynamodb --precise 1.68.0
+          cargo update -p aws-config --precise 1.6.0
+          cargo update -p aws-sdk-kms --precise 1.63.0
+          cargo update -p aws-sdk-s3 --precise 1.79.0
+          cargo update -p aws-sdk-sso --precise 1.62.0
+          cargo update -p aws-sdk-ssooidc --precise 1.63.0
+          cargo update -p aws-sdk-sts --precise 1.63.0
           cargo update -p home --precise 0.5.9
       - name: cargo +${{ matrix.msrv }} check
         env:

.github/workflows/update_package_lock_run.yml

@@ -3,6 +3,9 @@ name: Update package-lock.json
 on:
   workflow_dispatch:
 
+permissions:
+  contents: read
+
 jobs:
   publish:
     runs-on: ubuntu-latest

.github/workflows/update_package_lock_run_nodejs.yml

@@ -3,6 +3,9 @@ name: Update NodeJs package-lock.json
 on:
   workflow_dispatch:
 
+permissions:
+  contents: read
+
 jobs:
   publish:
     runs-on: ubuntu-latest

.github/workflows/upload_wheel/action.yml

@@ -2,9 +2,6 @@ name: upload-wheel
 
 description: "Upload wheels to Pypi"
 inputs:
-  pypi_token:
-    required: true
-    description: "release token for the repo"
   fury_token:
     required: true
     description: "release token for the fury repo"
@@ -12,12 +9,6 @@ inputs:
 runs:
   using: "composite"
   steps:
-  - name: Install dependencies
-    shell: bash
-    run: |
-      python -m pip install --upgrade pip
-      pip install twine
-      python3 -m pip install --upgrade pkginfo
   - name: Choose repo
     shell: bash
     id: choose_repo
@@ -27,19 +18,17 @@ runs:
       else
         echo "repo=pypi" >> $GITHUB_OUTPUT
       fi
-  - name: Publish to PyPI
+  - name: Publish to Fury
+    if: steps.choose_repo.outputs.repo == 'fury'
     shell: bash
     env:
       FURY_TOKEN: ${{ inputs.fury_token }}
-      PYPI_TOKEN: ${{ inputs.pypi_token }}
     run: |
-      if [[ ${{ steps.choose_repo.outputs.repo }} == fury ]]; then
-        WHEEL=$(ls target/wheels/lancedb-*.whl 2> /dev/null | head -n 1)
-        echo "Uploading $WHEEL to Fury"
-        curl -f -F package=@$WHEEL https://$FURY_TOKEN@push.fury.io/lancedb/
-      else
-        twine upload --repository ${{ steps.choose_repo.outputs.repo }} \
-          --username __token__ \
-          --password $PYPI_TOKEN \
-          target/wheels/lancedb-*.whl
-      fi
+      WHEEL=$(ls target/wheels/lancedb-*.whl 2> /dev/null | head -n 1)
+      echo "Uploading $WHEEL to Fury"
+      curl -f -F package=@$WHEEL https://$FURY_TOKEN@push.fury.io/lancedb/
+  - name: Publish to PyPI
+    if: steps.choose_repo.outputs.repo == 'pypi'
+    uses: pypa/gh-action-pypi-publish@release/v1
+    with:
+      packages-dir: target/wheels/

Cargo.toml

@@ -1,7 +1,5 @@
 [workspace]
 members = ["rust/lancedb", "nodejs", "python"]
-# Python package needs to be built by maturin.
-exclude = ["python"]
 resolver = "2"
 
 [workspace.package]
@@ -15,20 +13,20 @@ categories = ["database-implementations"]
 rust-version = "1.91.0"
 
 [workspace.dependencies]
-lance = { "version" = "=3.0.0-rc.3", default-features = false, "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-core = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-datagen = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-file = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-io = { "version" = "=3.0.0-rc.3", default-features = false, "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-index = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-linalg = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-namespace = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-namespace-impls = { "version" = "=3.0.0-rc.3", default-features = false, "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-table = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-testing = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-datafusion = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-encoding = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
-lance-arrow = { "version" = "=3.0.0-rc.3", "tag" = "v3.0.0-rc.3", "git" = "https://github.com/lance-format/lance.git" }
+lance = { "version" = "=6.0.0-beta.4", default-features = false, "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-core = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-datagen = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-file = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-io = { "version" = "=6.0.0-beta.4", default-features = false, "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-index = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-linalg = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace-impls = { "version" = "=6.0.0-beta.4", default-features = false, "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-table = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-testing = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-datafusion = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-encoding = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
+lance-arrow = { "version" = "=6.0.0-beta.4", "tag" = "v6.0.0-beta.4", "git" = "https://github.com/lance-format/lance.git" }
 ahash = "0.8"
 # Note that this one does not include pyarrow
 arrow = { version = "57.2", optional = false }

README.md

@@ -15,7 +15,7 @@
 
 # **The Multimodal AI Lakehouse**
 
-[**How to Install** ](#how-to-install) ✦ [**Detailed Documentation**](https://lancedb.com/docs) ✦ [**Tutorials and Recipes**](https://github.com/lancedb/vectordb-recipes/tree/main) ✦  [**Contributors**](#contributors) 
+[**How to Install** ](#how-to-install) ✦ [**Detailed Documentation**](https://docs.lancedb.com) ✦ [**Tutorials and Recipes**](https://github.com/lancedb/vectordb-recipes/tree/main) ✦  [**Contributors**](#contributors) 
 
 **The ultimate multimodal data platform for AI/ML applications.** 
 
@@ -57,7 +57,7 @@ LanceDB is a central location where developers can build, train and analyze thei
 
 ## **How to Install**:
 
-Follow the [Quickstart](https://lancedb.com/docs/quickstart/) doc to set up LanceDB locally. 
+Follow the [Quickstart](https://docs.lancedb.com/quickstart) doc to set up LanceDB locally. 
 
 **API & SDK:** We also support Python, Typescript and Rust SDKs
 

ci/check_lance_release.py

@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 import argparse
+import functools
 import json
 import os
 import re
@@ -26,6 +27,7 @@ SEMVER_RE = re.compile(
 )
 
 
+@functools.total_ordering
 @dataclass(frozen=True)
 class SemVer:
     major: int
@@ -156,7 +158,9 @@ def read_current_version(repo_root: Path) -> str:
 
 
 def determine_latest_tag(tags: Iterable[TagInfo]) -> TagInfo:
-    return max(tags, key=lambda tag: tag.semver)
+    # Stable releases (no prerelease) are always preferred over pre-releases.
+    # Within each group, standard semver ordering applies.
+    return max(tags, key=lambda tag: (not tag.semver.prerelease, tag.semver))
 
 
 def write_outputs(args: argparse.Namespace, payload: dict) -> None:

deny.toml

@@ -0,0 +1,172 @@
+# cargo-deny configuration for LanceDB.
+#
+# Run locally with `cargo deny check`. See
+# https://embarkstudios.github.io/cargo-deny/ for the full reference.
+
+# The set of target triples we care about. cargo-deny will only consider
+# dependencies that are used on at least one of these targets. Keeping this
+# explicit avoids noise from platform-specific crates (e.g. wasm, android,
+# ios) that we never actually ship.
+[graph]
+targets = [
+    "x86_64-unknown-linux-gnu",
+    "aarch64-unknown-linux-gnu",
+    "x86_64-apple-darwin",
+    "aarch64-apple-darwin",
+    "x86_64-pc-windows-msvc",
+    "aarch64-pc-windows-msvc",
+]
+all-features = true
+
+[output]
+feature-depth = 1
+
+# ---------------------------------------------------------------------------
+# Advisories: security vulnerabilities and yanked crates.
+# ---------------------------------------------------------------------------
+[advisories]
+version = 2
+# Fail the check if any crate in the lockfile has been yanked from crates.io.
+# Yanked crates are a signal the author retracted the release (often due to
+# bugs or security issues) and should not be depended on.
+yanked = "deny"
+# Advisory IDs we have explicitly reviewed and chosen to accept. Every
+# entry must include a rationale and, where possible, an upstream issue
+# pointing to a fix. Revisit this list whenever dependencies are updated.
+ignore = [
+    # rsa: Marvin Attack timing side-channel in PKCS#1 v1.5 decryption.
+    # Reached only through opendal → reqsign → rsa. We do not use RSA
+    # decryption in LanceDB ourselves; this is dormant in the signing path.
+    # No fixed release exists upstream as of this writing.
+    # https://rustsec.org/advisories/RUSTSEC-2023-0071
+    { id = "RUSTSEC-2023-0071", reason = "rsa crate via opendal/reqsign; no fixed upstream release" },
+
+    # instant: unmaintained. Pulled in via backoff → instant. Upstream
+    # recommends switching to `web-time`; fix has to come from backoff.
+    # https://rustsec.org/advisories/RUSTSEC-2024-0384
+    { id = "RUSTSEC-2024-0384", reason = "transitive via backoff; waiting on backoff replacement" },
+
+    # paste: unmaintained (author archived the repo). Used transitively by
+    # datafusion and the arrow ecosystem; widespread, no drop-in replacement.
+    # https://rustsec.org/advisories/RUSTSEC-2024-0436
+    { id = "RUSTSEC-2024-0436", reason = "transitive via datafusion; awaiting ecosystem migration" },
+
+    # tantivy: segfault on malformed input due to missing bounds check.
+    # Pulled in via lance for full-text search. We only feed tantivy
+    # documents we construct ourselves, not attacker-controlled bytes.
+    # Tracked for a lance dependency bump.
+    # https://rustsec.org/advisories/RUSTSEC-2025-0003
+    { id = "RUSTSEC-2025-0003", reason = "tantivy via lance; inputs are internally produced, not user-supplied bytes" },
+
+    # backoff: unmaintained. Reached only via async-openai. Replacement
+    # requires async-openai to migrate (or us to drop async-openai).
+    # https://rustsec.org/advisories/RUSTSEC-2025-0012
+    { id = "RUSTSEC-2025-0012", reason = "transitive via async-openai; waiting on upstream migration" },
+
+    # number_prefix: unmaintained. Transitive via indicatif → hf-hub.
+    # No security impact, just maintenance status.
+    # https://rustsec.org/advisories/RUSTSEC-2025-0119
+    { id = "RUSTSEC-2025-0119", reason = "transitive via hf-hub/indicatif; cosmetic formatting crate" },
+
+    # rustls-pemfile: unmaintained. Reached from two separate chains:
+    # rustls-native-certs 0.6 (via hyper-rustls 0.24) and object_store 0.12.
+    # Both upstream dependencies need to move before we can drop it.
+    # https://rustsec.org/advisories/RUSTSEC-2025-0134
+    { id = "RUSTSEC-2025-0134", reason = "transitive via rustls-native-certs/object_store; waiting on upstream migration" },
+
+    # rustls-webpki 0.101.7 (old major line): name-constraint checks for
+    # URI / wildcard names. Pulled in only via the legacy rustls 0.21 chain
+    # from aws-smithy-http-client. The 0.103 line we actively use is patched.
+    # Clearing the 0.101 copy requires the aws-sdk chain to migrate off
+    # rustls 0.21.
+    # https://rustsec.org/advisories/RUSTSEC-2026-0098
+    # https://rustsec.org/advisories/RUSTSEC-2026-0099
+    { id = "RUSTSEC-2026-0098", reason = "only affects rustls-webpki 0.101 from legacy aws-smithy/rustls 0.21 chain" },
+    { id = "RUSTSEC-2026-0099", reason = "only affects rustls-webpki 0.101 from legacy aws-smithy/rustls 0.21 chain" },
+
+    # rustls-webpki 0.101.7: reachable panic in CRL parsing. Same legacy
+    # rustls 0.21 chain from aws-smithy-http-client as above. The 0.103 line
+    # we actively use is upgraded to 0.103.13 which contains the fix.
+    # https://rustsec.org/advisories/RUSTSEC-2026-0104
+    { id = "RUSTSEC-2026-0104", reason = "only affects rustls-webpki 0.101 from legacy aws-smithy/rustls 0.21 chain" },
+]
+
+# ---------------------------------------------------------------------------
+# Licenses: only allow licenses we've reviewed as compatible with Apache-2.0.
+# ---------------------------------------------------------------------------
+[licenses]
+version = 2
+# SPDX identifiers for licenses that are compatible with our Apache-2.0
+# distribution. Additions require legal review.
+allow = [
+    "Apache-2.0",
+    "Apache-2.0 WITH LLVM-exception",
+    "MIT",
+    "BSD-2-Clause",
+    "BSD-3-Clause",
+    "ISC",
+    "Unicode-3.0",
+    "Unicode-DFS-2016",
+    "Zlib",
+    "CC0-1.0",
+    "MPL-2.0",
+    "BSL-1.0",
+    "OpenSSL",
+    # 0BSD ("BSD Zero Clause") is effectively public domain — no attribution
+    # required. Pulled in by `mock_instant`.
+    "0BSD",
+    # bzip2-1.0.6 is the permissive upstream bzip2 license (BSD-like). Pulled
+    # in by `libbz2-rs-sys`, the pure-Rust bzip2 implementation.
+    "bzip2-1.0.6",
+    # CDLA-Permissive-2.0 is a permissive data license used by `webpki-roots`
+    # for the Mozilla CA root bundle. Data-only, distribution-compatible.
+    "CDLA-Permissive-2.0",
+]
+confidence-threshold = 0.8
+# Crates whose license cannot be determined from Cargo metadata but whose
+# license we've manually confirmed from upstream. Keep this list minimal.
+[[licenses.clarify]]
+# polars-arrow-format omits the `license` field in its Cargo.toml, but the
+# upstream repo (pola-rs/polars-arrow-format) is dual-licensed Apache-2.0 OR
+# MIT. See https://github.com/pola-rs/polars-arrow-format/blob/main/LICENSE
+crate = "polars-arrow-format"
+expression = "Apache-2.0 OR MIT"
+license-files = []
+
+# ---------------------------------------------------------------------------
+# Bans: disallow specific crates and flag dependency hygiene issues.
+# ---------------------------------------------------------------------------
+[bans]
+# Warn (not deny) on duplicate versions of the same crate. In a large
+# workspace like this one, duplicates are common and often unavoidable
+# transitively. We surface them to discourage growth, but don't fail CI.
+multiple-versions = "warn"
+# Wildcard version requirements (`foo = "*"`) are a footgun — they let any
+# future release in without review. Ban them outright.
+wildcards = "deny"
+# Internal workspace crates reference each other via `path = "..."`, which
+# cargo-deny sees as a wildcard version. That's fine for private workspace
+# members (not published to crates.io), so allow it specifically for paths.
+allow-wildcard-paths = true
+# Features that, if enabled, should cause the check to fail.
+deny = []
+# Crates to skip when checking for duplicate versions.
+skip = []
+# Similar to `skip`, but also skips the entire transitive subtree.
+skip-tree = []
+
+# ---------------------------------------------------------------------------
+# Sources: restrict where crates can come from.
+# ---------------------------------------------------------------------------
+[sources]
+# Deny any registry other than the ones explicitly listed below.
+unknown-registry = "deny"
+# Deny any git dependency whose host isn't in the allow-list below. This
+# prevents accidental pulls from arbitrary forks.
+unknown-git = "deny"
+allow-registry = ["https://github.com/rust-lang/crates.io-index"]
+# Lance is developed in a sibling repo and pulled as a git dependency until
+# releases are cut to crates.io. Allow that specific host.
+allow-git = [
+    "https://github.com/lance-format/lance",
+]

docker-compose.yml

@@ -1,7 +1,7 @@
 version: "3.9"
 services:
   localstack:
-    image: localstack/localstack:3.3
+    image: localstack/localstack:4.0
     ports:
       - 4566:4566
     environment:

dockerfiles/Dockerfile

@@ -24,4 +24,4 @@ RUN python --version && \
   rustc --version && \
   protoc --version
 
-RUN pip install --no-cache-dir tantivy lancedb
+RUN pip install --no-cache-dir lancedb

docs/README.md

@@ -1,6 +1,6 @@
 # LanceDB Documentation
 
-LanceDB docs are available at [lancedb.com/docs](https://lancedb.com/docs).
+LanceDB docs are available at [docs.lancedb.com](https://docs.lancedb.com).
 
 The SDK docs are built and deployed automatically by [Github Actions](../.github/workflows/docs.yml)
 whenever a commit is pushed to the `main` branch. So it is possible for the docs to show

docs/requirements.txt

@@ -1,8 +1,8 @@
-mkdocs==1.5.3
+mkdocs==1.6.1
 mkdocs-jupyter==0.24.1
-mkdocs-material==9.5.3
+mkdocs-material==9.6.23
 mkdocs-autorefs>=0.5,<=1.0
-mkdocstrings[python]==0.25.2
+mkdocstrings[python]>=0.24,<1.0
 griffe>=0.40,<1.0
 mkdocs-render-swagger-plugin>=0.1.0
 pydantic>=2.0,<3.0

docs/src/java/java.md

@@ -14,7 +14,7 @@ Add the following dependency to your `pom.xml`:
 <dependency>
     <groupId>com.lancedb</groupId>
     <artifactId>lancedb-core</artifactId>
-    <version>0.27.0-beta.5</version>
+    <version>0.28.0-beta.10</version>
 </dependency>
 ```
 
@@ -57,32 +57,32 @@ LanceNamespace namespaceClient = LanceDbNamespaceClientBuilder.newBuilder()
 
 ## Metadata Operations
 
-### Creating a Namespace
+### Creating a Namespace Path
 
-Namespaces organize tables hierarchically. Create a namespace before creating tables within it:
+Namespace paths organize tables hierarchically. Create the desired namespace path before creating tables within it:
 
 ```java
 import org.lance.namespace.model.CreateNamespaceRequest;
 import org.lance.namespace.model.CreateNamespaceResponse;
 
-// Create a child namespace
+// Create a child namespace path
 CreateNamespaceRequest request = new CreateNamespaceRequest();
 request.setId(Arrays.asList("my_namespace"));
 
 CreateNamespaceResponse response = namespaceClient.createNamespace(request);
 ```
 
-You can also create nested namespaces:
+You can also create nested namespace paths:
 
 ```java
-// Create a nested namespace: parent/child
+// Create a nested namespace path: parent/child
 CreateNamespaceRequest request = new CreateNamespaceRequest();
 request.setId(Arrays.asList("parent_namespace", "child_namespace"));
 
 CreateNamespaceResponse response = namespaceClient.createNamespace(request);
 ```
 
-### Describing a Namespace
+### Describing a Namespace Path
 
 ```java
 import org.lance.namespace.model.DescribeNamespaceRequest;
@@ -95,22 +95,22 @@ DescribeNamespaceResponse response = namespaceClient.describeNamespace(request);
 System.out.println("Namespace properties: " + response.getProperties());
 ```
 
-### Listing Namespaces
+### Listing Namespace Paths
 
 ```java
 import org.lance.namespace.model.ListNamespacesRequest;
 import org.lance.namespace.model.ListNamespacesResponse;
 
-// List all namespaces at root level
+// List all namespace paths at the root level
 ListNamespacesRequest request = new ListNamespacesRequest();
 request.setId(Arrays.asList());  // Empty for root
 
 ListNamespacesResponse response = namespaceClient.listNamespaces(request);
 for (String ns : response.getNamespaces()) {
-    System.out.println("Namespace: " + ns);
+    System.out.println("Namespace path: " + ns);
 }
 
-// List child namespaces under a parent
+// List child namespace paths under a parent path
 ListNamespacesRequest childRequest = new ListNamespacesRequest();
 childRequest.setId(Arrays.asList("parent_namespace"));
 
@@ -123,7 +123,7 @@ ListNamespacesResponse childResponse = namespaceClient.listNamespaces(childReque
 import org.lance.namespace.model.ListTablesRequest;
 import org.lance.namespace.model.ListTablesResponse;
 
-// List tables in a namespace
+// List tables in a namespace path
 ListTablesRequest request = new ListTablesRequest();
 request.setId(Arrays.asList("my_namespace"));
 
@@ -133,7 +133,7 @@ for (String table : response.getTables()) {
 }
 ```
 
-### Dropping a Namespace
+### Dropping a Namespace Path
 
 ```java
 import org.lance.namespace.model.DropNamespaceRequest;
@@ -175,7 +175,7 @@ DropTableResponse response = namespaceClient.dropTable(request);
 
 ### Creating a Table
 
-Tables are created within a namespace by providing data in Apache Arrow IPC format:
+Tables are created within a namespace path by providing data in Apache Arrow IPC format:
 
 ```java
 import org.lance.namespace.LanceNamespace;
@@ -242,7 +242,7 @@ try (BufferAllocator allocator = new RootAllocator();
     }
     byte[] tableData = out.toByteArray();
 
-    // Create table in a namespace
+    // Create a table in a namespace path
     CreateTableRequest request = new CreateTableRequest();
     request.setId(Arrays.asList("my_namespace", "my_table"));
     CreateTableResponse response = namespaceClient.createTable(request, tableData);

docs/src/js/README.md

@@ -34,7 +34,7 @@ const results = await table.vectorSearch([0.1, 0.3]).limit(20).toArray();
 console.log(results);
 ```
 
-The [quickstart](https://lancedb.com/docs/quickstart/basic-usage/) contains more complete examples.
+The [quickstart](https://docs.lancedb.com/quickstart/) contains more complete examples.
 
 ## Development
 

docs/src/js/classes/Connection.md

@@ -61,8 +61,8 @@ sharing the same data, deletion, and index files.
 * **options.sourceVersion?**: `number`
     The version of the source table to clone.
 
-* **options.targetNamespace?**: `string`[]
-    The namespace for the target table (defaults to root namespace).
+* **options.targetNamespacePath?**: `string`[]
+    The namespace path for the target table (defaults to root namespace).
 
 #### Returns
 
@@ -116,13 +116,13 @@ Creates a new empty Table
 
 `Promise`<[`Table`](Table.md)>
 
-#### createEmptyTable(name, schema, namespace, options)
+#### createEmptyTable(name, schema, namespacePath, options)
 
 ```ts
 abstract createEmptyTable(
    name,
    schema,
-   namespace?,
+   namespacePath?,
    options?): Promise<Table>
 ```
 
@@ -136,8 +136,8 @@ Creates a new empty Table
 * **schema**: [`SchemaLike`](../type-aliases/SchemaLike.md)
     The schema of the table
 
-* **namespace?**: `string`[]
-    The namespace to create the table in (defaults to root namespace)
+* **namespacePath?**: `string`[]
+    The namespace path to create the table in (defaults to root namespace)
 
 * **options?**: `Partial`&lt;[`CreateTableOptions`](../interfaces/CreateTableOptions.md)&gt;
     Additional options
@@ -150,10 +150,10 @@ Creates a new empty Table
 
 ### createTable()
 
-#### createTable(options, namespace)
+#### createTable(options, namespacePath)
 
 ```ts
-abstract createTable(options, namespace?): Promise<Table>
+abstract createTable(options, namespacePath?): Promise<Table>
 ```
 
 Creates a new Table and initialize it with new data.
@@ -163,8 +163,8 @@ Creates a new Table and initialize it with new data.
 * **options**: `object` & `Partial`&lt;[`CreateTableOptions`](../interfaces/CreateTableOptions.md)&gt;
     The options object.
 
-* **namespace?**: `string`[]
-    The namespace to create the table in (defaults to root namespace)
+* **namespacePath?**: `string`[]
+    The namespace path to create the table in (defaults to root namespace)
 
 ##### Returns
 
@@ -197,13 +197,13 @@ Creates a new Table and initialize it with new data.
 
 `Promise`&lt;[`Table`](Table.md)&gt;
 
-#### createTable(name, data, namespace, options)
+#### createTable(name, data, namespacePath, options)
 
 ```ts
 abstract createTable(
    name,
    data,
-   namespace?,
+   namespacePath?,
    options?): Promise<Table>
 ```
 
@@ -218,8 +218,8 @@ Creates a new Table and initialize it with new data.
     Non-empty Array of Records
     to be inserted into the table
 
-* **namespace?**: `string`[]
-    The namespace to create the table in (defaults to root namespace)
+* **namespacePath?**: `string`[]
+    The namespace path to create the table in (defaults to root namespace)
 
 * **options?**: `Partial`&lt;[`CreateTableOptions`](../interfaces/CreateTableOptions.md)&gt;
     Additional options
@@ -247,15 +247,15 @@ Return a brief description of the connection
 ### dropAllTables()
 
 ```ts
-abstract dropAllTables(namespace?): Promise<void>
+abstract dropAllTables(namespacePath?): Promise<void>
 ```
 
 Drop all tables in the database.
 
 #### Parameters
 
-* **namespace?**: `string`[]
-    The namespace to drop tables from (defaults to root namespace).
+* **namespacePath?**: `string`[]
+    The namespace path to drop tables from (defaults to root namespace).
 
 #### Returns
 
@@ -266,7 +266,7 @@ Drop all tables in the database.
 ### dropTable()
 
 ```ts
-abstract dropTable(name, namespace?): Promise<void>
+abstract dropTable(name, namespacePath?): Promise<void>
 ```
 
 Drop an existing table.
@@ -276,8 +276,8 @@ Drop an existing table.
 * **name**: `string`
     The name of the table to drop.
 
-* **namespace?**: `string`[]
-    The namespace of the table (defaults to root namespace).
+* **namespacePath?**: `string`[]
+    The namespace path of the table (defaults to root namespace).
 
 #### Returns
 
@@ -304,7 +304,7 @@ Return true if the connection has not been closed
 ```ts
 abstract openTable(
    name,
-   namespace?,
+   namespacePath?,
    options?): Promise<Table>
 ```
 
@@ -315,8 +315,8 @@ Open a table in the database.
 * **name**: `string`
     The name of the table
 
-* **namespace?**: `string`[]
-    The namespace of the table (defaults to root namespace)
+* **namespacePath?**: `string`[]
+    The namespace path of the table (defaults to root namespace)
 
 * **options?**: `Partial`&lt;[`OpenTableOptions`](../interfaces/OpenTableOptions.md)&gt;
     Additional options
@@ -349,10 +349,10 @@ Tables will be returned in lexicographical order.
 
 `Promise`&lt;`string`[]&gt;
 
-#### tableNames(namespace, options)
+#### tableNames(namespacePath, options)
 
 ```ts
-abstract tableNames(namespace?, options?): Promise<string[]>
+abstract tableNames(namespacePath?, options?): Promise<string[]>
 ```
 
 List all the table names in this database.
@@ -361,8 +361,8 @@ Tables will be returned in lexicographical order.
 
 ##### Parameters
 
-* **namespace?**: `string`[]
-    The namespace to list tables from (defaults to root namespace)
+* **namespacePath?**: `string`[]
+    The namespace path to list tables from (defaults to root namespace)
 
 * **options?**: `Partial`&lt;[`TableNamesOptions`](../interfaces/TableNamesOptions.md)&gt;
     options to control the

docs/src/js/classes/Table.md

@@ -485,19 +485,7 @@ Modeled after ``VACUUM`` in PostgreSQL.
  - Prune: Removes old versions of the dataset
  - Index: Optimizes the indices, adding new data to existing indices
 
- Experimental API
- ----------------
-
- The optimization process is undergoing active development and may change.
- Our goal with these changes is to improve the performance of optimization and
- reduce the complexity.
-
- That being said, it is essential today to run optimize if you want the best
- performance.  It should be stable and safe to use in production, but it our
- hope that the API may be simplified (or not even need to be called) in the
- future.
-
- The frequency an application shoudl call optimize is based on the frequency of
+ The frequency an application should call optimize is based on the frequency of
  data modifications.  If data is frequently added, deleted, or updated then
  optimize should be run frequently.  A good rule of thumb is to run optimize if
  you have added or modified 100,000 or more records or run more than 20 data

docs/src/js/interfaces/ClientConfig.md

@@ -53,3 +53,18 @@ optional tlsConfig: TlsConfig;
 ```ts
 optional userAgent: string;
 ```
+
+***
+
+### userId?
+
+```ts
+optional userId: string;
+```
+
+User identifier for tracking purposes.
+
+This is sent as the `x-lancedb-user-id` header in requests to LanceDB Cloud/Enterprise.
+It can be set directly, or via the `LANCEDB_USER_ID` environment variable.
+Alternatively, set `LANCEDB_USER_ID_ENV_KEY` to specify another environment
+variable that contains the user ID value.

docs/src/js/interfaces/ConnectionOptions.md

@@ -41,6 +41,29 @@ for testing purposes.
 
 ***
 
+### manifestEnabled?
+
+```ts
+optional manifestEnabled: boolean;
+```
+
+(For LanceDB OSS only): use directory namespace manifests as the source
+of truth for table metadata. Existing directory-listed root tables are
+migrated into the manifest on access.
+
+***
+
+### namespaceClientProperties?
+
+```ts
+optional namespaceClientProperties: Record<string, string>;
+```
+
+(For LanceDB OSS only): extra properties for the backing namespace
+client used by manifest-enabled native connections.
+
+***
+
 ### readConsistencyInterval?
 
 ```ts
@@ -89,4 +112,4 @@ optional storageOptions: Record<string, string>;
 
 (For LanceDB OSS only): configuration for object storage.
 
-The available options are described at https://lancedb.com/docs/storage/
+The available options are described at https://docs.lancedb.com/storage/

docs/src/js/interfaces/CreateTableOptions.md

@@ -97,4 +97,4 @@ Configuration for object storage.
 Options already set on the connection will be inherited by the table,
 but can be overridden here.
 
-The available options are described at https://lancedb.com/docs/storage/
+The available options are described at https://docs.lancedb.com/storage/

docs/src/js/interfaces/OpenTableOptions.md

@@ -42,4 +42,4 @@ Configuration for object storage.
 Options already set on the connection will be inherited by the table,
 but can be overridden here.
 
-The available options are described at https://lancedb.com/docs/storage/
+The available options are described at https://docs.lancedb.com/storage/

docs/src/js/interfaces/OptimizeOptions.md

@@ -37,3 +37,12 @@ tbl.optimize({cleanupOlderThan: new Date()});
 ```ts
 deleteUnverified: boolean;
 ```
+
+Because they may be part of an in-progress transaction, files newer than
+7 days old are not deleted by default. If you are sure that there are no
+in-progress transactions, then you can set this to true to delete all
+files older than `cleanupOlderThan`.
+
+**WARNING**: This should only be set to true if you can guarantee that
+no other process is currently working on this dataset. Otherwise the
+dataset could be put into a corrupted state.

docs/src/js/namespaces/embedding/classes/EmbeddingFunction.md

@@ -52,7 +52,7 @@ new EmbeddingFunction<T, M>(): EmbeddingFunction<T, M>
 ### computeQueryEmbeddings()
 
 ```ts
-computeQueryEmbeddings(data): Promise<number[] | Float32Array | Float64Array>
+computeQueryEmbeddings(data): Promise<number[] | Uint8Array | Float32Array | Float64Array>
 ```
 
 Compute the embeddings for a single query
@@ -63,7 +63,7 @@ Compute the embeddings for a single query
 
 #### Returns
 
-`Promise`&lt;`number`[] \| `Float32Array` \| `Float64Array`&gt;
+`Promise`&lt;`number`[] \| `Uint8Array` \| `Float32Array` \| `Float64Array`&gt;
 
 ***
 

docs/src/js/namespaces/embedding/classes/TextEmbeddingFunction.md

@@ -37,7 +37,7 @@ new TextEmbeddingFunction<M>(): TextEmbeddingFunction<M>
 ### computeQueryEmbeddings()
 
 ```ts
-computeQueryEmbeddings(data): Promise<number[] | Float32Array | Float64Array>
+computeQueryEmbeddings(data): Promise<number[] | Uint8Array | Float32Array | Float64Array>
 ```
 
 Compute the embeddings for a single query
@@ -48,7 +48,7 @@ Compute the embeddings for a single query
 
 #### Returns
 
-`Promise`&lt;`number`[] \| `Float32Array` \| `Float64Array`&gt;
+`Promise`&lt;`number`[] \| `Uint8Array` \| `Float32Array` \| `Float64Array`&gt;
 
 #### Overrides
 

docs/src/js/type-aliases/IntoVector.md

@@ -7,5 +7,10 @@
 # Type Alias: IntoVector
 
 ```ts
-type IntoVector: Float32Array | Float64Array | number[] | Promise<Float32Array | Float64Array | number[]>;
+type IntoVector:
+  | Float32Array
+  | Float64Array
+  | Uint8Array
+  | number[]
+  | Promise<Float32Array | Float64Array | Uint8Array | number[]>;
 ```

docs/src/python/python.md

@@ -36,6 +36,20 @@ is also an [asynchronous API client](#connections-asynchronous).
 
 ::: lancedb.table.Tags
 
+## Expressions
+
+Type-safe expression builder for filters and projections. Use these instead
+of raw SQL strings with [where][lancedb.query.LanceQueryBuilder.where] and
+[select][lancedb.query.LanceQueryBuilder.select].
+
+::: lancedb.expr.Expr
+
+::: lancedb.expr.col
+
+::: lancedb.expr.lit
+
+::: lancedb.expr.func
+
 ## Querying (Synchronous)
 
 ::: lancedb.query.Query

java/lancedb-core/pom.xml

@@ -8,7 +8,7 @@
     <parent>
       <groupId>com.lancedb</groupId>
       <artifactId>lancedb-parent</artifactId>
-      <version>0.27.0-beta.5</version>
+      <version>0.28.0-beta.10</version>
       <relativePath>../pom.xml</relativePath>
     </parent>
 

java/pom.xml

@@ -6,7 +6,7 @@
 
     <groupId>com.lancedb</groupId>
     <artifactId>lancedb-parent</artifactId>
-    <version>0.27.0-beta.5</version>
+    <version>0.28.0-beta.10</version>
     <packaging>pom</packaging>
     <name>${project.artifactId}</name>
     <description>LanceDB Java SDK Parent POM</description>
@@ -28,7 +28,7 @@
     <properties>
         <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
         <arrow.version>15.0.0</arrow.version>
-        <lance-core.version>3.1.0-beta.2</lance-core.version>
+        <lance-core.version>6.0.0-beta.4</lance-core.version>
         <spotless.skip>false</spotless.skip>
         <spotless.version>2.30.0</spotless.version>
         <spotless.java.googlejavaformat.version>1.7</spotless.java.googlejavaformat.version>

nodejs/Cargo.toml

@@ -1,7 +1,8 @@
 [package]
 name = "lancedb-nodejs"
 edition.workspace = true
-version = "0.27.0-beta.5"
+version = "0.28.0-beta.10"
+publish = false
 license.workspace = true
 description.workspace = true
 repository.workspace = true
@@ -15,6 +16,8 @@ crate-type = ["cdylib"]
 async-trait.workspace = true
 arrow-ipc.workspace = true
 arrow-array.workspace = true
+arrow-buffer = "57.2"
+half.workspace = true
 arrow-schema.workspace = true
 env_logger.workspace = true
 futures.workspace = true
@@ -29,8 +32,8 @@ lzma-sys = { version = "0.1", features = ["static"] }
 log.workspace = true
 
 # Pin to resolve build failures; update periodically for security patches.
-aws-lc-sys = "=0.38.0"
-aws-lc-rs = "=1.16.1"
+aws-lc-sys = "=0.40.0"
+aws-lc-rs = "=1.16.3"
 
 [build-dependencies]
 napi-build = "2.3.1"

nodejs/README.md

@@ -30,7 +30,7 @@ const results = await table.vectorSearch([0.1, 0.3]).limit(20).toArray();
 console.log(results);
 ```
 
-The [quickstart](https://lancedb.com/docs/quickstart/basic-usage/) contains more complete examples.
+The [quickstart](https://docs.lancedb.com/quickstart/) contains more complete examples.
 
 ## Development
 

nodejs/__test__/rerankers.test.ts

@@ -1,6 +1,8 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright The LanceDB Authors
 
+import { spawn } from "node:child_process";
+import * as path from "node:path";
 import { RecordBatch } from "apache-arrow";
 import * as tmp from "tmp";
 import { Connection, Index, Table, connect, makeArrowTable } from "../lancedb";
@@ -76,4 +78,91 @@ describe("rerankers", function () {
 
     expect(result).toHaveLength(2);
   });
+
+  it("does not keep process alive after rerank query", async function () {
+    const script = `
+import * as lancedb from "./dist/index.js";
+import * as os from "node:os";
+import * as path from "node:path";
+import * as fs from "node:fs/promises";
+
+const dir = await fs.mkdtemp(path.join(os.tmpdir(), "lancedb-rerank-exit-"));
+const db = await lancedb.connect(dir);
+const table = await db.createTable("test", [{ text: "hello", vector: [1, 2, 3] }], {
+  mode: "overwrite",
+});
+await table.createIndex("text", { config: lancedb.Index.fts() });
+await table.waitForIndex(["text_idx"], 30);
+
+const reranker = await lancedb.rerankers.RRFReranker.create();
+await table
+  .query()
+  .nearestTo([1, 2, 3])
+  .fullTextSearch("hello")
+  .rerank(reranker)
+  .toArray();
+
+table.close();
+db.close();
+`;
+
+    await new Promise<void>((resolve, reject) => {
+      const child = spawn(
+        process.execPath,
+        ["--input-type=module", "-e", script],
+        {
+          cwd: path.resolve(__dirname, ".."),
+          stdio: ["ignore", "pipe", "pipe"],
+        },
+      );
+
+      let stdout = "";
+      let stderr = "";
+
+      child.stdout.on("data", (chunk) => {
+        stdout += chunk.toString();
+      });
+
+      child.stderr.on("data", (chunk) => {
+        stderr += chunk.toString();
+      });
+
+      const timeout = setTimeout(() => {
+        child.kill();
+        reject(
+          new Error(
+            `child process did not exit in time\nstdout:\n${stdout}\nstderr:\n${stderr}`,
+          ),
+        );
+      }, 20_000);
+
+      child.on("error", (err) => {
+        clearTimeout(timeout);
+        reject(err);
+      });
+
+      child.on("exit", (code, signal) => {
+        clearTimeout(timeout);
+        if (signal !== null) {
+          reject(
+            new Error(
+              `child process exited with signal ${signal}\nstdout:\n${stdout}\nstderr:\n${stderr}`,
+            ),
+          );
+          return;
+        }
+
+        if (code !== 0) {
+          reject(
+            new Error(
+              `child process exited with code ${code}\nstdout:\n${stdout}\nstderr:\n${stderr}`,
+            ),
+          );
+          return;
+        }
+
+        resolve();
+      });
+    });
+  });
 });

nodejs/__test__/table.test.ts

@@ -103,7 +103,7 @@ describe.each([arrow15, arrow16, arrow17, arrow18])(
         },
         numIndices: 0,
         numRows: 3,
-        totalBytes: 24,
+        totalBytes: 44,
       });
     });
 

nodejs/__test__/vector_types.test.ts

@@ -0,0 +1,110 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+import * as tmp from "tmp";
+
+import { type Table, connect } from "../lancedb";
+import {
+  Field,
+  FixedSizeList,
+  Float32,
+  Int64,
+  Schema,
+  makeArrowTable,
+} from "../lancedb/arrow";
+
+describe("Vector query with different typed arrays", () => {
+  let tmpDir: tmp.DirResult;
+
+  afterEach(() => {
+    tmpDir?.removeCallback();
+  });
+
+  async function createFloat32Table(): Promise<Table> {
+    tmpDir = tmp.dirSync({ unsafeCleanup: true });
+    const db = await connect(tmpDir.name);
+    const schema = new Schema([
+      new Field("id", new Int64(), true),
+      new Field(
+        "vec",
+        new FixedSizeList(2, new Field("item", new Float32())),
+        true,
+      ),
+    ]);
+    const data = makeArrowTable(
+      [
+        { id: 1n, vec: [1.0, 0.0] },
+        { id: 2n, vec: [0.0, 1.0] },
+        { id: 3n, vec: [1.0, 1.0] },
+      ],
+      { schema },
+    );
+    return db.createTable("test_f32", data);
+  }
+
+  it("should search with Float32Array (baseline)", async () => {
+    const table = await createFloat32Table();
+    const results = await table
+      .query()
+      .nearestTo(new Float32Array([1.0, 0.0]))
+      .limit(1)
+      .toArray();
+
+    expect(results.length).toBe(1);
+    expect(Number(results[0].id)).toBe(1);
+  });
+
+  it("should search with number[] (backward compat)", async () => {
+    const table = await createFloat32Table();
+    const results = await table
+      .query()
+      .nearestTo([1.0, 0.0])
+      .limit(1)
+      .toArray();
+
+    expect(results.length).toBe(1);
+    expect(Number(results[0].id)).toBe(1);
+  });
+
+  it("should search with Float64Array via raw path", async () => {
+    const table = await createFloat32Table();
+    const results = await table
+      .query()
+      .nearestTo(new Float64Array([1.0, 0.0]))
+      .limit(1)
+      .toArray();
+
+    expect(results.length).toBe(1);
+    expect(Number(results[0].id)).toBe(1);
+  });
+
+  it("should add multiple query vectors with Float64Array", async () => {
+    const table = await createFloat32Table();
+    const results = await table
+      .query()
+      .nearestTo(new Float64Array([1.0, 0.0]))
+      .addQueryVector(new Float64Array([0.0, 1.0]))
+      .limit(2)
+      .toArray();
+
+    expect(results.length).toBeGreaterThanOrEqual(2);
+  });
+
+  // Float16Array is only available in Node 22+; not in TypeScript's standard lib yet
+  const float16ArrayCtor = (globalThis as unknown as Record<string, unknown>)
+    .Float16Array as (new (values: number[]) => unknown) | undefined;
+  const hasFloat16 = float16ArrayCtor !== undefined;
+  const f16it = hasFloat16 ? it : it.skip;
+
+  f16it("should search with Float16Array via raw path", async () => {
+    const table = await createFloat32Table();
+    const results = await table
+      .query()
+      .nearestTo(new float16ArrayCtor!([1.0, 0.0]) as Float32Array)
+      .limit(1)
+      .toArray();
+
+    expect(results.length).toBe(1);
+    expect(Number(results[0].id)).toBe(1);
+  });
+});

nodejs/lancedb/arrow.ts

@@ -117,8 +117,9 @@ export type TableLike =
 export type IntoVector =
   | Float32Array
   | Float64Array
+  | Uint8Array
   | number[]
-  | Promise<Float32Array | Float64Array | number[]>;
+  | Promise<Float32Array | Float64Array | Uint8Array | number[]>;
 
 export type MultiVector = IntoVector[];
 
@@ -126,14 +127,48 @@ export function isMultiVector(value: unknown): value is MultiVector {
   return Array.isArray(value) && isIntoVector(value[0]);
 }
 
+// Float16Array is not in TypeScript's standard lib yet; access dynamically
+type Float16ArrayCtor = new (
+  ...args: unknown[]
+) => { buffer: ArrayBuffer; byteOffset: number; byteLength: number };
+const float16ArrayCtor = (globalThis as unknown as Record<string, unknown>)
+  .Float16Array as Float16ArrayCtor | undefined;
+
 export function isIntoVector(value: unknown): value is IntoVector {
   return (
     value instanceof Float32Array ||
     value instanceof Float64Array ||
+    value instanceof Uint8Array ||
+    (float16ArrayCtor !== undefined && value instanceof float16ArrayCtor) ||
     (Array.isArray(value) && !Array.isArray(value[0]))
   );
 }
 
+/**
+ * Extract the underlying byte buffer and data type from a typed array
+ * for passing to the Rust NAPI layer without precision loss.
+ */
+export function extractVectorBuffer(
+  vector: Float32Array | Float64Array | Uint8Array,
+): { data: Uint8Array; dtype: string } | null {
+  if (float16ArrayCtor !== undefined && vector instanceof float16ArrayCtor) {
+    return {
+      data: new Uint8Array(vector.buffer, vector.byteOffset, vector.byteLength),
+      dtype: "float16",
+    };
+  }
+  if (vector instanceof Float64Array) {
+    return {
+      data: new Uint8Array(vector.buffer, vector.byteOffset, vector.byteLength),
+      dtype: "float64",
+    };
+  }
+  if (vector instanceof Uint8Array && !(vector instanceof Float32Array)) {
+    return { data: vector, dtype: "uint8" };
+  }
+  return null;
+}
+
 export function isArrowTable(value: object): value is TableLike {
   if (value instanceof ArrowTable) return true;
   return "schema" in value && "batches" in value;

nodejs/lancedb/connection.ts

@@ -42,7 +42,7 @@ export interface CreateTableOptions {
    * Options already set on the connection will be inherited by the table,
    * but can be overridden here.
    *
-   * The available options are described at https://lancedb.com/docs/storage/
+   * The available options are described at https://docs.lancedb.com/storage/
    */
   storageOptions?: Record<string, string>;
 
@@ -78,7 +78,7 @@ export interface OpenTableOptions {
    * Options already set on the connection will be inherited by the table,
    * but can be overridden here.
    *
-   * The available options are described at https://lancedb.com/docs/storage/
+   * The available options are described at https://docs.lancedb.com/storage/
    */
   storageOptions?: Record<string, string>;
   /**
@@ -166,25 +166,25 @@ export abstract class Connection {
    * List all the table names in this database.
    *
    * Tables will be returned in lexicographical order.
-   * @param {string[]} namespace - The namespace to list tables from (defaults to root namespace)
+   * @param {string[]} namespacePath - The namespace path to list tables from (defaults to root namespace)
    * @param {Partial<TableNamesOptions>} options - options to control the
    * paging / start point
    *
    */
   abstract tableNames(
-    namespace?: string[],
+    namespacePath?: string[],
     options?: Partial<TableNamesOptions>,
   ): Promise<string[]>;
 
   /**
    * Open a table in the database.
    * @param {string} name - The name of the table
-   * @param {string[]} namespace - The namespace of the table (defaults to root namespace)
+   * @param {string[]} namespacePath - The namespace path of the table (defaults to root namespace)
    * @param {Partial<OpenTableOptions>} options - Additional options
    */
   abstract openTable(
     name: string,
-    namespace?: string[],
+    namespacePath?: string[],
     options?: Partial<OpenTableOptions>,
   ): Promise<Table>;
 
@@ -193,7 +193,7 @@ export abstract class Connection {
    * @param {object} options - The options object.
    * @param {string} options.name - The name of the table.
    * @param {Data} options.data - Non-empty Array of Records to be inserted into the table
-   * @param {string[]} namespace - The namespace to create the table in (defaults to root namespace)
+   * @param {string[]} namespacePath - The namespace path to create the table in (defaults to root namespace)
    *
    */
   abstract createTable(
@@ -201,7 +201,7 @@ export abstract class Connection {
       name: string;
       data: Data;
     } & Partial<CreateTableOptions>,
-    namespace?: string[],
+    namespacePath?: string[],
   ): Promise<Table>;
   /**
    * Creates a new Table and initialize it with new data.
@@ -220,13 +220,13 @@ export abstract class Connection {
    * @param {string} name - The name of the table.
    * @param {Record<string, unknown>[] | TableLike} data - Non-empty Array of Records
    * to be inserted into the table
-   * @param {string[]} namespace - The namespace to create the table in (defaults to root namespace)
+   * @param {string[]} namespacePath - The namespace path to create the table in (defaults to root namespace)
    * @param {Partial<CreateTableOptions>} options - Additional options
    */
   abstract createTable(
     name: string,
     data: Record<string, unknown>[] | TableLike,
-    namespace?: string[],
+    namespacePath?: string[],
     options?: Partial<CreateTableOptions>,
   ): Promise<Table>;
 
@@ -245,28 +245,28 @@ export abstract class Connection {
    * Creates a new empty Table
    * @param {string} name - The name of the table.
    * @param {Schema} schema - The schema of the table
-   * @param {string[]} namespace - The namespace to create the table in (defaults to root namespace)
+   * @param {string[]} namespacePath - The namespace path to create the table in (defaults to root namespace)
    * @param {Partial<CreateTableOptions>} options - Additional options
    */
   abstract createEmptyTable(
     name: string,
     schema: import("./arrow").SchemaLike,
-    namespace?: string[],
+    namespacePath?: string[],
     options?: Partial<CreateTableOptions>,
   ): Promise<Table>;
 
   /**
    * Drop an existing table.
    * @param {string} name The name of the table to drop.
-   * @param {string[]} namespace The namespace of the table (defaults to root namespace).
+   * @param {string[]} namespacePath The namespace path of the table (defaults to root namespace).
    */
-  abstract dropTable(name: string, namespace?: string[]): Promise<void>;
+  abstract dropTable(name: string, namespacePath?: string[]): Promise<void>;
 
   /**
    * Drop all tables in the database.
-   * @param {string[]} namespace The namespace to drop tables from (defaults to root namespace).
+   * @param {string[]} namespacePath The namespace path to drop tables from (defaults to root namespace).
    */
-  abstract dropAllTables(namespace?: string[]): Promise<void>;
+  abstract dropAllTables(namespacePath?: string[]): Promise<void>;
 
   /**
    * Clone a table from a source table.
@@ -279,7 +279,7 @@ export abstract class Connection {
    * @param {string} targetTableName - The name of the target table to create.
    * @param {string} sourceUri - The URI of the source table to clone from.
    * @param {object} options - Clone options.
-   * @param {string[]} options.targetNamespace - The namespace for the target table (defaults to root namespace).
+   * @param {string[]} options.targetNamespacePath - The namespace path for the target table (defaults to root namespace).
    * @param {number} options.sourceVersion - The version of the source table to clone.
    * @param {string} options.sourceTag - The tag of the source table to clone.
    * @param {boolean} options.isShallow - Whether to perform a shallow clone (defaults to true).
@@ -288,7 +288,7 @@ export abstract class Connection {
     targetTableName: string,
     sourceUri: string,
     options?: {
-      targetNamespace?: string[];
+      targetNamespacePath?: string[];
       sourceVersion?: number;
       sourceTag?: string;
       isShallow?: boolean;
@@ -319,25 +319,25 @@ export class LocalConnection extends Connection {
   }
 
   async tableNames(
-    namespaceOrOptions?: string[] | Partial<TableNamesOptions>,
+    namespacePathOrOptions?: string[] | Partial<TableNamesOptions>,
     options?: Partial<TableNamesOptions>,
   ): Promise<string[]> {
-    // Detect if first argument is namespace array or options object
-    let namespace: string[] | undefined;
+    // Detect if first argument is namespacePath array or options object
+    let namespacePath: string[] | undefined;
     let tableNamesOptions: Partial<TableNamesOptions> | undefined;
 
-    if (Array.isArray(namespaceOrOptions)) {
-      // First argument is namespace array
-      namespace = namespaceOrOptions;
+    if (Array.isArray(namespacePathOrOptions)) {
+      // First argument is namespacePath array
+      namespacePath = namespacePathOrOptions;
       tableNamesOptions = options;
     } else {
       // First argument is options object (backwards compatibility)
-      namespace = undefined;
-      tableNamesOptions = namespaceOrOptions;
+      namespacePath = undefined;
+      tableNamesOptions = namespacePathOrOptions;
     }
 
     return this.inner.tableNames(
-      namespace ?? [],
+      namespacePath ?? [],
       tableNamesOptions?.startAfter,
       tableNamesOptions?.limit,
     );
@@ -345,12 +345,12 @@ export class LocalConnection extends Connection {
 
   async openTable(
     name: string,
-    namespace?: string[],
+    namespacePath?: string[],
     options?: Partial<OpenTableOptions>,
   ): Promise<Table> {
     const innerTable = await this.inner.openTable(
       name,
-      namespace ?? [],
+      namespacePath ?? [],
       cleanseStorageOptions(options?.storageOptions),
       options?.indexCacheSize,
     );
@@ -362,7 +362,7 @@ export class LocalConnection extends Connection {
     targetTableName: string,
     sourceUri: string,
     options?: {
-      targetNamespace?: string[];
+      targetNamespacePath?: string[];
       sourceVersion?: number;
       sourceTag?: string;
       isShallow?: boolean;
@@ -371,7 +371,7 @@ export class LocalConnection extends Connection {
     const innerTable = await this.inner.cloneTable(
       targetTableName,
       sourceUri,
-      options?.targetNamespace ?? [],
+      options?.targetNamespacePath ?? [],
       options?.sourceVersion ?? null,
       options?.sourceTag ?? null,
       options?.isShallow ?? true,
@@ -406,42 +406,42 @@ export class LocalConnection extends Connection {
     nameOrOptions:
       | string
       | ({ name: string; data: Data } & Partial<CreateTableOptions>),
-    dataOrNamespace?: Record<string, unknown>[] | TableLike | string[],
-    namespaceOrOptions?: string[] | Partial<CreateTableOptions>,
+    dataOrNamespacePath?: Record<string, unknown>[] | TableLike | string[],
+    namespacePathOrOptions?: string[] | Partial<CreateTableOptions>,
     options?: Partial<CreateTableOptions>,
   ): Promise<Table> {
     if (typeof nameOrOptions !== "string" && "name" in nameOrOptions) {
-      // First overload: createTable(options, namespace?)
+      // First overload: createTable(options, namespacePath?)
       const { name, data, ...createOptions } = nameOrOptions;
-      const namespace = dataOrNamespace as string[] | undefined;
-      return this._createTableImpl(name, data, namespace, createOptions);
+      const namespacePath = dataOrNamespacePath as string[] | undefined;
+      return this._createTableImpl(name, data, namespacePath, createOptions);
     }
 
-    // Second overload: createTable(name, data, namespace?, options?)
+    // Second overload: createTable(name, data, namespacePath?, options?)
     const name = nameOrOptions;
-    const data = dataOrNamespace as Record<string, unknown>[] | TableLike;
+    const data = dataOrNamespacePath as Record<string, unknown>[] | TableLike;
 
-    // Detect if third argument is namespace array or options object
-    let namespace: string[] | undefined;
+    // Detect if third argument is namespacePath array or options object
+    let namespacePath: string[] | undefined;
     let createOptions: Partial<CreateTableOptions> | undefined;
 
-    if (Array.isArray(namespaceOrOptions)) {
-      // Third argument is namespace array
-      namespace = namespaceOrOptions;
+    if (Array.isArray(namespacePathOrOptions)) {
+      // Third argument is namespacePath array
+      namespacePath = namespacePathOrOptions;
       createOptions = options;
     } else {
       // Third argument is options object (backwards compatibility)
-      namespace = undefined;
-      createOptions = namespaceOrOptions;
+      namespacePath = undefined;
+      createOptions = namespacePathOrOptions;
     }
 
-    return this._createTableImpl(name, data, namespace, createOptions);
+    return this._createTableImpl(name, data, namespacePath, createOptions);
   }
 
   private async _createTableImpl(
     name: string,
     data: Data,
-    namespace?: string[],
+    namespacePath?: string[],
     options?: Partial<CreateTableOptions>,
   ): Promise<Table> {
     if (data === undefined) {
@@ -455,7 +455,7 @@ export class LocalConnection extends Connection {
       name,
       buf,
       mode,
-      namespace ?? [],
+      namespacePath ?? [],
       storageOptions,
     );
 
@@ -465,21 +465,21 @@ export class LocalConnection extends Connection {
   async createEmptyTable(
     name: string,
     schema: import("./arrow").SchemaLike,
-    namespaceOrOptions?: string[] | Partial<CreateTableOptions>,
+    namespacePathOrOptions?: string[] | Partial<CreateTableOptions>,
     options?: Partial<CreateTableOptions>,
   ): Promise<Table> {
-    // Detect if third argument is namespace array or options object
-    let namespace: string[] | undefined;
+    // Detect if third argument is namespacePath array or options object
+    let namespacePath: string[] | undefined;
     let createOptions: Partial<CreateTableOptions> | undefined;
 
-    if (Array.isArray(namespaceOrOptions)) {
-      // Third argument is namespace array
-      namespace = namespaceOrOptions;
+    if (Array.isArray(namespacePathOrOptions)) {
+      // Third argument is namespacePath array
+      namespacePath = namespacePathOrOptions;
       createOptions = options;
     } else {
       // Third argument is options object (backwards compatibility)
-      namespace = undefined;
-      createOptions = namespaceOrOptions;
+      namespacePath = undefined;
+      createOptions = namespacePathOrOptions;
     }
 
     let mode: string = createOptions?.mode ?? "create";
@@ -502,18 +502,18 @@ export class LocalConnection extends Connection {
       name,
       buf,
       mode,
-      namespace ?? [],
+      namespacePath ?? [],
       storageOptions,
     );
     return new LocalTable(innerTable);
   }
 
-  async dropTable(name: string, namespace?: string[]): Promise<void> {
-    return this.inner.dropTable(name, namespace ?? []);
+  async dropTable(name: string, namespacePath?: string[]): Promise<void> {
+    return this.inner.dropTable(name, namespacePath ?? []);
   }
 
-  async dropAllTables(namespace?: string[]): Promise<void> {
-    return this.inner.dropAllTables(namespace ?? []);
+  async dropAllTables(namespacePath?: string[]): Promise<void> {
+    return this.inner.dropAllTables(namespacePath ?? []);
   }
 }
 

nodejs/lancedb/query.ts

@@ -5,6 +5,7 @@ import {
   Table as ArrowTable,
   type IntoVector,
   RecordBatch,
+  extractVectorBuffer,
   fromBufferToRecordBatch,
   fromRecordBatchToBuffer,
   tableFromIPC,
@@ -661,10 +662,8 @@ export class VectorQuery extends StandardQueryBase<NativeVectorQuery> {
       const res = (async () => {
         try {
           const v = await vector;
-          const arr = Float32Array.from(v);
-          //
           // biome-ignore lint/suspicious/noExplicitAny: we need to get the `inner`, but js has no package scoping
-          const value: any = this.addQueryVector(arr);
+          const value: any = this.addQueryVector(v);
           const inner = value.inner as
             | NativeVectorQuery
             | Promise<NativeVectorQuery>;
@@ -676,7 +675,12 @@ export class VectorQuery extends StandardQueryBase<NativeVectorQuery> {
       return new VectorQuery(res);
     } else {
       super.doCall((inner) => {
-        inner.addQueryVector(Float32Array.from(vector));
+        const raw = Array.isArray(vector) ? null : extractVectorBuffer(vector);
+        if (raw) {
+          inner.addQueryVectorRaw(raw.data, raw.dtype);
+        } else {
+          inner.addQueryVector(Float32Array.from(vector as number[]));
+        }
       });
       return this;
     }
@@ -765,14 +769,23 @@ export class Query extends StandardQueryBase<NativeQuery> {
    * a default `limit` of 10 will be used.  @see {@link Query#limit}
    */
   nearestTo(vector: IntoVector): VectorQuery {
+    const callNearestTo = (
+      inner: NativeQuery,
+      resolved: Float32Array | Float64Array | Uint8Array | number[],
+    ): NativeVectorQuery => {
+      const raw = Array.isArray(resolved)
+        ? null
+        : extractVectorBuffer(resolved);
+      if (raw) {
+        return inner.nearestToRaw(raw.data, raw.dtype);
+      }
+      return inner.nearestTo(Float32Array.from(resolved as number[]));
+    };
+
     if (this.inner instanceof Promise) {
       const nativeQuery = this.inner.then(async (inner) => {
-        if (vector instanceof Promise) {
-          const arr = await vector.then((v) => Float32Array.from(v));
-          return inner.nearestTo(arr);
-        } else {
-          return inner.nearestTo(Float32Array.from(vector));
-        }
+        const resolved = vector instanceof Promise ? await vector : vector;
+        return callNearestTo(inner, resolved);
       });
       return new VectorQuery(nativeQuery);
     }
@@ -780,10 +793,8 @@ export class Query extends StandardQueryBase<NativeQuery> {
       const res = (async () => {
         try {
           const v = await vector;
-          const arr = Float32Array.from(v);
-          //
           // biome-ignore lint/suspicious/noExplicitAny: we need to get the `inner`, but js has no package scoping
-          const value: any = this.nearestTo(arr);
+          const value: any = this.nearestTo(v);
           const inner = value.inner as
             | NativeVectorQuery
             | Promise<NativeVectorQuery>;
@@ -794,7 +805,7 @@ export class Query extends StandardQueryBase<NativeQuery> {
       })();
       return new VectorQuery(res);
     } else {
-      const vectorQuery = this.inner.nearestTo(Float32Array.from(vector));
+      const vectorQuery = callNearestTo(this.inner, vector);
       return new VectorQuery(vectorQuery);
     }
   }

nodejs/npm/darwin-arm64/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-darwin-arm64",
-	"version": "0.27.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["darwin"],
 	"cpu": ["arm64"],
 	"main": "lancedb.darwin-arm64.node",

nodejs/npm/linux-arm64-gnu/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-gnu",
-	"version": "0.27.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-gnu.node",

nodejs/npm/linux-arm64-musl/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-musl",
-	"version": "0.27.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-musl.node",

nodejs/npm/linux-x64-gnu/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-gnu",
-	"version": "0.27.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-gnu.node",

nodejs/npm/linux-x64-musl/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-musl",
-	"version": "0.27.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-musl.node",

nodejs/npm/win32-arm64-msvc/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lancedb/lancedb-win32-arm64-msvc",
-  "version": "0.27.0-beta.5",
+  "version": "0.28.0-beta.10",
   "os": [
     "win32"
   ],

nodejs/npm/win32-x64-msvc/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-win32-x64-msvc",
-	"version": "0.27.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["win32"],
 	"cpu": ["x64"],
 	"main": "lancedb.win32-x64-msvc.node",

nodejs/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@lancedb/lancedb",
-  "version": "0.27.0-beta.5",
+  "version": "0.28.0-beta.10",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "@lancedb/lancedb",
-      "version": "0.27.0-beta.5",
+      "version": "0.28.0-beta.10",
       "cpu": [
         "x64",
         "arm64"

nodejs/package.json

@@ -11,7 +11,7 @@
     "ann"
   ],
   "private": false,
-  "version": "0.27.0-beta.5",
+  "version": "0.28.0-beta.10",
   "main": "dist/index.js",
   "exports": {
     ".": "./dist/index.js",

nodejs/src/connection.rs

@@ -67,6 +67,12 @@ impl Connection {
                 builder = builder.storage_option(key, value);
             }
         }
+        if let Some(manifest_enabled) = options.manifest_enabled {
+            builder = builder.manifest_enabled(manifest_enabled);
+        }
+        if let Some(namespace_client_properties) = options.namespace_client_properties {
+            builder = builder.namespace_client_properties(namespace_client_properties);
+        }
 
         // Create client config, optionally with header provider
         let client_config = options.client_config.unwrap_or_default();
@@ -119,12 +125,12 @@ impl Connection {
     #[napi(catch_unwind)]
     pub async fn table_names(
         &self,
-        namespace: Vec<String>,
+        namespace_path: Option<Vec<String>>,
         start_after: Option<String>,
         limit: Option<u32>,
     ) -> napi::Result<Vec<String>> {
         let mut op = self.get_inner()?.table_names();
-        op = op.namespace(namespace);
+        op = op.namespace(namespace_path.unwrap_or_default());
         if let Some(start_after) = start_after {
             op = op.start_after(start_after);
         }
@@ -146,7 +152,7 @@ impl Connection {
         name: String,
         buf: Buffer,
         mode: String,
-        namespace: Vec<String>,
+        namespace_path: Option<Vec<String>>,
         storage_options: Option<HashMap<String, String>>,
     ) -> napi::Result<Table> {
         let batches = ipc_file_to_batches(buf.to_vec())
@@ -154,7 +160,7 @@ impl Connection {
         let mode = Self::parse_create_mode_str(&mode)?;
         let mut builder = self.get_inner()?.create_table(&name, batches).mode(mode);
 
-        builder = builder.namespace(namespace);
+        builder = builder.namespace(namespace_path.unwrap_or_default());
 
         if let Some(storage_options) = storage_options {
             for (key, value) in storage_options {
@@ -171,7 +177,7 @@ impl Connection {
         name: String,
         schema_buf: Buffer,
         mode: String,
-        namespace: Vec<String>,
+        namespace_path: Option<Vec<String>>,
         storage_options: Option<HashMap<String, String>>,
     ) -> napi::Result<Table> {
         let schema = ipc_file_to_schema(schema_buf.to_vec()).map_err(|e| {
@@ -183,7 +189,7 @@ impl Connection {
             .create_empty_table(&name, schema)
             .mode(mode);
 
-        builder = builder.namespace(namespace);
+        builder = builder.namespace(namespace_path.unwrap_or_default());
 
         if let Some(storage_options) = storage_options {
             for (key, value) in storage_options {
@@ -198,13 +204,13 @@ impl Connection {
     pub async fn open_table(
         &self,
         name: String,
-        namespace: Vec<String>,
+        namespace_path: Option<Vec<String>>,
         storage_options: Option<HashMap<String, String>>,
         index_cache_size: Option<u32>,
     ) -> napi::Result<Table> {
         let mut builder = self.get_inner()?.open_table(&name);
 
-        builder = builder.namespace(namespace);
+        builder = builder.namespace(namespace_path.unwrap_or_default());
 
         if let Some(storage_options) = storage_options {
             for (key, value) in storage_options {
@@ -223,7 +229,7 @@ impl Connection {
         &self,
         target_table_name: String,
         source_uri: String,
-        target_namespace: Vec<String>,
+        target_namespace_path: Option<Vec<String>>,
         source_version: Option<i64>,
         source_tag: Option<String>,
         is_shallow: bool,
@@ -232,7 +238,7 @@ impl Connection {
             .get_inner()?
             .clone_table(&target_table_name, &source_uri);
 
-        builder = builder.target_namespace(target_namespace);
+        builder = builder.target_namespace(target_namespace_path.unwrap_or_default());
 
         if let Some(version) = source_version {
             builder = builder.source_version(version as u64);
@@ -250,18 +256,21 @@ impl Connection {
 
     /// Drop table with the name. Or raise an error if the table does not exist.
     #[napi(catch_unwind)]
-    pub async fn drop_table(&self, name: String, namespace: Vec<String>) -> napi::Result<()> {
+    pub async fn drop_table(
+        &self,
+        name: String,
+        namespace_path: Option<Vec<String>>,
+    ) -> napi::Result<()> {
+        let ns = namespace_path.unwrap_or_default();
         self.get_inner()?
-            .drop_table(&name, &namespace)
+            .drop_table(&name, &ns)
             .await
             .default_error()
     }
 
     #[napi(catch_unwind)]
-    pub async fn drop_all_tables(&self, namespace: Vec<String>) -> napi::Result<()> {
-        self.get_inner()?
-            .drop_all_tables(&namespace)
-            .await
-            .default_error()
+    pub async fn drop_all_tables(&self, namespace_path: Option<Vec<String>>) -> napi::Result<()> {
+        let ns = namespace_path.unwrap_or_default();
+        self.get_inner()?.drop_all_tables(&ns).await.default_error()
     }
 }

nodejs/src/lib.rs

@@ -35,8 +35,15 @@ pub struct ConnectionOptions {
     pub read_consistency_interval: Option<f64>,
     /// (For LanceDB OSS only): configuration for object storage.
     ///
-    /// The available options are described at https://lancedb.com/docs/storage/
+    /// The available options are described at https://docs.lancedb.com/storage/
     pub storage_options: Option<HashMap<String, String>>,
+    /// (For LanceDB OSS only): use directory namespace manifests as the source
+    /// of truth for table metadata. Existing directory-listed root tables are
+    /// migrated into the manifest on access.
+    pub manifest_enabled: Option<bool>,
+    /// (For LanceDB OSS only): extra properties for the backing namespace
+    /// client used by manifest-enabled native connections.
+    pub namespace_client_properties: Option<HashMap<String, String>>,
     /// (For LanceDB OSS only): the session to use for this connection. Holds
     /// shared caches and other session-specific state.
     pub session: Option<session::Session>,

nodejs/src/query.rs

@@ -3,6 +3,12 @@
 
 use std::sync::Arc;
 
+use arrow_array::{
+    Array, Float16Array as ArrowFloat16Array, Float32Array as ArrowFloat32Array,
+    Float64Array as ArrowFloat64Array, UInt8Array as ArrowUInt8Array,
+};
+use arrow_buffer::ScalarBuffer;
+use half::f16;
 use lancedb::index::scalar::{
     BooleanQuery, BoostQuery, FtsQuery, FullTextSearchQuery, MatchQuery, MultiMatchQuery, Occur,
     Operator, PhraseQuery,
@@ -24,6 +30,33 @@ use crate::rerankers::RerankHybridCallbackArgs;
 use crate::rerankers::Reranker;
 use crate::util::{parse_distance_type, schema_to_buffer};
 
+fn bytes_to_arrow_array(data: Uint8Array, dtype: String) -> napi::Result<Arc<dyn Array>> {
+    let buf = arrow_buffer::Buffer::from(data.to_vec());
+    let num_bytes = buf.len();
+    match dtype.as_str() {
+        "float16" => {
+            let scalar_buf = ScalarBuffer::<f16>::new(buf, 0, num_bytes / 2);
+            Ok(Arc::new(ArrowFloat16Array::new(scalar_buf, None)))
+        }
+        "float32" => {
+            let scalar_buf = ScalarBuffer::<f32>::new(buf, 0, num_bytes / 4);
+            Ok(Arc::new(ArrowFloat32Array::new(scalar_buf, None)))
+        }
+        "float64" => {
+            let scalar_buf = ScalarBuffer::<f64>::new(buf, 0, num_bytes / 8);
+            Ok(Arc::new(ArrowFloat64Array::new(scalar_buf, None)))
+        }
+        "uint8" => {
+            let scalar_buf = ScalarBuffer::<u8>::new(buf, 0, num_bytes);
+            Ok(Arc::new(ArrowUInt8Array::new(scalar_buf, None)))
+        }
+        _ => Err(napi::Error::from_reason(format!(
+            "Unsupported vector dtype: {}. Expected one of: float16, float32, float64, uint8",
+            dtype
+        ))),
+    }
+}
+
 #[napi]
 pub struct Query {
     inner: LanceDbQuery,
@@ -78,6 +111,13 @@ impl Query {
         Ok(VectorQuery { inner })
     }
 
+    #[napi]
+    pub fn nearest_to_raw(&mut self, data: Uint8Array, dtype: String) -> Result<VectorQuery> {
+        let array = bytes_to_arrow_array(data, dtype)?;
+        let inner = self.inner.clone().nearest_to(array).default_error()?;
+        Ok(VectorQuery { inner })
+    }
+
     #[napi]
     pub fn fast_search(&mut self) {
         self.inner = self.inner.clone().fast_search();
@@ -163,6 +203,13 @@ impl VectorQuery {
         Ok(())
     }
 
+    #[napi]
+    pub fn add_query_vector_raw(&mut self, data: Uint8Array, dtype: String) -> Result<()> {
+        let array = bytes_to_arrow_array(data, dtype)?;
+        self.inner = self.inner.clone().add_query_vector(array).default_error()?;
+        Ok(())
+    }
+
     #[napi]
     pub fn distance_type(&mut self, distance_type: String) -> napi::Result<()> {
         let distance_type = parse_distance_type(distance_type)?;

nodejs/src/remote.rs

@@ -92,6 +92,13 @@ pub struct ClientConfig {
     pub extra_headers: Option<HashMap<String, String>>,
     pub id_delimiter: Option<String>,
     pub tls_config: Option<TlsConfig>,
+    /// User identifier for tracking purposes.
+    ///
+    /// This is sent as the `x-lancedb-user-id` header in requests to LanceDB Cloud/Enterprise.
+    /// It can be set directly, or via the `LANCEDB_USER_ID` environment variable.
+    /// Alternatively, set `LANCEDB_USER_ID_ENV_KEY` to specify another environment
+    /// variable that contains the user ID value.
+    pub user_id: Option<String>,
 }
 
 impl From<TimeoutConfig> for lancedb::remote::TimeoutConfig {
@@ -145,6 +152,7 @@ impl From<ClientConfig> for lancedb::remote::ClientConfig {
             id_delimiter: config.id_delimiter,
             tls_config: config.tls_config.map(Into::into),
             header_provider: None, // the header provider is set separately later
+            user_id: config.user_id,
         }
     }
 }

nodejs/src/rerankers.rs

@@ -18,6 +18,7 @@ type RerankHybridFn = ThreadsafeFunction<
     RerankHybridCallbackArgs,
     Status,
     false,
+    true,
 >;
 
 /// Reranker implementation that "wraps" a NodeJS Reranker implementation.
@@ -32,7 +33,10 @@ impl Reranker {
     pub fn new(
         rerank_hybrid: Function<RerankHybridCallbackArgs, Promise<Buffer>>,
     ) -> napi::Result<Self> {
-        let rerank_hybrid = rerank_hybrid.build_threadsafe_function().build()?;
+        let rerank_hybrid = rerank_hybrid
+            .build_threadsafe_function()
+            .weak::<true>()
+            .build()?;
         Ok(Self { rerank_hybrid })
     }
 }

python/.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.30.0-beta.5"
+current_version = "0.31.0-beta.10"
 parse = """(?x)
     (?P<major>0|[1-9]\\d*)\\.
     (?P<minor>0|[1-9]\\d*)\\.

python/.gitignore

@@ -1,3 +1,5 @@
 # Test data created by some example tests
 data/
 _lancedb.pyd
+# macOS debug symbols bundle generated during build
+*.dSYM/

python/Cargo.toml

@@ -1,6 +1,7 @@
 [package]
 name = "lancedb-python"
-version = "0.30.0-beta.5"
+version = "0.31.0-beta.10"
+publish = false
 edition.workspace = true
 description = "Python bindings for LanceDB"
 license.workspace = true
@@ -23,6 +24,7 @@ lance-namespace.workspace = true
 lance-namespace-impls.workspace = true
 lance-io.workspace = true
 env_logger.workspace = true
+log.workspace = true
 pyo3 = { version = "0.26", features = ["extension-module", "abi3-py39"] }
 pyo3-async-runtimes = { version = "0.26", features = [
     "attributes",

python/PYTHON_THIRD_PARTY_LICENSES.md

@@ -183,7 +183,6 @@
 | stack-data                     | 0.6.3           | MIT License                                                                                      | http://github.com/alexmojaki/stack_data                                                           |
 | sympy                          | 1.14.0          | BSD License                                                                                      | https://sympy.org                                                                                 |
 | tabulate                       | 0.9.0           | MIT License                                                                                      | https://github.com/astanin/python-tabulate                                                        |
-| tantivy                        | 0.25.1          | UNKNOWN                                                                                          | UNKNOWN                                                                                           |
 | threadpoolctl                  | 3.6.0           | BSD License                                                                                      | https://github.com/joblib/threadpoolctl                                                           |
 | timm                           | 1.0.24          | Apache Software License                                                                          | https://github.com/huggingface/pytorch-image-models                                               |
 | tinycss2                       | 1.4.0           | BSD License                                                                                      | https://www.courtbouillon.org/tinycss2                                                            |

python/pyproject.toml

@@ -45,7 +45,7 @@ repository = "https://github.com/lancedb/lancedb"
 
 [project.optional-dependencies]
 pylance = [
-    "pylance>=4.0.0b7",
+    "pylance>=5.0.0b5",
 ]
 tests = [
     "aiohttp>=3.9.0",
@@ -57,9 +57,8 @@ tests = [
     "duckdb>=0.9.0",
     "pytz>=2023.3",
     "polars>=0.19, <=1.3.0",
-    "tantivy>=0.20.0",
     "pyarrow-stubs>=16.0",
-    "pylance>=4.0.0b7",
+    "pylance>=5.0.0b5",
     "requests>=2.31.0",
     "datafusion>=52,<53",
 ]
@@ -83,7 +82,7 @@ embeddings = [
     "colpali-engine>=0.3.10",
     "huggingface_hub>=0.19.0",
     "InstructorEmbedding>=1.0.1",
-    "google.generativeai>=0.3.0",
+    "google-genai>=1.0.0",
     "boto3>=1.28.57",
     "awscli>=1.44.38",
     "botocore>=1.31.57",

python/python/lancedb/__init__.py

@@ -6,7 +6,7 @@ import importlib.metadata
 import os
 from concurrent.futures import ThreadPoolExecutor
 from datetime import timedelta
-from typing import Dict, Optional, Union, Any
+from typing import Dict, Optional, Union, Any, List
 import warnings
 
 __version__ = importlib.metadata.version("lancedb")
@@ -15,9 +15,9 @@ from ._lancedb import connect as lancedb_connect
 from .common import URI, sanitize_uri
 from urllib.parse import urlparse
 from .db import AsyncConnection, DBConnection, LanceDBConnection
-from .io import StorageOptionsProvider
 from .remote import ClientConfig
 from .remote.db import RemoteDBConnection
+from .expr import Expr, col, lit, func
 from .schema import vector
 from .table import AsyncTable, Table
 from ._lancedb import Session
@@ -63,7 +63,7 @@ def _check_s3_bucket_with_dots(
 
 
 def connect(
-    uri: URI,
+    uri: Optional[URI] = None,
     *,
     api_key: Optional[str] = None,
     region: str = "us-east-1",
@@ -73,14 +73,19 @@ def connect(
     client_config: Union[ClientConfig, Dict[str, Any], None] = None,
     storage_options: Optional[Dict[str, str]] = None,
     session: Optional[Session] = None,
+    manifest_enabled: bool = False,
+    namespace_client_impl: Optional[str] = None,
+    namespace_client_properties: Optional[Dict[str, str]] = None,
+    namespace_client_pushdown_operations: Optional[List[str]] = None,
     **kwargs: Any,
 ) -> DBConnection:
     """Connect to a LanceDB database.
 
     Parameters
     ----------
-    uri: str or Path
-        The uri of the database.
+    uri: str or Path, optional
+        The uri of the database. When ``namespace_client_impl`` is provided you may
+        omit ``uri`` and connect through a namespace client instead.
     api_key: str, optional
         If presented, connect to LanceDB cloud.
         Otherwise, connect to a database on file system or cloud storage.
@@ -106,13 +111,29 @@ def connect(
         default configuration is used.
     storage_options: dict, optional
         Additional options for the storage backend. See available options at
-        <https://lancedb.com/docs/storage/>
+        <https://docs.lancedb.com/storage/>
+    manifest_enabled : bool, default False
+        When true for local/native connections, use directory namespace
+        manifests as the source of truth for table metadata. Existing
+        directory-listed root tables are migrated into the manifest on access.
     session: Session, optional
         (For LanceDB OSS only)
         A session to use for this connection. Sessions allow you to configure
         cache sizes for index and metadata caches, which can significantly
         impact memory use and performance. They can also be re-used across
         multiple connections to share the same cache state.
+    namespace_client_impl : str, optional
+        When provided along with ``namespace_client_properties``, ``connect``
+        returns a namespace-backed connection by delegating to
+        :func:`connect_namespace`. The value identifies which namespace
+        implementation to load (e.g., ``"dir"`` or ``"rest"``).
+    namespace_client_properties : dict, optional
+        Configuration to pass to the namespace client implementation. Required
+        when ``namespace_client_impl`` is set.
+    namespace_client_pushdown_operations : list[str], optional
+        Only used when ``namespace_client_properties`` is provided. Forwards to
+        :func:`connect_namespace` to control which operations are executed on the
+        namespace service (e.g., ``["QueryTable", "CreateTable"]``).
 
     Examples
     --------
@@ -132,11 +153,48 @@ def connect(
     >>> db = lancedb.connect("db://my_database", api_key="ldb_...",
     ...                      client_config={"retry_config": {"retries": 5}})
 
+    Connect to a namespace-backed database:
+
+    >>> db = lancedb.connect(namespace_client_impl="dir",
+    ...                      namespace_client_properties={"root": "/tmp/ns"})
+
     Returns
     -------
     conn : DBConnection
         A connection to a LanceDB database.
     """
+    if namespace_client_impl is not None:
+        if namespace_client_properties is None:
+            raise ValueError(
+                "namespace_client_properties must be provided when "
+                "namespace_client_impl is set"
+            )
+        if kwargs:
+            raise ValueError(f"Unknown keyword arguments: {kwargs}")
+        return connect_namespace(
+            namespace_client_impl,
+            namespace_client_properties,
+            read_consistency_interval=read_consistency_interval,
+            storage_options=storage_options,
+            session=session,
+            namespace_client_pushdown_operations=namespace_client_pushdown_operations,
+        )
+
+    if namespace_client_properties is not None and not manifest_enabled:
+        raise ValueError(
+            "namespace_client_impl must be provided when using "
+            "namespace_client_properties unless manifest_enabled=True"
+        )
+
+    if namespace_client_pushdown_operations is not None:
+        raise ValueError(
+            "namespace_client_pushdown_operations is only valid when "
+            "connecting through a namespace"
+        )
+    if uri is None:
+        raise ValueError(
+            "uri is required when not connecting through a namespace client"
+        )
     if isinstance(uri, str) and uri.startswith("db://"):
         if api_key is None:
             api_key = os.environ.get("LANCEDB_API_KEY")
@@ -165,9 +223,92 @@ def connect(
         read_consistency_interval=read_consistency_interval,
         storage_options=storage_options,
         session=session,
+        manifest_enabled=manifest_enabled,
+        namespace_client_properties=namespace_client_properties,
     )
 
 
+WORKER_PROPERTY_PREFIX = "_lancedb_worker_"
+
+
+def _apply_worker_overrides(props: dict[str, str]) -> dict[str, str]:
+    """Apply worker property overrides.
+
+    Any key starting with ``_lancedb_worker_`` is extracted, the prefix
+    is stripped, and the resulting key-value pair is put back into the
+    map (overriding the existing value if present).  The original
+    prefixed key is removed.
+    """
+    worker_keys = [k for k in props if k.startswith(WORKER_PROPERTY_PREFIX)]
+    if not worker_keys:
+        return props
+    result = dict(props)
+    for key in worker_keys:
+        value = result.pop(key)
+        real_key = key[len(WORKER_PROPERTY_PREFIX) :]
+        result[real_key] = value
+    return result
+
+
+def deserialize_conn(
+    data: str,
+    *,
+    for_worker: bool = False,
+) -> DBConnection:
+    """Reconstruct a DBConnection from a serialized string.
+
+    The string must have been produced by
+    :meth:`DBConnection.serialize`.
+
+    Parameters
+    ----------
+    data : str
+        String produced by ``serialize()``.
+    for_worker : bool, default False
+        When ``True``, any namespace client property whose key starts
+        with ``_lancedb_worker_`` has that prefix stripped and the
+        value overrides the corresponding property.  For example,
+        ``_lancedb_worker_uri`` replaces ``uri``.
+
+    Returns
+    -------
+    DBConnection
+        A new connection matching the serialized state.
+    """
+    import json
+
+    parsed = json.loads(data)
+    connection_type = parsed.get("connection_type")
+
+    rci_secs = parsed.get("read_consistency_interval_seconds")
+    rci = timedelta(seconds=rci_secs) if rci_secs is not None else None
+    storage_options = parsed.get("storage_options")
+
+    if connection_type == "namespace":
+        props = dict(parsed.get("namespace_client_properties") or {})
+        if for_worker:
+            props = _apply_worker_overrides(props)
+        return connect_namespace(
+            namespace_client_impl=parsed["namespace_client_impl"],
+            namespace_client_properties=props,
+            read_consistency_interval=rci,
+            storage_options=storage_options,
+            namespace_client_pushdown_operations=parsed.get(
+                "namespace_client_pushdown_operations"
+            ),
+        )
+    elif connection_type == "local":
+        return LanceDBConnection(
+            parsed["uri"],
+            read_consistency_interval=rci,
+            storage_options=storage_options,
+            manifest_enabled=parsed.get("manifest_enabled", False),
+            namespace_client_properties=parsed.get("namespace_client_properties"),
+        )
+    else:
+        raise ValueError(f"Unknown connection_type: {connection_type}")
+
+
 async def connect_async(
     uri: URI,
     *,
@@ -178,6 +319,8 @@ async def connect_async(
     client_config: Optional[Union[ClientConfig, Dict[str, Any]]] = None,
     storage_options: Optional[Dict[str, str]] = None,
     session: Optional[Session] = None,
+    manifest_enabled: bool = False,
+    namespace_client_properties: Optional[Dict[str, str]] = None,
 ) -> AsyncConnection:
     """Connect to a LanceDB database.
 
@@ -210,13 +353,20 @@ async def connect_async(
         default configuration is used.
     storage_options: dict, optional
         Additional options for the storage backend. See available options at
-        <https://lancedb.com/docs/storage/>
+        <https://docs.lancedb.com/storage/>
     session: Session, optional
         (For LanceDB OSS only)
         A session to use for this connection. Sessions allow you to configure
         cache sizes for index and metadata caches, which can significantly
         impact memory use and performance. They can also be re-used across
         multiple connections to share the same cache state.
+    manifest_enabled : bool, default False
+        When true for local/native connections, use directory namespace
+        manifests as the source of truth for table metadata. Existing
+        directory-listed root tables are migrated into the manifest on access.
+    namespace_client_properties : dict, optional
+        Additional directory namespace client properties to use with
+        ``manifest_enabled=True``.
 
     Examples
     --------
@@ -259,6 +409,8 @@ async def connect_async(
             client_config,
             storage_options,
             session,
+            manifest_enabled,
+            namespace_client_properties,
         )
     )
 
@@ -271,6 +423,10 @@ __all__ = [
     "AsyncConnection",
     "AsyncLanceNamespaceDBConnection",
     "AsyncTable",
+    "col",
+    "Expr",
+    "func",
+    "lit",
     "URI",
     "sanitize_uri",
     "vector",
@@ -279,7 +435,6 @@ __all__ = [
     "LanceNamespaceDBConnection",
     "RemoteDBConnection",
     "Session",
-    "StorageOptionsProvider",
     "Table",
     "__version__",
 ]

python/python/lancedb/_lancedb.pyi

@@ -14,7 +14,6 @@ from .index import (
     HnswSq,
     FTS,
 )
-from .io import StorageOptionsProvider
 from lance_namespace import (
     ListNamespacesResponse,
     CreateNamespaceResponse,
@@ -27,6 +26,32 @@ from .remote import ClientConfig
 IvfHnswPq: type[HnswPq] = HnswPq
 IvfHnswSq: type[HnswSq] = HnswSq
 
+class PyExpr:
+    """A type-safe DataFusion expression node (Rust-side handle)."""
+
+    def eq(self, other: "PyExpr") -> "PyExpr": ...
+    def ne(self, other: "PyExpr") -> "PyExpr": ...
+    def lt(self, other: "PyExpr") -> "PyExpr": ...
+    def lte(self, other: "PyExpr") -> "PyExpr": ...
+    def gt(self, other: "PyExpr") -> "PyExpr": ...
+    def gte(self, other: "PyExpr") -> "PyExpr": ...
+    def and_(self, other: "PyExpr") -> "PyExpr": ...
+    def or_(self, other: "PyExpr") -> "PyExpr": ...
+    def not_(self) -> "PyExpr": ...
+    def add(self, other: "PyExpr") -> "PyExpr": ...
+    def sub(self, other: "PyExpr") -> "PyExpr": ...
+    def mul(self, other: "PyExpr") -> "PyExpr": ...
+    def div(self, other: "PyExpr") -> "PyExpr": ...
+    def lower(self) -> "PyExpr": ...
+    def upper(self) -> "PyExpr": ...
+    def contains(self, substr: "PyExpr") -> "PyExpr": ...
+    def cast(self, data_type: pa.DataType) -> "PyExpr": ...
+    def to_sql(self) -> str: ...
+
+def expr_col(name: str) -> PyExpr: ...
+def expr_lit(value: Union[bool, int, float, str]) -> PyExpr: ...
+def expr_func(name: str, args: List[PyExpr]) -> PyExpr: ...
+
 class Session:
     def __init__(
         self,
@@ -46,35 +71,35 @@ class Connection(object):
     async def close(self): ...
     async def list_namespaces(
         self,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         page_token: Optional[str] = None,
         limit: Optional[int] = None,
     ) -> ListNamespacesResponse: ...
     async def create_namespace(
         self,
-        namespace: List[str],
+        namespace_path: List[str],
         mode: Optional[str] = None,
         properties: Optional[Dict[str, str]] = None,
     ) -> CreateNamespaceResponse: ...
     async def drop_namespace(
         self,
-        namespace: List[str],
+        namespace_path: List[str],
         mode: Optional[str] = None,
         behavior: Optional[str] = None,
     ) -> DropNamespaceResponse: ...
     async def describe_namespace(
         self,
-        namespace: List[str],
+        namespace_path: List[str],
     ) -> DescribeNamespaceResponse: ...
     async def list_tables(
         self,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         page_token: Optional[str] = None,
         limit: Optional[int] = None,
     ) -> ListTablesResponse: ...
     async def table_names(
         self,
-        namespace: Optional[List[str]],
+        namespace_path: Optional[List[str]],
         start_after: Optional[str],
         limit: Optional[int],
     ) -> list[str]: ...  # Deprecated: Use list_tables instead
@@ -83,9 +108,8 @@ class Connection(object):
         name: str,
         mode: str,
         data: pa.RecordBatchReader,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         storage_options: Optional[Dict[str, str]] = None,
-        storage_options_provider: Optional[StorageOptionsProvider] = None,
         location: Optional[str] = None,
     ) -> Table: ...
     async def create_empty_table(
@@ -93,17 +117,15 @@ class Connection(object):
         name: str,
         mode: str,
         schema: pa.Schema,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         storage_options: Optional[Dict[str, str]] = None,
-        storage_options_provider: Optional[StorageOptionsProvider] = None,
         location: Optional[str] = None,
     ) -> Table: ...
     async def open_table(
         self,
         name: str,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         storage_options: Optional[Dict[str, str]] = None,
-        storage_options_provider: Optional[StorageOptionsProvider] = None,
         index_cache_size: Optional[int] = None,
         location: Optional[str] = None,
     ) -> Table: ...
@@ -111,7 +133,7 @@ class Connection(object):
         self,
         target_table_name: str,
         source_uri: str,
-        target_namespace: Optional[List[str]] = None,
+        target_namespace_path: Optional[List[str]] = None,
         source_version: Optional[int] = None,
         source_tag: Optional[str] = None,
         is_shallow: bool = True,
@@ -120,13 +142,18 @@ class Connection(object):
         self,
         cur_name: str,
         new_name: str,
-        cur_namespace: Optional[List[str]] = None,
-        new_namespace: Optional[List[str]] = None,
+        cur_namespace_path: Optional[List[str]] = None,
+        new_namespace_path: Optional[List[str]] = None,
     ) -> None: ...
     async def drop_table(
-        self, name: str, namespace: Optional[List[str]] = None
+        self, name: str, namespace_path: Optional[List[str]] = None
     ) -> None: ...
-    async def drop_all_tables(self, namespace: Optional[List[str]] = None) -> None: ...
+    async def drop_all_tables(
+        self, namespace_path: Optional[List[str]] = None
+    ) -> None: ...
+    async def namespace_client_config(
+        self,
+    ) -> Dict[str, Any]: ...
 
 class Table:
     def name(self) -> str: ...
@@ -135,7 +162,10 @@ class Table:
     def close(self) -> None: ...
     async def schema(self) -> pa.Schema: ...
     async def add(
-        self, data: pa.RecordBatchReader, mode: Literal["append", "overwrite"]
+        self,
+        data: pa.RecordBatchReader,
+        mode: Literal["append", "overwrite"],
+        progress: Optional[Any] = None,
     ) -> AddResult: ...
     async def update(
         self, updates: Dict[str, str], where: Optional[str]
@@ -212,6 +242,8 @@ async def connect(
     client_config: Optional[Union[ClientConfig, Dict[str, Any]]],
     storage_options: Optional[Dict[str, str]],
     session: Optional[Session],
+    manifest_enabled: bool = False,
+    namespace_client_properties: Optional[Dict[str, str]] = None,
 ) -> Connection: ...
 
 class RecordBatchStream:
@@ -222,7 +254,9 @@ class RecordBatchStream:
 
 class Query:
     def where(self, filter: str): ...
-    def select(self, columns: Tuple[str, str]): ...
+    def where_expr(self, expr: PyExpr): ...
+    def select(self, columns: List[Tuple[str, str]]): ...
+    def select_expr(self, columns: List[Tuple[str, PyExpr]]): ...
     def select_columns(self, columns: List[str]): ...
     def limit(self, limit: int): ...
     def offset(self, offset: int): ...
@@ -248,7 +282,9 @@ class TakeQuery:
 
 class FTSQuery:
     def where(self, filter: str): ...
-    def select(self, columns: List[str]): ...
+    def where_expr(self, expr: PyExpr): ...
+    def select(self, columns: List[Tuple[str, str]]): ...
+    def select_expr(self, columns: List[Tuple[str, PyExpr]]): ...
     def limit(self, limit: int): ...
     def offset(self, offset: int): ...
     def fast_search(self): ...
@@ -267,7 +303,9 @@ class VectorQuery:
     async def output_schema(self) -> pa.Schema: ...
     async def execute(self) -> RecordBatchStream: ...
     def where(self, filter: str): ...
-    def select(self, columns: List[str]): ...
+    def where_expr(self, expr: PyExpr): ...
+    def select(self, columns: List[Tuple[str, str]]): ...
+    def select_expr(self, columns: List[Tuple[str, PyExpr]]): ...
     def select_with_projection(self, columns: Tuple[str, str]): ...
     def limit(self, limit: int): ...
     def offset(self, offset: int): ...
@@ -284,7 +322,9 @@ class VectorQuery:
 
 class HybridQuery:
     def where(self, filter: str): ...
-    def select(self, columns: List[str]): ...
+    def where_expr(self, expr: PyExpr): ...
+    def select(self, columns: List[Tuple[str, str]]): ...
+    def select_expr(self, columns: List[Tuple[str, PyExpr]]): ...
     def limit(self, limit: int): ...
     def offset(self, offset: int): ...
     def fast_search(self): ...

python/python/lancedb/common.py

@@ -96,7 +96,7 @@ def data_to_reader(
             f"Unknown data type {type(data)}. "
             "Supported types: list of dicts, pandas DataFrame, polars DataFrame, "
             "pyarrow Table/RecordBatch, or Pydantic models. "
-            "See https://lancedb.com/docs/tables/ for examples."
+            "See https://docs.lancedb.com/tables/ for examples."
         )
 
 

python/python/lancedb/embeddings/gemini_text.py

@@ -19,10 +19,10 @@ from .utils import TEXT, api_key_not_found_help
 @register("gemini-text")
 class GeminiText(TextEmbeddingFunction):
     """
-    An embedding function that uses the Google's Gemini API. Requires GOOGLE_API_KEY to
+    An embedding function that uses Google's Gemini API. Requires GOOGLE_API_KEY to
     be set.
 
-    https://ai.google.dev/docs/embeddings_guide
+    https://ai.google.dev/gemini-api/docs/embeddings
 
     Supports various tasks types:
     | Task Type               | Description                                            |
@@ -46,9 +46,12 @@ class GeminiText(TextEmbeddingFunction):
 
     Parameters
     ----------
-    name: str, default "models/embedding-001"
-        The name of the model to use. See the Gemini documentation for a list of
-        available models.
+    name: str, default "gemini-embedding-001"
+        The name of the model to use. Supported models include:
+        - "gemini-embedding-001" (768 dimensions)
+
+        Note: The legacy "models/embedding-001" format is also supported but
+        "gemini-embedding-001" is recommended.
 
     query_task_type: str, default "retrieval_query"
         Sets the task type for the queries.
@@ -77,7 +80,7 @@ class GeminiText(TextEmbeddingFunction):
 
     """
 
-    name: str = "models/embedding-001"
+    name: str = "gemini-embedding-001"
     query_task_type: str = "retrieval_query"
     source_task_type: str = "retrieval_document"
 
@@ -114,23 +117,48 @@ class GeminiText(TextEmbeddingFunction):
         texts: list[str] or np.ndarray (of str)
             The texts to embed
         """
-        if (
-            kwargs.get("task_type") == "retrieval_document"
-        ):  # Provide a title to use existing API design
-            title = "Embedding of a document"
-            kwargs["title"] = title
+        from google.genai import types
 
-        return [
-            self.client.embed_content(model=self.name, content=text, **kwargs)[
-                "embedding"
-            ]
-            for text in texts
-        ]
+        task_type = kwargs.get("task_type")
+
+        # Build content objects for embed_content
+        contents = []
+        for text in texts:
+            if task_type == "retrieval_document":
+                # Provide a title for retrieval_document task
+                contents.append(
+                    {"parts": [{"text": "Embedding of a document"}, {"text": text}]}
+                )
+            else:
+                contents.append({"parts": [{"text": text}]})
+
+        # Build config
+        config_kwargs = {}
+        if task_type:
+            config_kwargs["task_type"] = task_type.upper()  # API expects uppercase
+
+        # Call embed_content for each content
+        embeddings = []
+        for content in contents:
+            config = (
+                types.EmbedContentConfig(**config_kwargs) if config_kwargs else None
+            )
+            response = self.client.models.embed_content(
+                model=self.name,
+                contents=content,
+                config=config,
+            )
+            embeddings.append(response.embeddings[0].values)
+
+        return embeddings
 
     @cached_property
     def client(self):
-        genai = attempt_import_or_raise("google.generativeai", "google.generativeai")
+        attempt_import_or_raise("google.genai", "google-genai")
 
         if not os.environ.get("GOOGLE_API_KEY"):
             api_key_not_found_help("google")
-        return genai
+
+        from google import genai as genai_module
+
+        return genai_module.Client(api_key=os.environ.get("GOOGLE_API_KEY"))

python/python/lancedb/embeddings/utils.py

@@ -10,6 +10,7 @@ import sys
 import threading
 import time
 import urllib.error
+import urllib.request
 import weakref
 import logging
 from functools import wraps

python/python/lancedb/expr.py

@@ -0,0 +1,298 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+"""Type-safe expression builder for filters and projections.
+
+Instead of writing raw SQL strings you can build expressions with Python
+operators::
+
+    from lancedb.expr import col, lit
+
+    # filter: age > 18 AND status = 'active'
+    filt = (col("age") > lit(18)) & (col("status") == lit("active"))
+
+    # projection: compute a derived column
+    proj = {"score": col("raw_score") * lit(1.5)}
+
+    table.search().where(filt).select(proj).to_list()
+"""
+
+from __future__ import annotations
+
+from typing import Union
+
+import pyarrow as pa
+
+from lancedb._lancedb import PyExpr, expr_col, expr_lit, expr_func
+
+__all__ = ["Expr", "col", "lit", "func"]
+
+_STR_TO_PA_TYPE: dict = {
+    "bool": pa.bool_(),
+    "boolean": pa.bool_(),
+    "int8": pa.int8(),
+    "int16": pa.int16(),
+    "int32": pa.int32(),
+    "int64": pa.int64(),
+    "uint8": pa.uint8(),
+    "uint16": pa.uint16(),
+    "uint32": pa.uint32(),
+    "uint64": pa.uint64(),
+    "float16": pa.float16(),
+    "float32": pa.float32(),
+    "float": pa.float32(),
+    "float64": pa.float64(),
+    "double": pa.float64(),
+    "string": pa.string(),
+    "utf8": pa.string(),
+    "str": pa.string(),
+    "large_string": pa.large_utf8(),
+    "large_utf8": pa.large_utf8(),
+    "date32": pa.date32(),
+    "date": pa.date32(),
+    "date64": pa.date64(),
+}
+
+
+def _coerce(value: "ExprLike") -> "Expr":
+    """Return *value* as an :class:`Expr`, wrapping plain Python values via
+    :func:`lit` if needed."""
+    if isinstance(value, Expr):
+        return value
+    return lit(value)
+
+
+# Type alias used in annotations.
+ExprLike = Union["Expr", bool, int, float, str]
+
+
+class Expr:
+    """A type-safe expression node.
+
+    Construct instances with :func:`col` and :func:`lit`, then combine them
+    using Python operators or the named methods below.
+
+    Examples
+    --------
+    >>> from lancedb.expr import col, lit
+    >>> filt = (col("age") > lit(18)) & (col("name").lower() == lit("alice"))
+    >>> proj = {"double": col("x") * lit(2)}
+    """
+
+    # Make Expr unhashable so that == returns an Expr rather than being used
+    # for dict keys / set membership.
+    __hash__ = None  # type: ignore[assignment]
+
+    def __init__(self, inner: PyExpr) -> None:
+        self._inner = inner
+
+    # ── comparisons ──────────────────────────────────────────────────────────
+
+    def __eq__(self, other: ExprLike) -> "Expr":  # type: ignore[override]
+        """Equal to (``col("x") == 1``)."""
+        return Expr(self._inner.eq(_coerce(other)._inner))
+
+    def __ne__(self, other: ExprLike) -> "Expr":  # type: ignore[override]
+        """Not equal to (``col("x") != 1``)."""
+        return Expr(self._inner.ne(_coerce(other)._inner))
+
+    def __lt__(self, other: ExprLike) -> "Expr":
+        """Less than (``col("x") < 1``)."""
+        return Expr(self._inner.lt(_coerce(other)._inner))
+
+    def __le__(self, other: ExprLike) -> "Expr":
+        """Less than or equal to (``col("x") <= 1``)."""
+        return Expr(self._inner.lte(_coerce(other)._inner))
+
+    def __gt__(self, other: ExprLike) -> "Expr":
+        """Greater than (``col("x") > 1``)."""
+        return Expr(self._inner.gt(_coerce(other)._inner))
+
+    def __ge__(self, other: ExprLike) -> "Expr":
+        """Greater than or equal to (``col("x") >= 1``)."""
+        return Expr(self._inner.gte(_coerce(other)._inner))
+
+    # ── logical ──────────────────────────────────────────────────────────────
+
+    def __and__(self, other: "Expr") -> "Expr":
+        """Logical AND (``expr_a & expr_b``)."""
+        return Expr(self._inner.and_(_coerce(other)._inner))
+
+    def __or__(self, other: "Expr") -> "Expr":
+        """Logical OR (``expr_a | expr_b``)."""
+        return Expr(self._inner.or_(_coerce(other)._inner))
+
+    def __invert__(self) -> "Expr":
+        """Logical NOT (``~expr``)."""
+        return Expr(self._inner.not_())
+
+    # ── arithmetic ───────────────────────────────────────────────────────────
+
+    def __add__(self, other: ExprLike) -> "Expr":
+        """Add (``col("x") + 1``)."""
+        return Expr(self._inner.add(_coerce(other)._inner))
+
+    def __radd__(self, other: ExprLike) -> "Expr":
+        """Right-hand add (``1 + col("x")``)."""
+        return Expr(_coerce(other)._inner.add(self._inner))
+
+    def __sub__(self, other: ExprLike) -> "Expr":
+        """Subtract (``col("x") - 1``)."""
+        return Expr(self._inner.sub(_coerce(other)._inner))
+
+    def __rsub__(self, other: ExprLike) -> "Expr":
+        """Right-hand subtract (``1 - col("x")``)."""
+        return Expr(_coerce(other)._inner.sub(self._inner))
+
+    def __mul__(self, other: ExprLike) -> "Expr":
+        """Multiply (``col("x") * 2``)."""
+        return Expr(self._inner.mul(_coerce(other)._inner))
+
+    def __rmul__(self, other: ExprLike) -> "Expr":
+        """Right-hand multiply (``2 * col("x")``)."""
+        return Expr(_coerce(other)._inner.mul(self._inner))
+
+    def __truediv__(self, other: ExprLike) -> "Expr":
+        """Divide (``col("x") / 2``)."""
+        return Expr(self._inner.div(_coerce(other)._inner))
+
+    def __rtruediv__(self, other: ExprLike) -> "Expr":
+        """Right-hand divide (``1 / col("x")``)."""
+        return Expr(_coerce(other)._inner.div(self._inner))
+
+    # ── string methods ───────────────────────────────────────────────────────
+
+    def lower(self) -> "Expr":
+        """Convert string column values to lowercase."""
+        return Expr(self._inner.lower())
+
+    def upper(self) -> "Expr":
+        """Convert string column values to uppercase."""
+        return Expr(self._inner.upper())
+
+    def contains(self, substr: "ExprLike") -> "Expr":
+        """Return True where the string contains *substr*."""
+        return Expr(self._inner.contains(_coerce(substr)._inner))
+
+    # ── type cast ────────────────────────────────────────────────────────────
+
+    def cast(self, data_type: Union[str, "pa.DataType"]) -> "Expr":
+        """Cast values to *data_type*.
+
+        Parameters
+        ----------
+        data_type:
+            A PyArrow ``DataType`` (e.g. ``pa.int32()``) or one of the type
+            name strings: ``"bool"``, ``"int8"``, ``"int16"``, ``"int32"``,
+            ``"int64"``, ``"uint8"``–``"uint64"``, ``"float32"``,
+            ``"float64"``, ``"string"``, ``"date32"``, ``"date64"``.
+        """
+        if isinstance(data_type, str):
+            try:
+                data_type = _STR_TO_PA_TYPE[data_type]
+            except KeyError:
+                raise ValueError(
+                    f"unsupported data type: '{data_type}'. Supported: "
+                    f"{', '.join(_STR_TO_PA_TYPE)}"
+                )
+        return Expr(self._inner.cast(data_type))
+
+    # ── named comparison helpers (alternative to operators) ──────────────────
+
+    def eq(self, other: ExprLike) -> "Expr":
+        """Equal to."""
+        return self.__eq__(other)
+
+    def ne(self, other: ExprLike) -> "Expr":
+        """Not equal to."""
+        return self.__ne__(other)
+
+    def lt(self, other: ExprLike) -> "Expr":
+        """Less than."""
+        return self.__lt__(other)
+
+    def lte(self, other: ExprLike) -> "Expr":
+        """Less than or equal to."""
+        return self.__le__(other)
+
+    def gt(self, other: ExprLike) -> "Expr":
+        """Greater than."""
+        return self.__gt__(other)
+
+    def gte(self, other: ExprLike) -> "Expr":
+        """Greater than or equal to."""
+        return self.__ge__(other)
+
+    def and_(self, other: "Expr") -> "Expr":
+        """Logical AND."""
+        return self.__and__(other)
+
+    def or_(self, other: "Expr") -> "Expr":
+        """Logical OR."""
+        return self.__or__(other)
+
+    # ── utilities ────────────────────────────────────────────────────────────
+
+    def to_sql(self) -> str:
+        """Render the expression as a SQL string (useful for debugging)."""
+        return self._inner.to_sql()
+
+    def __repr__(self) -> str:
+        return f"Expr({self._inner.to_sql()})"
+
+
+# ── free functions ────────────────────────────────────────────────────────────
+
+
+def col(name: str) -> Expr:
+    """Reference a table column by name.
+
+    Parameters
+    ----------
+    name:
+        The column name.
+
+    Examples
+    --------
+    >>> from lancedb.expr import col, lit
+    >>> col("age") > lit(18)
+    Expr((age > 18))
+    """
+    return Expr(expr_col(name))
+
+
+def lit(value: Union[bool, int, float, str]) -> Expr:
+    """Create a literal (constant) value expression.
+
+    Parameters
+    ----------
+    value:
+        A Python ``bool``, ``int``, ``float``, or ``str``.
+
+    Examples
+    --------
+    >>> from lancedb.expr import col, lit
+    >>> col("price") * lit(1.1)
+    Expr((price * 1.1))
+    """
+    return Expr(expr_lit(value))
+
+
+def func(name: str, *args: ExprLike) -> Expr:
+    """Call an arbitrary SQL function by name.
+
+    Parameters
+    ----------
+    name:
+        The SQL function name (e.g. ``"lower"``, ``"upper"``).
+    *args:
+        The function arguments as :class:`Expr` or plain Python literals.
+
+    Examples
+    --------
+    >>> from lancedb.expr import col, func
+    >>> func("lower", col("name"))
+    Expr(lower(name))
+    """
+    inner_args = [_coerce(a)._inner for a in args]
+    return Expr(expr_func(name, inner_args))

python/python/lancedb/fts.py

@@ -1,201 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright The LanceDB Authors
-
-"""Full text search index using tantivy-py"""
-
-import os
-from typing import List, Tuple, Optional
-
-import pyarrow as pa
-
-try:
-    import tantivy
-except ImportError:
-    raise ImportError(
-        "Please install tantivy-py `pip install tantivy` to use the full text search feature."  # noqa: E501
-    )
-
-from .table import LanceTable
-
-
-def create_index(
-    index_path: str,
-    text_fields: List[str],
-    ordering_fields: Optional[List[str]] = None,
-    tokenizer_name: str = "default",
-) -> tantivy.Index:
-    """
-    Create a new Index (not populated)
-
-    Parameters
-    ----------
-    index_path : str
-        Path to the index directory
-    text_fields : List[str]
-        List of text fields to index
-    ordering_fields: List[str]
-        List of unsigned type fields to order by at search time
-    tokenizer_name : str, default "default"
-        The tokenizer to use
-
-    Returns
-    -------
-    index : tantivy.Index
-        The index object (not yet populated)
-    """
-    if ordering_fields is None:
-        ordering_fields = []
-    # Declaring our schema.
-    schema_builder = tantivy.SchemaBuilder()
-    # special field that we'll populate with row_id
-    schema_builder.add_integer_field("doc_id", stored=True)
-    # data fields
-    for name in text_fields:
-        schema_builder.add_text_field(name, stored=True, tokenizer_name=tokenizer_name)
-    if ordering_fields:
-        for name in ordering_fields:
-            schema_builder.add_unsigned_field(name, fast=True)
-    schema = schema_builder.build()
-    os.makedirs(index_path, exist_ok=True)
-    index = tantivy.Index(schema, path=index_path)
-    return index
-
-
-def populate_index(
-    index: tantivy.Index,
-    table: LanceTable,
-    fields: List[str],
-    writer_heap_size: Optional[int] = None,
-    ordering_fields: Optional[List[str]] = None,
-) -> int:
-    """
-    Populate an index with data from a LanceTable
-
-    Parameters
-    ----------
-    index : tantivy.Index
-        The index object
-    table : LanceTable
-        The table to index
-    fields : List[str]
-        List of fields to index
-    writer_heap_size : int
-        The writer heap size in bytes, defaults to 1GB
-
-    Returns
-    -------
-    int
-        The number of rows indexed
-    """
-    if ordering_fields is None:
-        ordering_fields = []
-    writer_heap_size = writer_heap_size or 1024 * 1024 * 1024
-    # first check the fields exist and are string or large string type
-    nested = []
-
-    for name in fields:
-        try:
-            f = table.schema.field(name)  # raises KeyError if not found
-        except KeyError:
-            f = resolve_path(table.schema, name)
-            nested.append(name)
-
-        if not pa.types.is_string(f.type) and not pa.types.is_large_string(f.type):
-            raise TypeError(f"Field {name} is not a string type")
-
-    # create a tantivy writer
-    writer = index.writer(heap_size=writer_heap_size)
-    # write data into index
-    dataset = table.to_lance()
-    row_id = 0
-
-    max_nested_level = 0
-    if len(nested) > 0:
-        max_nested_level = max([len(name.split(".")) for name in nested])
-
-    for b in dataset.to_batches(columns=fields + ordering_fields):
-        if max_nested_level > 0:
-            b = pa.Table.from_batches([b])
-            for _ in range(max_nested_level - 1):
-                b = b.flatten()
-        for i in range(b.num_rows):
-            doc = tantivy.Document()
-            for name in fields:
-                value = b[name][i].as_py()
-                if value is not None:
-                    doc.add_text(name, value)
-            for name in ordering_fields:
-                value = b[name][i].as_py()
-                if value is not None:
-                    doc.add_unsigned(name, value)
-            if not doc.is_empty:
-                doc.add_integer("doc_id", row_id)
-                writer.add_document(doc)
-            row_id += 1
-    # commit changes
-    writer.commit()
-    return row_id
-
-
-def resolve_path(schema, field_name: str) -> pa.Field:
-    """
-    Resolve a nested field path to a list of field names
-
-    Parameters
-    ----------
-    field_name : str
-        The field name to resolve
-
-    Returns
-    -------
-    List[str]
-        The resolved path
-    """
-    path = field_name.split(".")
-    field = schema.field(path.pop(0))
-    for segment in path:
-        if pa.types.is_struct(field.type):
-            field = field.type.field(segment)
-        else:
-            raise KeyError(f"field {field_name} not found in schema {schema}")
-    return field
-
-
-def search_index(
-    index: tantivy.Index, query: str, limit: int = 10, ordering_field=None
-) -> Tuple[Tuple[int], Tuple[float]]:
-    """
-    Search an index for a query
-
-    Parameters
-    ----------
-    index : tantivy.Index
-        The index object
-    query : str
-        The query string
-    limit : int
-        The maximum number of results to return
-
-    Returns
-    -------
-    ids_and_score: list[tuple[int], tuple[float]]
-        A tuple of two tuples, the first containing the document ids
-        and the second containing the scores
-    """
-    searcher = index.searcher()
-    query = index.parse_query(query)
-    # get top results
-    if ordering_field:
-        results = searcher.search(query, limit, order_by_field=ordering_field)
-    else:
-        results = searcher.search(query, limit)
-    if results.count == 0:
-        return tuple(), tuple()
-    return tuple(
-        zip(
-            *[
-                (searcher.doc(doc_address)["doc_id"][0], score)
-                for score, doc_address in results.hits
-            ]
-        )
-    )

python/python/lancedb/integrations/torch.py

@@ -0,0 +1,230 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+"""
+PyTorch integration for LanceDB.
+
+Exposes ``LanceTorchDataset`` (map-style) and ``LanceIterableTorchDataset``
+(iterable-style) wrappers that adapt a LanceDB table or permutation to the
+PyTorch ``torch.utils.data`` API, while transparently handling the bits
+that make a hand-rolled subclass tricky:
+
+* The underlying Lance reader holds Rust state that is not picklable, but
+  ``DataLoader(num_workers > 0)`` needs to fork the dataset to its workers.
+  These classes strip the reader on pickle and re-open it in the worker on
+  first read.
+* Constructing a permutation from a table involves several steps
+  (``permutation_builder``/``Permutation.from_tables``/``select_columns``
+  /``with_format``/...). The wrapper takes those as constructor arguments
+  and applies them once the dataset is opened in the worker.
+
+Example
+-------
+>>> import lancedb, torch                                   # doctest: +SKIP
+>>> from lancedb.integrations.torch import LanceTorchDataset
+>>> db = lancedb.connect(uri)                               # doctest: +SKIP
+>>> tbl = db.open_table("images_224")                       # doctest: +SKIP
+>>> ds = LanceTorchDataset(                                 # doctest: +SKIP
+...     tbl, columns=["image_bytes", "label"], format="torch"
+... )
+>>> loader = torch.utils.data.DataLoader(                   # doctest: +SKIP
+...     ds, batch_size=64, num_workers=4, shuffle=True,
+... )
+"""
+
+from typing import Any, Callable, Dict, List, Optional, Union
+
+import torch.utils.data as _torch_data
+
+from ..permutation import Permutation
+from ..table import LanceTable
+
+
+def _capture_table_state(table: LanceTable) -> Dict[str, Any]:
+    """Pull just enough state out of a LanceTable so we can re-open the same
+    table in a forked worker process where the Rust handle isn't valid."""
+    conn = table._conn
+    connect_kwargs: Dict[str, Any] = {}
+    storage_options = getattr(conn, "storage_options", None)
+    if storage_options is not None:
+        connect_kwargs["storage_options"] = storage_options
+    return {
+        "uri": conn.uri,
+        "table_name": table.name,
+        "connect_kwargs": connect_kwargs,
+    }
+
+
+def _open_permutation(state: Dict[str, Any]) -> Permutation:
+    """Reconstruct a Permutation from a captured state dict."""
+    import lancedb
+
+    db = lancedb.connect(state["uri"], **state["connect_kwargs"])
+    base = db.open_table(state["table_name"])
+
+    perm_table_name = state.get("perm_table_name")
+    if perm_table_name is not None:
+        perm_tbl = db.open_table(perm_table_name)
+        perm = Permutation.from_tables(base, perm_tbl, state.get("split"))
+    else:
+        perm = Permutation.identity(base)
+
+    columns = state.get("columns")
+    fmt = state.get("format")
+    transform = state.get("transform")
+    batch_size = state.get("batch_size")
+
+    if columns is not None:
+        perm = perm.select_columns(columns)
+    if fmt is not None:
+        perm = perm.with_format(fmt)
+    if transform is not None:
+        perm = perm.with_transform(transform)
+    if batch_size is not None:
+        perm = perm.with_batch_size(batch_size)
+    return perm
+
+
+class LanceTorchDataset(_torch_data.Dataset):
+    """
+    A PyTorch map-style ``Dataset`` backed by a LanceDB table or permutation.
+
+    Pass the same ``LanceTable`` you already opened (and, optionally, a
+    permutation table / split / column selection / output format) and use
+    the result anywhere a ``torch.utils.data.Dataset`` is expected.
+
+    The wrapper:
+
+    * Stores the URI / table name / storage options needed to re-open the
+      table, not the Rust reader handle. Pickling keeps only the rebuild
+      recipe, so ``DataLoader(num_workers > 0)`` works out of the box.
+    * Implements both ``__getitem__`` and PyTorch's ``__getitems__`` dunder
+      so the underlying batched ``Permutation.fetch`` is used when the
+      DataLoader fetches a batch of indices.
+
+    Parameters
+    ----------
+    table : LanceTable, optional
+        The base table to read from. Either ``table`` or both ``uri`` and
+        ``table_name`` must be provided.
+    uri : str, optional
+        Database URI to reconnect to. Required if ``table`` is not given.
+    table_name : str, optional
+        Name of the base table within ``uri``.
+    connect_kwargs : dict, optional
+        Extra keyword arguments forwarded to ``lancedb.connect`` when
+        re-opening the database in a worker.
+    permutation_table : LanceTable, optional
+        A pre-built permutation table (see ``permutation_builder``) used to
+        define the row ordering. If omitted, the identity permutation is
+        used (rows in physical order).
+    split : str or int, optional
+        Split selector when ``permutation_table`` defines splits.
+    columns : list[str], optional
+        Subset of columns to read.
+    format : str, optional
+        Output format, forwarded to ``Permutation.with_format`` (e.g.
+        ``"torch"`` for HuggingFace-style ``dict[str, Tensor]`` batches).
+    transform : Callable, optional
+        Custom batch transform, forwarded to ``Permutation.with_transform``.
+        Must be picklable to work with ``num_workers > 0``.
+    batch_size : int, optional
+        Forwarded to ``Permutation.with_batch_size`` for direct iteration.
+        DataLoader controls its own batching, so this only matters if the
+        dataset is iterated directly.
+    """
+
+    def __init__(
+        self,
+        table: Optional[LanceTable] = None,
+        *,
+        uri: Optional[str] = None,
+        table_name: Optional[str] = None,
+        connect_kwargs: Optional[Dict[str, Any]] = None,
+        permutation_table: Optional[LanceTable] = None,
+        split: Optional[Union[str, int]] = None,
+        columns: Optional[List[str]] = None,
+        format: Optional[str] = None,
+        transform: Optional[Callable] = None,
+        batch_size: Optional[int] = None,
+    ):
+        if table is None and (uri is None or table_name is None):
+            raise ValueError(
+                "Provide either `table` or both `uri` and `table_name`."
+            )
+
+        if table is not None:
+            state = _capture_table_state(table)
+            if connect_kwargs is not None:
+                state["connect_kwargs"] = connect_kwargs
+        else:
+            state = {
+                "uri": uri,
+                "table_name": table_name,
+                "connect_kwargs": connect_kwargs or {},
+            }
+
+        state["perm_table_name"] = (
+            permutation_table.name if permutation_table is not None else None
+        )
+        state["split"] = split
+        state["columns"] = columns
+        state["format"] = format
+        state["transform"] = transform
+        state["batch_size"] = batch_size
+
+        self._state: Dict[str, Any] = state
+        self._perm: Optional[Permutation] = None
+
+    def __getstate__(self) -> Dict[str, Any]:
+        # Strip the Rust-backed reader so the dataset is picklable. Workers
+        # rebuild it on first read via _ensure_open().
+        d = self.__dict__.copy()
+        d["_perm"] = None
+        return d
+
+    def __setstate__(self, d: Dict[str, Any]) -> None:
+        self.__dict__.update(d)
+
+    def _ensure_open(self) -> None:
+        if self._perm is None:
+            self._perm = _open_permutation(self._state)
+
+    def __len__(self) -> int:
+        self._ensure_open()
+        return len(self._perm)
+
+    def __getitem__(self, index: int) -> Any:
+        self._ensure_open()
+        return self._perm[index]
+
+    def __getitems__(self, indices: List[int]) -> Any:
+        self._ensure_open()
+        return self._perm.fetch(indices)
+
+
+class LanceIterableTorchDataset(_torch_data.IterableDataset):
+    """
+    PyTorch iterable-style ``IterableDataset`` over a LanceDB permutation.
+
+    Yields batches in the order defined by the underlying ``Permutation``.
+    With ``num_workers > 1`` each worker iterates the permutation
+    independently — for sharded iteration use the map-style
+    ``LanceTorchDataset`` together with a sampler.
+
+    Constructor arguments mirror ``LanceTorchDataset``.
+    """
+
+    def __init__(self, *args, **kwargs):
+        self._inner = LanceTorchDataset(*args, **kwargs)
+
+    def __getstate__(self) -> Dict[str, Any]:
+        return {"_inner": self._inner.__getstate__()}
+
+    def __setstate__(self, d: Dict[str, Any]) -> None:
+        self._inner = LanceTorchDataset.__new__(LanceTorchDataset)
+        self._inner.__setstate__(d["_inner"])
+
+    def __iter__(self):
+        self._inner._ensure_open()
+        return iter(self._inner._perm)

python/python/lancedb/io.py

@@ -2,70 +2,3 @@
 # SPDX-FileCopyrightText: Copyright The LanceDB Authors
 
 """I/O utilities and interfaces for LanceDB."""
-
-from abc import ABC, abstractmethod
-from typing import Dict
-
-
-class StorageOptionsProvider(ABC):
-    """Abstract base class for providing storage options to LanceDB tables.
-
-    Storage options providers enable automatic credential refresh for cloud
-    storage backends (e.g., AWS S3, Azure Blob Storage, GCS). When credentials
-    have an expiration time, the provider's fetch_storage_options() method will
-    be called periodically to get fresh credentials before they expire.
-
-    Example
-    -------
-    >>> class MyProvider(StorageOptionsProvider):
-    ...     def fetch_storage_options(self) -> Dict[str, str]:
-    ...         # Fetch fresh credentials from your credential manager
-    ...         return {
-    ...             "aws_access_key_id": "...",
-    ...             "aws_secret_access_key": "...",
-    ...             "expires_at_millis": "1234567890000"  # Optional
-    ...         }
-    """
-
-    @abstractmethod
-    def fetch_storage_options(self) -> Dict[str, str]:
-        """Fetch fresh storage credentials.
-
-        This method is called by LanceDB when credentials need to be refreshed.
-        If the returned dictionary contains an "expires_at_millis" key with a
-        Unix timestamp in milliseconds, LanceDB will automatically refresh the
-        credentials before that time. If the key is not present, credentials
-        are assumed to not expire.
-
-        Returns
-        -------
-        Dict[str, str]
-            Dictionary containing cloud storage credentials and optionally an
-            expiration time:
-            - "expires_at_millis" (optional): Unix timestamp in milliseconds when
-              credentials expire
-            - Provider-specific credential keys (e.g., aws_access_key_id,
-              aws_secret_access_key, etc.)
-
-        Raises
-        ------
-        RuntimeError
-            If credentials cannot be fetched or are invalid
-        """
-        pass
-
-    def provider_id(self) -> str:
-        """Return a human-readable unique identifier for this provider instance.
-
-        This identifier is used for caching and equality comparison. Two providers
-        with the same ID will share the same cached object store connection.
-
-        The default implementation uses the class name and string representation.
-        Override this method if you need custom identification logic.
-
-        Returns
-        -------
-        str
-            A unique identifier for this provider instance
-        """
-        return f"{self.__class__.__name__} {{ repr: {str(self)!r} }}"

python/python/lancedb/permutation.py

@@ -284,9 +284,8 @@ class Permutations:
         self.permutation_table = permutation_table
 
         if permutation_table.schema.metadata is not None:
-            split_names = permutation_table.schema.metadata.get(
-                b"split_names", None
-            ).decode("utf-8")
+            raw = permutation_table.schema.metadata.get(b"split_names")
+            split_names = raw.decode("utf-8") if raw is not None else None
             if split_names is not None:
                 self.split_names = json.loads(split_names)
                 self.split_dict = {
@@ -460,9 +459,8 @@ class Permutation:
                         f"Cannot create a permutation on split `{split}`"
                         " because no split names are defined in the permutation table"
                     )
-                split_names = permutation_table.schema.metadata.get(
-                    b"split_names", None
-                ).decode("utf-8")
+                raw = permutation_table.schema.metadata.get(b"split_names")
+                split_names = raw.decode("utf-8") if raw is not None else None
                 if split_names is None:
                     raise ValueError(
                         f"Cannot create a permutation on split `{split}`"
@@ -781,6 +779,25 @@ class Permutation:
         batch = LOOP.run(do_getitems())
         return self.transform_fn(batch)
 
+    def fetch(self, indices: list[int]) -> Any:
+        """
+        Fetch rows from the permutation by offset.
+
+        This is the public batch-access API. It returns the rows for the given
+        offsets in the same shape as configured by
+        [with_format](#with_format) / [with_transform](#with_transform).
+
+        Examples
+        --------
+        >>> import lancedb
+        >>> db = lancedb.connect("memory:///")
+        >>> tbl = db.create_table("tbl", data=[{"x": x} for x in range(10)])
+        >>> perm = Permutation.identity(tbl)
+        >>> perm.fetch([0, 5, 9])
+        [{'x': 0}, {'x': 5}, {'x': 9}]
+        """
+        return self.__getitems__(indices)
+
     @deprecated(details="Use with_skip instead")
     def skip(self, skip: int) -> "Permutation":
         """

python/python/lancedb/pydantic.py

@@ -10,6 +10,7 @@ import sys
 import types
 from abc import ABC, abstractmethod
 from datetime import date, datetime
+from enum import Enum
 from typing import (
     TYPE_CHECKING,
     Any,
@@ -314,6 +315,19 @@ def _pydantic_type_to_arrow_type(tp: Any, field: FieldInfo) -> pa.DataType:
                 return pa.list_(pa.list_(tp.value_arrow_type(), tp.dim()))
             # For regular Vector
             return pa.list_(tp.value_arrow_type(), tp.dim())
+        if _safe_issubclass(tp, Enum):
+            # Map Enum to the Arrow type of its value.
+            # For string-valued enums, use dictionary encoding for efficiency.
+            # For integer enums, use the native type.
+            # Fall back to utf8 for mixed-type or empty enums.
+            value_types = {type(m.value) for m in tp}
+            if len(value_types) == 1:
+                value_type = value_types.pop()
+                if value_type is str:
+                    # Use dictionary encoding for string enums
+                    return pa.dictionary(pa.int32(), pa.utf8())
+                return _py_type_to_arrow_type(value_type, field)
+            return pa.utf8()
     return _py_type_to_arrow_type(tp, field)
 
 

python/python/lancedb/query.py

@@ -25,7 +25,6 @@ import deprecation
 import numpy as np
 import pyarrow as pa
 import pyarrow.compute as pc
-import pyarrow.fs as pa_fs
 import pydantic
 
 from lancedb.pydantic import PYDANTIC_VERSION
@@ -38,6 +37,7 @@ from .rerankers.base import Reranker
 from .rerankers.rrf import RRFReranker
 from .rerankers.util import check_reranker_result
 from .util import flatten_columns
+from .expr import Expr
 from lancedb._lancedb import fts_query_to_json
 from typing_extensions import Annotated
 
@@ -70,7 +70,7 @@ def ensure_vector_query(
 ) -> Union[List[float], List[List[float]], pa.Array, List[pa.Array]]:
     if isinstance(val, list):
         if len(val) == 0:
-            return ValueError("Vector query must be a non-empty list")
+            raise ValueError("Vector query must be a non-empty list")
         sample = val[0]
     else:
         if isinstance(val, float):
@@ -83,7 +83,7 @@ def ensure_vector_query(
         return val
     if isinstance(sample, list):
         if len(sample) == 0:
-            return ValueError("Vector query must be a non-empty list")
+            raise ValueError("Vector query must be a non-empty list")
         if isinstance(sample[0], float):
             # val is list of list of floats
             return val
@@ -449,8 +449,8 @@ class Query(pydantic.BaseModel):
         ensure_vector_query,
     ] = None
 
-    # sql filter to refine the query with
-    filter: Optional[str] = None
+    # sql filter or type-safe Expr to refine the query with
+    filter: Optional[Union[str, Expr]] = None
 
     # if True then apply the filter after vector search
     postfilter: Optional[bool] = None
@@ -464,8 +464,8 @@ class Query(pydantic.BaseModel):
     # distance type to use for vector search
     distance_type: Optional[str] = None
 
-    # which columns to return in the results
-    columns: Optional[Union[List[str], Dict[str, str]]] = None
+    # which columns to return in the results (dict values may be str or Expr)
+    columns: Optional[Union[List[str], Dict[str, Union[str, Expr]]]] = None
 
     # minimum number of IVF partitions to search
     #
@@ -856,14 +856,15 @@ class LanceQueryBuilder(ABC):
             self._offset = offset
         return self
 
-    def select(self, columns: Union[list[str], dict[str, str]]) -> Self:
+    def select(self, columns: Union[list[str], dict[str, Union[str, Expr]]]) -> Self:
         """Set the columns to return.
 
         Parameters
         ----------
-        columns: list of str, or dict of str to str default None
+        columns: list of str, or dict of str to str or Expr
             List of column names to be fetched.
-            Or a dictionary of column names to SQL expressions.
+            Or a dictionary of column names to SQL expressions or
+            :class:`~lancedb.expr.Expr` objects.
             All columns are fetched if None or unspecified.
 
         Returns
@@ -877,15 +878,15 @@ class LanceQueryBuilder(ABC):
             raise ValueError("columns must be a list or a dictionary")
         return self
 
-    def where(self, where: str, prefilter: bool = True) -> Self:
+    def where(self, where: Union[str, Expr], prefilter: bool = True) -> Self:
         """Set the where clause.
 
         Parameters
         ----------
-        where: str
-            The where clause which is a valid SQL where clause. See
-            `Lance filter pushdown <https://lance.org/guide/read_and_write#filter-push-down>`_
-            for valid SQL expressions.
+        where: str or :class:`~lancedb.expr.Expr`
+            The filter condition.  Can be a SQL string or a type-safe
+            :class:`~lancedb.expr.Expr` built with :func:`~lancedb.expr.col`
+            and :func:`~lancedb.expr.lit`.
         prefilter: bool, default True
             If True, apply the filter before vector search, otherwise the
             filter is applied on the result of vector search.
@@ -1355,15 +1356,17 @@ class LanceVectorQueryBuilder(LanceQueryBuilder):
 
         return result_set
 
-    def where(self, where: str, prefilter: bool = None) -> LanceVectorQueryBuilder:
+    def where(
+        self, where: Union[str, Expr], prefilter: bool = None
+    ) -> LanceVectorQueryBuilder:
         """Set the where clause.
 
         Parameters
         ----------
-        where: str
-            The where clause which is a valid SQL where clause. See
-            `Lance filter pushdown <https://lance.org/guide/read_and_write#filter-push-down>`_
-            for valid SQL expressions.
+        where: str or :class:`~lancedb.expr.Expr`
+            The filter condition.  Can be a SQL string or a type-safe
+            :class:`~lancedb.expr.Expr` built with :func:`~lancedb.expr.col`
+            and :func:`~lancedb.expr.lit`.
         prefilter: bool, default True
             If True, apply the filter before vector search, otherwise the
             filter is applied on the result of vector search.
@@ -1522,9 +1525,7 @@ class LanceFtsQueryBuilder(LanceQueryBuilder):
         return self._table._output_schema(self.to_query_object())
 
     def to_arrow(self, *, timeout: Optional[timedelta] = None) -> pa.Table:
-        path, fs, exist = self._table._get_fts_index_path()
-        if exist:
-            return self.tantivy_to_arrow()
+        self._table._ensure_no_legacy_fts_index()
 
         query = self._query
         if self._phrase_query:
@@ -1548,90 +1549,6 @@ class LanceFtsQueryBuilder(LanceQueryBuilder):
     ):
         raise NotImplementedError("to_batches on an FTS query")
 
-    def tantivy_to_arrow(self) -> pa.Table:
-        try:
-            import tantivy
-        except ImportError:
-            raise ImportError(
-                "Please install tantivy-py `pip install tantivy` to use the full text search feature."  # noqa: E501
-            )
-
-        from .fts import search_index
-
-        # get the index path
-        path, fs, exist = self._table._get_fts_index_path()
-
-        # check if the index exist
-        if not exist:
-            raise FileNotFoundError(
-                "Fts index does not exist. "
-                "Please first call table.create_fts_index(['<field_names>']) to "
-                "create the fts index."
-            )
-
-        # Check that we are on local filesystem
-        if not isinstance(fs, pa_fs.LocalFileSystem):
-            raise NotImplementedError(
-                "Tantivy-based full text search "
-                "is only supported on the local filesystem"
-            )
-        # open the index
-        index = tantivy.Index.open(path)
-        # get the scores and doc ids
-        query = self._query
-        if self._phrase_query:
-            query = query.replace('"', "'")
-            query = f'"{query}"'
-        limit = self._limit if self._limit is not None else 10
-        row_ids, scores = search_index(
-            index, query, limit, ordering_field=self.ordering_field_name
-        )
-        if len(row_ids) == 0:
-            empty_schema = pa.schema([pa.field("_score", pa.float32())])
-            return pa.Table.from_batches([], schema=empty_schema)
-        scores = pa.array(scores)
-        output_tbl = self._table.to_lance().take(row_ids, columns=self._columns)
-        output_tbl = output_tbl.append_column("_score", scores)
-        # this needs to match vector search results which are uint64
-        row_ids = pa.array(row_ids, type=pa.uint64())
-
-        if self._where is not None:
-            tmp_name = "__lancedb__duckdb__indexer__"
-            output_tbl = output_tbl.append_column(
-                tmp_name, pa.array(range(len(output_tbl)))
-            )
-            try:
-                # TODO would be great to have Substrait generate pyarrow compute
-                # expressions or conversely have pyarrow support SQL expressions
-                # using Substrait
-                import duckdb
-
-                indexer = duckdb.sql(
-                    f"SELECT {tmp_name} FROM output_tbl WHERE {self._where}"
-                ).to_arrow_table()[tmp_name]
-                output_tbl = output_tbl.take(indexer).drop([tmp_name])
-                row_ids = row_ids.take(indexer)
-
-            except ImportError:
-                import tempfile
-
-                import lance
-
-                # TODO Use "memory://" instead once that's supported
-                with tempfile.TemporaryDirectory() as tmp:
-                    ds = lance.write_dataset(output_tbl, tmp)
-                    output_tbl = ds.to_table(filter=self._where)
-                    indexer = output_tbl[tmp_name]
-                    row_ids = row_ids.take(indexer)
-                    output_tbl = output_tbl.drop([tmp_name])
-
-        if self._with_row_id:
-            output_tbl = output_tbl.append_column("_rowid", row_ids)
-
-        if self._reranker is not None:
-            output_tbl = self._reranker.rerank_fts(self._query, output_tbl)
-        return output_tbl
-
     def rerank(self, reranker: Reranker) -> LanceFtsQueryBuilder:
         """Rerank the results using the specified reranker.
 
@@ -2205,8 +2122,8 @@ class LanceHybridQueryBuilder(LanceQueryBuilder):
             self._vector_query.select(self._columns)
             self._fts_query.select(self._columns)
         if self._where:
-            self._vector_query.where(self._where, self._postfilter)
-            self._fts_query.where(self._where, self._postfilter)
+            self._vector_query.where(self._where, not self._postfilter)
+            self._fts_query.where(self._where, not self._postfilter)
         if self._with_row_id:
             self._vector_query.with_row_id(True)
             self._fts_query.with_row_id(True)
@@ -2286,10 +2203,20 @@ class AsyncQueryBase(object):
         """
         if isinstance(columns, list) and all(isinstance(c, str) for c in columns):
             self._inner.select_columns(columns)
-        elif isinstance(columns, dict) and all(
-            isinstance(k, str) and isinstance(v, str) for k, v in columns.items()
-        ):
-            self._inner.select(list(columns.items()))
+        elif isinstance(columns, dict) and all(isinstance(k, str) for k in columns):
+            if any(isinstance(v, Expr) for v in columns.values()):
+                # At least one value is an Expr — use the type-safe path.
+                from .expr import _coerce
+
+                pairs = [(k, _coerce(v)._inner) for k, v in columns.items()]
+                self._inner.select_expr(pairs)
+            elif all(isinstance(v, str) for v in columns.values()):
+                self._inner.select(list(columns.items()))
+            else:
+                raise TypeError(
+                    "dict values must be str or Expr, got "
+                    + str({k: type(v) for k, v in columns.items()})
+                )
         else:
             raise TypeError("columns must be a list of column names or a dict")
         return self
@@ -2529,11 +2456,13 @@ class AsyncStandardQuery(AsyncQueryBase):
         """
         super().__init__(inner)
 
-    def where(self, predicate: str) -> Self:
+    def where(self, predicate: Union[str, Expr]) -> Self:
         """
         Only return rows matching the given predicate
 
-        The predicate should be supplied as an SQL query string.
+        The predicate can be a SQL string or a type-safe
+        :class:`~lancedb.expr.Expr` built with :func:`~lancedb.expr.col`
+        and :func:`~lancedb.expr.lit`.
 
         Examples
         --------
@@ -2545,7 +2474,10 @@ class AsyncStandardQuery(AsyncQueryBase):
         Filtering performance can often be improved by creating a scalar index
         on the filter column(s).
         """
-        self._inner.where(predicate)
+        if isinstance(predicate, Expr):
+            self._inner.where_expr(predicate._inner)
+        else:
+            self._inner.where(predicate)
         return self
 
     def limit(self, limit: int) -> Self:

python/python/lancedb/remote/__init__.py

@@ -145,6 +145,33 @@ class TlsConfig:
 
 @dataclass
 class ClientConfig:
+    """Configuration for the LanceDB Cloud HTTP client.
+
+    Attributes
+    ----------
+    user_agent: str
+        User agent string sent with requests.
+    retry_config: RetryConfig
+        Configuration for retrying failed requests.
+    timeout_config: Optional[TimeoutConfig]
+        Configuration for request timeouts.
+    extra_headers: Optional[dict]
+        Additional headers to include in requests.
+    id_delimiter: Optional[str]
+        The delimiter to use when constructing object identifiers.
+    tls_config: Optional[TlsConfig]
+        TLS/mTLS configuration for secure connections.
+    header_provider: Optional[HeaderProvider]
+        Provider for dynamic headers to be added to each request.
+    user_id: Optional[str]
+        User identifier for tracking purposes. This is sent as the
+        `x-lancedb-user-id` header in requests to LanceDB Cloud/Enterprise.
+
+        This can also be set via the `LANCEDB_USER_ID` environment variable.
+        Alternatively, set `LANCEDB_USER_ID_ENV_KEY` to specify another
+        environment variable that contains the user ID value.
+    """
+
     user_agent: str = f"LanceDB-Python-Client/{__version__}"
     retry_config: RetryConfig = field(default_factory=RetryConfig)
     timeout_config: Optional[TimeoutConfig] = field(default_factory=TimeoutConfig)
@@ -152,6 +179,7 @@ class ClientConfig:
     id_delimiter: Optional[str] = None
     tls_config: Optional[TlsConfig] = None
     header_provider: Optional["HeaderProvider"] = None
+    user_id: Optional[str] = None
 
     def __post_init__(self):
         if isinstance(self.retry_config, dict):

python/python/lancedb/remote/db.py

@@ -24,6 +24,7 @@ from ..common import DATA
 from ..db import DBConnection, LOOP
 from ..embeddings import EmbeddingFunctionConfig
 from lance_namespace import (
+    LanceNamespace,
     CreateNamespaceResponse,
     DescribeNamespaceResponse,
     DropNamespaceResponse,
@@ -111,7 +112,7 @@ class RemoteDBConnection(DBConnection):
     @override
     def list_namespaces(
         self,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         page_token: Optional[str] = None,
         limit: Optional[int] = None,
     ) -> ListNamespacesResponse:
@@ -119,7 +120,7 @@ class RemoteDBConnection(DBConnection):
 
         Parameters
         ----------
-        namespace: List[str], optional
+        namespace_path: List[str], optional
             The parent namespace to list namespaces in.
             None or empty list represents root namespace.
         page_token: str, optional
@@ -133,18 +134,18 @@ class RemoteDBConnection(DBConnection):
         ListNamespacesResponse
             Response containing namespace names and optional page_token for pagination.
         """
-        if namespace is None:
-            namespace = []
+        if namespace_path is None:
+            namespace_path = []
         return LOOP.run(
             self._conn.list_namespaces(
-                namespace=namespace, page_token=page_token, limit=limit
+                namespace_path=namespace_path, page_token=page_token, limit=limit
             )
         )
 
     @override
     def create_namespace(
         self,
-        namespace: List[str],
+        namespace_path: List[str],
         mode: Optional[str] = None,
         properties: Optional[Dict[str, str]] = None,
     ) -> CreateNamespaceResponse:
@@ -152,7 +153,7 @@ class RemoteDBConnection(DBConnection):
 
         Parameters
         ----------
-        namespace: List[str]
+        namespace_path: List[str]
             The namespace identifier to create.
         mode: str, optional
             Creation mode - "create" (fail if exists), "exist_ok" (skip if exists),
@@ -167,14 +168,14 @@ class RemoteDBConnection(DBConnection):
         """
         return LOOP.run(
             self._conn.create_namespace(
-                namespace=namespace, mode=mode, properties=properties
+                namespace_path=namespace_path, mode=mode, properties=properties
             )
         )
 
     @override
     def drop_namespace(
         self,
-        namespace: List[str],
+        namespace_path: List[str],
         mode: Optional[str] = None,
         behavior: Optional[str] = None,
     ) -> DropNamespaceResponse:
@@ -182,7 +183,7 @@ class RemoteDBConnection(DBConnection):
 
         Parameters
         ----------
-        namespace: List[str]
+        namespace_path: List[str]
             The namespace identifier to drop.
         mode: str, optional
             Whether to skip if not exists ("SKIP") or fail ("FAIL"). Case insensitive.
@@ -196,16 +197,20 @@ class RemoteDBConnection(DBConnection):
             Response containing properties and transaction_id if applicable.
         """
         return LOOP.run(
-            self._conn.drop_namespace(namespace=namespace, mode=mode, behavior=behavior)
+            self._conn.drop_namespace(
+                namespace_path=namespace_path, mode=mode, behavior=behavior
+            )
         )
 
     @override
-    def describe_namespace(self, namespace: List[str]) -> DescribeNamespaceResponse:
+    def describe_namespace(
+        self, namespace_path: List[str]
+    ) -> DescribeNamespaceResponse:
         """Describe a namespace.
 
         Parameters
         ----------
-        namespace: List[str]
+        namespace_path: List[str]
             The namespace identifier to describe.
 
         Returns
@@ -213,12 +218,12 @@ class RemoteDBConnection(DBConnection):
         DescribeNamespaceResponse
             Response containing the namespace properties.
         """
-        return LOOP.run(self._conn.describe_namespace(namespace=namespace))
+        return LOOP.run(self._conn.describe_namespace(namespace_path=namespace_path))
 
     @override
     def list_tables(
         self,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         page_token: Optional[str] = None,
         limit: Optional[int] = None,
     ) -> ListTablesResponse:
@@ -226,7 +231,7 @@ class RemoteDBConnection(DBConnection):
 
         Parameters
         ----------
-        namespace: List[str], optional
+        namespace_path: List[str], optional
             The namespace to list tables in.
             None or empty list represents root namespace.
         page_token: str, optional
@@ -240,11 +245,11 @@ class RemoteDBConnection(DBConnection):
         ListTablesResponse
             Response containing table names and optional page_token for pagination.
         """
-        if namespace is None:
-            namespace = []
+        if namespace_path is None:
+            namespace_path = []
         return LOOP.run(
             self._conn.list_tables(
-                namespace=namespace, page_token=page_token, limit=limit
+                namespace_path=namespace_path, page_token=page_token, limit=limit
             )
         )
 
@@ -254,7 +259,7 @@ class RemoteDBConnection(DBConnection):
         page_token: Optional[str] = None,
         limit: int = 10,
         *,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
     ) -> Iterable[str]:
         """List the names of all tables in the database.
 
@@ -263,7 +268,7 @@ class RemoteDBConnection(DBConnection):
 
         Parameters
         ----------
-        namespace: List[str], default []
+        namespace_path: List[str], default []
             The namespace to list tables in.
             Empty list represents root namespace.
         page_token: str
@@ -282,11 +287,11 @@ class RemoteDBConnection(DBConnection):
             DeprecationWarning,
             stacklevel=2,
         )
-        if namespace is None:
-            namespace = []
+        if namespace_path is None:
+            namespace_path = []
         return LOOP.run(
             self._conn.table_names(
-                namespace=namespace, start_after=page_token, limit=limit
+                namespace_path=namespace_path, start_after=page_token, limit=limit
             )
         )
 
@@ -295,7 +300,7 @@ class RemoteDBConnection(DBConnection):
         self,
         name: str,
         *,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         storage_options: Optional[Dict[str, str]] = None,
         index_cache_size: Optional[int] = None,
     ) -> Table:
@@ -305,7 +310,7 @@ class RemoteDBConnection(DBConnection):
         ----------
         name: str
             The name of the table.
-        namespace: List[str], optional
+        namespace_path: List[str], optional
             The namespace to open the table from.
             None or empty list represents root namespace.
 
@@ -315,15 +320,15 @@ class RemoteDBConnection(DBConnection):
         """
         from .table import RemoteTable
 
-        if namespace is None:
-            namespace = []
+        if namespace_path is None:
+            namespace_path = []
         if index_cache_size is not None:
             logging.info(
                 "index_cache_size is ignored in LanceDb Cloud"
                 " (there is no local cache to configure)"
             )
 
-        table = LOOP.run(self._conn.open_table(name, namespace=namespace))
+        table = LOOP.run(self._conn.open_table(name, namespace_path=namespace_path))
         return RemoteTable(table, self.db_name)
 
     def clone_table(
@@ -331,7 +336,7 @@ class RemoteDBConnection(DBConnection):
         target_table_name: str,
         source_uri: str,
         *,
-        target_namespace: Optional[List[str]] = None,
+        target_namespace_path: Optional[List[str]] = None,
         source_version: Optional[int] = None,
         source_tag: Optional[str] = None,
         is_shallow: bool = True,
@@ -344,7 +349,7 @@ class RemoteDBConnection(DBConnection):
             The name of the target table to create.
         source_uri: str
             The URI of the source table to clone from.
-        target_namespace: List[str], optional
+        target_namespace_path: List[str], optional
             The namespace for the target table.
             None or empty list represents root namespace.
         source_version: int, optional
@@ -361,13 +366,13 @@ class RemoteDBConnection(DBConnection):
         """
         from .table import RemoteTable
 
-        if target_namespace is None:
-            target_namespace = []
+        if target_namespace_path is None:
+            target_namespace_path = []
         table = LOOP.run(
             self._conn.clone_table(
                 target_table_name,
                 source_uri,
-                target_namespace=target_namespace,
+                target_namespace_path=target_namespace_path,
                 source_version=source_version,
                 source_tag=source_tag,
                 is_shallow=is_shallow,
@@ -387,7 +392,7 @@ class RemoteDBConnection(DBConnection):
         exist_ok: bool = False,
         embedding_functions: Optional[List[EmbeddingFunctionConfig]] = None,
         *,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
     ) -> Table:
         """Create a [Table][lancedb.table.Table] in the database.
 
@@ -395,7 +400,7 @@ class RemoteDBConnection(DBConnection):
         ----------
         name: str
             The name of the table.
-        namespace: List[str], optional
+        namespace_path: List[str], optional
             The namespace to create the table in.
             None or empty list represents root namespace.
         data: The data to initialize the table, *optional*
@@ -495,8 +500,8 @@ class RemoteDBConnection(DBConnection):
                 mode = "exist_ok"
             elif not mode:
                 mode = "exist_ok"
-        if namespace is None:
-            namespace = []
+        if namespace_path is None:
+            namespace_path = []
         validate_table_name(name)
         if embedding_functions is not None:
             logging.warning(
@@ -511,7 +516,7 @@ class RemoteDBConnection(DBConnection):
             self._conn.create_table(
                 name,
                 data,
-                namespace=namespace,
+                namespace_path=namespace_path,
                 mode=mode,
                 schema=schema,
                 on_bad_vectors=on_bad_vectors,
@@ -521,28 +526,28 @@ class RemoteDBConnection(DBConnection):
         return RemoteTable(table, self.db_name)
 
     @override
-    def drop_table(self, name: str, namespace: Optional[List[str]] = None):
+    def drop_table(self, name: str, namespace_path: Optional[List[str]] = None):
         """Drop a table from the database.
 
         Parameters
         ----------
         name: str
             The name of the table.
-        namespace: List[str], optional
+        namespace_path: List[str], optional
             The namespace to drop the table from.
             None or empty list represents root namespace.
         """
-        if namespace is None:
-            namespace = []
-        LOOP.run(self._conn.drop_table(name, namespace=namespace))
+        if namespace_path is None:
+            namespace_path = []
+        LOOP.run(self._conn.drop_table(name, namespace_path=namespace_path))
 
     @override
     def rename_table(
         self,
         cur_name: str,
         new_name: str,
-        cur_namespace: Optional[List[str]] = None,
-        new_namespace: Optional[List[str]] = None,
+        cur_namespace_path: Optional[List[str]] = None,
+        new_namespace_path: Optional[List[str]] = None,
     ):
         """Rename a table in the database.
 
@@ -553,19 +558,32 @@ class RemoteDBConnection(DBConnection):
         new_name: str
             The new name of the table.
         """
-        if cur_namespace is None:
-            cur_namespace = []
-        if new_namespace is None:
-            new_namespace = []
+        if cur_namespace_path is None:
+            cur_namespace_path = []
+        if new_namespace_path is None:
+            new_namespace_path = []
         LOOP.run(
             self._conn.rename_table(
                 cur_name,
                 new_name,
-                cur_namespace=cur_namespace,
-                new_namespace=new_namespace,
+                cur_namespace_path=cur_namespace_path,
+                new_namespace_path=new_namespace_path,
             )
         )
 
+    @override
+    def namespace_client(self) -> LanceNamespace:
+        """Get the equivalent namespace client for this connection.
+
+        Returns a RestNamespace with the same URI and authentication headers.
+
+        Returns
+        -------
+        LanceNamespace
+            The namespace client for this connection.
+        """
+        return LOOP.run(self._conn.namespace_client())
+
     async def close(self):
         """Close the connection to the database."""
-        self._client.close()
+        self._conn.close()

python/python/lancedb/remote/table.py

@@ -4,7 +4,7 @@
 from datetime import timedelta
 import logging
 from functools import cached_property
-from typing import Dict, Iterable, List, Optional, Union, Literal
+from typing import Any, Callable, Dict, Iterable, List, Optional, Union, Literal
 import warnings
 
 from lancedb._lancedb import (
@@ -35,6 +35,7 @@ import pyarrow as pa
 from lancedb.common import DATA, VEC, VECTOR_COLUMN_NAME
 from lancedb.merge import LanceMergeInsertBuilder
 from lancedb.embeddings import EmbeddingFunctionRegistry
+from lancedb.table import _normalize_progress
 
 from ..query import LanceVectorQueryBuilder, LanceQueryBuilder, LanceTakeQueryBuilder
 from ..table import AsyncTable, IndexStatistics, Query, Table, Tags
@@ -308,6 +309,7 @@ class RemoteTable(Table):
         mode: str = "append",
         on_bad_vectors: str = "error",
         fill_value: float = 0.0,
+        progress: Optional[Union[bool, Callable, Any]] = None,
     ) -> AddResult:
         """Add more data to the [Table](Table). It has the same API signature as
         the OSS version.
@@ -330,17 +332,29 @@ class RemoteTable(Table):
             One of "error", "drop", "fill".
         fill_value: float, default 0.
             The value to use when filling vectors. Only used if on_bad_vectors="fill".
+        progress: bool, callable, or tqdm-like, optional
+            A callback or tqdm-compatible progress bar. See
+            :meth:`Table.add` for details.
 
         Returns
         -------
         AddResult
             An object containing the new version number of the table after adding data.
         """
-        return LOOP.run(
-            self._table.add(
-                data, mode=mode, on_bad_vectors=on_bad_vectors, fill_value=fill_value
+        progress, owns = _normalize_progress(progress)
+        try:
+            return LOOP.run(
+                self._table.add(
+                    data,
+                    mode=mode,
+                    on_bad_vectors=on_bad_vectors,
+                    fill_value=fill_value,
+                    progress=progress,
+                )
             )
-        )
+        finally:
+            if owns:
+                progress.close()
 
     def search(
         self,

python/python/lancedb/table.py

@@ -14,6 +14,7 @@ from functools import cached_property
 from typing import (
     TYPE_CHECKING,
     Any,
+    Callable,
     Dict,
     Iterable,
     List,
@@ -88,7 +89,6 @@ from .index import lang_mapping
 
 if TYPE_CHECKING:
     from .db import LanceDBConnection
-    from .io import StorageOptionsProvider
     from ._lancedb import (
         Table as LanceDBTable,
         OptimizeStats,
@@ -191,7 +191,7 @@ def _into_pyarrow_reader(
             f"Unknown data type {type(data)}. "
             "Supported types: list of dicts, pandas DataFrame, polars DataFrame, "
             "pyarrow Table/RecordBatch, or Pydantic models. "
-            "See https://lancedb.com/docs/tables/ for examples."
+            "See https://docs.lancedb.com/tables/ for examples."
         )
 
 
@@ -270,15 +270,17 @@ def _sanitize_data(
         reader,
         on_bad_vectors=on_bad_vectors,
         fill_value=fill_value,
+        target_schema=target_schema,
+        metadata=metadata,
     )
 
     if target_schema is None:
         target_schema, reader = _infer_target_schema(reader)
 
     if metadata:
-        new_metadata = target_schema.metadata or {}
-        new_metadata = new_metadata.update(metadata)
-        target_schema = target_schema.with_metadata(new_metadata)
+        target_schema = target_schema.with_metadata(
+            _merge_metadata(target_schema.metadata, metadata)
+        )
 
     _validate_schema(target_schema)
     reader = _cast_to_target_schema(reader, target_schema, allow_subschema)
@@ -294,7 +296,7 @@ def _cast_to_target_schema(
     # pa.Table.cast expects field order not to be changed.
     # Lance doesn't care about field order, so we don't need to rearrange fields
     # to match the target schema. We just need to correctly cast the fields.
-    if reader.schema == target_schema:
+    if reader.schema.equals(target_schema, check_metadata=True):
         # Fast path when the schemas are already the same
         return reader
 
@@ -314,7 +316,13 @@ def _cast_to_target_schema(
     def gen():
         for batch in reader:
             # Table but not RecordBatch has cast.
-            yield pa.Table.from_batches([batch]).cast(reordered_schema).to_batches()[0]
+            cast_batches = (
+                pa.Table.from_batches([batch]).cast(reordered_schema).to_batches()
+            )
+            if cast_batches:
+                yield pa.RecordBatch.from_arrays(
+                    cast_batches[0].columns, schema=reordered_schema
+                )
 
     return pa.RecordBatchReader.from_batches(reordered_schema, gen())
 
@@ -332,37 +340,51 @@ def _align_field_types(
         if target_field is None:
             raise ValueError(f"Field '{field.name}' not found in target schema")
         if pa.types.is_struct(target_field.type):
-            new_type = pa.struct(
-                _align_field_types(
-                    field.type.fields,
-                    target_field.type.fields,
+            if pa.types.is_struct(field.type):
+                new_type = pa.struct(
+                    _align_field_types(
+                        field.type.fields,
+                        target_field.type.fields,
+                    )
                 )
-            )
+            else:
+                new_type = target_field.type
         elif pa.types.is_list(target_field.type):
-            new_type = pa.list_(
-                _align_field_types(
-                    [field.type.value_field],
-                    [target_field.type.value_field],
-                )[0]
-            )
+            if _is_list_like(field.type):
+                new_type = pa.list_(
+                    _align_field_types(
+                        [field.type.value_field],
+                        [target_field.type.value_field],
+                    )[0]
+                )
+            else:
+                new_type = target_field.type
         elif pa.types.is_large_list(target_field.type):
-            new_type = pa.large_list(
-                _align_field_types(
-                    [field.type.value_field],
-                    [target_field.type.value_field],
-                )[0]
-            )
+            if _is_list_like(field.type):
+                new_type = pa.large_list(
+                    _align_field_types(
+                        [field.type.value_field],
+                        [target_field.type.value_field],
+                    )[0]
+                )
+            else:
+                new_type = target_field.type
         elif pa.types.is_fixed_size_list(target_field.type):
-            new_type = pa.list_(
-                _align_field_types(
-                    [field.type.value_field],
-                    [target_field.type.value_field],
-                )[0],
-                target_field.type.list_size,
-            )
+            if _is_list_like(field.type):
+                new_type = pa.list_(
+                    _align_field_types(
+                        [field.type.value_field],
+                        [target_field.type.value_field],
+                    )[0],
+                    target_field.type.list_size,
+                )
+            else:
+                new_type = target_field.type
         else:
             new_type = target_field.type
-        new_fields.append(pa.field(field.name, new_type, field.nullable))
+        new_fields.append(
+            pa.field(field.name, new_type, field.nullable, target_field.metadata)
+        )
     return new_fields
 
 
@@ -440,6 +462,7 @@ def sanitize_create_table(
             schema = data.schema
 
     if metadata:
+        metadata = _merge_metadata(schema.metadata, metadata)
         schema = schema.with_metadata(metadata)
         # Need to apply metadata to the data as well
         if isinstance(data, pa.Table):
@@ -492,9 +515,9 @@ def _append_vector_columns(
     vector columns to the table.
     """
     if schema is None:
-        metadata = metadata or {}
+        metadata = _merge_metadata(metadata)
     else:
-        metadata = schema.metadata or metadata or {}
+        metadata = _merge_metadata(schema.metadata, metadata)
     functions = EmbeddingFunctionRegistry.get_instance().parse_functions(metadata)
 
     if not functions:
@@ -556,6 +579,21 @@ def _table_uri(base: str, table_name: str) -> str:
     return join_uri(base, f"{table_name}.lance")
 
 
+def _normalize_progress(progress):
+    """Normalize a ``progress`` parameter for :meth:`Table.add`.
+
+    Returns ``(progress_obj, owns)`` where *owns* is True when we created a
+    tqdm bar that the caller must close.
+    """
+    if progress is True:
+        from tqdm.auto import tqdm
+
+        return tqdm(unit=" rows"), True
+    if progress is False or progress is None:
+        return None, False
+    return progress, False
+
+
 class Table(ABC):
     """
     A Table is a collection of Records in a LanceDB Database.
@@ -905,29 +943,26 @@ class Table(ABC):
         Parameters
         ----------
         field_names: str or list of str
-            The name(s) of the field to index.
-            If ``use_tantivy`` is False (default), only a single field name
-            (str) is supported. To index multiple fields, create a separate
-            FTS index for each field.
+            The name of the field to index. Native FTS indexes can only be
+            created on a single field at a time. To search over multiple text
+            fields, create a separate FTS index for each field.
         replace: bool, default False
             If True, replace the existing index if it exists. Note that this is
             not yet an atomic operation; the index will be temporarily
             unavailable while the new index is being created.
         writer_heap_size: int, default 1GB
-            Only available with use_tantivy=True
+            Deprecated legacy Tantivy parameter. Any value other than the
+            default raises an error.
         ordering_field_names:
-            A list of unsigned type fields to index to optionally order
-            results on at search time.
-            only available with use_tantivy=True
+            Deprecated legacy Tantivy parameter. Setting this raises an error.
         tokenizer_name: str, default "default"
-            The tokenizer to use for the index. Can be "raw", "default" or the 2 letter
-            language code followed by "_stem". So for english it would be "en_stem".
-            For available languages see: https://docs.rs/tantivy/latest/tantivy/tokenizer/enum.Language.html
+            A compatibility alias for native tokenizer configs. Can be "raw",
+            "default" or the 2 letter language code followed by "_stem". So
+            for english it would be "en_stem".
         use_tantivy: bool, default False
-            If True, use the legacy full-text search implementation based on tantivy.
-            If False, use the new full-text search implementation based on lance-index.
+            Deprecated legacy Tantivy parameter. Setting this to True raises an
+            error.
         with_position: bool, default False
-            Only available with use_tantivy=False
             If False, do not store the positions of the terms in the text.
             This can reduce the size of the index and improve indexing speed.
             But it will raise an exception for phrase queries.
@@ -974,6 +1009,7 @@ class Table(ABC):
         mode: AddMode = "append",
         on_bad_vectors: OnBadVectorsType = "error",
         fill_value: float = 0.0,
+        progress: Optional[Union[bool, Callable, Any]] = None,
     ) -> AddResult:
         """Add more data to the [Table](Table).
 
@@ -995,6 +1031,29 @@ class Table(ABC):
             One of "error", "drop", "fill".
         fill_value: float, default 0.
             The value to use when filling vectors. Only used if on_bad_vectors="fill".
+        progress: bool, callable, or tqdm-like, optional
+            Progress reporting during the add operation. Can be:
+
+            - ``True`` to automatically create and display a tqdm progress
+              bar (requires ``tqdm`` to be installed)::
+
+                table.add(data, progress=True)
+
+            - A **callable** that receives a dict with keys ``output_rows``,
+              ``output_bytes``, ``total_rows``, ``elapsed_seconds``,
+              ``active_tasks``, ``total_tasks``, and ``done``::
+
+                def on_progress(p):
+                    print(f"{p['output_rows']}/{p['total_rows']} rows, "
+                          f"{p['active_tasks']}/{p['total_tasks']} workers")
+                table.add(data, progress=on_progress)
+
+            - A **tqdm-compatible** progress bar whose ``total`` and
+              ``update()`` will be called automatically. The postfix shows
+              write throughput (MB/s) and active worker count::
+
+                with tqdm() as pbar:
+                    table.add(data, progress=pbar)
 
         Returns
         -------
@@ -1684,6 +1743,16 @@ class Table(ABC):
         index_exists = fs.get_file_info(path).type != pa_fs.FileType.NotFound
         return (path, fs, index_exists)
 
+    def _ensure_no_legacy_fts_index(self):
+        path, _, exists = self._get_fts_index_path()
+        if exists:
+            raise ValueError(
+                "Legacy Tantivy FTS index detected at "
+                f"{path}. Tantivy-based FTS has been removed. "
+                "Delete the legacy index and recreate it with "
+                "table.create_fts_index(...)."
+            )
+
     @abstractmethod
     def uses_v2_manifest_paths(self) -> bool:
         """
@@ -1736,30 +1805,30 @@ class LanceTable(Table):
         connection: "LanceDBConnection",
         name: str,
         *,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         storage_options: Optional[Dict[str, str]] = None,
-        storage_options_provider: Optional["StorageOptionsProvider"] = None,
         index_cache_size: Optional[int] = None,
         location: Optional[str] = None,
         namespace_client: Optional[Any] = None,
         managed_versioning: Optional[bool] = None,
+        pushdown_operations: Optional[set] = None,
         _async: AsyncTable = None,
     ):
-        if namespace is None:
-            namespace = []
+        if namespace_path is None:
+            namespace_path = []
         self._conn = connection
-        self._namespace = namespace
+        self._namespace_path = namespace_path
         self._location = location  # Store location for use in _dataset_path
         self._namespace_client = namespace_client
+        self._pushdown_operations = pushdown_operations or set()
         if _async is not None:
             self._table = _async
         else:
             self._table = LOOP.run(
                 connection._conn.open_table(
                     name,
-                    namespace=namespace,
+                    namespace_path=namespace_path,
                     storage_options=storage_options,
-                    storage_options_provider=storage_options_provider,
                     index_cache_size=index_cache_size,
                     location=location,
                     namespace_client=namespace_client,
@@ -1774,13 +1843,13 @@ class LanceTable(Table):
     @property
     def namespace(self) -> List[str]:
         """Return the namespace path of the table."""
-        return self._namespace
+        return self._namespace_path
 
     @property
     def id(self) -> str:
         """Return the full identifier of the table (namespace$name)."""
-        if self._namespace:
-            return "$".join(self._namespace + [self.name])
+        if self._namespace_path:
+            return "$".join(self._namespace_path + [self.name])
         return self.name
 
     @classmethod
@@ -1801,26 +1870,26 @@ class LanceTable(Table):
         db,
         name,
         *,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         storage_options: Optional[Dict[str, str]] = None,
-        storage_options_provider: Optional["StorageOptionsProvider"] = None,
         index_cache_size: Optional[int] = None,
         location: Optional[str] = None,
         namespace_client: Optional[Any] = None,
         managed_versioning: Optional[bool] = None,
+        pushdown_operations: Optional[set] = None,
     ):
-        if namespace is None:
-            namespace = []
+        if namespace_path is None:
+            namespace_path = []
         tbl = cls(
             db,
             name,
-            namespace=namespace,
+            namespace_path=namespace_path,
             storage_options=storage_options,
-            storage_options_provider=storage_options_provider,
             index_cache_size=index_cache_size,
             location=location,
             namespace_client=namespace_client,
             managed_versioning=managed_versioning,
+            pushdown_operations=pushdown_operations,
         )
 
         # check the dataset exists
@@ -1853,11 +1922,11 @@ class LanceTable(Table):
             )
 
         if self._namespace_client is not None:
-            table_id = self._namespace + [self.name]
+            table_id = self._namespace_path + [self.name]
             return lance.dataset(
                 version=self.version,
                 storage_options=self._conn.storage_options,
-                namespace=self._namespace_client,
+                namespace_client=self._namespace_client,
                 table_id=table_id,
                 **kwargs,
             )
@@ -2343,84 +2412,63 @@ class LanceTable(Table):
         prefix_only: bool = False,
         name: Optional[str] = None,
     ):
-        if not use_tantivy:
-            if not isinstance(field_names, str):
-                raise ValueError(
-                    "Native FTS indexes can only be created on a single field "
-                    "at a time. To search over multiple text fields, create a "
-                    "separate FTS index for each field."
-                )
+        self._ensure_no_legacy_fts_index()
 
-            if tokenizer_name is None:
-                tokenizer_configs = {
-                    "base_tokenizer": base_tokenizer,
-                    "language": language,
-                    "with_position": with_position,
-                    "max_token_length": max_token_length,
-                    "lower_case": lower_case,
-                    "stem": stem,
-                    "remove_stop_words": remove_stop_words,
-                    "ascii_folding": ascii_folding,
-                    "ngram_min_length": ngram_min_length,
-                    "ngram_max_length": ngram_max_length,
-                    "prefix_only": prefix_only,
-                }
-            else:
-                tokenizer_configs = self.infer_tokenizer_configs(tokenizer_name)
-
-            config = FTS(
-                **tokenizer_configs,
+        if use_tantivy:
+            raise ValueError(
+                "Tantivy-based FTS has been removed. "
+                "Remove use_tantivy and recreate the index with native FTS."
             )
-
-            # delete the existing legacy index if it exists
-            if replace:
-                path, fs, exist = self._get_fts_index_path()
-                if exist:
-                    fs.delete_dir(path)
-
-            LOOP.run(
-                self._table.create_index(
-                    field_names,
-                    replace=replace,
-                    config=config,
-                    name=name,
-                )
+        if ordering_field_names is not None:
+            raise ValueError(
+                "ordering_field_names was only supported by the removed "
+                "Tantivy-based FTS implementation."
             )
-            return
-
-        from .fts import create_index, populate_index
-
-        if isinstance(field_names, str):
-            field_names = [field_names]
-
-        if isinstance(ordering_field_names, str):
-            ordering_field_names = [ordering_field_names]
-
-        path, fs, exist = self._get_fts_index_path()
-        if exist:
-            if not replace:
-                raise ValueError("Index already exists. Use replace=True to overwrite.")
-            fs.delete_dir(path)
-
-        if not isinstance(fs, pa_fs.LocalFileSystem):
-            raise NotImplementedError(
-                "Full-text search is only supported on the local filesystem"
+        if writer_heap_size != 1024 * 1024 * 1024:
+            raise ValueError(
+                "writer_heap_size was only supported by the removed "
+                "Tantivy-based FTS implementation."
+            )
+        if not isinstance(field_names, str):
+            raise ValueError(
+                "Native FTS indexes can only be created on a single field "
+                "at a time. To search over multiple text fields, create a "
+                "separate FTS index for each field."
+            )
+        if "." in field_names:
+            raise ValueError(
+                "Native FTS indexes can only be created on top-level fields. "
+                f"Received nested field path: {field_names!r}."
             )
 
         if tokenizer_name is None:
-            tokenizer_name = "default"
-        index = create_index(
-            path,
-            field_names,
-            ordering_fields=ordering_field_names,
-            tokenizer_name=tokenizer_name,
+            tokenizer_configs = {
+                "base_tokenizer": base_tokenizer,
+                "language": language,
+                "with_position": with_position,
+                "max_token_length": max_token_length,
+                "lower_case": lower_case,
+                "stem": stem,
+                "remove_stop_words": remove_stop_words,
+                "ascii_folding": ascii_folding,
+                "ngram_min_length": ngram_min_length,
+                "ngram_max_length": ngram_max_length,
+                "prefix_only": prefix_only,
+            }
+        else:
+            tokenizer_configs = self.infer_tokenizer_configs(tokenizer_name)
+
+        config = FTS(
+            **tokenizer_configs,
         )
-        populate_index(
-            index,
-            self,
-            field_names,
-            ordering_fields=ordering_field_names,
-            writer_heap_size=writer_heap_size,
+
+        LOOP.run(
+            self._table.create_index(
+                field_names,
+                replace=replace,
+                config=config,
+                name=name,
+            )
         )
 
     @staticmethod
@@ -2492,6 +2540,7 @@ class LanceTable(Table):
         mode: AddMode = "append",
         on_bad_vectors: OnBadVectorsType = "error",
         fill_value: float = 0.0,
+        progress: Optional[Union[bool, Callable, Any]] = None,
     ) -> AddResult:
         """Add data to the table.
         If vector columns are missing and the table
@@ -2510,17 +2559,29 @@ class LanceTable(Table):
             One of "error", "drop", "fill", "null".
         fill_value: float, default 0.
             The value to use when filling vectors. Only used if on_bad_vectors="fill".
+        progress: bool, callable, or tqdm-like, optional
+            A callback or tqdm-compatible progress bar. See
+            :meth:`Table.add` for details.
 
         Returns
         -------
         int
             The number of vectors in the table.
         """
-        return LOOP.run(
-            self._table.add(
-                data, mode=mode, on_bad_vectors=on_bad_vectors, fill_value=fill_value
+        progress, owns = _normalize_progress(progress)
+        try:
+            return LOOP.run(
+                self._table.add(
+                    data,
+                    mode=mode,
+                    on_bad_vectors=on_bad_vectors,
+                    fill_value=fill_value,
+                    progress=progress,
+                )
             )
-        )
+        finally:
+            if owns:
+                progress.close()
 
     def merge(
         self,
@@ -2750,13 +2811,13 @@ class LanceTable(Table):
         fill_value: float = 0.0,
         embedding_functions: Optional[List[EmbeddingFunctionConfig]] = None,
         *,
-        namespace: Optional[List[str]] = None,
+        namespace_path: Optional[List[str]] = None,
         storage_options: Optional[Dict[str, str | bool]] = None,
-        storage_options_provider: Optional["StorageOptionsProvider"] = None,
         data_storage_version: Optional[str] = None,
         enable_v2_manifest_paths: Optional[bool] = None,
         location: Optional[str] = None,
         namespace_client: Optional[Any] = None,
+        pushdown_operations: Optional[set] = None,
     ):
         """
         Create a new table.
@@ -2811,13 +2872,14 @@ class LanceTable(Table):
             Deprecated.  Set `storage_options` when connecting to the database and set
             `new_table_enable_v2_manifest_paths` in the options.
         """
-        if namespace is None:
-            namespace = []
+        if namespace_path is None:
+            namespace_path = []
         self = cls.__new__(cls)
         self._conn = db
-        self._namespace = namespace
+        self._namespace_path = namespace_path
         self._location = location
         self._namespace_client = namespace_client
+        self._pushdown_operations = pushdown_operations or set()
 
         if data_storage_version is not None:
             warnings.warn(
@@ -2850,10 +2912,10 @@ class LanceTable(Table):
                 on_bad_vectors=on_bad_vectors,
                 fill_value=fill_value,
                 embedding_functions=embedding_functions,
-                namespace=namespace,
+                namespace_path=namespace_path,
                 storage_options=storage_options,
-                storage_options_provider=storage_options_provider,
                 location=location,
+                namespace_client=namespace_client,
             )
         )
         return self
@@ -2921,6 +2983,15 @@ class LanceTable(Table):
         batch_size: Optional[int] = None,
         timeout: Optional[timedelta] = None,
     ) -> pa.RecordBatchReader:
+        if (
+            "QueryTable" in self._pushdown_operations
+            and self._namespace_client is not None
+        ):
+            from lancedb.namespace import _execute_server_side_query
+
+            table_id = self._namespace_path + [self.name]
+            return _execute_server_side_query(self._namespace_client, table_id, query)
+
         async_iter = LOOP.run(
             self._table._execute_query(query, batch_size=batch_size, timeout=timeout)
         )
@@ -3150,43 +3221,157 @@ def _handle_bad_vectors(
     reader: pa.RecordBatchReader,
     on_bad_vectors: Literal["error", "drop", "fill", "null"] = "error",
     fill_value: float = 0.0,
+    target_schema: Optional[pa.Schema] = None,
+    metadata: Optional[dict] = None,
 ) -> pa.RecordBatchReader:
-    vector_columns = []
+    vector_columns = _find_vector_columns(reader.schema, target_schema, metadata)
+    if not vector_columns:
+        return reader
 
-    for field in reader.schema:
-        # They can provide a 'vector' column that isn't yet a FSL
-        named_vector_col = (
-            (
-                pa.types.is_list(field.type)
-                or pa.types.is_large_list(field.type)
-                or pa.types.is_fixed_size_list(field.type)
-            )
-            and pa.types.is_floating(field.type.value_type)
-            and field.name == VECTOR_COLUMN_NAME
-        )
-        # TODO: we're making an assumption that fixed size list of 10 or more
-        # is a vector column. This is definitely a bit hacky.
-        likely_vector_col = (
-            pa.types.is_fixed_size_list(field.type)
-            and pa.types.is_floating(field.type.value_type)
-            and (field.type.list_size >= 10)
-        )
-
-        if named_vector_col or likely_vector_col:
-            vector_columns.append(field.name)
+    output_schema = _vector_output_schema(reader.schema, vector_columns)
 
     def gen():
         for batch in reader:
-            for name in vector_columns:
+            pending_dims = []
+            for vector_column in vector_columns:
+                dim = vector_column["expected_dim"]
+                if target_schema is not None and dim is None:
+                    dim = _infer_vector_dim(batch[vector_column["name"]])
+                    pending_dims.append(vector_column)
                 batch = _handle_bad_vector_column(
                     batch,
-                    vector_column_name=name,
+                    vector_column_name=vector_column["name"],
                     on_bad_vectors=on_bad_vectors,
                     fill_value=fill_value,
+                    expected_dim=dim,
+                    expected_value_type=vector_column["expected_value_type"],
                 )
-            yield batch
+            for vector_column in pending_dims:
+                if vector_column["expected_dim"] is None:
+                    vector_column["expected_dim"] = _infer_vector_dim(
+                        batch[vector_column["name"]]
+                    )
+            if batch.schema.equals(output_schema, check_metadata=True):
+                yield batch
+                continue
 
-    return pa.RecordBatchReader.from_batches(reader.schema, gen())
+            cast_batches = (
+                pa.Table.from_batches([batch]).cast(output_schema).to_batches()
+            )
+            if cast_batches:
+                yield pa.RecordBatch.from_arrays(
+                    cast_batches[0].columns,
+                    schema=output_schema,
+                )
+
+    return pa.RecordBatchReader.from_batches(output_schema, gen())
+
+
+def _find_vector_columns(
+    reader_schema: pa.Schema,
+    target_schema: Optional[pa.Schema],
+    metadata: Optional[dict],
+) -> List[dict]:
+    if target_schema is None:
+        vector_columns = []
+        for field in reader_schema:
+            named_vector_col = (
+                _is_list_like(field.type)
+                and pa.types.is_floating(field.type.value_type)
+                and field.name == VECTOR_COLUMN_NAME
+            )
+            likely_vector_col = (
+                pa.types.is_fixed_size_list(field.type)
+                and pa.types.is_floating(field.type.value_type)
+                and (field.type.list_size >= 10)
+            )
+            if named_vector_col or likely_vector_col:
+                vector_columns.append(
+                    {
+                        "name": field.name,
+                        "expected_dim": None,
+                        "expected_value_type": None,
+                    }
+                )
+        return vector_columns
+
+    reader_column_names = set(reader_schema.names)
+    active_metadata = _merge_metadata(target_schema.metadata, metadata)
+    embedding_function_columns = set(
+        EmbeddingFunctionRegistry.get_instance().parse_functions(active_metadata).keys()
+    )
+    vector_columns = []
+    for field in target_schema:
+        if field.name not in reader_column_names:
+            continue
+        if not _is_list_like(field.type) or not pa.types.is_floating(
+            field.type.value_type
+        ):
+            continue
+
+        reader_field = reader_schema.field(field.name)
+        named_vector_col = (
+            field.name in embedding_function_columns
+            or field.name == VECTOR_COLUMN_NAME
+            or (field.name == "embedding" and pa.types.is_fixed_size_list(field.type))
+        )
+        typed_fixed_vector_col = (
+            pa.types.is_fixed_size_list(reader_field.type)
+            and pa.types.is_floating(reader_field.type.value_type)
+            and reader_field.type.list_size >= 10
+        )
+
+        if named_vector_col or typed_fixed_vector_col:
+            vector_columns.append(
+                {
+                    "name": field.name,
+                    "expected_dim": (
+                        field.type.list_size
+                        if pa.types.is_fixed_size_list(field.type)
+                        else None
+                    ),
+                    "expected_value_type": field.type.value_type,
+                }
+            )
+
+    return vector_columns
+
+
+def _vector_output_schema(
+    reader_schema: pa.Schema,
+    vector_columns: List[dict],
+) -> pa.Schema:
+    columns_by_name = {column["name"]: column for column in vector_columns}
+    fields = []
+    for field in reader_schema:
+        column = columns_by_name.get(field.name)
+        if column is None:
+            output_type = field.type
+        else:
+            output_type = _vector_output_type(field, column)
+        fields.append(pa.field(field.name, output_type, field.nullable, field.metadata))
+    return pa.schema(fields, metadata=reader_schema.metadata)
+
+
+def _vector_output_type(field: pa.Field, vector_column: dict) -> pa.DataType:
+    if not _is_list_like(field.type):
+        return field.type
+
+    if vector_column["expected_value_type"] is not None and (
+        pa.types.is_null(field.type.value_type)
+        or pa.types.is_integer(field.type.value_type)
+        or pa.types.is_unsigned_integer(field.type.value_type)
+    ):
+        return pa.list_(vector_column["expected_value_type"])
+
+    if (
+        vector_column["expected_dim"] is not None
+        and pa.types.is_fixed_size_list(field.type)
+        and field.type.list_size != vector_column["expected_dim"]
+    ):
+        return pa.list_(field.type.value_type)
+
+    return field.type
 
 
 def _handle_bad_vector_column(
@@ -3194,6 +3379,8 @@ def _handle_bad_vector_column(
     vector_column_name: str,
     on_bad_vectors: str = "error",
     fill_value: float = 0.0,
+    expected_dim: Optional[int] = None,
+    expected_value_type: Optional[pa.DataType] = None,
 ) -> pa.RecordBatch:
     """
     Ensure that the vector column exists and has type fixed_size_list(float)
@@ -3210,14 +3397,39 @@ def _handle_bad_vector_column(
     fill_value: float, default 0.0
         The value to use when filling vectors. Only used if on_bad_vectors="fill".
     """
+    position = data.column_names.index(vector_column_name)
     vec_arr = data[vector_column_name]
+    if not _is_list_like(vec_arr.type):
+        return data
 
-    has_nan = has_nan_values(vec_arr)
+    if (
+        expected_dim is not None
+        and pa.types.is_fixed_size_list(vec_arr.type)
+        and vec_arr.type.list_size != expected_dim
+    ):
+        vec_arr = pa.array(vec_arr.to_pylist(), type=pa.list_(vec_arr.type.value_type))
+        data = data.set_column(position, vector_column_name, vec_arr)
 
-    if pa.types.is_fixed_size_list(vec_arr.type):
+    if expected_value_type is not None and (
+        pa.types.is_integer(vec_arr.type.value_type)
+        or pa.types.is_unsigned_integer(vec_arr.type.value_type)
+    ):
+        vec_arr = pa.array(vec_arr.to_pylist(), type=pa.list_(expected_value_type))
+        data = data.set_column(position, vector_column_name, vec_arr)
+
+    if pa.types.is_floating(vec_arr.type.value_type):
+        has_nan = has_nan_values(vec_arr)
+    else:
+        has_nan = pa.array([False] * len(vec_arr))
+
+    if expected_dim is not None:
+        dim = expected_dim
+    elif pa.types.is_fixed_size_list(vec_arr.type):
         dim = vec_arr.type.list_size
     else:
-        dim = _modal_list_size(vec_arr)
+        dim = _infer_vector_dim(vec_arr)
+        if dim is None:
+            return data
     has_wrong_dim = pc.not_equal(pc.list_value_length(vec_arr), dim)
 
     has_bad_vectors = pc.any(has_nan).as_py() or pc.any(has_wrong_dim).as_py()
@@ -3255,13 +3467,12 @@ def _handle_bad_vector_column(
                 )
             vec_arr = pc.if_else(
                 is_bad,
-                pa.scalar([fill_value] * dim),
+                pa.scalar([fill_value] * dim, type=vec_arr.type),
                 vec_arr,
             )
         else:
             raise ValueError(f"Invalid value for on_bad_vectors: {on_bad_vectors}")
 
-    position = data.column_names.index(vector_column_name)
     return data.set_column(position, vector_column_name, vec_arr)
 
 
@@ -3282,6 +3493,28 @@ def has_nan_values(arr: Union[pa.ListArray, pa.ChunkedArray]) -> pa.BooleanArray
     return pc.is_in(indices, has_nan_indices)
 
 
+def _is_list_like(data_type: pa.DataType) -> bool:
+    return (
+        pa.types.is_list(data_type)
+        or pa.types.is_large_list(data_type)
+        or pa.types.is_fixed_size_list(data_type)
+    )
+
+
+def _merge_metadata(*metadata_dicts: Optional[dict]) -> dict:
+    merged = {}
+    for metadata in metadata_dicts:
+        if metadata is None:
+            continue
+        for key, value in metadata.items():
+            if isinstance(key, str):
+                key = key.encode("utf-8")
+            if isinstance(value, str):
+                value = value.encode("utf-8")
+            merged[key] = value
+    return merged
+
+
 def _name_suggests_vector_column(field_name: str) -> bool:
     """Check if a field name indicates a vector column."""
     name_lower = field_name.lower()
@@ -3349,6 +3582,16 @@ def _modal_list_size(arr: Union[pa.ListArray, pa.ChunkedArray]) -> int:
     return pc.mode(pc.list_value_length(arr))[0].as_py()["mode"]
 
 
+def _infer_vector_dim(arr: Union[pa.Array, pa.ChunkedArray]) -> Optional[int]:
+    if not _is_list_like(arr.type):
+        return None
+    lengths = pc.list_value_length(arr)
+    lengths = pc.filter(lengths, pc.greater(lengths, 0))
+    if len(lengths) == 0:
+        return None
+    return pc.mode(lengths)[0].as_py()["mode"]
+
+
 def _validate_schema(schema: pa.Schema):
     """
     Make sure the metadata is valid utf8
@@ -3769,6 +4012,7 @@ class AsyncTable:
         mode: Optional[Literal["append", "overwrite"]] = "append",
         on_bad_vectors: Optional[OnBadVectorsType] = None,
         fill_value: Optional[float] = None,
+        progress: Optional[Union[bool, Callable, Any]] = None,
     ) -> AddResult:
         """Add more data to the [Table](Table).
 
@@ -3790,6 +4034,9 @@ class AsyncTable:
             One of "error", "drop", "fill", "null".
         fill_value: float, default 0.
             The value to use when filling vectors. Only used if on_bad_vectors="fill".
+        progress: callable or tqdm-like, optional
+            A callback or tqdm-compatible progress bar. See
+            :meth:`Table.add` for details.
 
         """
         schema = await self.schema()
@@ -3800,7 +4047,13 @@ class AsyncTable:
 
         # _santitize_data is an old code path, but we will use it until the
         # new code path is ready.
-        if on_bad_vectors != "error" or (
+        if mode == "overwrite":
+            # For overwrite, apply the same preprocessing as create_table
+            # so vector columns are inferred as FixedSizeList.
+            data, _ = sanitize_create_table(
+                data, None, on_bad_vectors=on_bad_vectors, fill_value=fill_value
+            )
+        elif on_bad_vectors != "error" or (
             schema.metadata is not None and b"embedding_functions" in schema.metadata
         ):
             data = _sanitize_data(
@@ -3813,8 +4066,9 @@ class AsyncTable:
             )
         _register_optional_converters()
         data = to_scannable(data)
+        progress, owns = _normalize_progress(progress)
         try:
-            return await self._inner.add(data, mode or "append")
+            return await self._inner.add(data, mode or "append", progress=progress)
         except RuntimeError as e:
             if "Cast error" in str(e):
                 raise ValueError(e)
@@ -3822,6 +4076,9 @@ class AsyncTable:
                 raise ValueError(e)
             else:
                 raise
+        finally:
+            if owns:
+                progress.close()
 
     def merge_insert(self, on: Union[str, Iterable[str]]) -> LanceMergeInsertBuilder:
         """
@@ -4144,7 +4401,7 @@ class AsyncTable:
             async_query = async_query.offset(query.offset)
         if query.columns:
             async_query = async_query.select(query.columns)
-        if query.filter:
+        if query.filter is not None:
             async_query = async_query.where(query.filter)
         if query.fast_search:
             async_query = async_query.fast_search()
@@ -4751,7 +5008,16 @@ class IndexStatistics:
     num_indexed_rows: int
     num_unindexed_rows: int
     index_type: Literal[
-        "IVF_PQ", "IVF_HNSW_PQ", "IVF_HNSW_SQ", "FTS", "BTREE", "BITMAP", "LABEL_LIST"
+        "IVF_FLAT",
+        "IVF_SQ",
+        "IVF_PQ",
+        "IVF_RQ",
+        "IVF_HNSW_SQ",
+        "IVF_HNSW_PQ",
+        "FTS",
+        "BTREE",
+        "BITMAP",
+        "LABEL_LIST",
     ]
     distance_type: Optional[Literal["l2", "cosine", "dot"]] = None
     num_indices: Optional[int] = None

python/python/tests/docs/test_search.py

@@ -180,7 +180,7 @@ def test_fts_fuzzy_query():
         ),
         mode="overwrite",
     )
-    table.create_fts_index("text", use_tantivy=False, replace=True)
+    table.create_fts_index("text", replace=True)
 
     results = table.search(MatchQuery("foo", "text", fuzziness=1)).to_pandas()
     assert len(results) == 4
@@ -230,7 +230,7 @@ def test_fts_boost_query():
         ),
         mode="overwrite",
     )
-    table.create_fts_index("desc", use_tantivy=False, replace=True)
+    table.create_fts_index("desc", replace=True)
 
     results = table.search(
         BoostQuery(
@@ -265,7 +265,7 @@ def test_fts_boolean_query(tmp_path):
         ],
         mode="overwrite",
     )
-    table.create_fts_index("text", use_tantivy=False, replace=True)
+    table.create_fts_index("text", replace=True)
 
     # SHOULD
     results = table.search(
@@ -319,9 +319,7 @@ def test_fts_native():
         ],
     )
 
-    # passing `use_tantivy=False` to use lance FTS index
-    # `use_tantivy=True` by default
-    table.create_fts_index("text", use_tantivy=False)
+    table.create_fts_index("text")
     table.search("puppy").limit(10).select(["text"]).to_list()
     # [{'text': 'Frodo was a happy puppy', '_score': 0.6931471824645996}]
     # ...
@@ -332,7 +330,6 @@ def test_fts_native():
     # --8<-- [start:fts_config_folding]
     table.create_fts_index(
         "text",
-        use_tantivy=False,
         language="French",
         stem=True,
         ascii_folding=True,
@@ -346,7 +343,7 @@ def test_fts_native():
     table.search("puppy").limit(10).where("text='foo'", prefilter=False).to_list()
     # --8<-- [end:fts_postfiltering]
     # --8<-- [start:fts_with_position]
-    table.create_fts_index("text", use_tantivy=False, with_position=True, replace=True)
+    table.create_fts_index("text", with_position=True, replace=True)
     # --8<-- [end:fts_with_position]
     # --8<-- [start:fts_incremental_index]
     table.add([{"vector": [3.1, 4.1], "text": "Frodo was a happy puppy"}])

python/python/tests/test_db.py

@@ -3,6 +3,7 @@
 
 
 import re
+import sys
 from datetime import timedelta
 import os
 
@@ -14,8 +15,7 @@ import pytest
 from lancedb.pydantic import LanceModel, Vector
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_basic(tmp_path, use_tantivy):
+def test_basic(tmp_path):
     db = lancedb.connect(tmp_path)
 
     assert db.uri == str(tmp_path)
@@ -48,7 +48,7 @@ def test_basic(tmp_path, use_tantivy):
     assert len(rs) == 1
     assert rs["item"].iloc[0] == "foo"
 
-    table.create_fts_index("item", use_tantivy=use_tantivy)
+    table.create_fts_index("item")
     rs = table.search("bar", query_type="fts").to_pandas()
     assert len(rs) == 1
     assert rs["item"].iloc[0] == "bar"
@@ -183,8 +183,8 @@ def test_table_names(tmp_db: lancedb.DBConnection):
     result = list(tmp_db.table_names("test2", limit=2))
     assert result == ["test3"], f"Expected ['test3'], got {result}"
 
-    # Test that namespace parameter can be passed as keyword
-    result = list(tmp_db.table_names(namespace=[]))
+    # Test that namespace_path parameter can be passed as keyword
+    result = list(tmp_db.table_names(namespace_path=[]))
     assert len(result) == 3
 
 
@@ -896,42 +896,22 @@ def test_bypass_vector_index_sync(tmp_db: lancedb.DBConnection):
 
 
 def test_local_namespace_operations(tmp_path):
-    """Test that local mode namespace operations behave as expected."""
-    # Create a local database connection
+    """Test that local mode namespace operations work via directory namespace."""
     db = lancedb.connect(tmp_path)
 
-    # Test list_namespaces returns empty list for root namespace
-    namespaces = db.list_namespaces().namespaces
-    assert namespaces == []
+    # Root namespace starts empty
+    assert db.list_namespaces().namespaces == []
 
-    # Test list_namespaces with non-empty namespace raises NotImplementedError
-    with pytest.raises(
-        NotImplementedError,
-        match="Namespace operations are not supported for listing database",
-    ):
-        db.list_namespaces(namespace=["test"])
+    # Create and list child namespace
+    db.create_namespace(["child"])
+    assert "child" in db.list_namespaces().namespaces
 
+    # List namespaces under child
+    assert db.list_namespaces(namespace_path=["child"]).namespaces == []
 
-def test_local_create_namespace_not_supported(tmp_path):
-    """Test that create_namespace is not supported in local mode."""
-    db = lancedb.connect(tmp_path)
-
-    with pytest.raises(
-        NotImplementedError,
-        match="Namespace operations are not supported for listing database",
-    ):
-        db.create_namespace(["test_namespace"])
-
-
-def test_local_drop_namespace_not_supported(tmp_path):
-    """Test that drop_namespace is not supported in local mode."""
-    db = lancedb.connect(tmp_path)
-
-    with pytest.raises(
-        NotImplementedError,
-        match="Namespace operations are not supported for listing database",
-    ):
-        db.drop_namespace(["test_namespace"])
+    # Drop namespace
+    db.drop_namespace(["child"])
+    assert db.list_namespaces().namespaces == []
 
 
 def test_clone_table_latest_version(tmp_path):
@@ -1048,3 +1028,59 @@ def test_clone_table_deep_clone_fails(tmp_path):
     source_uri = os.path.join(tmp_path, "source.lance")
     with pytest.raises(Exception, match="Deep clone is not yet implemented"):
         db.clone_table("cloned", source_uri, is_shallow=False)
+
+
+@pytest.mark.skipif(sys.platform == "win32", reason="Namespace client issues")
+def test_namespace_client_native_storage(tmp_path):
+    """Test namespace_client() returns DirectoryNamespace for native storage."""
+    from lance.namespace import DirectoryNamespace
+
+    db = lancedb.connect(tmp_path)
+    ns_client = db.namespace_client()
+
+    assert isinstance(ns_client, DirectoryNamespace)
+    assert str(tmp_path) in ns_client.namespace_id()
+
+
+@pytest.mark.skipif(sys.platform == "win32", reason="Namespace client issues")
+def test_namespace_client_with_storage_options(tmp_path):
+    """Test namespace_client() preserves storage options."""
+    from lance.namespace import DirectoryNamespace
+
+    storage_options = {"timeout": "10s"}
+    db = lancedb.connect(tmp_path, storage_options=storage_options)
+    ns_client = db.namespace_client()
+
+    assert isinstance(ns_client, DirectoryNamespace)
+
+
+@pytest.mark.skipif(sys.platform == "win32", reason="Namespace client issues")
+def test_namespace_client_operations(tmp_path):
+    """Test that namespace_client() returns a functional namespace client."""
+    db = lancedb.connect(tmp_path)
+    ns_client = db.namespace_client()
+
+    # Create a table through the main db connection
+    data = [{"id": 1, "text": "hello", "vector": [1.0, 2.0]}]
+    db.create_table("test_table", data=data)
+
+    # Verify the namespace client can see the table
+    from lance_namespace import ListTablesRequest
+
+    # id=[] means root namespace
+    response = ns_client.list_tables(ListTablesRequest(id=[]))
+    # Tables can be strings or objects with name attribute
+    table_names = [t.name if hasattr(t, "name") else t for t in response.tables]
+    assert "test_table" in table_names
+
+
+@pytest.mark.skipif(sys.platform == "win32", reason="Namespace client issues")
+def test_namespace_client_namespace_connection(tmp_path):
+    """Test namespace_client() returns the backing client for namespace connections."""
+    from lance.namespace import DirectoryNamespace
+
+    db = lancedb.connect_namespace("dir", {"root": str(tmp_path)})
+    ns_client = db.namespace_client()
+
+    assert isinstance(ns_client, DirectoryNamespace)
+    assert str(tmp_path) in ns_client.namespace_id()

python/python/tests/test_embeddings.py

@@ -546,3 +546,24 @@ def test_openai_no_retry_on_401(mock_sleep):
     assert mock_func.call_count == 1
     # Verify that sleep was never called (no retries)
     assert mock_sleep.call_count == 0
+
+
+def test_url_retrieve_downloads_image():
+    """
+    Embedding functions like open-clip, siglip, and jinaai use url_retrieve()
+    to download images from HTTP URLs. For example, open_clip._to_pil() calls:
+
+        PIL_Image.open(io.BytesIO(url_retrieve(image)))
+
+    Verify that url_retrieve() can download an image and open it as PIL Image,
+    matching the real usage pattern in embedding functions.
+    """
+    import io
+
+    Image = pytest.importorskip("PIL.Image")
+    from lancedb.embeddings.utils import url_retrieve
+
+    image_url = "http://farm1.staticflickr.com/53/167798175_7c7845bbbd_z.jpg"
+    image_bytes = url_retrieve(image_url)
+    img = Image.open(io.BytesIO(image_bytes))
+    assert img.size[0] > 0 and img.size[1] > 0

python/python/tests/test_fts.py

@@ -36,9 +36,6 @@ import pytest
 import pytest_asyncio
 from utils import exception_output
 
-pytest.importorskip("lancedb.fts")
-tantivy = pytest.importorskip("tantivy")
-
 
 @pytest.fixture
 def table(tmp_path) -> ldb.table.LanceTable:
@@ -144,58 +141,53 @@ async def async_table(tmp_path) -> ldb.table.AsyncTable:
     return table
 
 
-def test_create_index(tmp_path):
-    index = ldb.fts.create_index(str(tmp_path / "index"), ["text"])
-    assert isinstance(index, tantivy.Index)
-    assert os.path.exists(str(tmp_path / "index"))
+@pytest.mark.parametrize(
+    ("kwargs", "match"),
+    [
+        (
+            {"use_tantivy": True},
+            "Tantivy-based FTS has been removed",
+        ),
+        (
+            {"ordering_field_names": ["count"]},
+            "ordering_field_names was only supported",
+        ),
+        (
+            {"writer_heap_size": 128},
+            "writer_heap_size was only supported",
+        ),
+    ],
+)
+def test_reject_removed_tantivy_parameters(table, kwargs, match):
+    with pytest.raises(ValueError, match=match):
+        table.create_fts_index("text", **kwargs)
 
 
-def test_create_index_with_stemming(tmp_path, table):
-    index = ldb.fts.create_index(
-        str(tmp_path / "index"), ["text"], tokenizer_name="en_stem"
-    )
-    assert isinstance(index, tantivy.Index)
-    assert os.path.exists(str(tmp_path / "index"))
+def test_reject_legacy_tantivy_index(table):
+    path, _, _ = table._get_fts_index_path()
+    os.makedirs(path, exist_ok=True)
 
-    # Check stemming by running tokenizer on non empty table
-    table.create_fts_index("text", tokenizer_name="en_stem", use_tantivy=True)
+    with pytest.raises(ValueError, match="Legacy Tantivy FTS index detected"):
+        table.search("puppy").limit(5).to_list()
+
+    with pytest.raises(ValueError, match="Legacy Tantivy FTS index detected"):
+        table.create_fts_index("text")
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
 @pytest.mark.parametrize("with_position", [True, False])
-def test_create_inverted_index(table, use_tantivy, with_position):
-    if use_tantivy and not with_position:
-        pytest.skip("we don't support building a tantivy index without position")
+def test_create_inverted_index(table, with_position):
     table.create_fts_index(
         "text",
-        use_tantivy=use_tantivy,
         with_position=with_position,
         name="custom_fts_index",
     )
-    if not use_tantivy:
-        indices = table.list_indices()
-        fts_indices = [i for i in indices if i.index_type == "FTS"]
-        assert any(i.name == "custom_fts_index" for i in fts_indices)
+    indices = table.list_indices()
+    fts_indices = [i for i in indices if i.index_type == "FTS"]
+    assert any(i.name == "custom_fts_index" for i in fts_indices)
 
 
-def test_populate_index(tmp_path, table):
-    index = ldb.fts.create_index(str(tmp_path / "index"), ["text"])
-    assert ldb.fts.populate_index(index, table, ["text"]) == len(table)
-
-
-def test_search_index(tmp_path, table):
-    index = ldb.fts.create_index(str(tmp_path / "index"), ["text"])
-    ldb.fts.populate_index(index, table, ["text"])
-    index.reload()
-    results = ldb.fts.search_index(index, query="puppy", limit=5)
-    assert len(results) == 2
-    assert len(results[0]) == 5  # row_ids
-    assert len(results[1]) == 5  # _score
-
-
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_search_fts(table, use_tantivy):
-    table.create_fts_index("text", use_tantivy=use_tantivy)
+def test_search_fts(table):
+    table.create_fts_index("text")
     results = table.search("puppy").select(["id", "text"]).limit(5).to_list()
     assert len(results) == 5
     assert len(results[0]) == 3  # id, text, _score
@@ -204,53 +196,52 @@ def test_search_fts(table, use_tantivy):
     results = table.search("puppy").select(["id", "text"]).to_list()
     assert len(results) == 10
 
-    if not use_tantivy:
-        # Test with a query
-        results = (
-            table.search(MatchQuery("puppy", "text"))
-            .select(["id", "text"])
-            .limit(5)
-            .to_list()
-        )
-        assert len(results) == 5
+    # Test with a query
+    results = (
+        table.search(MatchQuery("puppy", "text"))
+        .select(["id", "text"])
+        .limit(5)
+        .to_list()
+    )
+    assert len(results) == 5
 
-        # Test boost query
-        results = (
-            table.search(
-                BoostQuery(
-                    MatchQuery("puppy", "text"),
-                    MatchQuery("runs", "text"),
-                )
+    # Test boost query
+    results = (
+        table.search(
+            BoostQuery(
+                MatchQuery("puppy", "text"),
+                MatchQuery("runs", "text"),
             )
-            .select(["id", "text"])
-            .limit(5)
-            .to_list()
         )
-        assert len(results) == 5
+        .select(["id", "text"])
+        .limit(5)
+        .to_list()
+    )
+    assert len(results) == 5
 
-        # Test multi match query
-        table.create_fts_index("text2", use_tantivy=use_tantivy)
-        results = (
-            table.search(MultiMatchQuery("puppy", ["text", "text2"]))
-            .select(["id", "text"])
-            .limit(5)
-            .to_list()
-        )
-        assert len(results) == 5
-        assert len(results[0]) == 3  # id, text, _score
+    # Test multi match query
+    table.create_fts_index("text2")
+    results = (
+        table.search(MultiMatchQuery("puppy", ["text", "text2"]))
+        .select(["id", "text"])
+        .limit(5)
+        .to_list()
+    )
+    assert len(results) == 5
+    assert len(results[0]) == 3  # id, text, _score
 
-        # Test boolean query
-        results = (
-            table.search(MatchQuery("puppy", "text") & MatchQuery("runs", "text"))
-            .select(["id", "text"])
-            .limit(5)
-            .to_list()
-        )
-        assert len(results) == 5
-        assert len(results[0]) == 3  # id, text, _score
-        for r in results:
-            assert "puppy" in r["text"]
-            assert "runs" in r["text"]
+    # Test boolean query
+    results = (
+        table.search(MatchQuery("puppy", "text") & MatchQuery("runs", "text"))
+        .select(["id", "text"])
+        .limit(5)
+        .to_list()
+    )
+    assert len(results) == 5
+    assert len(results[0]) == 3  # id, text, _score
+    for r in results:
+        assert "puppy" in r["text"]
+        assert "runs" in r["text"]
 
 
 @pytest.mark.asyncio
@@ -318,13 +309,13 @@ async def test_fts_select_async(async_table):
 
 
 def test_search_fts_phrase_query(table):
-    table.create_fts_index("text", use_tantivy=False, with_position=False)
+    table.create_fts_index("text", with_position=False)
     try:
         phrase_results = table.search('"puppy runs"').limit(100).to_list()
         assert False
     except Exception:
         pass
-    table.create_fts_index("text", use_tantivy=False, with_position=True, replace=True)
+    table.create_fts_index("text", with_position=True, replace=True)
     results = table.search("puppy").limit(100).to_list()
 
     # Test with quotation marks
@@ -375,8 +366,8 @@ async def test_search_fts_phrase_query_async(async_table):
 
 
 def test_search_fts_specify_column(table):
-    table.create_fts_index("text", use_tantivy=False)
-    table.create_fts_index("text2", use_tantivy=False)
+    table.create_fts_index("text")
+    table.create_fts_index("text2")
 
     results = table.search("puppy", fts_columns="text").limit(5).to_list()
     assert len(results) == 5
@@ -470,42 +461,8 @@ async def test_search_fts_specify_column_async(async_table):
         pass
 
 
-def test_search_ordering_field_index_table(tmp_path, table):
-    table.create_fts_index("text", ordering_field_names=["count"], use_tantivy=True)
-    rows = (
-        table.search("puppy", ordering_field_name="count")
-        .limit(20)
-        .select(["text", "count"])
-        .to_list()
-    )
-    for r in rows:
-        assert "puppy" in r["text"]
-    assert sorted(rows, key=lambda x: x["count"], reverse=True) == rows
-
-
-def test_search_ordering_field_index(tmp_path, table):
-    index = ldb.fts.create_index(
-        str(tmp_path / "index"), ["text"], ordering_fields=["count"]
-    )
-
-    ldb.fts.populate_index(index, table, ["text"], ordering_fields=["count"])
-    index.reload()
-    results = ldb.fts.search_index(
-        index, query="puppy", limit=5, ordering_field="count"
-    )
-    assert len(results) == 2
-    assert len(results[0]) == 5  # row_ids
-    assert len(results[1]) == 5  # _distance
-    rows = table.to_lance().take(results[0]).to_pylist()
-
-    for r in rows:
-        assert "puppy" in r["text"]
-    assert sorted(rows, key=lambda x: x["count"], reverse=True) == rows
-
-
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_create_index_from_table(tmp_path, table, use_tantivy):
-    table.create_fts_index("text", use_tantivy=use_tantivy)
+def test_create_index_from_table(tmp_path, table):
+    table.create_fts_index("text")
     df = table.search("puppy").limit(5).select(["text"]).to_pandas()
     assert len(df) <= 5
     assert "text" in df.columns
@@ -525,36 +482,24 @@ def test_create_index_from_table(tmp_path, table, use_tantivy):
     )
 
     with pytest.raises(Exception, match="already exists"):
-        table.create_fts_index("text", use_tantivy=use_tantivy)
+        table.create_fts_index("text")
 
-    table.create_fts_index("text", replace=True, use_tantivy=use_tantivy)
+    table.create_fts_index("text", replace=True)
     assert len(table.search("gorilla").limit(1).to_pandas()) == 1
 
 
 def test_create_index_multiple_columns(tmp_path, table):
-    table.create_fts_index(["text", "text2"], use_tantivy=True)
-    df = table.search("puppy").limit(5).to_pandas()
-    assert len(df) == 5
-    assert "text" in df.columns
-    assert "text2" in df.columns
-
-
-def test_empty_rs(tmp_path, table, mocker):
-    table.create_fts_index(["text", "text2"], use_tantivy=True)
-    mocker.patch("lancedb.fts.search_index", return_value=([], []))
-    df = table.search("puppy").limit(5).to_pandas()
-    assert len(df) == 0
+    with pytest.raises(ValueError, match="Native FTS indexes can only be created"):
+        table.create_fts_index(["text", "text2"])
 
 
 def test_nested_schema(tmp_path, table):
-    table.create_fts_index("nested.text", use_tantivy=True)
-    rs = table.search("puppy").limit(5).to_list()
-    assert len(rs) == 5
+    with pytest.raises(ValueError, match="top-level fields"):
+        table.create_fts_index("nested.text")
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_search_index_with_filter(table, use_tantivy):
-    table.create_fts_index("text", use_tantivy=use_tantivy)
+def test_search_index_with_filter(table):
+    table.create_fts_index("text")
     orig_import = __import__
 
     def import_mock(name, *args):
@@ -584,8 +529,7 @@ def test_search_index_with_filter(table, use_tantivy):
         assert r["_rowid"] is not None
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_null_input(table, use_tantivy):
+def test_null_input(table):
     table.add(
         [
             {
@@ -598,14 +542,13 @@ def test_null_input(table, use_tantivy):
             }
         ]
     )
-    table.create_fts_index("text", use_tantivy=use_tantivy)
+    table.create_fts_index("text")
 
 
 def test_syntax(table):
     # https://github.com/lancedb/lancedb/issues/769
-    table.create_fts_index("text", use_tantivy=True)
-    with pytest.raises(ValueError, match="Syntax Error"):
-        table.search("they could have been dogs OR").limit(10).to_list()
+    table.create_fts_index("text")
+    table.search("they could have been dogs OR").limit(10).to_list()
 
     # these should work
 
@@ -616,6 +559,7 @@ def test_syntax(table):
     ).to_list()
 
     # phrase queries
+    table.create_fts_index("text", with_position=True, replace=True)
     table.search("they could have been dogs OR cats").phrase_query().limit(10).to_list()
     table.search('"they could have been dogs OR cats"').limit(10).to_list()
     table.search('''"the cats OR dogs were not really 'pets' at all"''').limit(
@@ -639,7 +583,7 @@ def test_language(mem_db: DBConnection):
     table = mem_db.create_table("test", data=data)
 
     with pytest.raises(ValueError) as e:
-        table.create_fts_index("text", use_tantivy=False, language="klingon")
+        table.create_fts_index("text", language="klingon")
 
     assert exception_output(e) == (
         "ValueError: LanceDB does not support the requested language: 'klingon'\n"
@@ -650,7 +594,6 @@ def test_language(mem_db: DBConnection):
 
     table.create_fts_index(
         "text",
-        use_tantivy=False,
         language="French",
         stem=True,
         ascii_folding=True,
@@ -690,7 +633,7 @@ def test_fts_on_list(mem_db: DBConnection):
         }
     )
     table = mem_db.create_table("test", data=data)
-    table.create_fts_index("text", use_tantivy=False, with_position=True)
+    table.create_fts_index("text", with_position=True)
 
     res = table.search("lance").limit(5).to_list()
     assert len(res) == 3
@@ -702,7 +645,7 @@ def test_fts_on_list(mem_db: DBConnection):
 def test_fts_ngram(mem_db: DBConnection):
     data = pa.table({"text": ["hello world", "lance database", "lance is cool"]})
     table = mem_db.create_table("test", data=data)
-    table.create_fts_index("text", use_tantivy=False, base_tokenizer="ngram")
+    table.create_fts_index("text", base_tokenizer="ngram")
 
     results = table.search("lan", query_type="fts").limit(10).to_list()
     assert len(results) == 2
@@ -721,7 +664,6 @@ def test_fts_ngram(mem_db: DBConnection):
     # test setting min_ngram_length and prefix_only
     table.create_fts_index(
         "text",
-        use_tantivy=False,
         base_tokenizer="ngram",
         replace=True,
         ngram_min_length=2,
@@ -886,7 +828,7 @@ def test_fts_query_to_json():
 
 
 def test_fts_fast_search(table):
-    table.create_fts_index("text", use_tantivy=False)
+    table.create_fts_index("text")
 
     # Insert some unindexed data
     table.add(

python/python/tests/test_hybrid_query.py

@@ -28,7 +28,7 @@ def sync_table(tmpdir_factory) -> Table:
         }
     )
     table = db.create_table("test", data)
-    table.create_fts_index("text", with_position=False, use_tantivy=False)
+    table.create_fts_index("text", with_position=False)
     return table
 
 
@@ -177,6 +177,60 @@ async def test_analyze_plan(table: AsyncTable):
     assert "metrics=" in res
 
 
+@pytest.fixture
+def table_with_id(tmpdir_factory) -> Table:
+    tmp_path = str(tmpdir_factory.mktemp("data"))
+    db = lancedb.connect(tmp_path)
+    data = pa.table(
+        {
+            "id": pa.array([1, 2, 3, 4], type=pa.int64()),
+            "text": pa.array(["a", "b", "cat", "dog"]),
+            "vector": pa.array(
+                [[0.1, 0.1], [2, 2], [-0.1, -0.1], [0.5, -0.5]],
+                type=pa.list_(pa.float32(), list_size=2),
+            ),
+        }
+    )
+    table = db.create_table("test_with_id", data)
+    table.create_fts_index("text", with_position=False)
+    return table
+
+
+def test_hybrid_prefilter_explain_plan(table_with_id: Table):
+    """
+    Verify that the prefilter logic is not inverted in LanceHybridQueryBuilder.
+    """
+    plan_prefilter = (
+        table_with_id.search(query_type="hybrid")
+        .vector([0.0, 0.0])
+        .text("dog")
+        .where("id = 1", prefilter=True)
+        .limit(2)
+        .explain_plan(verbose=True)
+    )
+
+    plan_postfilter = (
+        table_with_id.search(query_type="hybrid")
+        .vector([0.0, 0.0])
+        .text("dog")
+        .where("id = 1", prefilter=False)
+        .limit(2)
+        .explain_plan(verbose=True)
+    )
+
+    # prefilter=True: filter is pushed into the LanceRead scan.
+    # The FTS sub-plan exposes this as "full_filter=id = Int64(1)" inside LanceRead.
+    assert "full_filter=id = Int64(1)" in plan_prefilter, (
+        f"Should push the filter into the scan.\nPlan:\n{plan_prefilter}"
+    )
+
+    # prefilter=False: filter is applied as a separate FilterExec after the search.
+    # The filter must NOT be embedded in the scan.
+    assert "full_filter=id = Int64(1)" not in plan_postfilter, (
+        f"Should NOT push the filter into the scan.\nPlan:\n{plan_postfilter}"
+    )
+
+
 def test_normalize_scores():
     cases = [
         (pa.array([0.1, 0.4]), pa.array([0.0, 1.0])),

python/python/tests/test_index.py

@@ -3,6 +3,7 @@
 
 from datetime import timedelta
 import random
+from typing import get_args, get_type_hints
 
 import pyarrow as pa
 import pytest
@@ -22,6 +23,7 @@ from lancedb.index import (
     HnswSq,
     FTS,
 )
+from lancedb.table import IndexStatistics
 
 
 @pytest_asyncio.fixture
@@ -283,3 +285,23 @@ async def test_create_index_with_binary_vectors(binary_table: AsyncTable):
     for v in range(256):
         res = await binary_table.query().nearest_to([v] * 128).to_arrow()
         assert res["id"][0].as_py() == v
+
+
+def test_index_statistics_index_type_lists_all_supported_values():
+    expected_index_types = {
+        "IVF_FLAT",
+        "IVF_SQ",
+        "IVF_PQ",
+        "IVF_RQ",
+        "IVF_HNSW_SQ",
+        "IVF_HNSW_PQ",
+        "FTS",
+        "BTREE",
+        "BITMAP",
+        "LABEL_LIST",
+    }
+
+    assert (
+        set(get_args(get_type_hints(IndexStatistics)["index_type"]))
+        == expected_index_types
+    )

python/python/tests/test_namespace.py

@@ -8,6 +8,7 @@ import shutil
 import pytest
 import pyarrow as pa
 import lancedb
+from lance_namespace.errors import NamespaceNotEmptyError, TableNotFoundError
 
 
 class TestNamespaceConnection:
@@ -32,6 +33,16 @@ class TestNamespaceConnection:
         # Initially no tables in root
         assert len(list(db.table_names())) == 0
 
+    def test_connect_via_connect_helper(self):
+        """Connecting via lancedb.connect should delegate to namespace connection."""
+        db = lancedb.connect(
+            namespace_client_impl="dir",
+            namespace_client_properties={"root": self.temp_dir},
+        )
+
+        assert isinstance(db, lancedb.LanceNamespaceDBConnection)
+        assert len(list(db.table_names())) == 0
+
     def test_create_table_through_namespace(self):
         """Test creating a table through namespace."""
         db = lancedb.connect_namespace("dir", {"root": self.temp_dir})
@@ -49,14 +60,14 @@ class TestNamespaceConnection:
         )
 
         # Create empty table in child namespace
-        table = db.create_table("test_table", schema=schema, namespace=["test_ns"])
+        table = db.create_table("test_table", schema=schema, namespace_path=["test_ns"])
         assert table is not None
         assert table.name == "test_table"
         assert table.namespace == ["test_ns"]
         assert table.id == "test_ns$test_table"
 
         # Table should appear in child namespace
-        table_names = list(db.table_names(namespace=["test_ns"]))
+        table_names = list(db.table_names(namespace_path=["test_ns"]))
         assert "test_table" in table_names
         assert len(table_names) == 1
 
@@ -79,10 +90,10 @@ class TestNamespaceConnection:
                 pa.field("vector", pa.list_(pa.float32(), 2)),
             ]
         )
-        db.create_table("test_table", schema=schema, namespace=["test_ns"])
+        db.create_table("test_table", schema=schema, namespace_path=["test_ns"])
 
         # Open the table
-        table = db.open_table("test_table", namespace=["test_ns"])
+        table = db.open_table("test_table", namespace_path=["test_ns"])
         assert table is not None
         assert table.name == "test_table"
         assert table.namespace == ["test_ns"]
@@ -107,31 +118,31 @@ class TestNamespaceConnection:
                 pa.field("vector", pa.list_(pa.float32(), 2)),
             ]
         )
-        db.create_table("table1", schema=schema, namespace=["test_ns"])
-        db.create_table("table2", schema=schema, namespace=["test_ns"])
+        db.create_table("table1", schema=schema, namespace_path=["test_ns"])
+        db.create_table("table2", schema=schema, namespace_path=["test_ns"])
 
         # Verify both tables exist in child namespace
-        table_names = list(db.table_names(namespace=["test_ns"]))
+        table_names = list(db.table_names(namespace_path=["test_ns"]))
         assert "table1" in table_names
         assert "table2" in table_names
         assert len(table_names) == 2
 
         # Drop one table
-        db.drop_table("table1", namespace=["test_ns"])
+        db.drop_table("table1", namespace_path=["test_ns"])
 
         # Verify only table2 remains
-        table_names = list(db.table_names(namespace=["test_ns"]))
+        table_names = list(db.table_names(namespace_path=["test_ns"]))
         assert "table1" not in table_names
         assert "table2" in table_names
         assert len(table_names) == 1
 
         # Drop the second table
-        db.drop_table("table2", namespace=["test_ns"])
-        assert len(list(db.table_names(namespace=["test_ns"]))) == 0
+        db.drop_table("table2", namespace_path=["test_ns"])
+        assert len(list(db.table_names(namespace_path=["test_ns"]))) == 0
 
         # Should not be able to open dropped table
-        with pytest.raises(RuntimeError):
-            db.open_table("table1", namespace=["test_ns"])
+        with pytest.raises(TableNotFoundError):
+            db.open_table("table1", namespace_path=["test_ns"])
 
     def test_create_table_with_schema(self):
         """Test creating a table with explicit schema through namespace."""
@@ -150,7 +161,7 @@ class TestNamespaceConnection:
         )
 
         # Create table with schema in child namespace
-        table = db.create_table("test_table", schema=schema, namespace=["test_ns"])
+        table = db.create_table("test_table", schema=schema, namespace_path=["test_ns"])
         assert table is not None
         assert table.namespace == ["test_ns"]
 
@@ -174,7 +185,7 @@ class TestNamespaceConnection:
                 pa.field("vector", pa.list_(pa.float32(), 2)),
             ]
         )
-        db.create_table("old_name", schema=schema, namespace=["test_ns"])
+        db.create_table("old_name", schema=schema, namespace_path=["test_ns"])
 
         # Rename should raise NotImplementedError
         with pytest.raises(NotImplementedError, match="rename_table is not supported"):
@@ -195,20 +206,20 @@ class TestNamespaceConnection:
             ]
         )
         for i in range(3):
-            db.create_table(f"table{i}", schema=schema, namespace=["test_ns"])
+            db.create_table(f"table{i}", schema=schema, namespace_path=["test_ns"])
 
         # Verify tables exist in child namespace
-        assert len(list(db.table_names(namespace=["test_ns"]))) == 3
+        assert len(list(db.table_names(namespace_path=["test_ns"]))) == 3
 
         # Drop all tables in child namespace
-        db.drop_all_tables(namespace=["test_ns"])
+        db.drop_all_tables(namespace_path=["test_ns"])
 
         # Verify all tables are gone from child namespace
-        assert len(list(db.table_names(namespace=["test_ns"]))) == 0
+        assert len(list(db.table_names(namespace_path=["test_ns"]))) == 0
 
         # Test that table_names works with keyword-only namespace parameter
-        db.create_table("test_table", schema=schema, namespace=["test_ns"])
-        result = list(db.table_names(namespace=["test_ns"]))
+        db.create_table("test_table", schema=schema, namespace_path=["test_ns"])
+        result = list(db.table_names(namespace_path=["test_ns"]))
         assert "test_table" in result
 
     def test_table_operations(self):
@@ -226,7 +237,7 @@ class TestNamespaceConnection:
                 pa.field("text", pa.string()),
             ]
         )
-        table = db.create_table("test_table", schema=schema, namespace=["test_ns"])
+        table = db.create_table("test_table", schema=schema, namespace_path=["test_ns"])
 
         # Verify empty table was created
         result = table.to_pandas()
@@ -297,25 +308,25 @@ class TestNamespaceConnection:
             ]
         )
         table = db.create_table(
-            "test_table", schema=schema, namespace=["test_namespace"]
+            "test_table", schema=schema, namespace_path=["test_namespace"]
         )
         assert table is not None
 
         # Verify table exists in namespace
-        tables_in_namespace = list(db.table_names(namespace=["test_namespace"]))
+        tables_in_namespace = list(db.table_names(namespace_path=["test_namespace"]))
         assert "test_table" in tables_in_namespace
         assert len(tables_in_namespace) == 1
 
         # Open table from namespace
-        table = db.open_table("test_table", namespace=["test_namespace"])
+        table = db.open_table("test_table", namespace_path=["test_namespace"])
         assert table is not None
         assert table.name == "test_table"
 
         # Drop table from namespace
-        db.drop_table("test_table", namespace=["test_namespace"])
+        db.drop_table("test_table", namespace_path=["test_namespace"])
 
         # Verify table no longer exists in namespace
-        tables_in_namespace = list(db.table_names(namespace=["test_namespace"]))
+        tables_in_namespace = list(db.table_names(namespace_path=["test_namespace"]))
         assert len(tables_in_namespace) == 0
 
         # Drop namespace
@@ -337,14 +348,14 @@ class TestNamespaceConnection:
                 pa.field("vector", pa.list_(pa.float32(), 2)),
             ]
         )
-        db.create_table("test_table", schema=schema, namespace=["test_namespace"])
+        db.create_table("test_table", schema=schema, namespace_path=["test_namespace"])
 
         # Try to drop namespace with tables - should fail
-        with pytest.raises(RuntimeError, match="is not empty"):
+        with pytest.raises(NamespaceNotEmptyError):
             db.drop_namespace(["test_namespace"])
 
         # Drop table first
-        db.drop_table("test_table", namespace=["test_namespace"])
+        db.drop_table("test_table", namespace_path=["test_namespace"])
 
         # Now dropping namespace should work
         db.drop_namespace(["test_namespace"])
@@ -367,10 +378,10 @@ class TestNamespaceConnection:
 
         # Create table with same name in both namespaces
         table_a = db.create_table(
-            "same_name_table", schema=schema, namespace=["namespace_a"]
+            "same_name_table", schema=schema, namespace_path=["namespace_a"]
         )
         table_b = db.create_table(
-            "same_name_table", schema=schema, namespace=["namespace_b"]
+            "same_name_table", schema=schema, namespace_path=["namespace_b"]
         )
 
         # Add different data to each table
@@ -388,7 +399,9 @@ class TestNamespaceConnection:
         table_b.add(data_b)
 
         # Verify data in namespace_a table
-        opened_table_a = db.open_table("same_name_table", namespace=["namespace_a"])
+        opened_table_a = db.open_table(
+            "same_name_table", namespace_path=["namespace_a"]
+        )
         result_a = opened_table_a.to_pandas().sort_values("id").reset_index(drop=True)
         assert len(result_a) == 2
         assert result_a["id"].tolist() == [1, 2]
@@ -399,7 +412,9 @@ class TestNamespaceConnection:
         assert [v.tolist() for v in result_a["vector"]] == [[1.0, 2.0], [3.0, 4.0]]
 
         # Verify data in namespace_b table
-        opened_table_b = db.open_table("same_name_table", namespace=["namespace_b"])
+        opened_table_b = db.open_table(
+            "same_name_table", namespace_path=["namespace_b"]
+        )
         result_b = opened_table_b.to_pandas().sort_values("id").reset_index(drop=True)
         assert len(result_b) == 3
         assert result_b["id"].tolist() == [10, 20, 30]
@@ -419,8 +434,8 @@ class TestNamespaceConnection:
         assert "same_name_table" not in root_tables
 
         # Clean up
-        db.drop_table("same_name_table", namespace=["namespace_a"])
-        db.drop_table("same_name_table", namespace=["namespace_b"])
+        db.drop_table("same_name_table", namespace_path=["namespace_a"])
+        db.drop_table("same_name_table", namespace_path=["namespace_b"])
         db.drop_namespace(["namespace_a"])
         db.drop_namespace(["namespace_b"])
 
@@ -448,6 +463,8 @@ class TestAsyncNamespaceConnection:
         table_names = await db.table_names()
         assert len(list(table_names)) == 0
 
+    # Async connect via namespace helper is not enabled yet.
+
     async def test_create_table_async(self):
         """Test creating a table asynchronously through namespace."""
         db = lancedb.connect_namespace_async("dir", {"root": self.temp_dir})
@@ -466,13 +483,13 @@ class TestAsyncNamespaceConnection:
 
         # Create empty table in child namespace
         table = await db.create_table(
-            "test_table", schema=schema, namespace=["test_ns"]
+            "test_table", schema=schema, namespace_path=["test_ns"]
         )
         assert table is not None
         assert isinstance(table, lancedb.AsyncTable)
 
         # Table should appear in child namespace
-        table_names = await db.table_names(namespace=["test_ns"])
+        table_names = await db.table_names(namespace_path=["test_ns"])
         assert "test_table" in list(table_names)
 
     async def test_open_table_async(self):
@@ -489,10 +506,10 @@ class TestAsyncNamespaceConnection:
                 pa.field("vector", pa.list_(pa.float32(), 2)),
             ]
         )
-        await db.create_table("test_table", schema=schema, namespace=["test_ns"])
+        await db.create_table("test_table", schema=schema, namespace_path=["test_ns"])
 
         # Open the table
-        table = await db.open_table("test_table", namespace=["test_ns"])
+        table = await db.open_table("test_table", namespace_path=["test_ns"])
         assert table is not None
         assert isinstance(table, lancedb.AsyncTable)
 
@@ -546,20 +563,20 @@ class TestAsyncNamespaceConnection:
                 pa.field("vector", pa.list_(pa.float32(), 2)),
             ]
         )
-        await db.create_table("table1", schema=schema, namespace=["test_ns"])
-        await db.create_table("table2", schema=schema, namespace=["test_ns"])
+        await db.create_table("table1", schema=schema, namespace_path=["test_ns"])
+        await db.create_table("table2", schema=schema, namespace_path=["test_ns"])
 
         # Verify both tables exist in child namespace
-        table_names = list(await db.table_names(namespace=["test_ns"]))
+        table_names = list(await db.table_names(namespace_path=["test_ns"]))
         assert "table1" in table_names
         assert "table2" in table_names
         assert len(table_names) == 2
 
         # Drop one table
-        await db.drop_table("table1", namespace=["test_ns"])
+        await db.drop_table("table1", namespace_path=["test_ns"])
 
         # Verify only table2 remains
-        table_names = list(await db.table_names(namespace=["test_ns"]))
+        table_names = list(await db.table_names(namespace_path=["test_ns"]))
         assert "table1" not in table_names
         assert "table2" in table_names
         assert len(table_names) == 1
@@ -588,20 +605,24 @@ class TestAsyncNamespaceConnection:
             ]
         )
         table = await db.create_table(
-            "test_table", schema=schema, namespace=["test_namespace"]
+            "test_table", schema=schema, namespace_path=["test_namespace"]
         )
         assert table is not None
 
         # Verify table exists in namespace
-        tables_in_namespace = list(await db.table_names(namespace=["test_namespace"]))
+        tables_in_namespace = list(
+            await db.table_names(namespace_path=["test_namespace"])
+        )
         assert "test_table" in tables_in_namespace
         assert len(tables_in_namespace) == 1
 
         # Drop table from namespace
-        await db.drop_table("test_table", namespace=["test_namespace"])
+        await db.drop_table("test_table", namespace_path=["test_namespace"])
 
         # Verify table no longer exists in namespace
-        tables_in_namespace = list(await db.table_names(namespace=["test_namespace"]))
+        tables_in_namespace = list(
+            await db.table_names(namespace_path=["test_namespace"])
+        )
         assert len(tables_in_namespace) == 0
 
         # Drop namespace
@@ -626,15 +647,98 @@ class TestAsyncNamespaceConnection:
             ]
         )
         for i in range(3):
-            await db.create_table(f"table{i}", schema=schema, namespace=["test_ns"])
+            await db.create_table(
+                f"table{i}", schema=schema, namespace_path=["test_ns"]
+            )
 
         # Verify tables exist in child namespace
-        table_names = await db.table_names(namespace=["test_ns"])
+        table_names = await db.table_names(namespace_path=["test_ns"])
         assert len(list(table_names)) == 3
 
         # Drop all tables in child namespace
-        await db.drop_all_tables(namespace=["test_ns"])
+        await db.drop_all_tables(namespace_path=["test_ns"])
 
         # Verify all tables are gone from child namespace
-        table_names = await db.table_names(namespace=["test_ns"])
+        table_names = await db.table_names(namespace_path=["test_ns"])
         assert len(list(table_names)) == 0
+
+
+class TestPushdownOperations:
+    """Test pushdown operations on namespace connections."""
+
+    def setup_method(self):
+        """Set up test fixtures."""
+        self.temp_dir = tempfile.mkdtemp()
+
+    def teardown_method(self):
+        """Clean up test fixtures."""
+        shutil.rmtree(self.temp_dir, ignore_errors=True)
+
+    def test_query_table_pushdown_stored(self):
+        """Test that QueryTable pushdown is stored on sync connection."""
+        db = lancedb.connect_namespace(
+            "dir",
+            {"root": self.temp_dir},
+            namespace_client_pushdown_operations=["QueryTable"],
+        )
+        assert "QueryTable" in db._namespace_client_pushdown_operations
+
+    def test_create_table_pushdown_stored(self):
+        """Test that CreateTable pushdown is stored on sync connection."""
+        db = lancedb.connect_namespace(
+            "dir",
+            {"root": self.temp_dir},
+            namespace_client_pushdown_operations=["CreateTable"],
+        )
+        assert "CreateTable" in db._namespace_client_pushdown_operations
+
+    def test_both_pushdowns_stored(self):
+        """Test that both pushdown operations can be set together."""
+        db = lancedb.connect_namespace(
+            "dir",
+            {"root": self.temp_dir},
+            namespace_client_pushdown_operations=["QueryTable", "CreateTable"],
+        )
+        assert "QueryTable" in db._namespace_client_pushdown_operations
+        assert "CreateTable" in db._namespace_client_pushdown_operations
+
+    def test_pushdown_defaults_to_empty(self):
+        """Test that pushdown operations default to empty."""
+        db = lancedb.connect_namespace("dir", {"root": self.temp_dir})
+        assert len(db._namespace_client_pushdown_operations) == 0
+
+
+@pytest.mark.asyncio
+class TestAsyncPushdownOperations:
+    """Test pushdown operations on async namespace connections."""
+
+    def setup_method(self):
+        """Set up test fixtures."""
+        self.temp_dir = tempfile.mkdtemp()
+
+    def teardown_method(self):
+        """Clean up test fixtures."""
+        shutil.rmtree(self.temp_dir, ignore_errors=True)
+
+    async def test_async_query_table_pushdown_stored(self):
+        """Test that QueryTable pushdown is stored on async connection."""
+        db = lancedb.connect_namespace_async(
+            "dir",
+            {"root": self.temp_dir},
+            namespace_client_pushdown_operations=["QueryTable"],
+        )
+        assert "QueryTable" in db._namespace_client_pushdown_operations
+
+    async def test_async_create_table_pushdown_stored(self):
+        """Test that CreateTable pushdown is stored on async connection."""
+        db = lancedb.connect_namespace_async(
+            "dir",
+            {"root": self.temp_dir},
+            namespace_client_pushdown_operations=["CreateTable"],
+        )
+        assert "CreateTable" in db._namespace_client_pushdown_operations
+
+    async def test_async_pushdown_defaults_to_empty(self):
+        """Test that pushdown operations default to empty on async connection."""
+        db = lancedb.connect_namespace_async("dir", {"root": self.temp_dir})
+        assert len(db._namespace_client_pushdown_operations) == 0

python/python/tests/test_namespace_integration.py

@@ -4,9 +4,11 @@
 """
 Integration tests for LanceDB Namespace with S3 and credential refresh.
 
-This test simulates a namespace server that returns incrementing credentials
-and verifies that the credential refresh mechanism works correctly for both
-create_table and open_table operations.
+This test uses DirectoryNamespace with native ops_metrics and vend_input_storage_options
+features to track API calls and test credential refresh mechanisms.
+
+Tests are parameterized to run with both DirectoryNamespace and a CustomNamespace
+wrapper to verify Python-Rust binding works correctly for custom implementations.
 
 Tests verify:
 - Storage options provider is auto-created and used
@@ -16,24 +18,141 @@ Tests verify:
 """
 
 import copy
+import shutil
+import sys
+import tempfile
 import time
 import uuid
-from threading import Lock
-from typing import Dict
+from typing import Dict, Optional
 
 import pyarrow as pa
 import pytest
-from lance_namespace import (
-    CreateEmptyTableRequest,
-    CreateEmptyTableResponse,
+from lance.namespace import (
     DeclareTableRequest,
     DeclareTableResponse,
     DescribeTableRequest,
     DescribeTableResponse,
+    DirectoryNamespace,
     LanceNamespace,
 )
+from lance_namespace import (
+    CreateNamespaceRequest,
+    CreateNamespaceResponse,
+    CreateTableRequest,
+    CreateTableResponse,
+    CreateTableVersionRequest,
+    CreateTableVersionResponse,
+    DeregisterTableRequest,
+    DeregisterTableResponse,
+    DescribeNamespaceRequest,
+    DescribeNamespaceResponse,
+    DescribeTableVersionRequest,
+    DescribeTableVersionResponse,
+    DropNamespaceRequest,
+    DropNamespaceResponse,
+    DropTableRequest,
+    DropTableResponse,
+    ListNamespacesRequest,
+    ListNamespacesResponse,
+    ListTablesRequest,
+    ListTablesResponse,
+    ListTableVersionsRequest,
+    ListTableVersionsResponse,
+    NamespaceExistsRequest,
+    RegisterTableRequest,
+    RegisterTableResponse,
+    TableExistsRequest,
+)
 from lancedb.namespace import LanceNamespaceDBConnection
 
+
+class CustomNamespace(LanceNamespace):
+    """A custom namespace wrapper that delegates to DirectoryNamespace.
+
+    This class verifies that the Python-Rust binding works correctly for
+    custom namespace implementations that wrap the native DirectoryNamespace.
+    All methods simply delegate to the underlying DirectoryNamespace instance.
+    """
+
+    def __init__(self, inner: DirectoryNamespace):
+        self._inner = inner
+
+    def namespace_id(self) -> str:
+        return f"CustomNamespace[{self._inner.namespace_id()}]"
+
+    def create_namespace(
+        self, request: CreateNamespaceRequest
+    ) -> CreateNamespaceResponse:
+        return self._inner.create_namespace(request)
+
+    def describe_namespace(
+        self, request: DescribeNamespaceRequest
+    ) -> DescribeNamespaceResponse:
+        return self._inner.describe_namespace(request)
+
+    def namespace_exists(self, request: NamespaceExistsRequest) -> None:
+        return self._inner.namespace_exists(request)
+
+    def drop_namespace(self, request: DropNamespaceRequest) -> DropNamespaceResponse:
+        return self._inner.drop_namespace(request)
+
+    def list_namespaces(self, request: ListNamespacesRequest) -> ListNamespacesResponse:
+        return self._inner.list_namespaces(request)
+
+    def create_table(
+        self, request: CreateTableRequest, data: bytes
+    ) -> CreateTableResponse:
+        return self._inner.create_table(request, data)
+
+    def declare_table(self, request: DeclareTableRequest) -> DeclareTableResponse:
+        return self._inner.declare_table(request)
+
+    def describe_table(self, request: DescribeTableRequest) -> DescribeTableResponse:
+        return self._inner.describe_table(request)
+
+    def table_exists(self, request: TableExistsRequest) -> None:
+        return self._inner.table_exists(request)
+
+    def drop_table(self, request: DropTableRequest) -> DropTableResponse:
+        return self._inner.drop_table(request)
+
+    def list_tables(self, request: ListTablesRequest) -> ListTablesResponse:
+        return self._inner.list_tables(request)
+
+    def register_table(self, request: RegisterTableRequest) -> RegisterTableResponse:
+        return self._inner.register_table(request)
+
+    def deregister_table(
+        self, request: DeregisterTableRequest
+    ) -> DeregisterTableResponse:
+        return self._inner.deregister_table(request)
+
+    def list_table_versions(
+        self, request: ListTableVersionsRequest
+    ) -> ListTableVersionsResponse:
+        return self._inner.list_table_versions(request)
+
+    def describe_table_version(
+        self, request: DescribeTableVersionRequest
+    ) -> DescribeTableVersionResponse:
+        return self._inner.describe_table_version(request)
+
+    def create_table_version(
+        self, request: CreateTableVersionRequest
+    ) -> CreateTableVersionResponse:
+        return self._inner.create_table_version(request)
+
+    def retrieve_ops_metrics(self) -> Optional[Dict[str, int]]:
+        return self._inner.retrieve_ops_metrics()
+
+
+def _wrap_if_custom(ns_client: DirectoryNamespace, use_custom: bool):
+    """Wrap namespace client in CustomNamespace if use_custom is True."""
+    if use_custom:
+        return CustomNamespace(ns_client)
+    return ns_client
+
+
 # LocalStack S3 configuration
 CONFIG = {
     "allow_http": "true",
@@ -89,157 +208,88 @@ def delete_bucket(s3, bucket_name):
         pass
 
 
-class TrackingNamespace(LanceNamespace):
+def create_tracking_namespace(
+    bucket_name: str,
+    storage_options: dict,
+    credential_expires_in_seconds: int = 60,
+    use_custom: bool = False,
+):
+    """Create a DirectoryNamespace with ops metrics and credential vending enabled.
+
+    Uses native DirectoryNamespace features:
+    - ops_metrics_enabled=true: Tracks API call counts via retrieve_ops_metrics()
+    - vend_input_storage_options=true: Returns input storage options in responses
+    - vend_input_storage_options_refresh_interval_millis: Adds expires_at_millis
+
+    Args:
+        bucket_name: S3 bucket name or local path
+        storage_options: Storage options to pass through (credentials, endpoint, etc.)
+        credential_expires_in_seconds: Interval in seconds for credential expiration
+        use_custom: If True, wrap in CustomNamespace for testing custom implementations
+
+    Returns:
+        Tuple of (namespace_client, inner_namespace_client) where inner is always
+        the DirectoryNamespace (used for metrics retrieval)
     """
-    Mock namespace that wraps DirectoryNamespace and tracks API calls.
+    # Add refresh_offset_millis to storage options so that credentials are not
+    # considered expired immediately. Set to 1 second (1000ms) so that refresh
+    # checks work correctly with short-lived credentials in tests.
+    storage_options_with_refresh = dict(storage_options)
+    storage_options_with_refresh["refresh_offset_millis"] = "1000"
 
-    This namespace returns incrementing credentials with each API call to simulate
-    credential rotation. It also tracks the number of times each API is called
-    to verify caching behavior.
-    """
+    dir_props = {f"storage.{k}": v for k, v in storage_options_with_refresh.items()}
 
-    def __init__(
-        self,
-        bucket_name: str,
-        storage_options: Dict[str, str],
-        credential_expires_in_seconds: int = 60,
-    ):
-        from lance.namespace import DirectoryNamespace
+    if bucket_name.startswith("/") or bucket_name.startswith("file://"):
+        dir_props["root"] = f"{bucket_name}/namespace_root"
+    else:
+        dir_props["root"] = f"s3://{bucket_name}/namespace_root"
 
-        self.bucket_name = bucket_name
-        self.base_storage_options = storage_options
-        self.credential_expires_in_seconds = credential_expires_in_seconds
-        self.describe_call_count = 0
-        self.create_call_count = 0
-        self.lock = Lock()
+    # Enable ops metrics tracking
+    dir_props["ops_metrics_enabled"] = "true"
+    # Enable storage options vending
+    dir_props["vend_input_storage_options"] = "true"
+    # Set refresh interval in milliseconds
+    dir_props["vend_input_storage_options_refresh_interval_millis"] = str(
+        credential_expires_in_seconds * 1000
+    )
 
-        # Create underlying DirectoryNamespace with storage options
-        dir_props = {f"storage.{k}": v for k, v in storage_options.items()}
+    inner_ns_client = DirectoryNamespace(**dir_props)
+    ns_client = _wrap_if_custom(inner_ns_client, use_custom)
+    return ns_client, inner_ns_client
 
-        # Use S3 path for bucket name, local path for file paths
-        if bucket_name.startswith("/") or bucket_name.startswith("file://"):
-            dir_props["root"] = f"{bucket_name}/namespace_root"
-        else:
-            dir_props["root"] = f"s3://{bucket_name}/namespace_root"
 
-        self.inner = DirectoryNamespace(**dir_props)
+def get_describe_call_count(namespace_client) -> int:
+    """Get the number of describe_table calls made to the namespace client."""
+    return namespace_client.retrieve_ops_metrics().get("describe_table", 0)
 
-    def get_describe_call_count(self) -> int:
-        """Thread-safe getter for describe call count."""
-        with self.lock:
-            return self.describe_call_count
 
-    def get_create_call_count(self) -> int:
-        """Thread-safe getter for create call count."""
-        with self.lock:
-            return self.create_call_count
-
-    def namespace_id(self) -> str:
-        """Return namespace identifier."""
-        return f"TrackingNamespace {{ inner: {self.inner.namespace_id()} }}"
-
-    def _modify_storage_options(
-        self, storage_options: Dict[str, str], count: int
-    ) -> Dict[str, str]:
-        """
-        Add incrementing credentials with expiration timestamp.
-
-        This simulates a credential rotation system where each call returns
-        new credentials that expire after credential_expires_in_seconds.
-        """
-        modified = copy.deepcopy(storage_options) if storage_options else {}
-
-        # Increment credentials to simulate rotation
-        modified["aws_access_key_id"] = f"AKID_{count}"
-        modified["aws_secret_access_key"] = f"SECRET_{count}"
-        modified["aws_session_token"] = f"TOKEN_{count}"
-
-        # Set expiration time
-        expires_at_millis = int(
-            (time.time() + self.credential_expires_in_seconds) * 1000
-        )
-        modified["expires_at_millis"] = str(expires_at_millis)
-
-        return modified
-
-    def declare_table(self, request: DeclareTableRequest) -> DeclareTableResponse:
-        """Track declare_table calls and inject rotating credentials."""
-        with self.lock:
-            self.create_call_count += 1
-            count = self.create_call_count
-
-        response = self.inner.declare_table(request)
-        response.storage_options = self._modify_storage_options(
-            response.storage_options, count
-        )
-
-        return response
-
-    def create_empty_table(
-        self, request: CreateEmptyTableRequest
-    ) -> CreateEmptyTableResponse:
-        """Track create_empty_table calls and inject rotating credentials."""
-        with self.lock:
-            self.create_call_count += 1
-            count = self.create_call_count
-
-        response = self.inner.create_empty_table(request)
-        response.storage_options = self._modify_storage_options(
-            response.storage_options, count
-        )
-
-        return response
-
-    def describe_table(self, request: DescribeTableRequest) -> DescribeTableResponse:
-        """Track describe_table calls and inject rotating credentials."""
-        with self.lock:
-            self.describe_call_count += 1
-            count = self.describe_call_count
-
-        response = self.inner.describe_table(request)
-        response.storage_options = self._modify_storage_options(
-            response.storage_options, count
-        )
-
-        return response
-
-    # Pass through other methods to inner namespace
-    def list_tables(self, request):
-        return self.inner.list_tables(request)
-
-    def drop_table(self, request):
-        return self.inner.drop_table(request)
-
-    def list_namespaces(self, request):
-        return self.inner.list_namespaces(request)
-
-    def create_namespace(self, request):
-        return self.inner.create_namespace(request)
-
-    def drop_namespace(self, request):
-        return self.inner.drop_namespace(request)
+def get_declare_call_count(namespace_client) -> int:
+    """Get the number of declare_table calls made to the namespace client."""
+    return namespace_client.retrieve_ops_metrics().get("declare_table", 0)
 
 
 @pytest.mark.s3_test
-def test_namespace_create_table_with_provider(s3_bucket: str):
+@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
+def test_namespace_create_table_with_provider(s3_bucket: str, use_custom: bool):
     """
     Test creating a table through namespace with storage options provider.
 
     Verifies:
-    - create_empty_table is called once to reserve location
+    - declare_table is called once to reserve location
     - Storage options provider is auto-created
     - Table can be written successfully
     - Credentials are cached during write operations
     """
     storage_options = copy.deepcopy(CONFIG)
 
-    namespace = TrackingNamespace(
+    ns_client, inner_ns_client = create_tracking_namespace(
         bucket_name=s3_bucket,
         storage_options=storage_options,
         credential_expires_in_seconds=3600,  # 1 hour
+        use_custom=use_custom,
     )
 
-    db = LanceNamespaceDBConnection(namespace)
+    db = LanceNamespaceDBConnection(ns_client)
 
     # Create unique namespace for this test
     namespace_name = f"test_ns_{uuid.uuid4().hex[:8]}"
@@ -249,8 +299,8 @@ def test_namespace_create_table_with_provider(s3_bucket: str):
     namespace_path = [namespace_name]
 
     # Verify initial state
-    assert namespace.get_create_call_count() == 0
-    assert namespace.get_describe_call_count() == 0
+    assert get_declare_call_count(inner_ns_client) == 0
+    assert get_describe_call_count(inner_ns_client) == 0
 
     # Create table with data
     data = pa.table(
@@ -261,12 +311,12 @@ def test_namespace_create_table_with_provider(s3_bucket: str):
         }
     )
 
-    table = db.create_table(table_name, data, namespace=namespace_path)
+    table = db.create_table(table_name, data, namespace_path=namespace_path)
 
-    # Verify create_empty_table was called exactly once
-    assert namespace.get_create_call_count() == 1
+    # Verify declare_table was called exactly once
+    assert get_declare_call_count(inner_ns_client) == 1
     # describe_table should NOT be called during create in create mode
-    assert namespace.get_describe_call_count() == 0
+    assert get_describe_call_count(inner_ns_client) == 0
 
     # Verify table was created successfully
     assert table.name == table_name
@@ -276,7 +326,8 @@ def test_namespace_create_table_with_provider(s3_bucket: str):
 
 
 @pytest.mark.s3_test
-def test_namespace_open_table_with_provider(s3_bucket: str):
+@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
+def test_namespace_open_table_with_provider(s3_bucket: str, use_custom: bool):
     """
     Test opening a table through namespace with storage options provider.
 
@@ -288,13 +339,14 @@ def test_namespace_open_table_with_provider(s3_bucket: str):
     """
     storage_options = copy.deepcopy(CONFIG)
 
-    namespace = TrackingNamespace(
+    ns_client, inner_ns_client = create_tracking_namespace(
         bucket_name=s3_bucket,
         storage_options=storage_options,
         credential_expires_in_seconds=3600,
+        use_custom=use_custom,
     )
 
-    db = LanceNamespaceDBConnection(namespace)
+    db = LanceNamespaceDBConnection(ns_client)
 
     # Create unique namespace for this test
     namespace_name = f"test_ns_{uuid.uuid4().hex[:8]}"
@@ -312,21 +364,21 @@ def test_namespace_open_table_with_provider(s3_bucket: str):
         }
     )
 
-    db.create_table(table_name, data, namespace=namespace_path)
+    db.create_table(table_name, data, namespace_path=namespace_path)
 
-    initial_create_count = namespace.get_create_call_count()
-    assert initial_create_count == 1
+    initial_declare_count = get_declare_call_count(inner_ns_client)
+    assert initial_declare_count == 1
 
     # Open the table
-    opened_table = db.open_table(table_name, namespace=namespace_path)
+    opened_table = db.open_table(table_name, namespace_path=namespace_path)
 
     # Verify describe_table was called exactly once
-    assert namespace.get_describe_call_count() == 1
-    # create_empty_table should not be called again
-    assert namespace.get_create_call_count() == initial_create_count
+    assert get_describe_call_count(inner_ns_client) == 1
+    # declare_table should not be called again
+    assert get_declare_call_count(inner_ns_client) == initial_declare_count
 
     # Perform multiple read operations
-    describe_count_after_open = namespace.get_describe_call_count()
+    describe_count_after_open = get_describe_call_count(inner_ns_client)
 
     for _ in range(3):
         result = opened_table.to_pandas()
@@ -335,11 +387,72 @@ def test_namespace_open_table_with_provider(s3_bucket: str):
         assert count == 5
 
     # Verify credentials were cached (no additional describe_table calls)
-    assert namespace.get_describe_call_count() == describe_count_after_open
+    assert get_describe_call_count(inner_ns_client) == describe_count_after_open
+
+
+@pytest.mark.skipif(
+    sys.platform == "win32",
+    reason="TODO: fix schema-only namespace metrics test on Windows",
+)
+@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
+def test_namespace_create_schema_only_with_provider(use_custom: bool):
+    """
+    Test creating a schema-only table through namespace with storage options provider.
+
+    Verifies:
+    - declare_table is called once to reserve the location
+    - describe_table is not needed during create in create mode
+    - the table can be reopened successfully afterward
+    - opening the table triggers exactly one describe_table call
+    """
+    temp_dir = tempfile.mkdtemp()
+    try:
+        ns_client, inner_ns_client = create_tracking_namespace(
+            bucket_name=temp_dir,
+            storage_options={},
+            credential_expires_in_seconds=3600,
+            use_custom=use_custom,
+        )
+
+        db = LanceNamespaceDBConnection(ns_client)
+
+        namespace_name = f"test_ns_{uuid.uuid4().hex[:8]}"
+        db.create_namespace([namespace_name])
+
+        table_name = f"test_table_{uuid.uuid4().hex}"
+        namespace_path = [namespace_name]
+        schema = pa.schema(
+            [
+                pa.field("id", pa.int64()),
+                pa.field("vector", pa.list_(pa.float32(), 2)),
+                pa.field("text", pa.string()),
+            ]
+        )
+
+        assert get_declare_call_count(inner_ns_client) == 0
+        assert get_describe_call_count(inner_ns_client) == 0
+
+        table = db.create_table(
+            table_name, schema=schema, namespace_path=namespace_path
+        )
+
+        assert table.name == table_name
+        assert table.namespace == namespace_path
+        assert get_declare_call_count(inner_ns_client) == 1
+        assert get_describe_call_count(inner_ns_client) == 0
+
+        reopened_table = db.open_table(table_name, namespace_path=namespace_path)
+
+        assert reopened_table.schema == schema
+        assert get_declare_call_count(inner_ns_client) == 1
+        assert get_describe_call_count(inner_ns_client) == 1
+    finally:
+        shutil.rmtree(temp_dir, ignore_errors=True)
 
 
 @pytest.mark.s3_test
-def test_namespace_credential_refresh_on_read(s3_bucket: str):
+@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
+def test_namespace_credential_refresh_on_read(s3_bucket: str, use_custom: bool):
     """
     Test credential refresh when credentials expire during read operations.
 
@@ -350,13 +463,14 @@ def test_namespace_credential_refresh_on_read(s3_bucket: str):
     """
     storage_options = copy.deepcopy(CONFIG)
 
-    namespace = TrackingNamespace(
+    ns_client, inner_ns_client = create_tracking_namespace(
         bucket_name=s3_bucket,
         storage_options=storage_options,
         credential_expires_in_seconds=3,  # Short expiration for testing
+        use_custom=use_custom,
     )
 
-    db = LanceNamespaceDBConnection(namespace)
+    db = LanceNamespaceDBConnection(ns_client)
 
     # Create unique namespace for this test
     namespace_name = f"test_ns_{uuid.uuid4().hex[:8]}"
@@ -373,16 +487,16 @@ def test_namespace_credential_refresh_on_read(s3_bucket: str):
         }
     )
 
-    db.create_table(table_name, data, namespace=namespace_path)
+    db.create_table(table_name, data, namespace_path=namespace_path)
 
     # Open table (triggers describe_table)
-    opened_table = db.open_table(table_name, namespace=namespace_path)
+    opened_table = db.open_table(table_name, namespace_path=namespace_path)
 
     # Perform an immediate read (should use credentials from open)
     result = opened_table.to_pandas()
     assert len(result) == 3
 
-    describe_count_after_first_read = namespace.get_describe_call_count()
+    describe_count_after_first_read = get_describe_call_count(inner_ns_client)
 
     # Wait for credentials to expire (3 seconds + buffer)
     time.sleep(5)
@@ -391,7 +505,7 @@ def test_namespace_credential_refresh_on_read(s3_bucket: str):
     result = opened_table.to_pandas()
     assert len(result) == 3
 
-    describe_count_after_refresh = namespace.get_describe_call_count()
+    describe_count_after_refresh = get_describe_call_count(inner_ns_client)
     # Verify describe_table was called again (credential refresh)
     refresh_delta = describe_count_after_refresh - describe_count_after_first_read
 
@@ -404,7 +518,8 @@ def test_namespace_credential_refresh_on_read(s3_bucket: str):
 
 
 @pytest.mark.s3_test
-def test_namespace_credential_refresh_on_write(s3_bucket: str):
+@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
+def test_namespace_credential_refresh_on_write(s3_bucket: str, use_custom: bool):
     """
     Test credential refresh when credentials expire during write operations.
 
@@ -415,13 +530,14 @@ def test_namespace_credential_refresh_on_write(s3_bucket: str):
     """
     storage_options = copy.deepcopy(CONFIG)
 
-    namespace = TrackingNamespace(
+    ns_client, inner_ns_client = create_tracking_namespace(
         bucket_name=s3_bucket,
         storage_options=storage_options,
         credential_expires_in_seconds=3,  # Short expiration
+        use_custom=use_custom,
     )
 
-    db = LanceNamespaceDBConnection(namespace)
+    db = LanceNamespaceDBConnection(ns_client)
 
     # Create unique namespace for this test
     namespace_name = f"test_ns_{uuid.uuid4().hex[:8]}"
@@ -438,7 +554,7 @@ def test_namespace_credential_refresh_on_write(s3_bucket: str):
         }
     )
 
-    table = db.create_table(table_name, initial_data, namespace=namespace_path)
+    table = db.create_table(table_name, initial_data, namespace_path=namespace_path)
 
     # Add more data (should use cached credentials)
     new_data = pa.table(
@@ -466,24 +582,26 @@ def test_namespace_credential_refresh_on_write(s3_bucket: str):
 
 
 @pytest.mark.s3_test
-def test_namespace_overwrite_mode(s3_bucket: str):
+@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
+def test_namespace_overwrite_mode(s3_bucket: str, use_custom: bool):
     """
     Test creating table in overwrite mode with credential tracking.
 
     Verifies:
-    - First create calls create_empty_table exactly once
+    - First create calls declare_table exactly once
     - Overwrite mode calls describe_table exactly once to check existence
     - Storage options provider works in overwrite mode
     """
     storage_options = copy.deepcopy(CONFIG)
 
-    namespace = TrackingNamespace(
+    ns_client, inner_ns_client = create_tracking_namespace(
         bucket_name=s3_bucket,
         storage_options=storage_options,
         credential_expires_in_seconds=3600,
+        use_custom=use_custom,
     )
 
-    db = LanceNamespaceDBConnection(namespace)
+    db = LanceNamespaceDBConnection(ns_client)
 
     # Create unique namespace for this test
     namespace_name = f"test_ns_{uuid.uuid4().hex[:8]}"
@@ -500,11 +618,11 @@ def test_namespace_overwrite_mode(s3_bucket: str):
         }
     )
 
-    table = db.create_table(table_name, data1, namespace=namespace_path)
-    # Exactly one create_empty_table call for initial create
-    assert namespace.get_create_call_count() == 1
+    table = db.create_table(table_name, data1, namespace_path=namespace_path)
+    # Exactly one declare_table call for initial create
+    assert get_declare_call_count(inner_ns_client) == 1
     # No describe_table calls in create mode
-    assert namespace.get_describe_call_count() == 0
+    assert get_describe_call_count(inner_ns_client) == 0
     assert table.count_rows() == 2
 
     # Overwrite the table
@@ -516,14 +634,14 @@ def test_namespace_overwrite_mode(s3_bucket: str):
     )
 
     table2 = db.create_table(
-        table_name, data2, namespace=namespace_path, mode="overwrite"
+        table_name, data2, namespace_path=namespace_path, mode="overwrite"
     )
 
-    # Should still have only 1 create_empty_table call
+    # Should still have only 1 declare_table call
     # (overwrite reuses location from describe_table)
-    assert namespace.get_create_call_count() == 1
+    assert get_declare_call_count(inner_ns_client) == 1
     # Should have called describe_table exactly once to get existing table location
-    assert namespace.get_describe_call_count() == 1
+    assert get_describe_call_count(inner_ns_client) == 1
 
     # Verify new data
     assert table2.count_rows() == 3
@@ -532,7 +650,8 @@ def test_namespace_overwrite_mode(s3_bucket: str):
 
 
 @pytest.mark.s3_test
-def test_namespace_multiple_tables(s3_bucket: str):
+@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
+def test_namespace_multiple_tables(s3_bucket: str, use_custom: bool):
     """
     Test creating and opening multiple tables in the same namespace.
 
@@ -543,13 +662,14 @@ def test_namespace_multiple_tables(s3_bucket: str):
     """
     storage_options = copy.deepcopy(CONFIG)
 
-    namespace = TrackingNamespace(
+    ns_client, inner_ns_client = create_tracking_namespace(
         bucket_name=s3_bucket,
         storage_options=storage_options,
         credential_expires_in_seconds=3600,
+        use_custom=use_custom,
     )
 
-    db = LanceNamespaceDBConnection(namespace)
+    db = LanceNamespaceDBConnection(ns_client)
 
     # Create unique namespace for this test
     namespace_name = f"test_ns_{uuid.uuid4().hex[:8]}"
@@ -559,22 +679,22 @@ def test_namespace_multiple_tables(s3_bucket: str):
     # Create first table
     table1_name = f"table1_{uuid.uuid4().hex}"
     data1 = pa.table({"id": [1, 2], "value": [10, 20]})
-    db.create_table(table1_name, data1, namespace=namespace_path)
+    db.create_table(table1_name, data1, namespace_path=namespace_path)
 
     # Create second table
     table2_name = f"table2_{uuid.uuid4().hex}"
     data2 = pa.table({"id": [3, 4], "value": [30, 40]})
-    db.create_table(table2_name, data2, namespace=namespace_path)
+    db.create_table(table2_name, data2, namespace_path=namespace_path)
 
-    # Should have 2 create calls (one per table)
-    assert namespace.get_create_call_count() == 2
+    # Should have 2 declare calls (one per table)
+    assert get_declare_call_count(inner_ns_client) == 2
 
     # Open both tables
-    opened1 = db.open_table(table1_name, namespace=namespace_path)
-    opened2 = db.open_table(table2_name, namespace=namespace_path)
+    opened1 = db.open_table(table1_name, namespace_path=namespace_path)
+    opened2 = db.open_table(table2_name, namespace_path=namespace_path)
 
     # Should have 2 describe calls (one per open)
-    assert namespace.get_describe_call_count() == 2
+    assert get_describe_call_count(inner_ns_client) == 2
 
     # Verify both tables work independently
     assert opened1.count_rows() == 2
@@ -588,7 +708,8 @@ def test_namespace_multiple_tables(s3_bucket: str):
 
 
 @pytest.mark.s3_test
-def test_namespace_with_schema_only(s3_bucket: str):
+@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
+def test_namespace_with_schema_only(s3_bucket: str, use_custom: bool):
     """
     Test creating empty table with schema only (no data).
 
@@ -599,13 +720,14 @@ def test_namespace_with_schema_only(s3_bucket: str):
     """
     storage_options = copy.deepcopy(CONFIG)
 
-    namespace = TrackingNamespace(
+    ns_client, inner_ns_client = create_tracking_namespace(
         bucket_name=s3_bucket,
         storage_options=storage_options,
         credential_expires_in_seconds=3600,
+        use_custom=use_custom,
     )
 
-    db = LanceNamespaceDBConnection(namespace)
+    db = LanceNamespaceDBConnection(ns_client)
 
     # Create unique namespace for this test
     namespace_name = f"test_ns_{uuid.uuid4().hex[:8]}"
@@ -623,12 +745,12 @@ def test_namespace_with_schema_only(s3_bucket: str):
         ]
     )
 
-    table = db.create_table(table_name, schema=schema, namespace=namespace_path)
+    table = db.create_table(table_name, schema=schema, namespace_path=namespace_path)
 
-    # Should have called create_empty_table once
-    assert namespace.get_create_call_count() == 1
+    # Should have called declare_table once
+    assert get_declare_call_count(inner_ns_client) == 1
     # Should NOT have called describe_table in create mode
-    assert namespace.get_describe_call_count() == 0
+    assert get_describe_call_count(inner_ns_client) == 0
 
     # Verify empty table
     assert table.count_rows() == 0

python/python/tests/test_permutation.py

@@ -522,6 +522,50 @@ def test_no_split_names(some_table: Table):
     assert permutations[1].num_rows == 500
 
 
+def test_permutations_metadata_without_split_names_key(mem_db: DBConnection):
+    """Regression: schema metadata present but missing split_names key must not crash.
+
+    Previously, `.get(b"split_names", None).decode()` was called unconditionally,
+    so any permutation table whose metadata dict had other keys but no split_names
+    raised AttributeError: 'NoneType' has no attribute 'decode'.
+    """
+    base = mem_db.create_table("base_nosplit", pa.table({"x": range(10)}))
+
+    # Build a permutation-like table that carries some metadata but NOT split_names.
+    raw = pa.table(
+        {
+            "row_id": pa.array(range(10), type=pa.uint64()),
+            "split_id": pa.array([0] * 10, type=pa.uint32()),
+        }
+    ).replace_schema_metadata({b"other_key": b"other_value"})
+    perm_tbl = mem_db.create_table("perm_nosplit", raw)
+
+    permutations = Permutations(base, perm_tbl)
+    assert permutations.split_names == []
+    assert permutations.split_dict == {}
+
+
+def test_from_tables_string_split_missing_names_key(mem_db: DBConnection):
+    """Regression: from_tables() with a string split must raise ValueError, not
+    AttributeError.
+
+    Previously, `.get(b"split_names", None).decode()` crashed with AttributeError
+    when the metadata dict existed but had no split_names key.
+    """
+    base = mem_db.create_table("base_strsplit", pa.table({"x": range(10)}))
+
+    raw = pa.table(
+        {
+            "row_id": pa.array(range(10), type=pa.uint64()),
+            "split_id": pa.array([0] * 10, type=pa.uint32()),
+        }
+    ).replace_schema_metadata({b"other_key": b"other_value"})
+    perm_tbl = mem_db.create_table("perm_strsplit", raw)
+
+    with pytest.raises(ValueError, match="no split names are defined"):
+        Permutation.from_tables(base, perm_tbl, split="train")
+
+
 @pytest.fixture
 def some_perm_table(some_table: Table) -> Table:
     return (
@@ -1051,3 +1095,23 @@ def test_getitems_invalid_offset(some_permutation: Permutation):
     """Test __getitems__ with an out-of-range offset raises an error."""
     with pytest.raises(Exception):
         some_permutation.__getitems__([999999])
+
+
+def test_fetch_matches_getitems(some_permutation: Permutation):
+    """Public fetch() should be equivalent to __getitems__."""
+    indices = [0, 1, 2, 10, 100]
+    assert some_permutation.fetch(indices) == some_permutation.__getitems__(indices)
+
+
+def test_fetch_respects_format(some_permutation: Permutation):
+    """fetch() applies the configured format/transform."""
+    arrow_perm = some_permutation.with_format("arrow")
+    result = arrow_perm.fetch([0, 1, 2])
+    assert isinstance(result, pa.RecordBatch)
+    assert result.num_rows == 3
+
+
+def test_fetch_invalid_offset(some_permutation: Permutation):
+    """fetch() with an out-of-range offset raises an error."""
+    with pytest.raises(Exception):
+        some_permutation.fetch([999999])

python/python/tests/test_pydantic.py

@@ -3,6 +3,7 @@
 
 import json
 from datetime import date, datetime
+from enum import Enum
 from typing import List, Optional, Tuple
 
 import pyarrow as pa
@@ -673,3 +674,29 @@ async def test_aliases_in_lance_model_async(mem_db_async):
     assert hasattr(model, "name")
     assert hasattr(model, "distance")
     assert model.distance < 0.01
+
+
+def test_enum_types():
+    """Enum fields should map to the Arrow type of their value (issue #1846)."""
+
+    class StrStatus(str, Enum):
+        PENDING = "pending"
+        RUNNING = "running"
+        DONE = "done"
+
+    class IntPriority(int, Enum):
+        LOW = 1
+        MEDIUM = 2
+        HIGH = 3
+
+    class TestModel(pydantic.BaseModel):
+        status: StrStatus
+        priority: IntPriority
+        opt_status: Optional[StrStatus] = None
+
+    schema = pydantic_to_schema(TestModel)
+
+    assert schema.field("status").type == pa.dictionary(pa.int32(), pa.utf8())
+    assert schema.field("priority").type == pa.int64()
+    assert schema.field("opt_status").type == pa.dictionary(pa.int32(), pa.utf8())
+    assert schema.field("opt_status").nullable

python/python/tests/test_query.py

@@ -30,6 +30,7 @@ from lancedb.query import (
     PhraseQuery,
     Query,
     FullTextSearchQuery,
+    ensure_vector_query,
 )
 from lancedb.rerankers.cross_encoder import CrossEncoderReranker
 from lancedb.table import AsyncTable, LanceTable
@@ -1384,7 +1385,7 @@ def test_query_timeout(tmp_path):
         }
     )
     table = db.create_table("test", data)
-    table.create_fts_index("text", use_tantivy=False)
+    table.create_fts_index("text")
 
     with pytest.raises(Exception, match="Query timeout"):
         table.search().where("text = 'a'").to_list(timeout=timedelta(0))
@@ -1501,6 +1502,18 @@ def test_search_empty_table(mem_db):
     assert results == []
 
 
+def test_ensure_vector_query_empty_list():
+    """Regression: ensure_vector_query used to return instead of raise ValueError."""
+    with pytest.raises(ValueError, match="non-empty"):
+        ensure_vector_query([])
+
+
+def test_ensure_vector_query_nested_empty_list():
+    """Regression: ensure_vector_query used to return instead of raise ValueError."""
+    with pytest.raises(ValueError, match="non-empty"):
+        ensure_vector_query([[]])
+
+
 def test_fast_search(tmp_path):
     db = lancedb.connect(tmp_path)
 

python/python/tests/test_remote_db.py

@@ -1201,6 +1201,18 @@ async def test_header_provider_overrides_static_headers():
         await db.table_names()
 
 
+def test_close():
+    """Test that close() works without AttributeError."""
+    import asyncio
+
+    def handler(req):
+        req.send_response(200)
+        req.end_headers()
+
+    with mock_lancedb_connection(handler) as db:
+        asyncio.run(db.close())
+
+
 @pytest.mark.parametrize("exception", [KeyboardInterrupt, SystemExit, GeneratorExit])
 def test_background_loop_cancellation(exception):
     """Test that BackgroundEventLoop.run() cancels the future on interrupt."""

python/python/tests/test_rerankers.py

@@ -26,11 +26,8 @@ from lancedb.rerankers import (
 )
 from lancedb.table import LanceTable
 
-# Tests rely on FTS index
-pytest.importorskip("lancedb.fts")
 
-
-def get_test_table(tmp_path, use_tantivy):
+def get_test_table(tmp_path):
     db = lancedb.connect(tmp_path)
     # Create a LanceDB table schema with a vector and a text column
     emb = EmbeddingFunctionRegistry.get_instance().get("test").create()
@@ -98,7 +95,7 @@ def get_test_table(tmp_path, use_tantivy):
     )
 
     # Create a fts index
-    table.create_fts_index("text", use_tantivy=use_tantivy, replace=True)
+    table.create_fts_index("text", replace=True)
 
     return table, MyTable
 
@@ -208,8 +205,8 @@ def _run_test_reranker(reranker, table, query, query_vector, schema):
     assert len(result) == 20 and result == result_arrow
 
 
-def _run_test_hybrid_reranker(reranker, tmp_path, use_tantivy):
-    table, schema = get_test_table(tmp_path, use_tantivy)
+def _run_test_hybrid_reranker(reranker, tmp_path):
+    table, schema = get_test_table(tmp_path)
     # The default reranker
     result1 = (
         table.search(
@@ -285,8 +282,7 @@ def _run_test_hybrid_reranker(reranker, tmp_path, use_tantivy):
     )
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_linear_combination(tmp_path, use_tantivy):
+def test_linear_combination(tmp_path):
     reranker = LinearCombinationReranker()
 
     vector_results = pa.Table.from_pydict(
@@ -313,22 +309,20 @@ def test_linear_combination(tmp_path, use_tantivy):
     assert "_score" not in combined_results.column_names
     assert "_relevance_score" in combined_results.column_names
 
-    _run_test_hybrid_reranker(reranker, tmp_path, use_tantivy)
+    _run_test_hybrid_reranker(reranker, tmp_path)
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_rrf_reranker(tmp_path, use_tantivy):
+def test_rrf_reranker(tmp_path):
     reranker = RRFReranker()
-    _run_test_hybrid_reranker(reranker, tmp_path, use_tantivy)
+    _run_test_hybrid_reranker(reranker, tmp_path)
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_mrr_reranker(tmp_path, use_tantivy):
+def test_mrr_reranker(tmp_path):
     reranker = MRRReranker()
-    _run_test_hybrid_reranker(reranker, tmp_path, use_tantivy)
+    _run_test_hybrid_reranker(reranker, tmp_path)
 
     # Test multi-vector part
-    table, schema = get_test_table(tmp_path, use_tantivy)
+    table, schema = get_test_table(tmp_path)
     query = "single player experience"
     rs1 = table.search(query, vector_column_name="vector").limit(10).with_row_id(True)
     rs2 = (
@@ -363,7 +357,7 @@ def test_rrf_reranker_distance():
     table = db.create_table("test", data)
 
     table.create_index(num_partitions=1, num_sub_vectors=2)
-    table.create_fts_index("text", use_tantivy=False)
+    table.create_fts_index("text")
 
     reranker = RRFReranker(return_score="all")
 
@@ -422,35 +416,31 @@ def test_rrf_reranker_distance():
 @pytest.mark.skipif(
     os.environ.get("COHERE_API_KEY") is None, reason="COHERE_API_KEY not set"
 )
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_cohere_reranker(tmp_path, use_tantivy):
+def test_cohere_reranker(tmp_path):
     pytest.importorskip("cohere")
     reranker = CohereReranker()
-    table, schema = get_test_table(tmp_path, use_tantivy)
+    table, schema = get_test_table(tmp_path)
     _run_test_reranker(reranker, table, "single player experience", None, schema)
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_cross_encoder_reranker(tmp_path, use_tantivy):
+def test_cross_encoder_reranker(tmp_path):
     pytest.importorskip("sentence_transformers")
     reranker = CrossEncoderReranker()
-    table, schema = get_test_table(tmp_path, use_tantivy)
+    table, schema = get_test_table(tmp_path)
     _run_test_reranker(reranker, table, "single player experience", None, schema)
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_colbert_reranker(tmp_path, use_tantivy):
+def test_colbert_reranker(tmp_path):
     pytest.importorskip("rerankers")
     reranker = ColbertReranker()
-    table, schema = get_test_table(tmp_path, use_tantivy)
+    table, schema = get_test_table(tmp_path)
     _run_test_reranker(reranker, table, "single player experience", None, schema)
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_answerdotai_reranker(tmp_path, use_tantivy):
+def test_answerdotai_reranker(tmp_path):
     pytest.importorskip("rerankers")
     reranker = AnswerdotaiRerankers()
-    table, schema = get_test_table(tmp_path, use_tantivy)
+    table, schema = get_test_table(tmp_path)
     _run_test_reranker(reranker, table, "single player experience", None, schema)
 
 
@@ -459,10 +449,9 @@ def test_answerdotai_reranker(tmp_path, use_tantivy):
     or os.environ.get("OPENAI_BASE_URL") is not None,
     reason="OPENAI_API_KEY not set",
 )
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_openai_reranker(tmp_path, use_tantivy):
+def test_openai_reranker(tmp_path):
     pytest.importorskip("openai")
-    table, schema = get_test_table(tmp_path, use_tantivy)
+    table, schema = get_test_table(tmp_path)
     reranker = OpenaiReranker()
     _run_test_reranker(reranker, table, "single player experience", None, schema)
 
@@ -470,10 +459,9 @@ def test_openai_reranker(tmp_path, use_tantivy):
 @pytest.mark.skipif(
     os.environ.get("JINA_API_KEY") is None, reason="JINA_API_KEY not set"
 )
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_jina_reranker(tmp_path, use_tantivy):
+def test_jina_reranker(tmp_path):
     pytest.importorskip("jina")
-    table, schema = get_test_table(tmp_path, use_tantivy)
+    table, schema = get_test_table(tmp_path)
     reranker = JinaReranker()
     _run_test_reranker(reranker, table, "single player experience", None, schema)
 
@@ -481,11 +469,10 @@ def test_jina_reranker(tmp_path, use_tantivy):
 @pytest.mark.skipif(
     os.environ.get("VOYAGE_API_KEY") is None, reason="VOYAGE_API_KEY not set"
 )
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_voyageai_reranker(tmp_path, use_tantivy):
+def test_voyageai_reranker(tmp_path):
     pytest.importorskip("voyageai")
     reranker = VoyageAIReranker(model_name="rerank-2.5")
-    table, schema = get_test_table(tmp_path, use_tantivy)
+    table, schema = get_test_table(tmp_path)
     _run_test_reranker(reranker, table, "single player experience", None, schema)
 
 
@@ -504,7 +491,7 @@ def test_empty_result_reranker():
 
     # Create empty table with schema
     empty_table = db.create_table("empty_table", schema=schema, mode="overwrite")
-    empty_table.create_fts_index("text", use_tantivy=False, replace=True)
+    empty_table.create_fts_index("text", replace=True)
     for reranker in [
         CrossEncoderReranker(),
         # ColbertReranker(),
@@ -603,11 +590,10 @@ def test_empty_hybrid_result_reranker():
     assert "_rowid" in result.column_names
 
 
-@pytest.mark.parametrize("use_tantivy", [True, False])
-def test_cross_encoder_reranker_return_all(tmp_path, use_tantivy):
+def test_cross_encoder_reranker_return_all(tmp_path):
     pytest.importorskip("sentence_transformers")
     reranker = CrossEncoderReranker(return_score="all")
-    table, schema = get_test_table(tmp_path, use_tantivy)
+    table, schema = get_test_table(tmp_path)
     query = "single player experience"
     result = (
         table.search(query, query_type="hybrid", vector_column_name="vector")

python/python/tests/test_s3.py

@@ -242,8 +242,8 @@ def test_s3_dynamodb_sync(s3_bucket: str, commit_table: str, monkeypatch):
 
     # FTS indices should error since they are not supported yet.
     with pytest.raises(
-        NotImplementedError,
-        match="Full-text search is only supported on the local filesystem",
+        ValueError,
+        match="Tantivy-based FTS has been removed",
     ):
         table.create_fts_index("x", use_tantivy=True)
 

python/python/tests/test_table.py

@@ -3,6 +3,7 @@
 
 
 import os
+import sys
 from datetime import date, datetime, timedelta
 from time import sleep
 from typing import List
@@ -527,6 +528,132 @@ async def test_add_async(mem_db_async: AsyncConnection):
     assert await table.count_rows() == 3
 
 
+def test_add_overwrite_infers_vector_schema(mem_db: DBConnection):
+    """Overwrite should infer vector columns the same way create_table does.
+
+    Regression test for https://github.com/lancedb/lancedb/issues/3183
+    """
+    table = mem_db.create_table(
+        "test_overwrite_vec",
+        data=[
+            {"vector": [1.0, 2.0, 3.0, 4.0], "item": "foo"},
+            {"vector": [5.0, 6.0, 7.0, 8.0], "item": "bar"},
+        ],
+    )
+    # create_table infers vector as fixed_size_list<float32, 4>
+    original_type = table.schema.field("vector").type
+    assert pa.types.is_fixed_size_list(original_type)
+
+    # overwrite with plain Python lists (PyArrow infers list<double>)
+    table.add(
+        [
+            {"vector": [10.0, 20.0, 30.0, 40.0], "item": "baz"},
+        ],
+        mode="overwrite",
+    )
+    # overwrite should infer vector column the same way as create_table
+    new_type = table.schema.field("vector").type
+    assert pa.types.is_fixed_size_list(new_type), (
+        f"Expected fixed_size_list after overwrite, got {new_type}"
+    )
+
+
+def test_add_progress_callback(mem_db: DBConnection):
+    table = mem_db.create_table(
+        "test",
+        data=[{"id": 1}, {"id": 2}],
+    )
+
+    updates = []
+    table.add([{"id": 3}, {"id": 4}], progress=lambda p: updates.append(dict(p)))
+
+    assert len(table) == 4
+    # The done callback always fires, so we should always get at least one.
+    assert len(updates) >= 1, "expected at least one progress callback"
+    for p in updates:
+        assert "output_rows" in p
+        assert "output_bytes" in p
+        assert "total_rows" in p
+        assert "elapsed_seconds" in p
+        assert "active_tasks" in p
+        assert "total_tasks" in p
+        assert "done" in p
+    # The last callback should have done=True.
+    assert updates[-1]["done"] is True
+
+
+def test_add_progress_tqdm_like(mem_db: DBConnection):
+    """Test that a tqdm-like object gets total set and update() called."""
+
+    class FakeBar:
+        def __init__(self):
+            self.total = None
+            self.n = 0
+            self.postfix = None
+
+        def update(self, n):
+            self.n += n
+
+        def set_postfix_str(self, s):
+            self.postfix = s
+
+        def refresh(self):
+            pass
+
+    table = mem_db.create_table(
+        "test",
+        data=[{"id": 1}, {"id": 2}],
+    )
+
+    bar = FakeBar()
+    table.add([{"id": 3}, {"id": 4}], progress=bar)
+
+    assert len(table) == 4
+    # Postfix should contain throughput and worker count
+    if bar.postfix is not None:
+        assert "MB/s" in bar.postfix
+        assert "workers" in bar.postfix
+
+
+def test_add_progress_bool(mem_db: DBConnection):
+    """Test that progress=True creates and closes a tqdm bar automatically."""
+    table = mem_db.create_table(
+        "test",
+        data=[{"id": 1}, {"id": 2}],
+    )
+
+    table.add([{"id": 3}, {"id": 4}], progress=True)
+    assert len(table) == 4
+
+    # progress=False should be the same as None
+    table.add([{"id": 5}], progress=False)
+    assert len(table) == 5
+
+
+@pytest.mark.asyncio
+async def test_add_progress_callback_async(mem_db_async: AsyncConnection):
+    """Progress callbacks work through the async path too."""
+    table = await mem_db_async.create_table("test", data=[{"id": 1}, {"id": 2}])
+
+    updates = []
+    await table.add([{"id": 3}, {"id": 4}], progress=lambda p: updates.append(dict(p)))
+
+    assert await table.count_rows() == 4
+    assert len(updates) >= 1
+    assert updates[-1]["done"] is True
+
+
+def test_add_progress_callback_error(mem_db: DBConnection):
+    """A failing callback must not prevent the write from succeeding."""
+    table = mem_db.create_table("test", data=[{"id": 1}, {"id": 2}])
+
+    def bad_callback(p):
+        raise RuntimeError("boom")
+
+    table.add([{"id": 3}, {"id": 4}], progress=bad_callback)
+    assert len(table) == 4
+
+
 def test_polars(mem_db: DBConnection):
     data = {
         "vector": [[3.1, 4.1], [5.9, 26.5]],
@@ -923,6 +1050,231 @@ def test_add_with_nans(mem_db: DBConnection):
     assert np.allclose(v, np.array([0.0, 0.0]))
 
 
+def test_add_with_empty_fixed_size_list_drops_bad_rows(mem_db: DBConnection):
+    class Schema(LanceModel):
+        text: str
+        embedding: Vector(16)
+
+    table = mem_db.create_table("test_empty_embeddings", schema=Schema)
+    table.add(
+        [
+            {"text": "hello", "embedding": []},
+            {"text": "bar", "embedding": [0.1] * 16},
+        ],
+        on_bad_vectors="drop",
+    )
+
+    data = table.to_arrow()
+    assert data["text"].to_pylist() == ["bar"]
+    assert np.allclose(data["embedding"].to_pylist()[0], np.array([0.1] * 16))
+
+
+def test_add_with_integer_embeddings_preserves_casting(mem_db: DBConnection):
+    class Schema(LanceModel):
+        text: str
+        embedding: Vector(4)
+
+    table = mem_db.create_table("test_integer_embeddings", schema=Schema)
+    table.add(
+        [{"text": "foo", "embedding": [1, 2, 3, 4]}],
+        on_bad_vectors="drop",
+    )
+
+    assert table.to_arrow()["embedding"].to_pylist() == [[1.0, 2.0, 3.0, 4.0]]
+
+
+def test_on_bad_vectors_does_not_handle_non_vector_fixed_size_lists(
+    mem_db: DBConnection,
+):
+    schema = pa.schema(
+        [
+            pa.field("vector", pa.list_(pa.float32(), 4)),
+            pa.field("bbox", pa.list_(pa.float32(), 4)),
+        ]
+    )
+    table = mem_db.create_table("test_bbox_schema", schema=schema)
+
+    with pytest.raises(RuntimeError, match="FixedSizeListType"):
+        table.add(
+            [{"vector": [1.0, 2.0, 3.0, 4.0], "bbox": [0.0, 1.0]}],
+            on_bad_vectors="drop",
+        )
+
+
+def test_on_bad_vectors_does_not_handle_custom_named_fixed_size_lists(
+    mem_db: DBConnection,
+):
+    schema = pa.schema([pa.field("features", pa.list_(pa.float32(), 16))])
+    table = mem_db.create_table("test_custom_named_fixed_size_vector", schema=schema)
+
+    with pytest.raises(RuntimeError, match="FixedSizeListType"):
+        table.add(
+            [
+                {"features": []},
+                {"features": [0.1] * 16},
+            ],
+            on_bad_vectors="drop",
+        )
+
+
+def test_on_bad_vectors_with_schema_list_vector_still_sanitizes(mem_db: DBConnection):
+    schema = pa.schema([pa.field("vector", pa.list_(pa.float32()))])
+    table = mem_db.create_table("test_schema_list_vector", schema=schema)
+    table.add(
+        [
+            {"vector": [1.0, 2.0]},
+            {"vector": [3.0]},
+            {"vector": [4.0, 5.0]},
+        ],
+        on_bad_vectors="drop",
+    )
+
+    assert table.to_arrow()["vector"].to_pylist() == [[1.0, 2.0], [4.0, 5.0]]
+
+
+def test_on_bad_vectors_handles_typed_custom_fixed_vectors_for_list_schema(
+    mem_db: DBConnection,
+):
+    schema = pa.schema([pa.field("vec", pa.list_(pa.float32()))])
+    table = mem_db.create_table("test_typed_custom_fixed_vector", schema=schema)
+    data = pa.table(
+        {
+            "vec": pa.array(
+                [[float("nan")] * 16, [1.0] * 16],
+                type=pa.list_(pa.float32(), 16),
+            )
+        }
+    )
+
+    table.add(data, on_bad_vectors="drop")
+
+    assert table.to_arrow()["vec"].to_pylist() == [[1.0] * 16]
+
+
+def test_on_bad_vectors_fill_preserves_arrow_nested_vector_type(mem_db: DBConnection):
+    schema = pa.schema([pa.field("vector", pa.list_(pa.float32()))])
+    table = mem_db.create_table("test_fill_arrow_nested_type", schema=schema)
+    data = pa.table(
+        {
+            "vector": pa.array(
+                [[1.0, 2.0], [float("nan"), 3.0]],
+                type=pa.list_(pa.float32(), 2),
+            )
+        }
+    )
+
+    table.add(
+        data,
+        on_bad_vectors="fill",
+        fill_value=0.0,
+    )
+
+    assert table.to_arrow()["vector"].to_pylist() == [[1.0, 2.0], [0.0, 0.0]]
+
+
+@pytest.mark.parametrize(
+    ("table_name", "batch1", "expected"),
+    [
+        (
+            "test_schema_list_vector_empty_prefix",
+            pa.record_batch({"vector": [[], []]}),
+            [[], [], [1.0, 2.0], [3.0, 4.0]],
+        ),
+        (
+            "test_schema_list_vector_all_bad_prefix",
+            pa.record_batch({"vector": [[float("nan")] * 3, [float("nan")] * 3]}),
+            [[1.0, 2.0], [3.0, 4.0]],
+        ),
+    ],
+)
+def test_on_bad_vectors_with_schema_list_vector_ignores_invalid_prefix_batches(
+    mem_db: DBConnection,
+    table_name: str,
+    batch1: pa.RecordBatch,
+    expected: list,
+):
+    schema = pa.schema([pa.field("vector", pa.list_(pa.float32()))])
+    table = mem_db.create_table(table_name, schema=schema)
+    batch2 = pa.record_batch({"vector": [[1.0, 2.0], [3.0, 4.0]]})
+    reader = pa.RecordBatchReader.from_batches(batch1.schema, [batch1, batch2])
+
+    table.add(reader, on_bad_vectors="drop")
+
+    assert table.to_arrow()["vector"].to_pylist() == expected
+
+
+def test_on_bad_vectors_with_multiple_vectors_locks_dim_after_final_drop(
+    mem_db: DBConnection,
+):
+    registry = EmbeddingFunctionRegistry.get_instance()
+    func = MockTextEmbeddingFunction.create()
+    metadata = registry.get_table_metadata(
+        [
+            EmbeddingFunctionConfig(
+                source_column="text1", vector_column="vec1", function=func
+            ),
+            EmbeddingFunctionConfig(
+                source_column="text2", vector_column="vec2", function=func
+            ),
+        ]
+    )
+    schema = pa.schema(
+        [
+            pa.field("vec1", pa.list_(pa.float32())),
+            pa.field("vec2", pa.list_(pa.float32())),
+        ],
+        metadata=metadata,
+    )
+    table = mem_db.create_table("test_multi_vector_dim_lock", schema=schema)
+    batch1 = pa.record_batch(
+        {
+            "vec1": [[1.0, 2.0, 3.0], [10.0, 11.0]],
+            "vec2": [[float("nan"), 0.0], [5.0, 6.0]],
+        }
+    )
+    batch2 = pa.record_batch(
+        {
+            "vec1": [[20.0, 21.0], [30.0, 31.0]],
+            "vec2": [[7.0, 8.0], [9.0, 10.0]],
+        }
+    )
+    reader = pa.RecordBatchReader.from_batches(batch1.schema, [batch1, batch2])
+
+    table.add(reader, on_bad_vectors="drop")
+
+    data = table.to_arrow()
+    assert data["vec1"].to_pylist() == [[10.0, 11.0], [20.0, 21.0], [30.0, 31.0]]
+    assert data["vec2"].to_pylist() == [[5.0, 6.0], [7.0, 8.0], [9.0, 10.0]]
+
+
+def test_on_bad_vectors_does_not_handle_non_vector_list_columns(mem_db: DBConnection):
+    schema = pa.schema([pa.field("embedding_history", pa.list_(pa.float32()))])
+    table = mem_db.create_table("test_non_vector_list_schema", schema=schema)
+    table.add(
+        [
+            {"embedding_history": [1.0, 2.0]},
+            {"embedding_history": [3.0]},
+        ],
+        on_bad_vectors="drop",
+    )
+
+    assert table.to_arrow()["embedding_history"].to_pylist() == [
+        [1.0, 2.0],
+        [3.0],
+    ]
+
+
+def test_on_bad_vectors_all_null_schema_vector_batches_do_not_crash(
+    mem_db: DBConnection,
+):
+    schema = pa.schema([pa.field("vector", pa.list_(pa.float32(), 2), nullable=True)])
+    table = mem_db.create_table("test_all_null_vector_batch", schema=schema)
+
+    table.add([{"vector": None}], on_bad_vectors="drop")
+
+    assert table.to_arrow()["vector"].to_pylist() == [None]
+
+
 def test_restore(mem_db: DBConnection):
     table = mem_db.create_table(
         "my_table",
@@ -1596,7 +1948,6 @@ def setup_hybrid_search_table(db: DBConnection, embedding_func):
 
 def test_hybrid_search(tmp_db: DBConnection):
     # This test uses an FTS index
-    pytest.importorskip("lancedb.fts")
     pytest.importorskip("lance")
 
     table, MyTable, emb = setup_hybrid_search_table(tmp_db, "test")
@@ -1667,7 +2018,6 @@ def test_hybrid_search(tmp_db: DBConnection):
 
 def test_hybrid_search_metric_type(tmp_db: DBConnection):
     # This test uses an FTS index
-    pytest.importorskip("lancedb.fts")
     pytest.importorskip("lance")
 
     # Need to use nonnorm as the embedding function so l2 and dot results
@@ -1689,6 +2039,13 @@ def test_hybrid_search_metric_type(tmp_db: DBConnection):
 @pytest.mark.parametrize(
     "consistency_interval", [None, timedelta(seconds=0), timedelta(seconds=0.1)]
 )
+@pytest.mark.skipif(
+    sys.platform == "win32",
+    reason=(
+        "TODO: directory namespace is not supported on Windows yet; "
+        "re-enable after that is fixed."
+    ),
+)
 def test_consistency(tmp_path, consistency_interval):
     db = lancedb.connect(tmp_path)
     table = db.create_table("my_table", data=[{"id": 0}])
@@ -1709,7 +2066,6 @@ def test_consistency(tmp_path, consistency_interval):
     elif consistency_interval == timedelta(seconds=0):
         assert table2.version == table.version
     else:
-        # (consistency_interval == timedelta(seconds=0.1)
         assert table2.version == table.version - 1
         sleep(0.1)
         assert table2.version == table.version
@@ -1982,7 +2338,7 @@ def test_stats(mem_db: DBConnection):
     stats = table.stats()
     print(f"{stats=}")
     assert stats == {
-        "total_bytes": 38,
+        "total_bytes": 60,
         "num_rows": 2,
         "num_indices": 0,
         "fragment_stats": {
@@ -2047,3 +2403,33 @@ def test_table_uri(tmp_path):
     db = lancedb.connect(tmp_path)
     table = db.create_table("my_table", data=[{"x": 0}])
     assert table.uri == str(tmp_path / "my_table.lance")
+
+
+def test_sanitize_data_metadata_not_stripped():
+    """Regression test: dict.update() returns None, so assigning its result
+    would silently replace metadata with None, causing with_metadata(None)
+    to strip all schema metadata from the target schema."""
+    from lancedb.table import _sanitize_data
+
+    schema = pa.schema(
+        [pa.field("x", pa.int64())],
+        metadata={b"existing_key": b"existing_value"},
+    )
+    batch = pa.record_batch([pa.array([1, 2, 3])], schema=schema)
+
+    # Use a different field type so the reader and target schemas differ,
+    # forcing _cast_to_target_schema to rebuild the schema with the
+    # target's metadata (instead of taking the fast-path).
+    target_schema = pa.schema(
+        [pa.field("x", pa.int32())],
+        metadata={b"existing_key": b"existing_value"},
+    )
+
+    reader = pa.RecordBatchReader.from_batches(schema, [batch])
+    metadata = {b"new_key": b"new_value"}
+    result = _sanitize_data(reader, target_schema=target_schema, metadata=metadata)
+
+    result_schema = result.schema
+    assert result_schema.metadata is not None
+    assert result_schema.metadata[b"existing_key"] == b"existing_value"
+    assert result_schema.metadata[b"new_key"] == b"new_value"