feat(python): align with_format("torch") with HuggingFace semantics

- New "torch" format yields list-of-per-row-dicts whose values are torch tensors. PyTorch DataLoader's default collate stacks these into a dict[str, Tensor] per batch — matching HuggingFace dataset.set_format("torch") and removing the need for a custom collate_fn. - The previous "torch" behavior (list of 1-D row tensors without column names) moves to a new "torch_row" format. - "torch_col" is unchanged (2-D tensor, requires custom collate). Closes lancedb/lancedb#3245
feat: add manifest-enabled directory namespace mode (#3332 )
2026-05-21 14:00:40 +00:00 · 2026-04-29 22:15:40 +05:30 · 2026-04-29 09:22:06 -07:00 · 2026-04-28 13:29:26 +00:00 · 2026-04-28 13:29:00 +00:00 · 2026-04-27 17:56:10 -07:00
69 changed files with 1727 additions and 1230 deletions
--- a/.bumpversion.toml
+++ b/.bumpversion.toml
@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.28.0-beta.5"
+current_version = "0.28.0-beta.10"
 parse = """(?x)
    (?P<major>0|[1-9]\\d*)\\.
    (?P<minor>0|[1-9]\\d*)\\.
--- a/.github/dependabot.yml
+++ b/.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
--- a/.github/workflows/dev.yml
+++ b/.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:
--- a/.github/workflows/java-publish.yml
+++ b/.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
--- a/.github/workflows/java.yml
+++ b/.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
--- a/.github/workflows/license-header-check.yml
+++ b/.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
--- a/.github/workflows/nodejs.yml
+++ b/.github/workflows/nodejs.yml
@@ -15,6 +15,9 @@ on:
      - .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
--- a/.github/workflows/pypi-publish.yml
+++ b/.github/workflows/pypi-publish.yml
@@ -14,10 +14,16 @@ on:
 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:
@@ -57,10 +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:
@@ -85,10 +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
@@ -107,7 +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')
--- a/.github/workflows/python.yml
+++ b/.github/workflows/python.yml
@@ -17,6 +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
@@ -108,7 +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
@@ -227,6 +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
--- a/.github/workflows/rust.yml
+++ b/.github/workflows/rust.yml
@@ -9,9 +9,15 @@ on:
      - 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
@@ -53,6 +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
--- a/.github/workflows/update_package_lock_run.yml
+++ b/.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
--- a/.github/workflows/update_package_lock_run_nodejs.yml
+++ b/.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
--- a/.github/workflows/upload_wheel/action.yml
+++ b/.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/
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -572,9 +572,9 @@ dependencies = [

 [[package]]
 name = "aws-lc-rs"
-version = "1.16.1"
+version = "1.16.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "94bffc006df10ac2a68c83692d734a465f8ee6c5b384d8545a636f81d858f4bf"
+checksum = "0ec6fb3fe69024a75fa7e1bfb48aa6cf59706a101658ea01bfd33b2b248a038f"
 dependencies = [
 "aws-lc-sys",
 "untrusted 0.7.1",
@@ -583,9 +583,9 @@ dependencies = [

 [[package]]
 name = "aws-lc-sys"
-version = "0.38.0"
+version = "0.40.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "4321e568ed89bb5a7d291a7f37997c2c0df89809d7b6d12062c81ddb54aa782e"
+checksum = "f50037ee5e1e41e7b8f9d161680a725bd1626cb6f8c7e901f91f942850852fe7"
 dependencies = [
 "cc",
 "cmake",
@@ -1373,7 +1373,7 @@ dependencies = [
 "memmap2 0.9.10",
 "num-traits",
 "num_cpus",
- "rand 0.9.2",
+ "rand 0.9.4",
 "rand_distr 0.5.1",
 "rayon",
 "safetensors",
@@ -1409,7 +1409,7 @@ dependencies = [
 "candle-nn",
 "fancy-regex",
 "num-traits",
- "rand 0.9.2",
+ "rand 0.9.4",
 "rayon",
 "serde",
 "serde_json",
@@ -1966,7 +1966,7 @@ dependencies = [
 "log",
 "object_store",
 "parking_lot",
- "rand 0.9.2",
+ "rand 0.9.4",
 "regex",
 "sqlparser 0.59.0",
 "tempfile",
@@ -2080,7 +2080,7 @@ dependencies = [
 "itertools 0.14.0",
 "log",
 "object_store",
- "rand 0.9.2",
+ "rand 0.9.4",
 "tokio",
 "url",
 ]
@@ -2176,7 +2176,7 @@ dependencies = [
 "log",
 "object_store",
 "parking_lot",
- "rand 0.9.2",
+ "rand 0.9.4",
 "tempfile",
 "url",
 ]
@@ -2240,7 +2240,7 @@ dependencies = [
 "log",
 "md-5",
 "num-traits",
- "rand 0.9.2",
+ "rand 0.9.4",
 "regex",
 "sha2",
 "unicode-segmentation",
@@ -2642,7 +2642,7 @@ dependencies = [
 "libc",
 "option-ext",
 "redox_users",
- "windows-sys 0.59.0",
+ "windows-sys 0.61.2",
 ]

 [[package]]
@@ -2830,7 +2830,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "39cab71617ae0d63f51a36d69f866391735b51691dbda63cf6f96d042b63efeb"
 dependencies = [
 "libc",
- "windows-sys 0.59.0",
+ "windows-sys 0.61.2",
 ]

 [[package]]
@@ -2965,7 +2965,7 @@ checksum = "719a903cc23e4a89e87962c2a80fdb45cdaad0983a89bd150bb57b4c8571a7d5"
 dependencies = [
 "half",
 "num-traits",
- "rand 0.9.2",
+ "rand 0.9.4",
 "rand_distr 0.5.1",
 ]

@@ -3010,11 +3010,11 @@ checksum = "42703706b716c37f96a77aea830392ad231f44c9e9a67872fa5548707e11b11c"

 [[package]]
 name = "fsst"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow-array",
- "rand 0.9.2",
+ "rand 0.9.4",
 ]

 [[package]]
@@ -3387,7 +3387,7 @@ dependencies = [
 "cfg-if",
 "crunchy",
 "num-traits",
- "rand 0.9.2",
+ "rand 0.9.4",
 "rand_distr 0.5.1",
 "zerocopy",
 ]
@@ -3470,7 +3470,7 @@ dependencies = [
 "libc",
 "log",
 "num_cpus",
- "rand 0.9.2",
+ "rand 0.9.4",
 "reqwest",
 "serde",
 "serde_json",
@@ -3980,7 +3980,7 @@ dependencies = [
 "portable-atomic",
 "portable-atomic-util",
 "serde_core",
- "windows-sys 0.59.0",
+ "windows-sys 0.61.2",
 ]

 [[package]]
@@ -4043,7 +4043,7 @@ dependencies = [
 "nom 8.0.0",
 "num-traits",
 "ordered-float",
- "rand 0.9.2",
+ "rand 0.9.4",
 "ryu",
 "serde",
 "serde_json",
@@ -4066,8 +4066,8 @@ dependencies = [

 [[package]]
 name = "lance"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow",
 "arrow-arith",
@@ -4119,7 +4119,7 @@ dependencies = [
 "prost",
 "prost-build",
 "prost-types",
- "rand 0.9.2",
+ "rand 0.9.4",
 "roaring",
 "semver",
 "serde",
@@ -4135,8 +4135,8 @@ dependencies = [

 [[package]]
 name = "lance-arrow"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -4152,13 +4152,13 @@ dependencies = [
 "half",
 "jsonb",
 "num-traits",
- "rand 0.9.2",
+ "rand 0.9.4",
 ]

 [[package]]
 name = "lance-bitpacking"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrayref",
 "paste",
@@ -4167,8 +4167,8 @@ dependencies = [

 [[package]]
 name = "lance-core"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -4191,7 +4191,7 @@ dependencies = [
 "object_store",
 "pin-project",
 "prost",
- "rand 0.9.2",
+ "rand 0.9.4",
 "roaring",
 "serde_json",
 "snafu 0.9.0",
@@ -4205,8 +4205,8 @@ dependencies = [

 [[package]]
 name = "lance-datafusion"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -4237,8 +4237,8 @@ dependencies = [

 [[package]]
 name = "lance-datagen"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -4248,7 +4248,7 @@ dependencies = [
 "futures",
 "half",
 "hex",
- "rand 0.9.2",
+ "rand 0.9.4",
 "rand_distr 0.5.1",
 "rand_xoshiro",
 "random_word 0.5.2",
@@ -4256,8 +4256,8 @@ dependencies = [

 [[package]]
 name = "lance-encoding"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow-arith",
 "arrow-array",
@@ -4283,7 +4283,7 @@ dependencies = [
 "prost",
 "prost-build",
 "prost-types",
- "rand 0.9.2",
+ "rand 0.9.4",
 "snafu 0.9.0",
 "strum",
 "tokio",
@@ -4294,8 +4294,8 @@ dependencies = [

 [[package]]
 name = "lance-file"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow-arith",
 "arrow-array",
@@ -4327,8 +4327,8 @@ dependencies = [

 [[package]]
 name = "lance-index"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow",
 "arrow-arith",
@@ -4374,7 +4374,7 @@ dependencies = [
 "prost",
 "prost-build",
 "prost-types",
- "rand 0.9.2",
+ "rand 0.9.4",
 "rand_distr 0.5.1",
 "rangemap",
 "rayon",
@@ -4392,8 +4392,8 @@ dependencies = [

 [[package]]
 name = "lance-io"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow",
 "arrow-arith",
@@ -4426,7 +4426,7 @@ dependencies = [
 "path_abs",
 "pin-project",
 "prost",
- "rand 0.9.2",
+ "rand 0.9.4",
 "serde",
 "snafu 0.9.0",
 "tempfile",
@@ -4437,8 +4437,8 @@ dependencies = [

 [[package]]
 name = "lance-linalg"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -4449,13 +4449,13 @@ dependencies = [
 "lance-arrow",
 "lance-core",
 "num-traits",
- "rand 0.9.2",
+ "rand 0.9.4",
 ]

 [[package]]
 name = "lance-namespace"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow",
 "async-trait",
@@ -4468,17 +4468,19 @@ dependencies = [

 [[package]]
 name = "lance-namespace-impls"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow",
 "arrow-ipc",
 "arrow-schema",
 "async-trait",
 "axum",
+ "base64 0.22.1",
 "bytes",
 "chrono",
 "futures",
+ "hmac",
 "lance",
 "lance-core",
 "lance-index",
@@ -4488,10 +4490,12 @@ dependencies = [
 "lance-table",
 "log",
 "object_store",
- "rand 0.9.2",
+ "quick-xml 0.38.4",
+ "rand 0.9.4",
 "reqwest",
 "serde",
 "serde_json",
+ "sha2",
 "snafu 0.9.0",
 "tokio",
 "tower",
@@ -4501,9 +4505,9 @@ dependencies = [

 [[package]]
 name = "lance-namespace-reqwest-client"
-version = "0.6.1"
+version = "0.7.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "ee2e48de899e2931afb67fcddd0a08e439bf5d8b6ea2a2ed9cb8f4df669bd5cc"
+checksum = "0f061dd6fe63e3ba4052702a9d40973ee4ac57f612f04222897a149576213832"
 dependencies = [
 "reqwest",
 "serde",
@@ -4514,8 +4518,8 @@ dependencies = [

 [[package]]
 name = "lance-table"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow",
 "arrow-array",
@@ -4539,7 +4543,7 @@ dependencies = [
 "prost",
 "prost-build",
 "prost-types",
- "rand 0.9.2",
+ "rand 0.9.4",
 "rangemap",
 "roaring",
 "semver",
@@ -4554,20 +4558,20 @@ dependencies = [

 [[package]]
 name = "lance-testing"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "arrow-array",
 "arrow-schema",
 "lance-arrow",
 "num-traits",
- "rand 0.9.2",
+ "rand 0.9.4",
 ]

 [[package]]
 name = "lance-tokenizer"
-version = "6.0.0-beta.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.1#c7a7d3a0e944646e793d297d4a2e2cf7e4fb28a3"
+version = "6.0.0-beta.4"
+source = "git+https://github.com/lance-format/lance.git?tag=v6.0.0-beta.4#226dafe0a00f995d9e3230c2d61cc06be51994d4"
 dependencies = [
 "rust-stemmers",
 "serde",
@@ -4576,7 +4580,7 @@ dependencies = [

 [[package]]
 name = "lancedb"
-version = "0.28.0-beta.5"
+version = "0.28.0-beta.10"
 dependencies = [
 "ahash",
 "anyhow",
@@ -4637,7 +4641,7 @@ dependencies = [
 "pin-project",
 "polars",
 "polars-arrow",
- "rand 0.9.2",
+ "rand 0.9.4",
 "random_word 0.4.3",
 "regex",
 "reqwest",
@@ -4658,7 +4662,7 @@ dependencies = [

 [[package]]
 name = "lancedb-nodejs"
-version = "0.28.0-beta.5"
+version = "0.28.0-beta.10"
 dependencies = [
 "arrow-array",
 "arrow-buffer",
@@ -4680,7 +4684,7 @@ dependencies = [

 [[package]]
 name = "lancedb-python"
-version = "0.31.0-beta.5"
+version = "0.31.0-beta.10"
 dependencies = [
 "arrow",
 "async-trait",
@@ -5235,7 +5239,7 @@ version = "0.50.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7957b9740744892f114936ab4a57b3f487491bbeafaf8083688b16841a4240e5"
 dependencies = [
- "windows-sys 0.59.0",
+ "windows-sys 0.61.2",
 ]

 [[package]]
@@ -5357,7 +5361,7 @@ dependencies = [
 "parking_lot",
 "percent-encoding",
 "quick-xml 0.38.4",
- "rand 0.9.2",
+ "rand 0.9.4",
 "reqwest",
 "ring",
 "rustls-pemfile",
@@ -6199,8 +6203,8 @@ version = "0.14.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "343d3bd7056eda839b03204e68deff7d1b13aba7af2b2fd16890697274262ee7"
 dependencies = [
- "heck 0.5.0",
- "itertools 0.11.0",
+ "heck 0.4.1",
+ "itertools 0.14.0",
 "log",
 "multimap",
 "petgraph",
@@ -6219,7 +6223,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "27c6023962132f4b30eb4c172c91ce92d933da334c59c23cddee82358ddafb0b"
 dependencies = [
 "anyhow",
- "itertools 0.11.0",
+ "itertools 0.14.0",
 "proc-macro2",
 "quote",
 "syn 2.0.117",
@@ -6402,7 +6406,7 @@ dependencies = [
 "bytes",
 "getrandom 0.3.4",
 "lru-slab",
- "rand 0.9.2",
+ "rand 0.9.4",
 "ring",
 "rustc-hash",
 "rustls 0.23.37",
@@ -6425,7 +6429,7 @@ dependencies = [
 "once_cell",
 "socket2 0.6.3",
 "tracing",
- "windows-sys 0.59.0",
+ "windows-sys 0.60.2",
 ]

 [[package]]
@@ -6468,9 +6472,9 @@ dependencies = [

 [[package]]
 name = "rand"
-version = "0.9.2"
+version = "0.9.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6db2770f06117d490610c7488547d543617b21bfa07796d7a12f6f1bd53850d1"
+checksum = "44c5af06bb1b7d3216d91932aed5265164bf384dc89cd6ba05cf59a35f5f76ea"
 dependencies = [
 "rand_chacha 0.9.0",
 "rand_core 0.9.5",
@@ -6531,7 +6535,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "6a8615d50dcf34fa31f7ab52692afec947c4dd0ab803cc87cb3b0b4570ff7463"
 dependencies = [
 "num-traits",
- "rand 0.9.2",
+ "rand 0.9.4",
 ]

 [[package]]
@@ -6566,7 +6570,7 @@ dependencies = [
 "ahash",
 "brotli 8.0.2",
 "paste",
- "rand 0.9.2",
+ "rand 0.9.4",
 "unicase",
 ]

@@ -6954,7 +6958,7 @@ dependencies = [
 "errno",
 "libc",
 "linux-raw-sys",
- "windows-sys 0.59.0",
+ "windows-sys 0.61.2",
 ]

 [[package]]
@@ -6980,7 +6984,7 @@ dependencies = [
 "once_cell",
 "ring",
 "rustls-pki-types",
- "rustls-webpki 0.103.10",
+ "rustls-webpki 0.103.13",
 "subtle",
 "zeroize",
 ]
@@ -7028,9 +7032,9 @@ dependencies = [

 [[package]]
 name = "rustls-webpki"
-version = "0.103.10"
+version = "0.103.13"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "df33b2b81ac578cabaf06b89b0631153a3f416b0a886e8a7a1707fb51abbd1ef"
+checksum = "61c429a8649f110dddef65e2a5ad240f747e85f7758a6bccc7e5777bd33f756e"
 dependencies = [
 "aws-lc-rs",
 "ring",
@@ -7465,7 +7469,7 @@ version = "0.8.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c1c97747dbf44bb1ca44a561ece23508e99cb592e862f22222dcf42f51d1e451"
 dependencies = [
- "heck 0.5.0",
+ "heck 0.4.1",
 "proc-macro2",
 "quote",
 "syn 2.0.117",
@@ -7477,7 +7481,7 @@ version = "0.9.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "54254b8531cafa275c5e096f62d48c81435d1015405a91198ddb11e967301d40"
 dependencies = [
- "heck 0.5.0",
+ "heck 0.4.1",
 "proc-macro2",
 "quote",
 "syn 2.0.117",
@@ -7818,7 +7822,7 @@ dependencies = [
 "getrandom 0.4.2",
 "once_cell",
 "rustix",
- "windows-sys 0.59.0",
+ "windows-sys 0.61.2",
 ]

 [[package]]
@@ -8242,7 +8246,7 @@ version = "2.1.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9ea3136b675547379c4bd395ca6b938e5ad3c3d20fad76e7fe85f9e0d011419c"
 dependencies = [
- "rand 0.9.2",
+ "rand 0.9.4",
 ]

 [[package]]
@@ -8298,9 +8302,9 @@ dependencies = [

 [[package]]
 name = "unicode-segmentation"
-version = "1.13.1"
+version = "1.13.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "da36089a805484bcccfffe0739803392c8298778a2d2f09febf76fac5ad9025b"
+checksum = "9629274872b2bfaf8d66f5f15725007f635594914870f65218920345aa11aa8c"

 [[package]]
 name = "unicode-width"
@@ -8632,7 +8636,7 @@ version = "0.1.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c2a7b1c03c876122aa43f3020e6c3c3ee5c05081c9a00739faf7503aeba10d22"
 dependencies = [
- "windows-sys 0.59.0",
+ "windows-sys 0.61.2",
 ]

 [[package]]
--- a/Cargo.toml
+++ b/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" = "=6.0.0-beta.1", default-features = false, "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-core = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-datagen = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-file = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-io = { "version" = "=6.0.0-beta.1", default-features = false, "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-index = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-linalg = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-namespace = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-namespace-impls = { "version" = "=6.0.0-beta.1", default-features = false, "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-table = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-testing = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-datafusion = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-encoding = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-arrow = { "version" = "=6.0.0-beta.1", "tag" = "v6.0.0-beta.1", "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 }
--- a/deny.toml
+++ b/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",
+]
--- a/dockerfiles/Dockerfile
+++ b/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
--- a/docs/src/java/java.md
+++ b/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>0.28.0-beta.5</version>
+    <version>0.28.0-beta.10</version>
 </dependency>
 ```

--- a/docs/src/js/interfaces/ConnectionOptions.md
+++ b/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
--- a/java/lancedb-core/pom.xml
+++ b/java/lancedb-core/pom.xml
@@ -8,7 +8,7 @@
    <parent>
      <groupId>com.lancedb</groupId>
      <artifactId>lancedb-parent</artifactId>
-      <version>0.28.0-beta.5</version>
+      <version>0.28.0-beta.10</version>
      <relativePath>../pom.xml</relativePath>
    </parent>

--- a/java/pom.xml
+++ b/java/pom.xml
@@ -6,7 +6,7 @@

    <groupId>com.lancedb</groupId>
    <artifactId>lancedb-parent</artifactId>
-    <version>0.28.0-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>6.0.0-beta.1</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>
--- a/nodejs/Cargo.toml
+++ b/nodejs/Cargo.toml
@@ -1,7 +1,8 @@
 [package]
 name = "lancedb-nodejs"
 edition.workspace = true
-version = "0.28.0-beta.5"
+version = "0.28.0-beta.10"
+publish = false
 license.workspace = true
 description.workspace = true
 repository.workspace = true
@@ -31,8 +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"
--- a/nodejs/test/rerankers.test.ts
+++ b/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();
+      });
+    });
+  });
 });
--- a/nodejs/npm/darwin-arm64/package.json
+++ b/nodejs/npm/darwin-arm64/package.json
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-darwin-arm64",
-	"version": "0.28.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["darwin"],
 	"cpu": ["arm64"],
 	"main": "lancedb.darwin-arm64.node",
--- a/nodejs/npm/linux-arm64-gnu/package.json
+++ b/nodejs/npm/linux-arm64-gnu/package.json
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-gnu",
-	"version": "0.28.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-gnu.node",
--- a/nodejs/npm/linux-arm64-musl/package.json
+++ b/nodejs/npm/linux-arm64-musl/package.json
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-musl",
-	"version": "0.28.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-musl.node",
--- a/nodejs/npm/linux-x64-gnu/package.json
+++ b/nodejs/npm/linux-x64-gnu/package.json
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-gnu",
-	"version": "0.28.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-gnu.node",
--- a/nodejs/npm/linux-x64-musl/package.json
+++ b/nodejs/npm/linux-x64-musl/package.json
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-musl",
-	"version": "0.28.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-musl.node",
--- a/nodejs/npm/win32-arm64-msvc/package.json
+++ b/nodejs/npm/win32-arm64-msvc/package.json
@@ -1,6 +1,6 @@
 {
  "name": "@lancedb/lancedb-win32-arm64-msvc",
-  "version": "0.28.0-beta.5",
+  "version": "0.28.0-beta.10",
  "os": [
    "win32"
  ],
--- a/nodejs/npm/win32-x64-msvc/package.json
+++ b/nodejs/npm/win32-x64-msvc/package.json
@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-win32-x64-msvc",
-	"version": "0.28.0-beta.5",
+	"version": "0.28.0-beta.10",
 	"os": ["win32"],
 	"cpu": ["x64"],
 	"main": "lancedb.win32-x64-msvc.node",
--- a/nodejs/package-lock.json
+++ b/nodejs/package-lock.json
@@ -1,12 +1,12 @@
 {
  "name": "@lancedb/lancedb",
-  "version": "0.28.0-beta.5",
+  "version": "0.28.0-beta.10",
  "lockfileVersion": 3,
  "requires": true,
  "packages": {
    "": {
      "name": "@lancedb/lancedb",
-      "version": "0.28.0-beta.5",
+      "version": "0.28.0-beta.10",
      "cpu": [
        "x64",
        "arm64"
--- a/nodejs/package.json
+++ b/nodejs/package.json
@@ -11,7 +11,7 @@
    "ann"
  ],
  "private": false,
-  "version": "0.28.0-beta.5",
+  "version": "0.28.0-beta.10",
  "main": "dist/index.js",
  "exports": {
    ".": "./dist/index.js",
--- a/nodejs/src/connection.rs
+++ b/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();
--- a/nodejs/src/lib.rs
+++ b/nodejs/src/lib.rs
@@ -37,6 +37,13 @@ pub struct ConnectionOptions {
    ///
    /// 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>,
--- a/nodejs/src/rerankers.rs
+++ b/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 })
    }
 }
--- a/python/.bumpversion.toml
+++ b/python/.bumpversion.toml
@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.31.0-beta.6"
+current_version = "0.31.0-beta.10"
 parse = """(?x)
    (?P<major>0|[1-9]\\d*)\\.
    (?P<minor>0|[1-9]\\d*)\\.
--- a/python/Cargo.toml
+++ b/python/Cargo.toml
@@ -1,6 +1,7 @@
 [package]
 name = "lancedb-python"
-version = "0.31.0-beta.6"
+version = "0.31.0-beta.10"
+publish = false
 edition.workspace = true
 description = "Python bindings for LanceDB"
 license.workspace = true
--- a/python/PYTHON_THIRD_PARTY_LICENSES.md
+++ b/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                                                            |
--- a/python/pyproject.toml
+++ b/python/pyproject.toml
@@ -57,7 +57,6 @@ tests = [
    "duckdb>=0.9.0",
    "pytz>=2023.3",
    "polars>=0.19, <=1.3.0",
-    "tantivy>=0.20.0",
    "pyarrow-stubs>=16.0",
    "pylance>=5.0.0b5",
    "requests>=2.31.0",
--- a/python/python/lancedb/init.py
+++ b/python/python/lancedb/init.py
@@ -73,6 +73,7 @@ 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,
@@ -111,6 +112,10 @@ def connect(
    storage_options: dict, optional
        Additional options for the storage backend. See available options at
        <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
@@ -158,11 +163,11 @@ def connect(
    conn : DBConnection
        A connection to a LanceDB database.
    """
-    if namespace_client_impl is not None or namespace_client_properties is not None:
-        if namespace_client_impl is None or namespace_client_properties is None:
+    if namespace_client_impl is not None:
+        if namespace_client_properties is None:
            raise ValueError(
-                "Both namespace_client_impl and "
-                "namespace_client_properties must be provided"
+                "namespace_client_properties must be provided when "
+                "namespace_client_impl is set"
            )
        if kwargs:
            raise ValueError(f"Unknown keyword arguments: {kwargs}")
@@ -175,6 +180,12 @@ def connect(
            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 "
@@ -212,6 +223,8 @@ def connect(
        read_consistency_interval=read_consistency_interval,
        storage_options=storage_options,
        session=session,
+        manifest_enabled=manifest_enabled,
+        namespace_client_properties=namespace_client_properties,
    )


@@ -289,6 +302,8 @@ def deserialize_conn(
            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}")
@@ -304,6 +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.

@@ -343,6 +360,13 @@ async def connect_async(
        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
    --------
@@ -385,6 +409,8 @@ async def connect_async(
            client_config,
            storage_options,
            session,
+            manifest_enabled,
+            namespace_client_properties,
        )
    )

--- a/python/python/lancedb/_lancedb.pyi
+++ b/python/python/lancedb/_lancedb.pyi
@@ -242,6 +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:
--- a/python/python/lancedb/db.py
+++ b/python/python/lancedb/db.py
@@ -590,8 +590,13 @@ class LanceDBConnection(DBConnection):
        read_consistency_interval: Optional[timedelta] = None,
        storage_options: Optional[Dict[str, str]] = None,
        session: Optional[Session] = None,
+        manifest_enabled: bool = False,
+        namespace_client_properties: Optional[Dict[str, str]] = None,
        _inner: Optional[LanceDbConnection] = None,
    ):
+        self.storage_options = storage_options
+        self._manifest_enabled = manifest_enabled
+        self._namespace_client_properties = namespace_client_properties
        if _inner is not None:
            self._conn = _inner
            self._cached_namespace_client = None
@@ -633,6 +638,8 @@ class LanceDBConnection(DBConnection):
                None,
                storage_options,
                session,
+                manifest_enabled,
+                namespace_client_properties,
            )

        # TODO: It would be nice if we didn't store self.storage_options but it is
@@ -640,7 +647,6 @@ class LanceDBConnection(DBConnection):
        # work because some paths like LanceDBConnection.from_inner will lose the
        # storage_options.  Also, this class really shouldn't be holding any state
        # beyond _conn.
-        self.storage_options = storage_options
        self._conn = AsyncConnection(LOOP.run(do_connect()))
        self._cached_namespace_client: Optional[LanceNamespace] = None

@@ -677,6 +683,8 @@ class LanceDBConnection(DBConnection):
                "connection_type": "local",
                "uri": self.uri,
                "storage_options": self.storage_options,
+                "manifest_enabled": self._manifest_enabled,
+                "namespace_client_properties": self._namespace_client_properties,
                "read_consistency_interval_seconds": (
                    rci.total_seconds() if rci else None
                ),
@@ -1390,6 +1398,7 @@ class AsyncConnection(object):
        namespace_path: Optional[List[str]] = None,
        embedding_functions: Optional[List[EmbeddingFunctionConfig]] = None,
        location: Optional[str] = None,
+        namespace_client: Optional[Any] = None,
    ) -> AsyncTable:
        """Create an [AsyncTable][lancedb.table.AsyncTable] in the database.

@@ -1587,6 +1596,7 @@ class AsyncConnection(object):
                namespace_path=namespace_path,
                storage_options=storage_options,
                location=location,
+                namespace_client=namespace_client,
            )
        else:
            data = data_to_reader(data, schema)
@@ -1597,6 +1607,7 @@ class AsyncConnection(object):
                namespace_path=namespace_path,
                storage_options=storage_options,
                location=location,
+                namespace_client=namespace_client,
            )

        return AsyncTable(new_table)
--- a/python/python/lancedb/fts.py
+++ b/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
-            ]
-        )
-    )
--- a/python/python/lancedb/namespace.py
+++ b/python/python/lancedb/namespace.py
@@ -10,7 +10,6 @@ through a namespace abstraction.

 from __future__ import annotations

-import asyncio
 import sys
 from typing import TYPE_CHECKING, Any, Dict, Iterable, List, Optional, Union

@@ -25,7 +24,24 @@ if TYPE_CHECKING:
 from datetime import timedelta
 import pyarrow as pa

-from lancedb.db import DBConnection, LanceDBConnection
+from lance_namespace_urllib3_client.models.json_arrow_data_type import JsonArrowDataType
+from lance_namespace_urllib3_client.models.json_arrow_field import JsonArrowField
+from lance_namespace_urllib3_client.models.json_arrow_schema import JsonArrowSchema
+from lance_namespace_urllib3_client.models.query_table_request import QueryTableRequest
+from lance_namespace_urllib3_client.models.query_table_request_columns import (
+    QueryTableRequestColumns,
+)
+from lance_namespace_urllib3_client.models.query_table_request_full_text_query import (
+    QueryTableRequestFullTextQuery,
+)
+from lance_namespace_urllib3_client.models.query_table_request_vector import (
+    QueryTableRequestVector,
+)
+from lance_namespace_urllib3_client.models.string_fts_query import StringFtsQuery
+from lance_namespace.errors import TableNotFoundError
+from lancedb._lancedb import connect_namespace_client as _connect_namespace_client
+from lancedb.background_loop import LOOP
+from lancedb.db import AsyncConnection, DBConnection
 from lancedb.namespace_utils import (
    _normalize_create_namespace_mode,
    _normalize_drop_namespace_mode,
@@ -40,14 +56,11 @@ from lance_namespace import (
    ListNamespacesResponse,
    ListTablesResponse,
    ListTablesRequest,
-    DescribeTableRequest,
    DescribeNamespaceRequest,
    DropTableRequest,
    ListNamespacesRequest,
    CreateNamespaceRequest,
    DropNamespaceRequest,
-    DeclareTableRequest,
-    CreateTableRequest,
 )
 from lancedb.table import AsyncTable, LanceTable, Table
 from lancedb.util import validate_table_name
@@ -56,21 +69,6 @@ from lancedb.pydantic import LanceModel
 from lancedb.embeddings import EmbeddingFunctionConfig
 from ._lancedb import Session

-from lance_namespace_urllib3_client.models.json_arrow_schema import JsonArrowSchema
-from lance_namespace_urllib3_client.models.json_arrow_field import JsonArrowField
-from lance_namespace_urllib3_client.models.json_arrow_data_type import JsonArrowDataType
-from lance_namespace_urllib3_client.models.query_table_request import QueryTableRequest
-from lance_namespace_urllib3_client.models.query_table_request_vector import (
-    QueryTableRequestVector,
-)
-from lance_namespace_urllib3_client.models.query_table_request_columns import (
-    QueryTableRequestColumns,
-)
-from lance_namespace_urllib3_client.models.query_table_request_full_text_query import (
-    QueryTableRequestFullTextQuery,
-)
-from lance_namespace_urllib3_client.models.string_fts_query import StringFtsQuery
-

 def _query_to_namespace_request(
    table_id: List[str],
@@ -424,6 +422,23 @@ class LanceNamespaceDBConnection(DBConnection):
        )
        self._namespace_client_impl = namespace_client_impl
        self._namespace_client_properties = namespace_client_properties
+        self._inner = AsyncConnection(
+            _connect_namespace_client(
+                namespace_client,
+                read_consistency_interval=(
+                    read_consistency_interval.total_seconds()
+                    if read_consistency_interval is not None
+                    else None
+                ),
+                storage_options=self.storage_options or None,
+                session=session,
+                namespace_client_pushdown_operations=(
+                    list(self._namespace_client_pushdown_operations)
+                ),
+                namespace_client_impl=namespace_client_impl,
+                namespace_client_properties=namespace_client_properties,
+            )
+        )

    @override
    def serialize(self) -> str:
@@ -497,13 +512,10 @@ class LanceNamespaceDBConnection(DBConnection):
        if mode.lower() not in ["create", "overwrite"]:
            raise ValueError("mode must be either 'create' or 'overwrite'")
        validate_table_name(name)
-
-        table_id = namespace_path + [name]
-
-        if "CreateTable" in self._namespace_client_pushdown_operations:
-            return self._create_table_server_side(
-                name=name,
-                data=data,
+        async_table = LOOP.run(
+            self._inner.create_table(
+                name,
+                data,
                schema=schema,
                mode=mode,
                exist_ok=exist_ok,
@@ -513,130 +525,15 @@ class LanceNamespaceDBConnection(DBConnection):
                namespace_path=namespace_path,
                storage_options=storage_options,
            )
-
-        # Local create path: declare_table + local write
-        # Step 1: Get the table location and storage options from namespace
-        # In overwrite mode, if table exists, use describe_table to get
-        # existing location. Otherwise, call create_empty_table to reserve
-        # a new location
-        location = None
-        namespace_storage_options = None
-        if mode.lower() == "overwrite":
-            # Try to describe the table first to see if it exists
-            try:
-                describe_request = DescribeTableRequest(id=table_id)
-                describe_response = self._namespace_client.describe_table(
-                    describe_request
-                )
-                location = describe_response.location
-                namespace_storage_options = describe_response.storage_options
-            except Exception:
-                # Table doesn't exist, will create a new one below
-                pass
-
-        if location is None:
-            # Table doesn't exist or mode is "create", reserve a new location
-            declare_request = DeclareTableRequest(
-                id=table_id,
-                location=None,
-                properties=self.storage_options if self.storage_options else None,
-            )
-            declare_response = self._namespace_client.declare_table(declare_request)
-
-            if not declare_response.location:
-                raise ValueError(
-                    "Table location is missing from declare_table response"
-                )
-
-            location = declare_response.location
-            namespace_storage_options = declare_response.storage_options
-
-        # Merge storage options: self.storage_options < user options < namespace options
-        merged_storage_options = dict(self.storage_options)
-        if storage_options:
-            merged_storage_options.update(storage_options)
-        if namespace_storage_options:
-            merged_storage_options.update(namespace_storage_options)
-
-        # Step 2: Create table using LanceTable.create with the location
-        # We need a temporary connection for the LanceTable.create method
-        temp_conn = LanceDBConnection(
-            location,  # Use the actual location as the connection URI
-            read_consistency_interval=self.read_consistency_interval,
-            storage_options=merged_storage_options,
-            session=self.session,
        )

-        # Note: storage_options_provider is auto-created in Rust from namespace_client
-        tbl = LanceTable.create(
-            temp_conn,
+        return LanceTable(
+            self,
            name,
-            data,
-            schema,
-            mode=mode,
-            exist_ok=exist_ok,
-            on_bad_vectors=on_bad_vectors,
-            fill_value=fill_value,
-            embedding_functions=embedding_functions,
            namespace_path=namespace_path,
-            storage_options=merged_storage_options,
-            location=location,
            namespace_client=self._namespace_client,
            pushdown_operations=self._namespace_client_pushdown_operations,
-        )
-
-        return tbl
-
-    def _create_table_server_side(
-        self,
-        name: str,
-        data: Optional[DATA],
-        schema: Optional[Union[pa.Schema, LanceModel]],
-        mode: str,
-        exist_ok: bool,
-        on_bad_vectors: str,
-        fill_value: float,
-        embedding_functions: Optional[List[EmbeddingFunctionConfig]],
-        namespace_path: Optional[List[str]],
-        storage_options: Optional[Dict[str, str]],
-    ) -> Table:
-        """Create a table using server-side namespace.create_table()."""
-        if namespace_path is None:
-            namespace_path = []
-        table_id = namespace_path + [name]
-
-        arrow_ipc_bytes = _data_to_arrow_ipc(
-            data=data,
-            schema=schema,
-            embedding_functions=embedding_functions,
-            on_bad_vectors=on_bad_vectors,
-            fill_value=fill_value,
-        )
-
-        merged = dict(self.storage_options or {})
-        if storage_options:
-            merged.update(storage_options)
-        request = CreateTableRequest(
-            id=table_id,
-            mode=_normalize_create_table_mode(mode),
-            properties=merged or None,
-        )
-
-        try:
-            self._namespace_client.create_table(request, arrow_ipc_bytes)
-        except Exception as e:
-            if exist_ok and "already exists" in str(e).lower():
-                return self.open_table(
-                    name,
-                    namespace_path=namespace_path,
-                    storage_options=storage_options,
-                )
-            raise
-
-        return self.open_table(
-            name,
-            namespace_path=namespace_path,
-            storage_options=storage_options,
+            _async=async_table,
        )

    @override
@@ -650,30 +547,28 @@ class LanceNamespaceDBConnection(DBConnection):
    ) -> Table:
        if namespace_path is None:
            namespace_path = []
-        table_id = namespace_path + [name]
-        request = DescribeTableRequest(id=table_id)
-        response = self._namespace_client.describe_table(request)
+        try:
+            async_table = LOOP.run(
+                self._inner.open_table(
+                    name,
+                    namespace_path=namespace_path,
+                    storage_options=storage_options,
+                    index_cache_size=index_cache_size,
+                )
+            )
+        except RuntimeError as e:
+            if "Table not found" in str(e):
+                table_id = namespace_path + [name]
+                raise TableNotFoundError(f"Table not found: {'$'.join(table_id)}")
+            raise

-        # Merge storage options: self.storage_options < user options < namespace options
-        merged_storage_options = dict(self.storage_options)
-        if storage_options:
-            merged_storage_options.update(storage_options)
-        if response.storage_options:
-            merged_storage_options.update(response.storage_options)
-
-        # Pass managed_versioning to avoid redundant describe_table call in Rust.
-        # Convert None to False since we already have the answer from describe_table.
-        managed_versioning = response.managed_versioning is True
-
-        # Note: storage_options_provider is auto-created in Rust from namespace_client
-        return self._lance_table_from_uri(
+        return LanceTable(
+            self,
            name,
-            response.location,
            namespace_path=namespace_path,
-            storage_options=merged_storage_options,
-            index_cache_size=index_cache_size,
            namespace_client=self._namespace_client,
-            managed_versioning=managed_versioning,
+            pushdown_operations=self._namespace_client_pushdown_operations,
+            _async=async_table,
        )

    @override
@@ -897,33 +792,34 @@ class LanceNamespaceDBConnection(DBConnection):
        namespace_client: Optional[Any] = None,
        managed_versioning: Optional[bool] = None,
    ) -> LanceTable:
-        # Open a table directly from a URI using the location parameter
-        # Note: storage_options should already be merged by the caller
-        # Note: storage_options_provider is auto-created in Rust from namespace_client
+        # Open a table directly from the namespace-resolved physical location.
+        #
+        # Open the table through the Rust namespace-backed connection.  The Rust
+        # layer keeps the logical namespace path and namespace client intact.
        if namespace_path is None:
            namespace_path = []
-        temp_conn = LanceDBConnection(
-            table_uri,  # Use the table location as the connection URI
-            read_consistency_interval=self.read_consistency_interval,
-            storage_options=storage_options if storage_options is not None else {},
-            session=self.session,
+
+        async_table = LOOP.run(
+            self._inner.open_table(
+                name,
+                namespace_path=namespace_path,
+                storage_options=storage_options,
+                index_cache_size=index_cache_size,
+                location=None,
+                namespace_client=namespace_client,
+                managed_versioning=managed_versioning,
+            )
        )

-        # Open the table using the temporary connection with the location parameter
-        # Pass namespace_client to enable managed versioning support and auto-create
-        # storage options provider
-        # Pass managed_versioning to avoid redundant describe_table call
-        # Pass pushdown_operations if configured on this connection
-        return LanceTable.open(
-            temp_conn,
+        return LanceTable(
+            self,
            name,
            namespace_path=namespace_path,
-            storage_options=storage_options,
-            index_cache_size=index_cache_size,
            location=table_uri,
            namespace_client=namespace_client,
            managed_versioning=managed_versioning,
            pushdown_operations=self._namespace_client_pushdown_operations,
+            _async=async_table,
        )

    @override
@@ -990,6 +886,23 @@ class AsyncLanceNamespaceDBConnection:
        self._namespace_client_pushdown_operations = set(
            namespace_client_pushdown_operations or []
        )
+        self._inner = AsyncConnection(
+            _connect_namespace_client(
+                namespace_client,
+                read_consistency_interval=(
+                    read_consistency_interval.total_seconds()
+                    if read_consistency_interval is not None
+                    else None
+                ),
+                storage_options=self.storage_options or None,
+                session=session,
+                namespace_client_pushdown_operations=(
+                    list(self._namespace_client_pushdown_operations)
+                ),
+                namespace_client_impl=None,
+                namespace_client_properties=None,
+            )
+        )

    async def table_names(
        self,
@@ -1041,148 +954,16 @@ class AsyncLanceNamespaceDBConnection:
        if mode.lower() not in ["create", "overwrite"]:
            raise ValueError("mode must be either 'create' or 'overwrite'")
        validate_table_name(name)
-
-        table_id = namespace_path + [name]
-
-        if "CreateTable" in self._namespace_client_pushdown_operations:
-            return await self._create_table_server_side(
-                name=name,
-                data=data,
-                schema=schema,
-                mode=mode,
-                exist_ok=exist_ok,
-                on_bad_vectors=on_bad_vectors,
-                fill_value=fill_value,
-                embedding_functions=embedding_functions,
-                namespace_path=namespace_path,
-                storage_options=storage_options,
-            )
-
-        # Local create path: declare_table + local write
-        # Step 1: Get the table location and storage options from namespace
-        location = None
-        namespace_storage_options = None
-        if mode.lower() == "overwrite":
-            # Try to describe the table first to see if it exists
-            try:
-                describe_request = DescribeTableRequest(id=table_id)
-                describe_response = self._namespace_client.describe_table(
-                    describe_request
-                )
-                location = describe_response.location
-                namespace_storage_options = describe_response.storage_options
-            except Exception:
-                # Table doesn't exist, will create a new one below
-                pass
-
-        if location is None:
-            # Table doesn't exist or mode is "create", reserve a new location
-            declare_request = DeclareTableRequest(
-                id=table_id,
-                location=None,
-                properties=self.storage_options if self.storage_options else None,
-            )
-            declare_response = self._namespace_client.declare_table(declare_request)
-
-            if not declare_response.location:
-                raise ValueError(
-                    "Table location is missing from declare_table response"
-                )
-
-            location = declare_response.location
-            namespace_storage_options = declare_response.storage_options
-
-        # Merge storage options: self.storage_options < user options < namespace options
-        merged_storage_options = dict(self.storage_options)
-        if storage_options:
-            merged_storage_options.update(storage_options)
-        if namespace_storage_options:
-            merged_storage_options.update(namespace_storage_options)
-
-        # Step 2: Create table using LanceTable.create with the location
-        # Run the sync operation in a thread
-        def _create_table():
-            temp_conn = LanceDBConnection(
-                location,
-                read_consistency_interval=self.read_consistency_interval,
-                storage_options=merged_storage_options,
-                session=self.session,
-            )
-
-            # storage_options_provider is auto-created in Rust from namespace_client
-            return LanceTable.create(
-                temp_conn,
-                name,
-                data,
-                schema,
-                mode=mode,
-                exist_ok=exist_ok,
-                on_bad_vectors=on_bad_vectors,
-                fill_value=fill_value,
-                embedding_functions=embedding_functions,
-                namespace_path=namespace_path,
-                storage_options=merged_storage_options,
-                location=location,
-                namespace_client=self._namespace_client,
-                pushdown_operations=self._namespace_client_pushdown_operations,
-            )
-
-        lance_table = await asyncio.to_thread(_create_table)
-        # Get the underlying async table from LanceTable
-        return lance_table._table
-
-    async def _create_table_server_side(
-        self,
-        name: str,
-        data: Optional[DATA],
-        schema: Optional[Union[pa.Schema, LanceModel]],
-        mode: str,
-        exist_ok: bool,
-        on_bad_vectors: str,
-        fill_value: float,
-        embedding_functions: Optional[List[EmbeddingFunctionConfig]],
-        namespace_path: Optional[List[str]],
-        storage_options: Optional[Dict[str, str]],
-    ) -> AsyncTable:
-        """Create a table using server-side namespace.create_table()."""
-        if namespace_path is None:
-            namespace_path = []
-        table_id = namespace_path + [name]
-
-        def _prepare_and_create():
-            arrow_ipc_bytes = _data_to_arrow_ipc(
-                data=data,
-                schema=schema,
-                embedding_functions=embedding_functions,
-                on_bad_vectors=on_bad_vectors,
-                fill_value=fill_value,
-            )
-
-            merged = dict(self.storage_options or {})
-            if storage_options:
-                merged.update(storage_options)
-            request = CreateTableRequest(
-                id=table_id,
-                mode=_normalize_create_table_mode(mode),
-                properties=merged or None,
-            )
-
-            self._namespace_client.create_table(request, arrow_ipc_bytes)
-
-        try:
-            await asyncio.to_thread(_prepare_and_create)
-        except Exception as e:
-            if exist_ok and "already exists" in str(e).lower():
-                return await self.open_table(
-                    name,
-                    namespace_path=namespace_path,
-                    storage_options=storage_options,
-                )
-            raise
-
-        return await self.open_table(
+        return await self._inner.create_table(
            name,
+            data,
+            schema=schema,
+            mode=mode,
+            exist_ok=exist_ok,
+            on_bad_vectors=on_bad_vectors,
+            fill_value=fill_value,
            namespace_path=namespace_path,
+            embedding_functions=embedding_functions,
            storage_options=storage_options,
        )

@@ -1197,45 +978,18 @@ class AsyncLanceNamespaceDBConnection:
        """Open an existing table from the namespace."""
        if namespace_path is None:
            namespace_path = []
-        table_id = namespace_path + [name]
-        request = DescribeTableRequest(id=table_id)
-        response = self._namespace_client.describe_table(request)
-
-        # Merge storage options: self.storage_options < user options < namespace options
-        merged_storage_options = dict(self.storage_options)
-        if storage_options:
-            merged_storage_options.update(storage_options)
-        if response.storage_options:
-            merged_storage_options.update(response.storage_options)
-
-        # Capture managed_versioning from describe response.
-        # Convert None to False since we already have the answer from describe_table.
-        managed_versioning = response.managed_versioning is True
-
-        # Open table in a thread
-        # Note: storage_options_provider is auto-created in Rust from namespace_client
-        def _open_table():
-            temp_conn = LanceDBConnection(
-                response.location,
-                read_consistency_interval=self.read_consistency_interval,
-                storage_options=merged_storage_options,
-                session=self.session,
-            )
-
-            return LanceTable.open(
-                temp_conn,
+        try:
+            return await self._inner.open_table(
                name,
                namespace_path=namespace_path,
-                storage_options=merged_storage_options,
+                storage_options=storage_options,
                index_cache_size=index_cache_size,
-                location=response.location,
-                namespace_client=self._namespace_client,
-                managed_versioning=managed_versioning,
-                pushdown_operations=self._namespace_client_pushdown_operations,
            )
-
-        lance_table = await asyncio.to_thread(_open_table)
-        return lance_table._table
+        except RuntimeError as e:
+            if "Table not found" in str(e):
+                table_id = namespace_path + [name]
+                raise TableNotFoundError(f"Table not found: {'$'.join(table_id)}")
+            raise

    async def drop_table(self, name: str, namespace_path: Optional[List[str]] = None):
        """Drop a table from the namespace."""
--- a/python/python/lancedb/permutation.py
+++ b/python/python/lancedb/permutation.py
@@ -9,7 +9,7 @@ import json
 from ._lancedb import async_permutation_builder, PermutationReader
 from .table import LanceTable
 from .background_loop import LOOP
-from .util import batch_to_tensor, batch_to_tensor_rows
+from .util import batch_to_tensor, batch_to_tensor_dict, batch_to_tensor_rows
 from typing import Any, Callable, Iterator, Literal, Optional, TYPE_CHECKING, Union

 if TYPE_CHECKING:
@@ -697,6 +697,7 @@ class Permutation:
            "pandas",
            "arrow",
            "torch",
+            "torch_row",
            "torch_col",
            "polars",
        ],
@@ -712,8 +713,17 @@ class Permutation:
        - "python_col" - the batch will be a dict of lists (one entry per column)
        - "pandas" - the batch will be a pandas DataFrame
        - "arrow" - the batch will be a pyarrow RecordBatch
-        - "torch" - the batch will be a list of tensors, one per row
-        - "torch_col" - the batch will be a 2D torch tensor (first dim indexes columns)
+        - "torch" - a list of per-row dicts whose values are torch tensors. When
+          used with ``torch.utils.data.DataLoader`` (default collate), each
+          batch yielded by the loader is ``dict[str, Tensor]`` — one tensor per
+          column, with column names preserved. This matches HuggingFace
+          ``dataset.set_format("torch")`` semantics.
+        - "torch_row" - a list of 1-D torch tensors, one per row. Each tensor
+          stacks all column values into a single row vector and column names
+          are not preserved. (This was the previous "torch" behavior.)
+        - "torch_col" - a 2-D torch tensor of shape ``(n_cols, n_rows)``. Column
+          names are not preserved. Requires ``collate_fn=lambda x: x`` if used
+          with ``DataLoader``.
        - "polars" - the batch will be a polars DataFrame

        Conversion may or may not involve a data copy.  Lance uses Arrow internally
@@ -741,6 +751,8 @@ class Permutation:
        elif format == "arrow":
            return self.with_transform(Transforms.arrow2arrow)
        elif format == "torch":
+            return self.with_transform(batch_to_tensor_dict)
+        elif format == "torch_row":
            return self.with_transform(batch_to_tensor_rows)
        elif format == "torch_col":
            return self.with_transform(batch_to_tensor)
--- a/python/python/lancedb/query.py
+++ b/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
@@ -1526,9 +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:
@@ -1552,90 +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.

--- a/python/python/lancedb/table.py
+++ b/python/python/lancedb/table.py
@@ -943,29 +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.
@@ -1746,6 +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:
        """
@@ -2405,84 +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
@@ -2929,6 +2915,7 @@ class LanceTable(Table):
                namespace_path=namespace_path,
                storage_options=storage_options,
                location=location,
+                namespace_client=namespace_client,
            )
        )
        return self
--- a/python/python/lancedb/util.py
+++ b/python/python/lancedb/util.py
@@ -448,3 +448,29 @@ def batch_to_tensor_rows(batch: pa.RecordBatch):
    stacked = torch.tensor(numpy.column_stack(columns))
    rows = list(stacked.unbind(dim=0))
    return rows
+
+
+def batch_to_tensor_dict(batch: pa.RecordBatch):
+    """
+    Convert a PyArrow RecordBatch into a list of per-row dicts whose values
+    are PyTorch tensors.
+
+    Each column is converted to a tensor in one shot (zero-copy via DLPack
+    when supported), then sliced per row. The result is shaped to work with
+    PyTorch's default DataLoader collate, which stacks the per-row dicts
+    into a single ``dict[str, Tensor]`` per batch — matching the
+    HuggingFace ``dataset.set_format("torch")`` convention.
+
+    Fails if torch is not installed.
+    Fails if a column's data type is not supported by PyTorch.
+    """
+    torch = attempt_import_or_raise("torch", "torch")
+    columns: dict[str, "torch.Tensor"] = {}
+    for i, name in enumerate(batch.schema.names):
+        col = batch.column(i)
+        try:
+            columns[name] = torch.from_dlpack(col)
+        except Exception:
+            columns[name] = torch.tensor(col.to_numpy(zero_copy_only=False))
+    n = batch.num_rows
+    return [{name: t[i] for name, t in columns.items()} for i in range(n)]
--- a/python/python/tests/docs/test_search.py
+++ b/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"}])
--- a/python/python/tests/test_db.py
+++ b/python/python/tests/test_db.py
@@ -15,8 +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)
@@ -49,7 +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"
--- a/python/python/tests/test_fts.py
+++ b/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(
--- a/python/python/tests/test_hybrid_query.py
+++ b/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


@@ -192,7 +192,7 @@ def table_with_id(tmpdir_factory) -> Table:
        }
    )
    table = db.create_table("test_with_id", data)
-    table.create_fts_index("text", with_position=False, use_tantivy=False)
+    table.create_fts_index("text", with_position=False)
    return table


--- a/python/python/tests/test_namespace_integration.py
+++ b/python/python/tests/test_namespace_integration.py
@@ -18,6 +18,9 @@ Tests verify:
 """

 import copy
+import shutil
+import sys
+import tempfile
 import time
 import uuid
 from typing import Dict, Optional
@@ -387,6 +390,66 @@ def test_namespace_open_table_with_provider(s3_bucket: str, use_custom: bool):
    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
@pytest.mark.parametrize("use_custom", [False, True], ids=["DirectoryNS", "CustomNS"])
 def test_namespace_credential_refresh_on_read(s3_bucket: str, use_custom: bool):
--- a/python/python/tests/test_permutation.py
+++ b/python/python/tests/test_permutation.py
@@ -950,14 +950,27 @@ def test_transform_fn(mem_db):
    try:
        import torch

+        # "torch" format: list of per-row dicts of tensors (HF-compatible).
        torch_result = list(
            permutation.with_format("torch").iter(10, skip_last_batch=False)
        )[0]
        assert isinstance(torch_result, list)
        assert len(torch_result) == 10
-        assert isinstance(torch_result[0], torch.Tensor)
-        assert torch_result[0].shape == (2,)
-        assert torch_result[0].dtype == torch.int64
+        assert isinstance(torch_result[0], dict)
+        assert set(torch_result[0].keys()) == {"id", "value"}
+        assert isinstance(torch_result[0]["id"], torch.Tensor)
+        assert torch_result[0]["id"].shape == ()
+        assert torch_result[0]["id"].dtype == torch.int64
+
+        # "torch_row" format: list of 1-D row tensors (previous "torch" behavior).
+        torch_row_result = list(
+            permutation.with_format("torch_row").iter(10, skip_last_batch=False)
+        )[0]
+        assert isinstance(torch_row_result, list)
+        assert len(torch_row_result) == 10
+        assert isinstance(torch_row_result[0], torch.Tensor)
+        assert torch_row_result[0].shape == (2,)
+        assert torch_row_result[0].dtype == torch.int64
    except ImportError:
        # Skip check if torch is not installed
        pass
--- a/python/python/tests/test_query.py
+++ b/python/python/tests/test_query.py
@@ -1385,7 +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))
--- a/python/python/tests/test_rerankers.py
+++ b/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")
--- a/python/python/tests/test_s3.py
+++ b/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)

--- a/python/python/tests/test_table.py
+++ b/python/python/tests/test_table.py
@@ -1948,7 +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")
@@ -2019,7 +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
--- a/python/python/tests/test_torch.py
+++ b/python/python/tests/test_torch.py
@@ -27,8 +27,18 @@ def test_permutation_dataloader(mem_db):
    for batch in dataloader:
        assert batch["a"].size(0) == 10

+    # New "torch" format: per-row dicts of tensors, default collate yields
+    # dict[str, Tensor] (HuggingFace style).
    permutation = permutation.with_format("torch")
    dataloader = torch.utils.data.DataLoader(permutation, batch_size=10, shuffle=True)
+    for batch in dataloader:
+        assert isinstance(batch, dict)
+        assert "a" in batch
+        assert batch["a"].size() == (10,)
+
+    # Previous "torch" semantics is preserved under the "torch_row" name.
+    permutation = permutation.with_format("torch_row")
+    dataloader = torch.utils.data.DataLoader(permutation, batch_size=10, shuffle=True)
    for batch in dataloader:
        assert batch.size(0) == 10
        assert batch.size(1) == 1
--- a/python/src/connection.rs
+++ b/python/src/connection.rs
@@ -1,11 +1,17 @@
 // SPDX-License-Identifier: Apache-2.0
 // SPDX-FileCopyrightText: Copyright The LanceDB Authors

-use std::{collections::HashMap, sync::Arc, time::Duration};
+use std::{
+    collections::{HashMap, HashSet},
+    sync::Arc,
+    time::Duration,
+};

 use arrow::{datatypes::Schema, ffi_stream::ArrowArrayStreamReader, pyarrow::FromPyArrow};
 use lancedb::{
    connection::Connection as LanceConnection,
+    connection::NamespaceClientPushdownOperation,
+    database::namespace::LanceNamespaceDatabase,
    database::{CreateTableMode, Database, ReadConsistency},
 };
 use pyo3::{
@@ -39,6 +45,29 @@ impl Connection {
    }
 }

+fn parse_namespace_client_pushdown_operations(
+    operations: Option<Vec<String>>,
+) -> PyResult<HashSet<NamespaceClientPushdownOperation>> {
+    let mut parsed = HashSet::new();
+    for operation in operations.unwrap_or_default() {
+        match operation.as_str() {
+            "QueryTable" => {
+                parsed.insert(NamespaceClientPushdownOperation::QueryTable);
+            }
+            "CreateTable" => {
+                parsed.insert(NamespaceClientPushdownOperation::CreateTable);
+            }
+            _ => {
+                return Err(PyValueError::new_err(format!(
+                    "Invalid pushdown operation: {}",
+                    operation
+                )));
+            }
+        }
+    }
+    Ok(parsed)
+}
+
 impl Connection {
    fn parse_create_mode_str(mode: &str) -> PyResult<CreateTableMode> {
        match mode {
@@ -496,7 +525,7 @@ impl Connection {
 }

 #[pyfunction]
-#[pyo3(signature = (uri, api_key=None, region=None, host_override=None, read_consistency_interval=None, client_config=None, storage_options=None, session=None))]
+#[pyo3(signature = (uri, api_key=None, region=None, host_override=None, read_consistency_interval=None, client_config=None, storage_options=None, session=None, manifest_enabled=false, namespace_client_properties=None))]
 #[allow(clippy::too_many_arguments)]
 pub fn connect(
    py: Python<'_>,
@@ -508,6 +537,8 @@ pub fn connect(
    client_config: Option<PyClientConfig>,
    storage_options: Option<HashMap<String, String>>,
    session: Option<crate::session::Session>,
+    manifest_enabled: bool,
+    namespace_client_properties: Option<HashMap<String, String>>,
 ) -> PyResult<Bound<'_, PyAny>> {
    future_into_py(py, async move {
        let mut builder = lancedb::connect(&uri);
@@ -527,6 +558,12 @@ pub fn connect(
        if let Some(storage_options) = storage_options {
            builder = builder.storage_options(storage_options);
        }
+        if manifest_enabled {
+            builder = builder.manifest_enabled(true);
+        }
+        if let Some(namespace_client_properties) = namespace_client_properties {
+            builder = builder.namespace_client_properties(namespace_client_properties);
+        }
        #[cfg(feature = "remote")]
        if let Some(client_config) = client_config {
            builder = builder.client_config(client_config.into());
@@ -538,6 +575,52 @@ pub fn connect(
    })
 }

+#[pyfunction]
+#[pyo3(signature = (
+    namespace_client,
+    read_consistency_interval=None,
+    storage_options=None,
+    session=None,
+    namespace_client_pushdown_operations=None,
+    namespace_client_impl=None,
+    namespace_client_properties=None,
+))]
+#[allow(clippy::too_many_arguments)]
+pub fn connect_namespace_client(
+    py: Python<'_>,
+    namespace_client: Py<PyAny>,
+    read_consistency_interval: Option<f64>,
+    storage_options: Option<HashMap<String, String>>,
+    session: Option<crate::session::Session>,
+    namespace_client_pushdown_operations: Option<Vec<String>>,
+    namespace_client_impl: Option<String>,
+    namespace_client_properties: Option<HashMap<String, String>>,
+) -> PyResult<Connection> {
+    let namespace_client = extract_namespace_arc(py, namespace_client)?;
+    let read_consistency_interval = read_consistency_interval.map(Duration::from_secs_f64);
+    let namespace_client_pushdown_operations =
+        parse_namespace_client_pushdown_operations(namespace_client_pushdown_operations)?;
+    let ns_impl = namespace_client_impl.unwrap_or_else(|| "python".to_string());
+    let ns_properties = namespace_client_properties.unwrap_or_default();
+    let storage_options = storage_options.unwrap_or_default();
+    let session = session.map(|s| s.inner.clone());
+
+    let database = LanceNamespaceDatabase::from_namespace_client(
+        namespace_client,
+        ns_impl,
+        ns_properties,
+        storage_options,
+        read_consistency_interval,
+        session,
+        namespace_client_pushdown_operations,
+    );
+
+    Ok(Connection::new(LanceConnection::new(
+        Arc::new(database),
+        Arc::new(lancedb::embeddings::MemoryRegistry::new()),
+    )))
+}
+
 #[derive(FromPyObject)]
 pub struct PyClientConfig {
    user_agent: String,
--- a/python/src/lib.rs
+++ b/python/src/lib.rs
@@ -2,7 +2,7 @@
 // SPDX-FileCopyrightText: Copyright The LanceDB Authors

 use arrow::RecordBatchStream;
-use connection::{Connection, connect};
+use connection::{Connection, connect, connect_namespace_client};
 use env_logger::Env;
 use expr::{PyExpr, expr_col, expr_func, expr_lit};
 use index::IndexConfig;
@@ -58,6 +58,7 @@ pub fn _lancedb(_py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
    m.add_class::<PyPermutationReader>()?;
    m.add_class::<PyExpr>()?;
    m.add_function(wrap_pyfunction!(connect, m)?)?;
+    m.add_function(wrap_pyfunction!(connect_namespace_client, m)?)?;
    m.add_function(wrap_pyfunction!(permutation::async_permutation_builder, m)?)?;
    m.add_function(wrap_pyfunction!(util::validate_table_name, m)?)?;
    m.add_function(wrap_pyfunction!(query::fts_query_to_json, m)?)?;
--- a/python/uv.lock
+++ b/python/uv.lock
@@ -1996,7 +1996,6 @@ tests = [
    { name = "pytest-mock" },
    { name = "pytz" },
    { name = "requests" },
-    { name = "tantivy" },
 ]

 [package.metadata]
@@ -2050,7 +2049,6 @@ requires-dist = [
    { name = "sentence-transformers", marker = "extra == 'embeddings'", specifier = ">=2.2.0" },
    { name = "sentencepiece", marker = "extra == 'embeddings'", specifier = ">=0.1.99" },
    { name = "sentencepiece", marker = "extra == 'siglip'" },
-    { name = "tantivy", marker = "extra == 'tests'", specifier = ">=0.20.0" },
    { name = "torch", marker = "extra == 'clip'" },
    { name = "torch", marker = "extra == 'embeddings'", specifier = ">=2.0.0" },
    { name = "torch", marker = "extra == 'siglip'" },
@@ -4779,44 +4777,6 @@ wheels = [
    { url = "https://files.pythonhosted.org/packages/40/44/4a5f08c96eb108af5cb50b41f76142f0afa346dfa99d5296fe7202a11854/tabulate-0.9.0-py3-none-any.whl", hash = "sha256:024ca478df22e9340661486f85298cff5f6dcdba14f3813e8830015b9ed1948f", size = 35252, upload-time = "2022-10-06T17:21:44.262Z" },
 ]

-[[package]]
-name = "tantivy"
-version = "0.25.1"
-source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/1b/f9/0cd3955d155d3e3ef74b864769514dd191e5dacba9f0beb7af2d914942ce/tantivy-0.25.1.tar.gz", hash = "sha256:68a3314699a7d18fcf338b52bae8ce46a97dde1128a3e47e33fa4db7f71f265e", size = 75120, upload-time = "2025-12-02T11:57:12.997Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/80/f7/2276bed3bed983ce2970dc70e3571f372587fe4f5f2bac1d6d617df08fa3/tantivy-0.25.1-cp310-cp310-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl", hash = "sha256:7aa587a3dc9470584cacf5e3640fee93d12ec5f10109669c1f47c4e90820b958", size = 7638510, upload-time = "2025-12-02T11:56:08.754Z" },
-    { url = "https://files.pythonhosted.org/packages/20/8c/078dc50570e243414356b05633f52fe544b85179281ffa9f1fe05d76bbd8/tantivy-0.25.1-cp310-cp310-macosx_10_12_x86_64.whl", hash = "sha256:56d77fe667595693d9fa5f0b4545776d84da9526bab0273b3fc6c7536dc0d8a2", size = 3932659, upload-time = "2025-12-02T11:56:10.621Z" },
-    { url = "https://files.pythonhosted.org/packages/bd/dc/281c48436a1e3178b58fe463af314434fe0f3a4ec0c7588a362900e0c69e/tantivy-0.25.1-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5ba8c347cd48595fcaeabb28a909ebce92cf9c5e5c84ab5ba1136a280a307b5c", size = 4197430, upload-time = "2025-12-02T11:56:12.65Z" },
-    { url = "https://files.pythonhosted.org/packages/7b/6c/61e6e0b0a350007d10a9b66a35703361d3345e14e7a7cc83494776b2a054/tantivy-0.25.1-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:aa7c4932e8fde1f09f2d46226060e827e197c2749abdc6129d73a752773adc38", size = 4184055, upload-time = "2025-12-02T11:56:14.647Z" },
-    { url = "https://files.pythonhosted.org/packages/5f/fd/0eb059b12f0b6f91623a54a46448a83b7f716d08f3bca68c095d697b85da/tantivy-0.25.1-cp310-cp310-win_amd64.whl", hash = "sha256:afcfc5dbb0bcd5d24531f4471737ae0896f33528426ab0b1dad3e427c19120f6", size = 3424134, upload-time = "2025-12-02T11:56:16.242Z" },
-    { url = "https://files.pythonhosted.org/packages/4e/7a/8a277f377e8a151fc0e71d4ffc1114aefb6e5e1c7dd609fed0955cf34ed8/tantivy-0.25.1-cp311-cp311-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl", hash = "sha256:d363d7b4207d3a5aa7f0d212420df35bed18bdb6bae26a2a8bd57428388b7c29", size = 7637033, upload-time = "2025-12-02T11:56:18.104Z" },
-    { url = "https://files.pythonhosted.org/packages/71/31/8b4acdedfc9f9a2d04b1340d07eef5213d6f151d1e18da0cb423e5f090d2/tantivy-0.25.1-cp311-cp311-macosx_10_12_x86_64.whl", hash = "sha256:8f4389cf1d889a1df7c5a3195806b4b56c37cee10d8a26faaa0dea35a867b5ff", size = 3932180, upload-time = "2025-12-02T11:56:19.833Z" },
-    { url = "https://files.pythonhosted.org/packages/2f/dc/3e8499c21b4b9795e8f2fc54c68ce5b92905aaeadadaa56ecfa9180b11b1/tantivy-0.25.1-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:99864c09fc54652c3c2486cdf13f86cdc8200f4b481569cb291e095ca5d496e5", size = 4197620, upload-time = "2025-12-02T11:56:21.496Z" },
-    { url = "https://files.pythonhosted.org/packages/f8/8e/f2ce62fffc811eb62bead92c7b23c2e218f817cbd54c4f3b802e03ba1438/tantivy-0.25.1-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:05abf37ddbc5063c575548be0d62931629c086bff7a5a1b67cf5a8f5ebf4cd8c", size = 4183794, upload-time = "2025-12-02T11:56:23.215Z" },
-    { url = "https://files.pythonhosted.org/packages/de/64/24e2891b0ba3fd9853e10c296095a33b89bf3efd65e29da1ee5dae736040/tantivy-0.25.1-cp311-cp311-win_amd64.whl", hash = "sha256:f307ee8ad21597b0be23af83008fd66cfd5f958cdfa24ec0aaa08a38e86bbef4", size = 3424235, upload-time = "2025-12-02T11:56:25.172Z" },
-    { url = "https://files.pythonhosted.org/packages/41/e7/6849c713ed0996c7628324c60512c4882006f0a62145e56c624a93407f90/tantivy-0.25.1-cp312-cp312-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl", hash = "sha256:90fd919e5f611809f746560ecf36eb9be824dec62e21ae17a27243759edb9aa1", size = 7621494, upload-time = "2025-12-02T11:56:27.069Z" },
-    { url = "https://files.pythonhosted.org/packages/c5/22/c3d8294600dc6e7fa350daef9ff337d3c06e132b81df727de9f7a50c692a/tantivy-0.25.1-cp312-cp312-macosx_10_12_x86_64.whl", hash = "sha256:4613c7cf6c23f3a97989819690a0f956d799354957de7a204abcc60083cebe02", size = 3925219, upload-time = "2025-12-02T11:56:29.403Z" },
-    { url = "https://files.pythonhosted.org/packages/41/fc/cbb1df71dd44c9110eff4eaaeda9d44f2d06182fe0452193be20ddfba93f/tantivy-0.25.1-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c477bd20b4df804d57dfc5033431bef27cde605695ae141b03abbf6ebc069129", size = 4198699, upload-time = "2025-12-02T11:56:31.359Z" },
-    { url = "https://files.pythonhosted.org/packages/47/4d/71abb78b774073c3ce12a4faa4351a9d910a71ffa3659526affba163873d/tantivy-0.25.1-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:f9b1a1ba1113c523c7ff7b10f282d6c4074006f7ef8d71e1d973d51bf7291ddb", size = 4183585, upload-time = "2025-12-02T11:56:33.317Z" },
-    { url = "https://files.pythonhosted.org/packages/be/16/3f00cd7ec458b92a0e977960af9ddfbeb762127d9acc68da9094a1fda556/tantivy-0.25.1-cp312-cp312-win_amd64.whl", hash = "sha256:9de0bafd3bd7ac9f8f82d53e17562e9db11a5af308fe5185c4bd86feaddbe4a6", size = 3424622, upload-time = "2025-12-02T11:56:34.788Z" },
-    { url = "https://files.pythonhosted.org/packages/3d/25/73cfbcf1a8ea49be6c42817431cac46b70a119fe64da903fcc2d92b5b511/tantivy-0.25.1-cp313-cp313-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl", hash = "sha256:f51ff7196c6f31719202080ed8372d5e3d51e92c749c032fb8234f012e99744c", size = 7622530, upload-time = "2025-12-02T11:56:36.839Z" },
-    { url = "https://files.pythonhosted.org/packages/12/c8/c0d7591cdf4f7e7a9fc4da786d1ca8cd1aacffaa2be16ea6d401a8e4a566/tantivy-0.25.1-cp313-cp313-macosx_10_12_x86_64.whl", hash = "sha256:550e63321bfcacc003859f2fa29c1e8e56450807b3c9a501c1add27cfb9236d9", size = 3925637, upload-time = "2025-12-02T11:56:38.425Z" },
-    { url = "https://files.pythonhosted.org/packages/3a/09/bedfc223bffec7641b417dd7ab071134b2ef8f8550e9b1fb6014657ef52e/tantivy-0.25.1-cp313-cp313-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:fde31cc8d6e122faf7902aeea32bc008a429a6e8904e34d3468126a3ec01b016", size = 4197322, upload-time = "2025-12-02T11:56:40.411Z" },
-    { url = "https://files.pythonhosted.org/packages/f5/f1/1fa5183500c8042200c9f2b840d34f5bbcfb434a1ee750e7132262d2a5c9/tantivy-0.25.1-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:b11bd5a518b0be645320b47af8493f6a40c4f3234313e37adcf4534a564d27dd", size = 4183143, upload-time = "2025-12-02T11:56:42.048Z" },
-    { url = "https://files.pythonhosted.org/packages/d5/74/a4c4f4eb95888ccb784da3b017aa0625ab1ac411bf5d022a9a797d9a2334/tantivy-0.25.1-cp313-cp313-win_amd64.whl", hash = "sha256:cc7fe88853e06b3251ee4fa42b7a2038727f850c8765bcc8167cfc73585dd24e", size = 3423491, upload-time = "2025-12-02T11:56:43.858Z" },
-    { url = "https://files.pythonhosted.org/packages/8b/2f/581519492226f97d23bd0adc95dad991ebeaa73ea6abc8bff389a3096d9a/tantivy-0.25.1-cp313-cp313t-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl", hash = "sha256:dae99e75b7eaa9bf5bd16ab106b416370f08c135aed0e117d62a3201cd1ffe36", size = 7610316, upload-time = "2025-12-02T11:56:45.927Z" },
-    { url = "https://files.pythonhosted.org/packages/91/40/5d7bc315ab9e6a22c5572656e8ada1c836cfa96dccf533377504fbc3c9d9/tantivy-0.25.1-cp313-cp313t-macosx_10_12_x86_64.whl", hash = "sha256:506e9533c5ef4d3df43bad64ffecc0aa97c76e361ea610815dc3a20a9d6b30b3", size = 3919882, upload-time = "2025-12-02T11:56:48.469Z" },
-    { url = "https://files.pythonhosted.org/packages/02/b9/e0ef2f57a6a72444cb66c2ffbc310ab33ffaace275f1c4b0319d84ea3f18/tantivy-0.25.1-cp313-cp313t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:5dbd4f8f264dacbcc9dee542832da2173fd53deaaea03f082d95214f8b5ed6bc", size = 4196031, upload-time = "2025-12-02T11:56:50.151Z" },
-    { url = "https://files.pythonhosted.org/packages/1e/02/bf3f8cacfd08642e14a73f7956a3fb95d58119132c98c121b9065a1f8615/tantivy-0.25.1-cp313-cp313t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:824c643ccb640dd9e35e00c5d5054ddf3323f56fe4219d57d428a9eeea13d22c", size = 4183437, upload-time = "2025-12-02T11:56:51.818Z" },
-    { url = "https://files.pythonhosted.org/packages/9c/83/afa90e570198e2d1139dd567bec3c9cf44d8c54f63a649f16d711ede02f5/tantivy-0.25.1-cp313-cp313t-win_amd64.whl", hash = "sha256:09c987b840afcebac817836ac08407eff17272d8aa60ce6e291f89c81830221d", size = 3419409, upload-time = "2025-12-02T11:56:53.451Z" },
-    { url = "https://files.pythonhosted.org/packages/ff/44/9f1d67aa5030f7eebc966c863d1316a510a971dd8bb45651df4acdfae9ed/tantivy-0.25.1-cp314-cp314-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl", hash = "sha256:7f5d29ae85dd0f23df8d15b3e7b341d4f9eb5a446bbb9640df48ac1f6d9e0c6c", size = 7623723, upload-time = "2025-12-02T11:56:55.066Z" },
-    { url = "https://files.pythonhosted.org/packages/db/30/6e085bd3ed9d12da3c91c185854abd70f9dfd35fb36a75ea98428d42c30b/tantivy-0.25.1-cp314-cp314-macosx_10_12_x86_64.whl", hash = "sha256:f2d2938fb69a74fc1bb36edfaf7f0d1596fa1264db0f377bda2195c58bcb6245", size = 3926243, upload-time = "2025-12-02T11:56:57.058Z" },
-    { url = "https://files.pythonhosted.org/packages/32/f5/a00d65433430f51718e5cc6938df571765d7c4e03aedec5aef4ab567aa9b/tantivy-0.25.1-cp314-cp314-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:4f5ff124c4802558e627091e780b362ca944169736caba5a372eef39a79d0ae0", size = 4207186, upload-time = "2025-12-02T11:56:58.803Z" },
-    { url = "https://files.pythonhosted.org/packages/19/63/61bdb12fc95f2a7f77bd419a5149bfa9f28caa76cb569bf2b6b06e1d033e/tantivy-0.25.1-cp314-cp314-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:43b80ef62a340416139c93d19264e5f808da48e04f9305f1092b8ed22be0a5be", size = 4187312, upload-time = "2025-12-02T11:57:00.595Z" },
-    { url = "https://files.pythonhosted.org/packages/b7/de/e39c0b01d59019bf5c38face8b81defbc4a68cebf5e0c53bcb2cd715a449/tantivy-0.25.1-cp314-cp314-win_amd64.whl", hash = "sha256:286b654f40c70c1e6b64b9bc7031ed0bf5c440f5bffeaeeee21a0ee6cc39f0e2", size = 3436535, upload-time = "2025-12-02T11:57:02.267Z" },
-]
-
 [[package]]
 name = "threadpoolctl"
 version = "3.6.0"
--- a/rust/lancedb/Cargo.toml
+++ b/rust/lancedb/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "lancedb"
-version = "0.28.0-beta.5"
+version = "0.28.0-beta.10"
 edition.workspace = true
 description = "LanceDB: A serverless, low-latency vector database for AI applications"
 license.workspace = true
@@ -111,7 +111,12 @@ default = []
 aws = ["lance/aws", "lance-io/aws", "lance-namespace-impls/dir-aws"]
 oss = ["lance/oss", "lance-io/oss", "lance-namespace-impls/dir-oss"]
 gcs = ["lance/gcp", "lance-io/gcp", "lance-namespace-impls/dir-gcp"]
-azure = ["lance/azure", "lance-io/azure", "lance-namespace-impls/dir-azure"]
+azure = [
+    "lance/azure",
+    "lance-io/azure",
+    "lance-namespace-impls/dir-azure",
+    "lance-namespace-impls/credential-vendor-azure",
+]
 huggingface = [
    "lance/huggingface",
    "lance-io/huggingface",
--- a/rust/lancedb/src/connection.rs
+++ b/rust/lancedb/src/connection.rs
@@ -590,6 +590,15 @@ pub struct ConnectRequest {
    /// storage options.
    pub namespace_client_properties: HashMap<String, String>,

+    /// Use directory namespace manifests as the source of truth for native
+    /// LanceDB table metadata.
+    ///
+    /// When enabled for a local/native connection, LanceDB returns a
+    /// namespace-backed database directly. Directory listing fallback remains
+    /// enabled for migration, and directory-listing-to-manifest migration is
+    /// forced on.
+    pub manifest_enabled: bool,
+
    /// The interval at which to check for updates from other processes.
    ///
    /// If None, then consistency is not checked. For performance
@@ -630,6 +639,7 @@ impl ConnectBuilder {
                read_consistency_interval: None,
                options: HashMap::new(),
                namespace_client_properties: HashMap::new(),
+                manifest_enabled: false,
                session: None,
            },
            embedding_registry: None,
@@ -791,6 +801,17 @@ impl ConnectBuilder {
        self
    }

+    /// Enable or disable manifest-backed directory namespace mode for local
+    /// native connections.
+    ///
+    /// When enabled, the connection uses the directory namespace database
+    /// directly for all table operations and forces
+    /// `dir_listing_to_manifest_migration_enabled=true`.
+    pub fn manifest_enabled(mut self, enabled: bool) -> Self {
+        self.request.manifest_enabled = enabled;
+        self
+    }
+
    /// The interval at which to check for updates from other processes. This
    /// only affects LanceDB OSS.
    ///
@@ -886,6 +907,16 @@ impl ConnectBuilder {
    pub async fn execute(self) -> Result<Connection> {
        if self.request.uri.starts_with("db") {
            self.execute_remote()
+        } else if self.request.manifest_enabled {
+            let internal = Arc::new(
+                ListingDatabase::connect_manifest_enabled_namespace_database(&self.request).await?,
+            );
+            Ok(Connection {
+                internal,
+                embedding_registry: self
+                    .embedding_registry
+                    .unwrap_or_else(|| Arc::new(MemoryRegistry::new())),
+            })
        } else {
            let internal = Arc::new(ListingDatabase::connect_with_options(&self.request).await?);
            Ok(Connection {
@@ -915,7 +946,7 @@ use std::collections::HashSet;
 /// These operations will be executed on the namespace server instead of locally
 /// when enabled via [`ConnectNamespaceBuilder::pushdown_operations`].
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
-pub enum PushdownOperation {
+pub enum NamespaceClientPushdownOperation {
    /// Execute queries on the namespace server via `query_table()` instead of locally.
    QueryTable,
    /// Execute table creation on the namespace server via `create_table()`
@@ -931,7 +962,7 @@ pub struct ConnectNamespaceBuilder {
    read_consistency_interval: Option<std::time::Duration>,
    embedding_registry: Option<Arc<dyn EmbeddingRegistry>>,
    session: Option<Arc<lance::session::Session>>,
-    pushdown_operations: HashSet<PushdownOperation>,
+    pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
 }

 impl ConnectNamespaceBuilder {
@@ -1029,11 +1060,11 @@ impl ConnectNamespaceBuilder {
    /// and leveraging server-side compute resources.
    ///
    /// Available operations:
-    /// - [`PushdownOperation::QueryTable`]: Execute queries via `namespace.query_table()`
-    /// - [`PushdownOperation::CreateTable`]: Execute table creation via `namespace.create_table()`
+    /// - [`NamespaceClientPushdownOperation::QueryTable`]: Execute queries via `namespace.query_table()`
+    /// - [`NamespaceClientPushdownOperation::CreateTable`]: Execute table creation via `namespace.create_table()`
    ///
    /// By default, no operations are pushed down (all executed locally).
-    pub fn pushdown_operation(mut self, operation: PushdownOperation) -> Self {
+    pub fn pushdown_operation(mut self, operation: NamespaceClientPushdownOperation) -> Self {
        self.pushdown_operations.insert(operation);
        self
    }
@@ -1043,7 +1074,7 @@ impl ConnectNamespaceBuilder {
    /// See [`Self::pushdown_operation`] for details.
    pub fn pushdown_operations(
        mut self,
-        operations: impl IntoIterator<Item = PushdownOperation>,
+        operations: impl IntoIterator<Item = NamespaceClientPushdownOperation>,
    ) -> Self {
        self.pushdown_operations.extend(operations);
        self
@@ -1132,6 +1163,9 @@ mod tests {
    use lance_testing::datagen::{BatchGenerator, IncrementingInt32};
    use tempfile::tempdir;

+    use crate::database::listing::{ListingDatabaseOptions, OPT_NEW_TABLE_V2_MANIFEST_PATHS};
+    use crate::database::namespace::LanceNamespaceDatabase;
+    use crate::table::NativeTable;
    use crate::test_utils::connection::new_test_connection;

    use super::*;
@@ -1204,6 +1238,147 @@ mod tests {
        );
    }

+    #[tokio::test]
+    async fn test_connect_with_manifest_enabled_uses_directory_namespace() {
+        let tmp_dir = tempdir().unwrap();
+        let uri = tmp_dir.path().to_str().unwrap();
+
+        let db = connect(uri)
+            .manifest_enabled(true)
+            .storage_option("timeout", "30s")
+            .namespace_client_property("manifest_enabled", "false")
+            .namespace_client_property("dir_listing_to_manifest_migration_enabled", "false")
+            .execute()
+            .await
+            .unwrap();
+
+        assert!(
+            db.database()
+                .as_any()
+                .downcast_ref::<LanceNamespaceDatabase>()
+                .is_some()
+        );
+        assert_eq!(db.uri(), uri);
+
+        let (ns_impl, properties) = db.namespace_client_config().await.unwrap();
+        assert_eq!(ns_impl, "dir");
+        assert_eq!(properties.get("root"), Some(&uri.to_string()));
+        assert_eq!(
+            properties.get("manifest_enabled"),
+            Some(&"true".to_string())
+        );
+        assert_eq!(
+            properties.get("dir_listing_to_manifest_migration_enabled"),
+            Some(&"true".to_string())
+        );
+        assert_eq!(properties.get("storage.timeout"), Some(&"30s".to_string()));
+    }
+
+    #[tokio::test]
+    async fn test_manifest_enabled_rejects_commit_engine_uri() {
+        let Err(err) = connect("s3+ddb://bucket/db?ddbTableName=manifest")
+            .manifest_enabled(true)
+            .execute()
+            .await
+        else {
+            panic!("expected manifest-enabled s3+ddb connection to fail");
+        };
+        assert!(
+            matches!(err, Error::NotSupported { message } if message.contains("commit engine URI schemes"))
+        );
+
+        let Err(err) = connect("s3://bucket/db?engine=ddb&ddbTableName=manifest")
+            .manifest_enabled(true)
+            .execute()
+            .await
+        else {
+            panic!("expected manifest-enabled engine query connection to fail");
+        };
+        assert!(
+            matches!(err, Error::NotSupported { message } if message.contains("commit engine"))
+        );
+    }
+
+    #[tokio::test]
+    async fn test_manifest_enabled_connection_migrates_root_listing_table() {
+        let tmp_dir = tempdir().unwrap();
+        let uri = tmp_dir.path().to_str().unwrap();
+        let schema = Arc::new(Schema::new(vec![Field::new("x", DataType::Int32, false)]));
+
+        connect(uri)
+            .execute()
+            .await
+            .unwrap()
+            .create_empty_table("legacy", schema)
+            .execute()
+            .await
+            .unwrap();
+
+        let db = connect(uri).manifest_enabled(true).execute().await.unwrap();
+        let tables = db.table_names().execute().await.unwrap();
+        assert_eq!(tables, vec!["legacy".to_string()]);
+        db.open_table("legacy").execute().await.unwrap();
+    }
+
+    #[tokio::test]
+    async fn test_manifest_enabled_preserves_new_table_options() {
+        let tmp_dir = tempdir().unwrap();
+        let uri = tmp_dir.path().to_str().unwrap();
+        let options = ListingDatabaseOptions::builder()
+            .enable_v2_manifest_paths(true)
+            .build();
+        let schema = Arc::new(Schema::new(vec![Field::new("x", DataType::Int32, false)]));
+
+        let table = connect(uri)
+            .manifest_enabled(true)
+            .database_options(&options)
+            .execute()
+            .await
+            .unwrap()
+            .create_empty_table("v1_manifest", schema)
+            .storage_option(OPT_NEW_TABLE_V2_MANIFEST_PATHS, "false")
+            .execute()
+            .await
+            .unwrap();
+
+        let native_table = table
+            .base_table()
+            .as_any()
+            .downcast_ref::<NativeTable>()
+            .unwrap();
+        assert!(!native_table.uses_v2_manifest_paths().await.unwrap());
+    }
+
+    #[tokio::test]
+    async fn test_manifest_enabled_vend_input_storage_options() {
+        let tmp_dir = tempdir().unwrap();
+        let uri = tmp_dir.path().to_str().unwrap();
+        let schema = Arc::new(Schema::new(vec![Field::new("x", DataType::Int32, false)]));
+
+        let table = connect(uri)
+            .manifest_enabled(true)
+            .storage_option("test_storage_option", "test_value")
+            .namespace_client_property("vend_input_storage_options", "true")
+            .namespace_client_property(
+                "vend_input_storage_options_refresh_interval_millis",
+                "60000",
+            )
+            .execute()
+            .await
+            .unwrap()
+            .create_empty_table("vended", schema)
+            .execute()
+            .await
+            .unwrap();
+
+        let storage_options = table.latest_storage_options().await.unwrap().unwrap();
+        assert_eq!(
+            storage_options.get("test_storage_option"),
+            Some(&"test_value".to_string())
+        );
+        assert!(storage_options.contains_key("expires_at_millis"));
+    }
+
    #[tokio::test]
    async fn test_table_names() {
        let tc = new_test_connection().await.unwrap();
--- a/rust/lancedb/src/database/listing.rs
+++ b/rust/lancedb/src/database/listing.rs
@@ -285,7 +285,7 @@ const MIRRORED_STORE: &str = "mirroredStore";

 /// A connection to LanceDB
 impl ListingDatabase {
-    fn build_namespace_client_properties(
+    pub(crate) fn build_namespace_client_properties(
        uri: &str,
        storage_options: &HashMap<String, String>,
        namespace_client_properties: HashMap<String, String>,
@@ -298,6 +298,24 @@ impl ListingDatabase {
        properties
    }

+    pub(crate) fn build_manifest_enabled_namespace_client_properties(
+        uri: &str,
+        storage_options: &HashMap<String, String>,
+        namespace_client_properties: HashMap<String, String>,
+    ) -> HashMap<String, String> {
+        let mut properties = Self::build_namespace_client_properties(
+            uri,
+            storage_options,
+            namespace_client_properties,
+        );
+        properties.insert("manifest_enabled".to_string(), "true".to_string());
+        properties.insert(
+            "dir_listing_to_manifest_migration_enabled".to_string(),
+            "true".to_string(),
+        );
+        properties
+    }
+
    async fn connect_namespace_database(
        uri: &str,
        storage_options: HashMap<String, String>,
@@ -323,6 +341,119 @@ impl ListingDatabase {
        ))
    }

+    async fn prepare_namespace_root(
+        uri: &str,
+        storage_options: &HashMap<String, String>,
+        session: Arc<lance::session::Session>,
+    ) -> Result<String> {
+        match url::Url::parse(uri) {
+            Ok(url) if url.scheme().len() == 1 && cfg!(windows) => {
+                let (object_store, _) = ObjectStore::from_uri_and_params(
+                    session.store_registry(),
+                    uri,
+                    &ObjectStoreParams::default(),
+                )
+                .await?;
+                if object_store.is_local() {
+                    Self::try_create_dir(uri).context(CreateDirSnafu { path: uri })?;
+                }
+                Ok(uri.to_string())
+            }
+            Ok(mut url) => {
+                if url.scheme().contains('+') {
+                    return Err(Error::NotSupported {
+                        message: "commit engine URI schemes are not supported for manifest-enabled namespace connections".to_string(),
+                    });
+                }
+
+                for (key, value) in url.query_pairs() {
+                    if key == ENGINE {
+                        return Err(Error::NotSupported {
+                            message: format!(
+                                "commit engine '{}' is not supported for manifest-enabled namespace connections",
+                                value
+                            ),
+                        });
+                    } else if key == MIRRORED_STORE {
+                        return Err(Error::NotSupported {
+                            message: "mirrored store is not supported for manifest-enabled namespace connections"
+                                .to_string(),
+                        });
+                    }
+                }
+
+                url.set_query(None);
+                let plain_uri = url.to_string();
+
+                let os_params = ObjectStoreParams {
+                    storage_options_accessor: if storage_options.is_empty() {
+                        None
+                    } else {
+                        Some(Arc::new(StorageOptionsAccessor::with_static_options(
+                            storage_options.clone(),
+                        )))
+                    },
+                    ..Default::default()
+                };
+                let (object_store, _) = ObjectStore::from_uri_and_params(
+                    session.store_registry(),
+                    &plain_uri,
+                    &os_params,
+                )
+                .await?;
+                if object_store.is_local() {
+                    Self::try_create_dir(&plain_uri).context(CreateDirSnafu {
+                        path: plain_uri.clone(),
+                    })?;
+                }
+
+                Ok(plain_uri)
+            }
+            Err(_) => {
+                let (object_store, _) = ObjectStore::from_uri_and_params(
+                    session.store_registry(),
+                    uri,
+                    &ObjectStoreParams::default(),
+                )
+                .await?;
+                if object_store.is_local() {
+                    Self::try_create_dir(uri).context(CreateDirSnafu { path: uri })?;
+                }
+                Ok(uri.to_string())
+            }
+        }
+    }
+
+    pub(crate) async fn connect_manifest_enabled_namespace_database(
+        request: &ConnectRequest,
+    ) -> Result<LanceNamespaceDatabase> {
+        let options = ListingDatabaseOptions::parse_from_map(&request.options)?;
+        let session = request
+            .session
+            .clone()
+            .unwrap_or_else(|| Arc::new(lance::session::Session::default()));
+        let namespace_root =
+            Self::prepare_namespace_root(&request.uri, &options.storage_options, session.clone())
+                .await?;
+        let ns_properties = Self::build_manifest_enabled_namespace_client_properties(
+            &namespace_root,
+            &options.storage_options,
+            request.namespace_client_properties.clone(),
+        );
+
+        LanceNamespaceDatabase::connect_with_new_table_config(
+            "dir",
+            ns_properties,
+            options.storage_options,
+            request.read_consistency_interval,
+            Some(session),
+            HashSet::new(),
+            options.new_table_config,
+        )
+        .await
+        .map(|db| db.with_uri(request.uri.clone()))
+    }
+
    /// Connect to a listing database
    ///
    /// The URI should be a path to a directory where the tables are stored.
@@ -690,15 +821,12 @@ impl ListingDatabase {
            store_params.storage_options_accessor = Some(Arc::new(accessor));
        }

-        write_params.data_storage_version = self
-            .new_table_config
-            .data_storage_version
-            .or(storage_version_override);
+        write_params.data_storage_version = storage_version_override
+            .or(write_params.data_storage_version)
+            .or(self.new_table_config.data_storage_version);

-        if let Some(enable_v2_manifest_paths) = self
-            .new_table_config
-            .enable_v2_manifest_paths
-            .or(v2_manifest_override)
+        if let Some(enable_v2_manifest_paths) =
+            v2_manifest_override.or(self.new_table_config.enable_v2_manifest_paths)
        {
            write_params.enable_v2_manifest_paths = enable_v2_manifest_paths;
        }
@@ -1158,6 +1286,7 @@ mod tests {
            client_config: Default::default(),
            options: Default::default(),
            namespace_client_properties: Default::default(),
+            manifest_enabled: false,
            read_consistency_interval: None,
            session: None,
        };
@@ -1292,6 +1421,7 @@ mod tests {
            client_config: Default::default(),
            options: options.clone(),
            namespace_client_properties: Default::default(),
+            manifest_enabled: false,
            read_consistency_interval: None,
            session: None,
        };
@@ -1827,6 +1957,7 @@ mod tests {
            client_config: Default::default(),
            options,
            namespace_client_properties: Default::default(),
+            manifest_enabled: false,
            read_consistency_interval: None,
            session: None,
        };
@@ -1933,6 +2064,7 @@ mod tests {
            client_config: Default::default(),
            options,
            namespace_client_properties: Default::default(),
+            manifest_enabled: false,
            read_consistency_interval: None,
            session: None,
        };
@@ -2005,6 +2137,7 @@ mod tests {
            client_config: Default::default(),
            options,
            namespace_client_properties: Default::default(),
+            manifest_enabled: false,
            read_consistency_interval: None,
            session: None,
        };
@@ -2202,6 +2335,7 @@ mod tests {
            client_config: Default::default(),
            options: Default::default(),
            namespace_client_properties,
+            manifest_enabled: false,
            read_consistency_interval: None,
            session: None,
        };
--- a/rust/lancedb/src/database/namespace.rs
+++ b/rust/lancedb/src/database/namespace.rs
@@ -22,10 +22,15 @@ use lance_namespace_impls::ConnectBuilder;
 use lance_table::io::commit::CommitHandler;
 use lance_table::io::commit::external_manifest::ExternalManifestCommitHandler;

-use crate::connection::PushdownOperation;
+use crate::connection::NamespaceClientPushdownOperation;
 use crate::database::ReadConsistency;
+use crate::database::listing::{
+    NewTableConfig, OPT_NEW_TABLE_ENABLE_STABLE_ROW_IDS, OPT_NEW_TABLE_STORAGE_VERSION,
+    OPT_NEW_TABLE_V2_MANIFEST_PATHS,
+};
 use crate::error::{Error, Result};
 use crate::table::NativeTable;
+use lance::dataset::WriteMode;

 use super::{
    BaseTable, CloneTableRequest, CreateTableMode, CreateTableRequest as DbCreateTableRequest,
@@ -44,21 +49,71 @@ pub struct LanceNamespaceDatabase {
    // database URI
    uri: String,
    // Operations to push down to the namespace server
-    pushdown_operations: HashSet<PushdownOperation>,
+    pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
    // Namespace implementation type (e.g., "dir", "rest")
    ns_impl: String,
    // Namespace properties used to construct the namespace client
    ns_properties: HashMap<String, String>,
+    // Options for tables created by this connection
+    new_table_config: NewTableConfig,
 }

 impl LanceNamespaceDatabase {
+    pub fn from_namespace_client(
+        namespace_client: Arc<dyn LanceNamespace>,
+        namespace_client_impl: String,
+        namespace_client_properties: HashMap<String, String>,
+        storage_options: HashMap<String, String>,
+        read_consistency_interval: Option<std::time::Duration>,
+        session: Option<Arc<lance::session::Session>>,
+        namespace_client_pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
+    ) -> Self {
+        Self {
+            namespace: namespace_client,
+            storage_options,
+            read_consistency_interval,
+            session,
+            uri: format!("namespace://{}", namespace_client_impl),
+            pushdown_operations: namespace_client_pushdown_operations,
+            ns_impl: namespace_client_impl,
+            ns_properties: namespace_client_properties,
+            new_table_config: NewTableConfig::default(),
+        }
+    }
+
+    pub(crate) fn with_uri(mut self, uri: impl Into<String>) -> Self {
+        self.uri = uri.into();
+        self
+    }
+
    pub async fn connect(
        ns_impl: &str,
        ns_properties: HashMap<String, String>,
        storage_options: HashMap<String, String>,
        read_consistency_interval: Option<std::time::Duration>,
        session: Option<Arc<lance::session::Session>>,
-        pushdown_operations: HashSet<PushdownOperation>,
+        pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
+    ) -> Result<Self> {
+        Self::connect_with_new_table_config(
+            ns_impl,
+            ns_properties,
+            storage_options,
+            read_consistency_interval,
+            session,
+            pushdown_operations,
+            NewTableConfig::default(),
+        )
+        .await
+    }
+
+    pub(crate) async fn connect_with_new_table_config(
+        ns_impl: &str,
+        ns_properties: HashMap<String, String>,
+        storage_options: HashMap<String, String>,
+        read_consistency_interval: Option<std::time::Duration>,
+        session: Option<Arc<lance::session::Session>>,
+        pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
+        new_table_config: NewTableConfig,
    ) -> Result<Self> {
        let mut builder = ConnectBuilder::new(ns_impl);
        for (key, value) in ns_properties.clone() {
@@ -80,8 +135,79 @@ impl LanceNamespaceDatabase {
            pushdown_operations,
            ns_impl: ns_impl.to_string(),
            ns_properties,
+            new_table_config,
        })
    }
+
+    fn extract_storage_overrides(
+        &self,
+        request: &DbCreateTableRequest,
+    ) -> Result<(
+        Option<lance_encoding::version::LanceFileVersion>,
+        Option<bool>,
+        Option<bool>,
+    )> {
+        let storage_options = request
+            .write_options
+            .lance_write_params
+            .as_ref()
+            .and_then(|p| p.store_params.as_ref())
+            .and_then(|sp| sp.storage_options());
+
+        let storage_version_override = storage_options
+            .and_then(|opts| opts.get(OPT_NEW_TABLE_STORAGE_VERSION))
+            .map(|s| s.parse::<lance_encoding::version::LanceFileVersion>())
+            .transpose()?;
+
+        let v2_manifest_override = storage_options
+            .and_then(|opts| opts.get(OPT_NEW_TABLE_V2_MANIFEST_PATHS))
+            .map(|s| s.parse::<bool>())
+            .transpose()
+            .map_err(|_| Error::InvalidInput {
+                message: "enable_v2_manifest_paths must be a boolean".to_string(),
+            })?;
+
+        let stable_row_ids_override = storage_options
+            .and_then(|opts| opts.get(OPT_NEW_TABLE_ENABLE_STABLE_ROW_IDS))
+            .map(|s| s.parse::<bool>())
+            .transpose()
+            .map_err(|_| Error::InvalidInput {
+                message: "enable_stable_row_ids must be a boolean".to_string(),
+            })?;
+
+        Ok((
+            storage_version_override,
+            v2_manifest_override,
+            stable_row_ids_override,
+        ))
+    }
+
+    fn apply_new_table_config(
+        &self,
+        params: &mut lance::dataset::WriteParams,
+        request: &DbCreateTableRequest,
+    ) -> Result<()> {
+        let (storage_version_override, v2_manifest_override, stable_row_ids_override) =
+            self.extract_storage_overrides(request)?;
+
+        params.data_storage_version = storage_version_override
+            .or(params.data_storage_version)
+            .or(self.new_table_config.data_storage_version);
+
+        if let Some(enable_v2_manifest_paths) =
+            v2_manifest_override.or(self.new_table_config.enable_v2_manifest_paths)
+        {
+            params.enable_v2_manifest_paths = enable_v2_manifest_paths;
+        }
+
+        if let Some(enable_stable_row_ids) =
+            stable_row_ids_override.or(self.new_table_config.enable_stable_row_ids)
+        {
+            params.enable_stable_row_ids = enable_stable_row_ids;
+        }
+
+        Ok(())
+    }
 }

 impl std::fmt::Debug for LanceNamespaceDatabase {
@@ -163,37 +289,23 @@ impl Database for LanceNamespaceDatabase {
    async fn create_table(&self, request: DbCreateTableRequest) -> Result<Arc<dyn BaseTable>> {
        let mut table_id = request.namespace_path.clone();
        table_id.push(request.name.clone());
-        let describe_request = DescribeTableRequest {
-            id: Some(table_id.clone()),
-            ..Default::default()
-        };
-
-        let describe_result = self.namespace.describe_table(describe_request).await;
+        let mut existing_table = None;

        match request.mode {
-            CreateTableMode::Create => {
-                if describe_result.is_ok() {
-                    return Err(Error::TableAlreadyExists {
-                        name: request.name.clone(),
-                    });
-                }
-            }
+            CreateTableMode::Create => {}
            CreateTableMode::Overwrite => {
-                if describe_result.is_ok() {
-                    // Drop the existing table - must succeed
-                    let drop_request = DropTableRequest {
-                        id: Some(table_id.clone()),
-                        ..Default::default()
-                    };
-                    self.namespace
-                        .drop_table(drop_request)
-                        .await
-                        .map_err(|e| Error::Runtime {
-                            message: format!("Failed to drop existing table for overwrite: {}", e),
-                        })?;
-                }
+                let describe_request = DescribeTableRequest {
+                    id: Some(table_id.clone()),
+                    ..Default::default()
+                };
+                existing_table = self.namespace.describe_table(describe_request).await.ok();
            }
            CreateTableMode::ExistOk(_) => {
+                let describe_request = DescribeTableRequest {
+                    id: Some(table_id.clone()),
+                    ..Default::default()
+                };
+                let describe_result = self.namespace.describe_table(describe_request).await;
                if describe_result.is_ok() {
                    let native_table = NativeTable::open_from_namespace(
                        self.namespace.clone(),
@@ -221,20 +333,86 @@ impl Database for LanceNamespaceDatabase {
        };

        let (location, initial_storage_options, managed_versioning) = {
-            let response = self.namespace.declare_table(declare_request).await?;
-            let loc = response.location.ok_or_else(|| Error::Runtime {
-                message: "Table location is missing from declare_table response".to_string(),
-            })?;
-            // Use storage options from response, fall back to self.storage_options
-            let opts = response
-                .storage_options
-                .or_else(|| Some(self.storage_options.clone()))
-                .filter(|o| !o.is_empty());
-            (loc, opts, response.managed_versioning)
+            if let Some(response) = existing_table {
+                let loc = response.location.ok_or_else(|| Error::Runtime {
+                    message: "Table location is missing from describe_table response".to_string(),
+                })?;
+                let opts = response
+                    .storage_options
+                    .or_else(|| Some(self.storage_options.clone()))
+                    .filter(|o| !o.is_empty());
+                (loc, opts, response.managed_versioning)
+            } else {
+                match self.namespace.declare_table(declare_request).await {
+                    Ok(response) => {
+                        let loc = response.location.ok_or_else(|| Error::Runtime {
+                            message: "Table location is missing from declare_table response"
+                                .to_string(),
+                        })?;
+                        let opts = response
+                            .storage_options
+                            .or_else(|| Some(self.storage_options.clone()))
+                            .filter(|o: &HashMap<String, String>| !o.is_empty());
+                        (loc, opts, response.managed_versioning)
+                    }
+                    Err(e)
+                        if matches!(request.mode, CreateTableMode::Create) && {
+                            let err_str = e.to_string();
+                            err_str.contains("already exists")
+                                || err_str.contains("TableAlreadyExists")
+                                || err_str.contains("table already exists")
+                        } =>
+                    {
+                        let response = self
+                            .namespace
+                            .describe_table(DescribeTableRequest {
+                                id: Some(table_id.clone()),
+                                ..Default::default()
+                            })
+                            .await
+                            .map_err(|describe_err| Error::Runtime {
+                                message: format!(
+                                    "Failed to describe existing declared table after declare conflict: {}",
+                                    describe_err
+                                ),
+                            })?;
+
+                        if response.version.is_some() && response.schema.is_some() {
+                            return Err(Error::TableAlreadyExists {
+                                name: request.name.clone(),
+                            });
+                        }
+
+                        let loc = response.location.ok_or_else(|| Error::Runtime {
+                            message: "Table location is missing from describe_table response"
+                                .to_string(),
+                        })?;
+                        let opts = response
+                            .storage_options
+                            .or_else(|| Some(self.storage_options.clone()))
+                            .filter(|o: &HashMap<String, String>| !o.is_empty());
+                        (loc, opts, response.managed_versioning)
+                    }
+                    Err(e) => {
+                        return Err(Error::Runtime {
+                            message: format!("Failed to declare table: {}", e),
+                        });
+                    }
+                }
+            }
        };

        // Build write params with storage options and commit handler
-        let mut params = request.write_options.lance_write_params.unwrap_or_default();
+        let mut params = request
+            .write_options
+            .lance_write_params
+            .clone()
+            .unwrap_or_default();
+        self.apply_new_table_config(&mut params, &request)?;
+
+        if matches!(request.mode, CreateTableMode::Overwrite) {
+            params.mode = WriteMode::Overwrite;
+        }

        // Set up storage options if provided
        if let Some(storage_opts) = initial_storage_options {
@@ -692,6 +870,58 @@ mod tests {
        assert_eq!(id_col.value(2), 30);
    }

+    #[tokio::test]
+    async fn test_namespace_create_table_after_declare_conflict() {
+        let tmp_dir = tempdir().unwrap();
+        let root_path = tmp_dir.path().to_str().unwrap().to_string();
+
+        let mut properties = HashMap::new();
+        properties.insert("root".to_string(), root_path);
+
+        let conn = connect_namespace("dir", properties)
+            .execute()
+            .await
+            .expect("Failed to connect to namespace");
+
+        conn.create_namespace(CreateNamespaceRequest {
+            id: Some(vec!["test_ns".into()]),
+            ..Default::default()
+        })
+        .await
+        .expect("Failed to create namespace");
+
+        let namespace_client = conn.namespace_client().await.unwrap();
+        namespace_client
+            .declare_table(DeclareTableRequest {
+                id: Some(vec!["test_ns".into(), "declared_test".into()]),
+                ..Default::default()
+            })
+            .await
+            .expect("Failed to declare table");
+
+        let test_data = create_test_data();
+        let table = conn
+            .create_table("declared_test", test_data)
+            .namespace(vec!["test_ns".into()])
+            .execute()
+            .await
+            .expect("Failed to create table after declare conflict");
+
+        let results = table
+            .query()
+            .execute()
+            .await
+            .expect("Failed to query table")
+            .try_collect::<Vec<_>>()
+            .await
+            .expect("Failed to collect results");
+
+        assert_eq!(results.len(), 1);
+        assert_eq!(results[0].num_rows(), 5);
+        assert_eq!(table.namespace(), &["test_ns"]);
+        assert_eq!(table.id(), "test_ns$declared_test");
+    }
+
    #[tokio::test]
    async fn test_namespace_create_table_exist_ok_mode() {
        // Setup: Create a temporary directory for the namespace
--- a/rust/lancedb/src/remote/client.rs
+++ b/rust/lancedb/src/remote/client.rs
@@ -16,7 +16,7 @@ use crate::remote::retry::{ResolvedRetryConfig, RetryCounter};
 const REQUEST_ID_HEADER: HeaderName = HeaderName::from_static("x-request-id");

 /// Configuration for TLS/mTLS settings.
-#[derive(Clone, Debug, Default)]
+#[derive(Clone, Debug)]
 pub struct TlsConfig {
    /// Path to the client certificate file (PEM format)
    pub cert_file: Option<String>,
@@ -24,10 +24,22 @@ pub struct TlsConfig {
    pub key_file: Option<String>,
    /// Path to the CA certificate file for server verification (PEM format)
    pub ssl_ca_cert: Option<String>,
-    /// Whether to verify the hostname in the server's certificate
+    /// Whether to verify the hostname in the server's certificate.
+    /// Defaults to `true`.
    pub assert_hostname: bool,
 }

+impl Default for TlsConfig {
+    fn default() -> Self {
+        Self {
+            cert_file: None,
+            key_file: None,
+            ssl_ca_cert: None,
+            assert_hostname: true,
+        }
+    }
+}
+
 /// Trait for providing custom headers for each request
 #[async_trait::async_trait]
 pub trait HeaderProvider: Send + Sync + std::fmt::Debug {
@@ -926,7 +938,7 @@ mod tests {
        assert!(config.cert_file.is_none());
        assert!(config.key_file.is_none());
        assert!(config.ssl_ca_cert.is_none());
-        assert!(!config.assert_hostname);
+        assert!(config.assert_hostname);
    }

    #[test]
--- a/rust/lancedb/src/table.rs
+++ b/rust/lancedb/src/table.rs
@@ -47,7 +47,7 @@ use std::format;
 use std::path::Path;
 use std::sync::Arc;

-use crate::connection::PushdownOperation;
+use crate::connection::NamespaceClientPushdownOperation;

 use crate::data::scannable::{PeekedScannable, Scannable, estimate_write_partitions};
 use crate::database::Database;
@@ -1272,7 +1272,7 @@ pub struct NativeTable {
    pub(crate) namespace_client: Option<Arc<dyn LanceNamespace>>,
    // Operations to push down to the namespace server.
    // pub(crate) so query.rs can access the field for server-side query execution.
-    pub(crate) pushdown_operations: HashSet<PushdownOperation>,
+    pub(crate) pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
 }

 impl std::fmt::Debug for NativeTable {
@@ -1359,7 +1359,7 @@ impl NativeTable {
        params: Option<ReadParams>,
        read_consistency_interval: Option<std::time::Duration>,
        namespace_client: Option<Arc<dyn LanceNamespace>>,
-        pushdown_operations: HashSet<PushdownOperation>,
+        pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
        managed_versioning: Option<bool>,
    ) -> Result<Self> {
        let params = params.unwrap_or_default();
@@ -1470,7 +1470,7 @@ impl NativeTable {
        write_store_wrapper: Option<Arc<dyn WrappingObjectStore>>,
        params: Option<ReadParams>,
        read_consistency_interval: Option<std::time::Duration>,
-        pushdown_operations: HashSet<PushdownOperation>,
+        pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
        session: Option<Arc<lance::session::Session>>,
    ) -> Result<Self> {
        let mut params = params.unwrap_or_default();
@@ -1518,7 +1518,7 @@ impl NativeTable {
        let id = Self::build_id(&namespace, name);

        let stored_namespace_client =
-            if pushdown_operations.contains(&PushdownOperation::QueryTable) {
+            if pushdown_operations.contains(&NamespaceClientPushdownOperation::QueryTable) {
                Some(namespace_client)
            } else {
                None
@@ -1588,7 +1588,7 @@ impl NativeTable {
        params: Option<WriteParams>,
        read_consistency_interval: Option<std::time::Duration>,
        namespace_client: Option<Arc<dyn LanceNamespace>>,
-        pushdown_operations: HashSet<PushdownOperation>,
+        pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
    ) -> Result<Self> {
        // Default params uses format v1.
        let params = params.unwrap_or(WriteParams {
@@ -1635,7 +1635,7 @@ impl NativeTable {
        params: Option<WriteParams>,
        read_consistency_interval: Option<std::time::Duration>,
        namespace_client: Option<Arc<dyn LanceNamespace>>,
-        pushdown_operations: HashSet<PushdownOperation>,
+        pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
    ) -> Result<Self> {
        let data: Box<dyn Scannable> = Box::new(RecordBatch::new_empty(schema));
        Self::create(
@@ -1685,7 +1685,7 @@ impl NativeTable {
        write_store_wrapper: Option<Arc<dyn WrappingObjectStore>>,
        params: Option<WriteParams>,
        read_consistency_interval: Option<std::time::Duration>,
-        pushdown_operations: HashSet<PushdownOperation>,
+        pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
        session: Option<Arc<lance::session::Session>>,
    ) -> Result<Self> {
        // Build table_id from namespace + name for the storage options provider
@@ -1738,7 +1738,7 @@ impl NativeTable {
        let id = Self::build_id(&namespace, name);

        let stored_namespace_client =
-            if pushdown_operations.contains(&PushdownOperation::QueryTable) {
+            if pushdown_operations.contains(&NamespaceClientPushdownOperation::QueryTable) {
                Some(namespace_client)
            } else {
                None
--- a/rust/lancedb/src/table/query.rs
+++ b/rust/lancedb/src/table/query.rs
@@ -4,7 +4,7 @@
 use std::sync::Arc;

 use super::NativeTable;
-use crate::connection::PushdownOperation;
+use crate::connection::NamespaceClientPushdownOperation;
 use crate::error::{Error, Result};
 use crate::expr::expr_to_sql_string;
 use crate::query::{
@@ -44,7 +44,7 @@ pub async fn execute_query(
    // If QueryTable pushdown is enabled and namespace client is configured, use server-side query execution
    if table
        .pushdown_operations
-        .contains(&PushdownOperation::QueryTable)
+        .contains(&NamespaceClientPushdownOperation::QueryTable)
        && let Some(ref namespace_client) = table.namespace_client
    {
        return execute_namespace_query(table, namespace_client.clone(), query, options).await;
Author	SHA1	Message	Date
Ayush Chaurasia	2fa4d04de7	feat(python): align with_format("torch") with HuggingFace semantics - New "torch" format yields list-of-per-row-dicts whose values are torch tensors. PyTorch DataLoader's default collate stacks these into a dict[str, Tensor] per batch — matching HuggingFace dataset.set_format("torch") and removing the need for a custom collate_fn. - The previous "torch" behavior (list of 1-D row tensors without column names) moves to a new "torch_row" format. - "torch_col" is unchanged (2-D tensor, requires custom collate). Closes lancedb/lancedb#3245	2026-04-29 22:15:40 +05:30
Jack Ye	25dfe2cfd4	feat: add manifest-enabled directory namespace mode (#3332 ) Adds manifest_enabled for local/native connections so directory namespace manifests can be the source of truth, including migration from directory listing and Azure credential vending feature wiring. Also exposes the option through Rust, Python, and Node bindings with focused validation.	2026-04-29 09:22:06 -07:00
Lance Release	4dcd7f4314	Bump version: 0.28.0-beta.9 → 0.28.0-beta.10	2026-04-28 13:29:26 +00:00
Lance Release	2e36cd9dad	Bump version: 0.31.0-beta.9 → 0.31.0-beta.10	2026-04-28 13:29:00 +00:00
Weston Pace	f31e27768a	fix: address RUSTSEC-2026-0104 cargo-deny advisory (#3326 ) ## Summary - Update `rustls-webpki` 0.103.10 → 0.103.13 to fix RUSTSEC-2026-0104 (reachable panic in CRL parsing) - Add advisory ignore for the legacy `rustls-webpki` 0.101.7 copy pinned to the aws-smithy/rustls 0.21 chain (same chain already exempted for RUSTSEC-2026-0098/0099) Fixes the `deny` CI job failure seen in #3325. ## Test plan - [x] `cargo deny check advisories` passes locally 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-27 17:56:10 -07:00
LanceDB Robot	b84150a53e	chore: update lance dependency to v6.0.0-beta.4 (#3325 ) ## Summary - Updates Lance Rust dependencies to `6.0.0-beta.4` using `ci/set_lance_version.py`. - Updates the Java `lance-core.version` property to `6.0.0-beta.4`. - Triggering Lance tag: https://github.com/lance-format/lance/releases/tag/v6.0.0-beta.4 ## Verification - `cargo clippy --workspace --tests --all-features -- -D warnings` - `cargo fmt --all`	2026-04-27 15:13:07 -07:00
Will Jones	d135c18db6	ci: add cargo-deny configuration and CI check (#3307 ) Adds a `deny.toml` at the workspace root and a `deny` CI job that runs `cargo deny check` on every PR. Catches yanked crates, license drift, banned or wildcard dependencies, unapproved sources, and new RUSTSEC advisories. As part of wiring this up: - Updated `aws-lc-rs` 1.13.0 → 1.16.3 / `aws-lc-sys` 0.28.0 → 0.40.0 to clear four 2026 AWS-LC advisories (timing side-channel, PKCS7 bypass, CRL scope). Removed the `=0.28.0` workaround pin; the original build failure no longer reproduces. - Updated `bytes`, `zlib-rs`, `rand`, `rustls-webpki`, `lz4_flex` to clear their current advisories. - Marked `lancedb-nodejs` and `lancedb-python` as `publish = false` and pinned `lzma-sys` from `*` to `0.1` so `bans.wildcards = "deny"` can be enforced. 10 remaining advisories have no safe upgrade available (transitive via opendal, lance, datafusion, async-openai, aws-sdk on the legacy rustls 0.21 chain). Each is ignored in `deny.toml` with a per-entry rationale and a link to the RUSTSEC advisory. New advisories still fail CI. Fixes #3297 --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-04-24 20:53:15 -07:00
Will Jones	ef399de092	ci: switch PyPI publish to OIDC trusted publishing (#3302 ) ## Summary - Replaces `LANCEDB_PYPI_API_TOKEN` (long-lived token) with OIDC trusted publishing via `pypa/gh-action-pypi-publish` - Adds `id-token: write` permission to linux/mac/windows jobs - Removes `twine`-based upload and the `pypi_token` input from `upload_wheel` composite action - Enables PEP 740 Sigstore attestations on published wheels as a bonus After merging, rotate/revoke the `LANCEDB_PYPI_API_TOKEN` secret. Closes #3294 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-04-24 20:53:06 -07:00
Will Jones	0d767abd0e	ci: add Dependabot config for shipped Rust binaries (#3300 ) Adds `.github/dependabot.yml` enabling weekly cargo update PRs for the root workspace, which produces the Rust binaries we ship: the Node.js and Python native extensions. The `rust/lancedb` library crate shares the same lockfile — its consumers pick versions themselves, but bumping transitive deps here keeps the shipped binaries current. Also removes the misleading `exclude = ["python"]` line from the root `Cargo.toml`: `python` is listed in `members`, and `cargo metadata` confirms it's a workspace member, so the exclude was dead code that implied the opposite. Minor/patch updates are grouped to reduce PR noise. Part of #3292. Only covers the cargo ecosystem; pip, npm, and github-actions can follow.	2026-04-24 20:52:54 -07:00
Jack Ye	a92ae0ded5	fix: enable hostname verification by default (#3304 ) ## Summary - make `TlsConfig::default()` enable hostname verification by default - align the Rust default with the documented Python and Node behavior - update the Rust unit test to lock in the safe default	2026-04-21 08:39:03 -07:00
Xuanwo	c54888a83a	refactor(python): remove legacy tantivy FTS support (#3282 ) This follows the Rust-side Tantivy removal by deleting the remaining Python Tantivy runtime, tests, and packaging references. It also turns the legacy Python-only Tantivy parameters into explicit errors and stops reading legacy `_indices/fts` directories so Python FTS is fully native-only.	2026-04-20 09:28:45 +08:00
Will Jones	ba6c44abc9	ci: add top-level permissions to GHA workflows (#3255 ) Adds `permissions: contents: read` to the 10 workflows that had no top-level permissions block. Workflows that already declared permissions, or individual jobs that need elevated permissions (`issues: write`, `pull-requests: write`, `contents: write`), are left unchanged. Affected workflows: `dev.yml`, `java-publish.yml`, `java.yml`, `license-header-check.yml`, `nodejs.yml`, `pypi-publish.yml`, `python.yml`, `rust.yml`, `update_package_lock_run.yml`, `update_package_lock_run_nodejs.yml`	2026-04-20 09:22:27 +08:00
Lance Release	75b0a8e0a3	Bump version: 0.28.0-beta.8 → 0.28.0-beta.9	2026-04-19 20:39:29 +00:00
Lance Release	2a886141f7	Bump version: 0.31.0-beta.8 → 0.31.0-beta.9	2026-04-19 20:39:04 +00:00
Jack Ye	2a1df8edcf	fix(rust): materialize declared namespace tables on create (#3288 ) ## Summary - handle `declare_table` already-exists conflicts in the Rust namespace database create path - reuse declared-but-not-materialized table metadata instead of failing create mode - preserve overwrite behavior while allowing declared Geneva system tables to be materialized	2026-04-19 13:25:53 -07:00
C Kaustubh	fd98b845ea	fix(node): prevent reranker from keeping process alive (#3270 ) Fixes #3269. ## What I observed Using a reranker in a hybrid query could keep the Node.js process alive even after `table.close()` and `db.close()`. ## Root cause The reranker callback bridge used a `ThreadsafeFunction` in referenced mode, which can keep the event loop alive longer than intended. ## Minimal fix - In `nodejs/src/rerankers.rs`, create the reranker callback TSFN in weak mode (`.weak::<true>()`). - Add a regression test in `nodejs/__test__/rerankers.test.ts` that spawns a child process, runs a rerank query, and asserts the process exits naturally. ## Validation - Built Node bindings successfully. - Ran targeted tests: `rerankers.test.ts` passes (including new regression test). - Pre-commit checks for changed files were run and clean.	2026-04-19 14:02:23 +08:00
Lance Release	be48ada352	Bump version: 0.28.0-beta.7 → 0.28.0-beta.8	2026-04-19 04:19:10 +00:00
Lance Release	9ad2dfe601	Bump version: 0.31.0-beta.7 → 0.31.0-beta.8	2026-04-19 04:18:45 +00:00
Jack Ye	f909df3e87	fix(python): use namespace-backed rust connection for namespace tables (#3286 ) So far, I have been using a hacky approach that creates and opens namespace-backed table, by getting its location and use a temporary lancedb connection to create or open it. This was working for features like credentials vending but is no longer fully working for the managed versioning feature, recently geneva tests have been failing here and there and various patches are not addressing the root cause. This PR fully fixes this and implements proper rust binding for it. Specifically: - build a real Rust namespace-backed connection from the Python namespace client - route namespace table create/open through that connection instead of resolved-location temp connections - keep namespace client naming consistent in the Rust bridge and preserve federated namespace + DuckDB behavior	2026-04-18 21:17:52 -07:00
Lance Release	d715bbb588	Bump version: 0.28.0-beta.6 → 0.28.0-beta.7	2026-04-17 08:12:27 +00:00
Lance Release	5ce3d8d141	Bump version: 0.31.0-beta.6 → 0.31.0-beta.7	2026-04-17 08:12:03 +00:00
Jack Ye	5eaac178b1	fix(python): pass namespace client on schema-only table create (#3283 ) ## Summary - pass `namespace_client` through the Python create-table path - ensure schema-only namespace table creation uses the namespace-aware empty-table flow - fix reopening namespace tables created without initial data	2026-04-17 01:11:18 -07:00
Lance Release	11af763fcd	Bump version: 0.28.0-beta.5 → 0.28.0-beta.6	2026-04-16 18:57:28 +00:00