chore: update lance dependency to v9.0.0-beta.16

### **Summary**
Closes #3212

Extends the Python `lit()` helper to natively support three additional
types (`date`, `datetime`, and `Decimal`) and implements reflexive
operators for the `Expr` class.

This implementation specifically addresses the blocking feedback
regarding precision loss, CI discovery, and query engine limitations:

* **Logic Refactoring**: Simplified `lit()` by combining `date` and
`datetime` normalization into ISO-8601 strings, ensuring stable SQL
parsing across different engine locales.
* **Precision Preservation**: `decimal.Decimal` objects are now passed
as high-precision strings to the Rust bridge, bypassing intermediate
float conversions and preserving full 128-bit decimal precision for
DataFusion.
* **Averted CI Failures**: Temporarily deferred `bytes` literal support
to a future PR to resolve a known DataFusion `expr_to_sql` limitation
that was crashing the `Doctest` runner.
* **Reflexive Operators**: Added support for "literal-first" arithmetic
and logical operations (e.g., `10 + col('a')` or `True &
col('active')`). Redundant reflexive comparisons (e.g., `__rlt__`) were
pruned as Python's data model handles them automatically.
* **Integration Verification**: Added dedicated integration tests in the
official test directory to ensure the query engine correctly handles the
new types and preserves bit-perfect fidelity.

### **Changes**  
####
[python/python/lancedb/expr.py](file:///c:/Users/Laksh/Documents/lancedb/python/python/lancedb/expr.py)
* Updated `lit()` to handle `date`, `datetime`, and `Decimal` natively.
* Implemented reflexive operators (`__radd__`, `__rand__`, `__rmul__`,
etc.) to support literals on the left-hand side.
* Removed the problematic `bytes` doctest example and `lit()` type
support to unblock CI.

####
[python/src/expr.rs](file:///c:/Users/Laksh/Documents/lancedb/python/src/expr.rs)
* Modified the Rust FFI bridge to extract `Decimal` objects as strings.
* Ensured the `expr_lit` handler is ready to receive normalized temporal
strings.
*   Consolidated imports and added missing operator documentation.

####
[python/python/lancedb/_lancedb.pyi](file:///c:/Users/Laksh/Documents/lancedb/python/python/lancedb/_lancedb.pyi)
* Updated type stubs for `expr_lit` to include `Any` (allowing for
`Decimal`).

### **Testing**  
Added several new advanced test cases in
[python/python/tests/test_expr.py](file:///c:/Users/Laksh/Documents/lancedb/python/python/tests/test_expr.py)
covering:
* **High-precision Decimal preservation**: Verified against 128-bit
boundaries with a "one point off" test case (`1.234567890123456789 <
1.234567890123456790`).
* **Reflexive operator positioning**: Verified successful query
construction with literals on the left.
* **Timezone-aware normalization**: Confirmed stable behavior for
`datetime` objects.
* **Integration Testing**: Confirmed Date32 and Decimal columns return
the correct Python types and values from the engine during `.to_arrow()`
calls.

---------

Co-authored-by: Will Jones <willjones127@gmail.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

.agents/skills/lancedb-branch-ops/SKILL.md

@@ -0,0 +1,137 @@
+---
+name: lancedb-branch-ops
+description: Branch management for LanceDB tables via the REST API. Use this skill whenever someone wants to create, delete, list, or switch branches on a LanceDB table — or needs to make sure a write (metadata update, index build, etc.) lands on a specific branch instead of main. Invoke it even without the word "branch" if context makes clear they want an experimental copy of a table, want to isolate changes, or want to confirm a mutation didn't touch main. Covers: branches/list, branches/create, branches/delete, and passing "branch" in describe/update_field_metadata/create_index to target a non-main version.
+---
+
+## Goal
+
+Manage branches on a LanceDB table: list what exists, create new ones, delete stale ones, and direct read/write operations at a specific branch without touching main.
+
+## Step 0: Establish the connection
+
+Use the `lancedb-connect` skill to resolve the base URL and auth headers (`x-api-key`, `x-lancedb-database`). Skip this only if the connection is already known from the current conversation.
+
+All examples below use `{base_url}` — substitute the resolved endpoint and include the auth headers on every request.
+
+## The branch model (important)
+
+LanceDB branches are named snapshots that diverge from the table's current state at creation time. There is **no checkout command** — you never switch the whole table to a branch. Instead, you **pass `"branch": "<name>"` in the request body** of any operation to target that branch. Omitting the key (or sending an empty body) always targets main.
+
+`branches/list` returns only non-main branches. Main always exists and is not listed.
+
+## List branches
+
+```http
+POST {base_url}/v1/table/{table_id}/branches/list
+Content-Type: application/json
+
+{}
+```
+
+Response:
+```json
+{
+  "branches": {
+    "experiment-reindex": {"parentVersion": 1, "createAt": 1782506085, "manifestSize": 1029}
+  }
+}
+```
+
+If `branches` is `{}`, the table has no branches besides main.
+
+## Create a branch
+
+```http
+POST {base_url}/v1/table/{table_id}/branches/create
+Content-Type: application/json
+
+{"name": "experiment-reindex"}
+```
+
+HTTP 200 with `{}` body = success. The branch is created off the table's current state on main.
+
+Verify by calling `branches/list` and confirming the new name appears.
+
+## Delete a branch
+
+```http
+POST {base_url}/v1/table/{table_id}/branches/delete
+Content-Type: application/json
+
+{"name": "stale-2024"}
+```
+
+HTTP 200 with `{}` body = success. Only the branch pointer is removed — main and all row data remain intact.
+
+Verify by calling `branches/list` (name gone) and `describe` with no branch param (main still responds).
+
+## Operate on a specific branch
+
+Pass `"branch": "<name>"` in the body of any operation to scope it to that branch:
+
+**Read schema on a branch:**
+```http
+POST {base_url}/v1/table/{table_id}/describe
+Content-Type: application/json
+
+{"branch": "wip-branch"}
+```
+
+**Write metadata to a branch (not main):**
+```http
+POST {base_url}/v1/table/{table_id}/update_field_metadata
+Content-Type: application/json
+
+{
+  "branch": "wip-branch",
+  "updates": [
+    {
+      "path": "category",
+      "metadata": {"lancedb:description": "Product category label."},
+      "replace": false
+    }
+  ]
+}
+```
+
+**Build an index on a branch:**
+```http
+POST {base_url}/v1/table/{table_id}/create_index
+Content-Type: application/json
+
+{
+  "branch": "wip-branch",
+  "column": "category",
+  "index_type": "BTREE"
+}
+```
+
+## Verifying isolation
+
+After writing to a branch, always confirm the change did NOT land on main:
+
+```bash
+# Should show the new metadata
+curl -s -X POST {base_url}/v1/table/{table_id}/describe \
+  -H "x-api-key: <key>" -H "x-lancedb-database: <db>" \
+  -H "content-type: application/json" \
+  -d '{"branch": "wip-branch"}'
+
+# Should NOT show the new metadata
+curl -s -X POST {base_url}/v1/table/{table_id}/describe \
+  -H "x-api-key: <key>" -H "x-lancedb-database: <db>" \
+  -H "content-type: application/json" \
+  -d '{}'
+```
+
+## Quick reference
+
+| Goal | Endpoint | Body |
+|------|----------|------|
+| List all branches | `branches/list` | `{}` |
+| Create a branch | `branches/create` | `{"name": "..."}` |
+| Delete a branch | `branches/delete` | `{"name": "..."}` |
+| Read schema on branch | `describe` | `{"branch": "..."}` |
+| Write metadata on branch | `update_field_metadata` | `{"branch": "...", "updates": [...]}` |
+| Build index on branch | `create_index` | `{"branch": "...", "column": ..., "index_type": ...}` |
+| Target main (default) | any endpoint | omit `"branch"` key |

.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.31.0-beta.2"
+current_version = "0.31.0-beta.6"
 parse = """(?x)
     (?P<major>0|[1-9]\\d*)\\.
     (?P<minor>0|[1-9]\\d*)\\.

.github/workflows/cargo-publish.yml

@@ -25,7 +25,7 @@ jobs:
     # Only runs on tags that matches the make-release action
     if: startsWith(github.ref, 'refs/tags/v')
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
       - uses: Swatinem/rust-cache@v2
         with:
           workspaces: rust
@@ -47,7 +47,7 @@ jobs:
       contents: read
       issues: write
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
       - uses: ./.github/actions/create-failure-issue
         with:
           job-results: ${{ toJSON(needs) }}

.github/workflows/codex-fix-ci.yml

@@ -36,14 +36,14 @@ jobs:
           echo "guidelines = ${{ inputs.guidelines }}"
 
       - name: Checkout Repo
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           ref: ${{ inputs.branch }}
           fetch-depth: 0
           persist-credentials: true
 
       - name: Set up Node.js
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@v6
         with:
           # pnpm 11 (used by the nodejs install step below) requires
           # Node >= 22.13; use 24 since 22 hits EOL in October.
@@ -82,7 +82,7 @@ jobs:
           cache: maven
 
       - name: Setup pnpm
-        uses: pnpm/action-setup@v4
+        uses: pnpm/action-setup@v6
         with:
           version: 11.1.1
       - name: Install Node.js dependencies for TypeScript bindings

.github/workflows/codex-update-lance-dependency.yml

@@ -30,13 +30,13 @@ jobs:
           echo "tag = ${{ inputs.tag || 'latest' }}"
 
       - name: Checkout Repo LanceDB
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           fetch-depth: 0
           persist-credentials: true
 
       - name: Set up Node.js
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@v6
         with:
           node-version: 20
 

.github/workflows/dev.yml

@@ -27,7 +27,7 @@ jobs:
     name: Verify PR title / description conforms to semantic-release
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/setup-node@v4
+      - uses: actions/setup-node@v6
         with:
           node-version: "18"
       # These rules are disabled because Github will always ensure there

.github/workflows/docs.yml

@@ -35,7 +35,7 @@ jobs:
     runs-on: ubuntu-24.04
     steps:
       - name: Checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
       - name: Install dependencies needed for ubuntu
         run: |
           sudo apt install -y protobuf-compiler libssl-dev
@@ -53,7 +53,7 @@ jobs:
           python -m pip install --extra-index-url https://pypi.fury.io/lance-format/ --extra-index-url https://pypi.fury.io/lancedb/ -e .
           python -m pip install --extra-index-url https://pypi.fury.io/lance-format/ --extra-index-url https://pypi.fury.io/lancedb/ -r ../docs/requirements.txt
       - name: Set up node
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@v6
         with:
           node-version: 20
           cache: 'npm'

.github/workflows/java-publish.yml

@@ -32,7 +32,7 @@ jobs:
         working-directory: ./java
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
       - name: Set up Java 8
         uses: actions/setup-java@v4
         with:
@@ -73,7 +73,7 @@ jobs:
       contents: read
       issues: write
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
       - uses: ./.github/actions/create-failure-issue
         with:
           job-results: ${{ toJSON(needs) }}

.github/workflows/java.yml

@@ -36,7 +36,7 @@ jobs:
         working-directory: ./java
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
       - name: Set up Java 17
         uses: actions/setup-java@v4
         with:

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

@@ -19,7 +19,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Check out code
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
       - name: Install license-header-checker
         working-directory: /tmp
         run: |

.github/workflows/make-release-commit.yml

@@ -49,7 +49,7 @@ jobs:
     steps:
       - name: Output Inputs
         run: echo "${{ toJSON(github.event.inputs) }}"
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true

.github/workflows/nodejs.yml

@@ -38,14 +38,14 @@ jobs:
       CC: gcc-12
       CXX: g++-12
     steps:
-    - uses: actions/checkout@v4
+    - uses: actions/checkout@v6
       with:
         fetch-depth: 0
         lfs: true
-    - uses: pnpm/action-setup@v4
+    - uses: pnpm/action-setup@v6
       with:
         version: 11.1.1
-    - uses: actions/setup-node@v4
+    - uses: actions/setup-node@v6
       with:
         # pnpm 11 requires Node >= 22.13; use 24 since 22 hits EOL
         # in October. The library itself still supports Node >= 18
@@ -86,14 +86,14 @@ jobs:
         shell: bash
         working-directory: nodejs
     steps:
-    - uses: actions/checkout@v4
+    - uses: actions/checkout@v6
       with:
         fetch-depth: 0
         lfs: true
-    - uses: pnpm/action-setup@v4
+    - uses: pnpm/action-setup@v6
       with:
         version: 11.1.1
-    - uses: actions/setup-node@v4
+    - uses: actions/setup-node@v6
       name: Setup Node.js 24 for build
       with:
         # pnpm 11 requires Node >= 22.13; use 24 since 22 hits EOL
@@ -130,7 +130,7 @@ jobs:
           echo "Run 'pnpm run docs', fix any warnings, and commit the changes."
           exit 1
         fi
-    - uses: actions/setup-node@v4
+    - uses: actions/setup-node@v6
       name: Setup Node.js ${{ matrix.node-version }} for test
       with:
         node-version: ${{ matrix.node-version }}
@@ -166,14 +166,14 @@ jobs:
         shell: bash
         working-directory: nodejs
     steps:
-    - uses: actions/checkout@v4
+    - uses: actions/checkout@v6
       with:
         fetch-depth: 0
         lfs: true
-    - uses: pnpm/action-setup@v4
+    - uses: pnpm/action-setup@v6
       with:
         version: 11.1.1
-    - uses: actions/setup-node@v4
+    - uses: actions/setup-node@v6
       with:
         # pnpm 11 requires Node >= 22.13; use 24 since 22 hits EOL
         # in October.

.github/workflows/npm-publish.yml

@@ -32,7 +32,7 @@ jobs:
     permissions:
       contents: write
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true
@@ -170,13 +170,13 @@ jobs:
       run:
         working-directory: nodejs
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
       - name: Setup pnpm
-        uses: pnpm/action-setup@v4
+        uses: pnpm/action-setup@v6
         with:
           version: 11.1.1
       - name: Setup node
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@v6
         with:
           # pnpm 11 requires Node >= 22.13; use 24 since 22 hits EOL
           # in October.
@@ -190,7 +190,7 @@ jobs:
           toolchain: stable
           targets: ${{ matrix.settings.target }}
       - name: Cache cargo
-        uses: actions/cache@v4
+        uses: actions/cache@v5
         with:
           path: |
             ~/.cargo/registry/index/
@@ -244,7 +244,7 @@ jobs:
         if: ${{ !matrix.settings.docker }}
         shell: bash
       - name: Upload artifact
-        uses: actions/upload-artifact@v4
+        uses: actions/upload-artifact@v7
         with:
           name: lancedb-${{ matrix.settings.target }}
           path: nodejs/dist/*.node
@@ -256,7 +256,7 @@ jobs:
         run: pnpm tsc
       - name: Upload Generic Artifacts
         if: ${{ matrix.settings.target == 'aarch64-apple-darwin' }}
-        uses: actions/upload-artifact@v4
+        uses: actions/upload-artifact@v7
         with:
           name: nodejs-dist
           path: |
@@ -287,13 +287,13 @@ jobs:
         shell: bash
         working-directory: nodejs
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
       - name: Setup pnpm
-        uses: pnpm/action-setup@v4
+        uses: pnpm/action-setup@v6
         with:
           version: 11.1.1
       - name: Setup Node.js 24 for install
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@v6
         with:
           # pnpm 11 requires Node >= 22.13; use 24 since 22 hits EOL
           # in October.
@@ -303,18 +303,18 @@ jobs:
       - name: Install dependencies
         run: pnpm install --frozen-lockfile
       - name: Setup Node.js ${{ matrix.node }} for test
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@v6
         with:
           node-version: ${{ matrix.node }}
       - name: Download artifacts
-        uses: actions/download-artifact@v4
+        uses: actions/download-artifact@v8
         with:
           name: lancedb-${{ matrix.settings.target }}
           path: nodejs/dist/
           # For testing purposes:
           # run-id: 13982782871
           # github-token: ${{ secrets.GITHUB_TOKEN }} # token with actions:read permissions on target repo
-      - uses: actions/download-artifact@v4
+      - uses: actions/download-artifact@v8
         with:
           name: nodejs-dist
           path: nodejs/dist
@@ -339,13 +339,13 @@ jobs:
     needs:
       - test-lancedb
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
       - name: Setup pnpm
-        uses: pnpm/action-setup@v4
+        uses: pnpm/action-setup@v6
         with:
           version: 11.1.1
       - name: Setup node
-        uses: actions/setup-node@v4
+        uses: actions/setup-node@v6
         with:
           node-version: 24
           cache: pnpm
@@ -353,14 +353,14 @@ jobs:
           registry-url: "https://registry.npmjs.org"
       - name: Install dependencies
         run: pnpm install --frozen-lockfile
-      - uses: actions/download-artifact@v4
+      - uses: actions/download-artifact@v8
         with:
           name: nodejs-dist
           path: nodejs/dist
           # For testing purposes:
           # run-id: 13982782871
           # github-token: ${{ secrets.GITHUB_TOKEN }} # token with actions:read permissions on target repo
-      - uses: actions/download-artifact@v4
+      - uses: actions/download-artifact@v8
         name: Download arch-specific binaries
         with:
           pattern: lancedb-*
@@ -398,7 +398,7 @@ jobs:
       contents: read
       issues: write
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
       - uses: ./.github/actions/create-failure-issue
         with:
           job-results: ${{ toJSON(needs) }}

.github/workflows/python.yml

@@ -41,7 +41,7 @@ jobs:
         shell: bash
         working-directory: python
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true
@@ -66,7 +66,7 @@ jobs:
         shell: bash
         working-directory: python
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true
@@ -95,7 +95,7 @@ jobs:
         shell: bash
         working-directory: python
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true
@@ -126,7 +126,7 @@ jobs:
         shell: bash
         working-directory: python
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true
@@ -160,7 +160,7 @@ jobs:
         shell: bash
         working-directory: python
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true
@@ -189,7 +189,7 @@ jobs:
         shell: bash
         working-directory: python
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true
@@ -212,7 +212,7 @@ jobs:
         shell: bash
         working-directory: python
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true

.github/workflows/rust.yml

@@ -40,7 +40,7 @@ jobs:
       CC: clang-18
       CXX: clang++-18
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true
@@ -65,7 +65,7 @@ jobs:
     timeout-minutes: 10
     runs-on: ubuntu-24.04
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
       - uses: EmbarkStudios/cargo-deny-action@v2
         with:
           command: check advisories bans licenses sources
@@ -78,7 +78,7 @@ jobs:
       CC: clang
       CXX: clang++
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
       # Building without a lock file often requires the latest Rust version since downstream
       # dependencies may have updated their minimum Rust version.
       - uses: actions-rust-lang/setup-rust-toolchain@v1
@@ -113,7 +113,7 @@ jobs:
       CXX: clang++-18
       GH_TOKEN: ${{ secrets.SOPHON_READ_TOKEN }}
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true
@@ -152,7 +152,7 @@ jobs:
         shell: bash
         working-directory: rust
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           fetch-depth: 0
           lfs: true
@@ -181,7 +181,7 @@ jobs:
       run:
         working-directory: rust/lancedb
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
       - name: Set target
         run: rustup target add ${{ matrix.target }}
       - uses: Swatinem/rust-cache@v2
@@ -210,7 +210,7 @@ jobs:
       CC: clang-18
       CXX: clang++-18
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
         with:
           submodules: true
       - name: Install dependencies

.github/workflows/update_package_lock_run.yml

@@ -11,7 +11,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           ref: main
           persist-credentials: false

.github/workflows/update_package_lock_run_nodejs.yml

@@ -11,7 +11,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           ref: main
           persist-credentials: false

Cargo.lock

@@ -157,9 +157,9 @@ dependencies = [
 
 [[package]]
 name = "anyhow"
-version = "1.0.102"
+version = "1.0.103"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7f202df86484c868dbad7eaa557ef785d5c66295e41b460ef922eca0723b842c"
+checksum = "2a4385e2e34eb35d6b3efe798b9eb88096925d87726c0798709bf56d9ed84af3"
 
 [[package]]
 name = "approx"
@@ -1297,15 +1297,6 @@ version = "2.11.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "c4512299f36f043ab09a583e57bceb5a5aab7a73db1805848e8fef3c9e8c78b3"
 
-[[package]]
-name = "bitpacking"
-version = "0.9.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "96a7139abd3d9cebf8cd6f920a389cf3dc9576172e32f4563f188cae3c3eb019"
-dependencies = [
- "crunchy",
-]
-
 [[package]]
 name = "bitvec"
 version = "1.0.1"
@@ -3186,9 +3177,9 @@ dependencies = [
 
 [[package]]
 name = "env_filter"
-version = "1.0.1"
+version = "2.0.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "32e90c2accc4b07a8456ea0debdc2e7587bdd890680d71173a15d4ae604f6eef"
+checksum = "900d271a03799a1ee8d1ca9b19893b48ca674a9284fefcfb85f05e74ed314217"
 dependencies = [
  "log",
  "regex",
@@ -3196,9 +3187,9 @@ dependencies = [
 
 [[package]]
 name = "env_logger"
-version = "0.11.10"
+version = "0.11.11"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0621c04f2196ac3f488dd583365b9c09be011a4ab8b9f37248ffcc8f6198b56a"
+checksum = "de671bd27a75a797dc9ae289ba1e77276e75e2026408aab65185384e2d5cd3f6"
 dependencies = [
  "anstream",
  "anstyle",
@@ -3432,8 +3423,8 @@ checksum = "42703706b716c37f96a77aea830392ad231f44c9e9a67872fa5548707e11b11c"
 
 [[package]]
 name = "fsst"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow-array",
  "rand 0.9.4",
@@ -4735,8 +4726,8 @@ checksum = "e037a2e1d8d5fdbd49b16a4ea09d5d6401c1f29eca5ff29d03d3824dba16256a"
 
 [[package]]
 name = "lance"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arc-swap",
  "arrow",
@@ -4754,7 +4745,6 @@ dependencies = [
  "async_cell",
  "aws-credential-types",
  "aws-sdk-dynamodb",
- "bitpacking",
  "byteorder",
  "bytes",
  "chrono",
@@ -4773,6 +4763,7 @@ dependencies = [
  "humantime",
  "itertools 0.14.0",
  "lance-arrow",
+ "lance-bitpacking",
  "lance-core",
  "lance-datafusion",
  "lance-encoding",
@@ -4810,8 +4801,8 @@ dependencies = [
 
 [[package]]
 name = "lance-arrow"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -4832,7 +4823,7 @@ dependencies = [
 [[package]]
 name = "lance-arrow-scalar"
 version = "58.0.0"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -4846,7 +4837,7 @@ dependencies = [
 [[package]]
 name = "lance-arrow-stats"
 version = "58.0.0"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow-array",
  "arrow-schema",
@@ -4855,18 +4846,19 @@ dependencies = [
 
 [[package]]
 name = "lance-bitpacking"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrayref",
+ "crunchy",
  "paste",
  "seq-macro",
 ]
 
 [[package]]
 name = "lance-core"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -4904,8 +4896,8 @@ dependencies = [
 
 [[package]]
 name = "lance-datafusion"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow",
  "arrow-array",
@@ -4935,8 +4927,8 @@ dependencies = [
 
 [[package]]
 name = "lance-datagen"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow",
  "arrow-array",
@@ -4953,8 +4945,8 @@ dependencies = [
 
 [[package]]
 name = "lance-derive"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -4963,8 +4955,8 @@ dependencies = [
 
 [[package]]
 name = "lance-encoding"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow-arith",
  "arrow-array",
@@ -4999,8 +4991,8 @@ dependencies = [
 
 [[package]]
 name = "lance-file"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow-arith",
  "arrow-array",
@@ -5030,8 +5022,8 @@ dependencies = [
 
 [[package]]
 name = "lance-index"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arc-swap",
  "arrow",
@@ -5043,7 +5035,6 @@ dependencies = [
  "async-channel",
  "async-recursion",
  "async-trait",
- "bitpacking",
  "bitvec",
  "bytes",
  "chrono",
@@ -5061,6 +5052,7 @@ dependencies = [
  "jsonb",
  "lance-arrow",
  "lance-arrow-stats",
+ "lance-bitpacking",
  "lance-core",
  "lance-datafusion",
  "lance-datagen",
@@ -5096,8 +5088,8 @@ dependencies = [
 
 [[package]]
 name = "lance-io"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow",
  "arrow-arith",
@@ -5138,8 +5130,8 @@ dependencies = [
 
 [[package]]
 name = "lance-linalg"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -5155,8 +5147,8 @@ dependencies = [
 
 [[package]]
 name = "lance-namespace"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow",
  "async-trait",
@@ -5168,8 +5160,8 @@ dependencies = [
 
 [[package]]
 name = "lance-namespace-impls"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow",
  "arrow-ipc",
@@ -5223,8 +5215,8 @@ dependencies = [
 
 [[package]]
 name = "lance-select"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -5239,8 +5231,8 @@ dependencies = [
 
 [[package]]
 name = "lance-table"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow",
  "arrow-array",
@@ -5279,22 +5271,22 @@ dependencies = [
 
 [[package]]
 name = "lance-testing"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "arrow-array",
  "arrow-schema",
  "criterion",
  "lance-arrow",
  "num-traits",
- "pprof",
+ "pprof 0.15.0",
  "rand 0.9.4",
 ]
 
 [[package]]
 name = "lance-tokenizer"
-version = "9.0.0-beta.8"
-source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.8#71c4aa2174971e98acb7e256fde1e1589024f5bc"
+version = "9.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v9.0.0-beta.16#78a814b6b448b433f5ebfdb9a8dbdec7b904c59f"
 dependencies = [
  "icu_segmenter",
  "jieba-rs",
@@ -5307,12 +5299,13 @@ dependencies = [
 
 [[package]]
 name = "lancedb"
-version = "0.31.0-beta.2"
+version = "0.31.0-beta.6"
 dependencies = [
  "ahash",
  "anyhow",
  "arrow",
  "arrow-array",
+ "arrow-buffer",
  "arrow-cast",
  "arrow-data",
  "arrow-ipc",
@@ -5368,6 +5361,7 @@ dependencies = [
  "pin-project",
  "polars",
  "polars-arrow",
+ "pprof 0.14.1",
  "rand 0.9.4",
  "random_word",
  "regex",
@@ -5384,13 +5378,14 @@ dependencies = [
  "tokenizers",
  "tokio",
  "url",
+ "urlencoding",
  "uuid",
  "walkdir",
 ]
 
 [[package]]
 name = "lancedb-nodejs"
-version = "0.31.0-beta.2"
+version = "0.31.0-beta.6"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -5415,7 +5410,7 @@ dependencies = [
 
 [[package]]
 name = "lancedb-python"
-version = "0.34.0-beta.2"
+version = "0.34.0-beta.6"
 dependencies = [
  "arrow",
  "async-trait",
@@ -5648,9 +5643,9 @@ dependencies = [
 
 [[package]]
 name = "log"
-version = "0.4.32"
+version = "0.4.33"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "953f07c43838f8e6f9758cab68bf5bed85465e7587ebe0b823f1bcd81978ad3a"
+checksum = "0ceec5bc11778974d1bcb055b18002eba7f4b3518b6a0081b3af5f21666da9ad"
 
 [[package]]
 name = "loom"
@@ -5958,9 +5953,9 @@ dependencies = [
 
 [[package]]
 name = "napi"
-version = "3.9.3"
+version = "3.10.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fbd9f9295f3ff5921e78a71222c3361a8216f7760b1a99a6ad4e8441de18bbb9"
+checksum = "0c71997d6f7ad4a756966e452426848ac27d3b37a295302d63afbbcce0270f93"
 dependencies = [
  "bitflags 2.11.1",
  "chrono",
@@ -5983,9 +5978,9 @@ checksum = "c9c366d2c8c60b86fa632df75f745509b52f9128f91a6bad4c796e44abb505e1"
 
 [[package]]
 name = "napi-derive"
-version = "3.5.6"
+version = "3.5.9"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "89b3f766e04667e6da0e181e2da4f85475d5a6513b7cf6a80bea184e224a5b42"
+checksum = "d4ba572deef53e2c386759a8c2014175a62679d74ff83adc205c8bc0e0285727"
 dependencies = [
  "convert_case",
  "ctor 1.0.5",
@@ -5997,9 +5992,9 @@ dependencies = [
 
 [[package]]
 name = "napi-derive-backend"
-version = "5.0.4"
+version = "5.1.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0d5af30503edf933ce7377cf6d4c877a62b0f1107ea05585f1b5e430e88d5baf"
+checksum = "ddd961eb2aa8965e3f29722d754f3a86907eb1984e2fbcbe3fe87b9a02d6bfba"
 dependencies = [
  "convert_case",
  "proc-macro2",
@@ -7332,6 +7327,28 @@ version = "0.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "439ee305def115ba05938db6eb1644ff94165c5ab5e9420d1c1bcedbba909391"
 
+[[package]]
+name = "pprof"
+version = "0.14.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "afad4d4df7b31280028245f152d5a575083e2abb822d05736f5e47653e77689f"
+dependencies = [
+ "aligned-vec",
+ "backtrace",
+ "cfg-if 1.0.4",
+ "findshlibs",
+ "inferno",
+ "libc",
+ "log",
+ "nix",
+ "once_cell",
+ "smallvec",
+ "spin 0.10.0",
+ "symbolic-demangle",
+ "tempfile",
+ "thiserror 1.0.69",
+]
+
 [[package]]
 name = "pprof"
 version = "0.15.0"
@@ -10127,9 +10144,9 @@ checksum = "06abde3611657adf66d383f00b093d7faecc7fa57071cce2578660c9f1010821"
 
 [[package]]
 name = "uuid"
-version = "1.23.3"
+version = "1.23.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "144d6b123cef80b301b8f72a9e2ca4370ddec21950d0a103dd22c437006d2db7"
+checksum = "bf80a72845275afea99e7f2b434723d3bc7e38470fcd1c7ed39a599c73319a53"
 dependencies = [
  "getrandom 0.4.2",
  "js-sys",

Cargo.toml

@@ -13,24 +13,25 @@ categories = ["database-implementations"]
 rust-version = "1.91.0"
 
 [workspace.dependencies]
-lance = { "version" = "=9.0.0-beta.8", default-features = false, "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-core = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-datagen = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-file = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-io = { "version" = "=9.0.0-beta.8", default-features = false, "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-index = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-linalg = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-namespace = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-namespace-impls = { "version" = "=9.0.0-beta.8", default-features = false, "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-table = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-testing = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-datafusion = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-encoding = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
-lance-arrow = { "version" = "=9.0.0-beta.8", "tag" = "v9.0.0-beta.8", "git" = "https://github.com/lance-format/lance.git" }
+lance = { "version" = "=9.0.0-beta.16", default-features = false, "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-core = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-datagen = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-file = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-io = { "version" = "=9.0.0-beta.16", default-features = false, "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-index = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-linalg = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace-impls = { "version" = "=9.0.0-beta.16", default-features = false, "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-table = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-testing = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-datafusion = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-encoding = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-arrow = { "version" = "=9.0.0-beta.16", "tag" = "v9.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
 ahash = "0.8"
 # Note that this one does not include pyarrow
 arrow = { version = "58.0.0", optional = false }
 arrow-array = "58.0.0"
+arrow-buffer = "58.0.0"
 arrow-data = "58.0.0"
 arrow-ipc = "58.0.0"
 arrow-ord = "58.0.0"

deny.toml

@@ -119,6 +119,18 @@ ignore = [
     # https://rustsec.org/advisories/RUSTSEC-2026-0177
     { id = "RUSTSEC-2026-0176", reason = "pyo3 in Python bindings; awaiting patched pyo3 release" },
     { id = "RUSTSEC-2026-0177", reason = "pyo3 in Python bindings; awaiting patched pyo3 release" },
+
+    # quick-xml < 0.41.0: quadratic runtime on duplicate attribute names (DoS).
+    # quick-xml < 0.41.0: unbounded namespace-declaration allocation in NsReader (DoS).
+    # Pulled in transitively by inferno (dev-only flame-graph dep), lance-namespace-impls
+    # (git dep from lance), and opendal/reqsign (cloud storage XML parsing). The XML
+    # parsed by opendal/reqsign comes from trusted cloud-storage endpoints (S3, GCS,
+    # Azure), not attacker-controlled input. Clearing requires upstream crates to migrate
+    # to quick-xml >= 0.41.0.
+    # https://rustsec.org/advisories/RUSTSEC-2026-0194
+    # https://rustsec.org/advisories/RUSTSEC-2026-0195
+    { id = "RUSTSEC-2026-0194", reason = "transitive via inferno/lance/opendal; XML from trusted cloud endpoints, not attacker-controlled" },
+    { id = "RUSTSEC-2026-0195", reason = "transitive via inferno/lance/opendal; XML from trusted cloud endpoints, not attacker-controlled" },
 ]
 
 # ---------------------------------------------------------------------------

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>0.31.0-beta.2</version>
+    <version>0.31.0-beta.6</version>
 </dependency>
 ```
 

docs/src/js/classes/Query.md

@@ -518,6 +518,9 @@ x > 5 OR y = 'test'
 
 Filtering performance can often be improved by creating a scalar index
 on the filter column(s).
+
+Calling this multiple times combines the filters with a logical AND rather
+than replacing the previous filter.
 ```
 
 #### Inherited from

docs/src/js/classes/VectorQuery.md

@@ -767,6 +767,9 @@ x > 5 OR y = 'test'
 
 Filtering performance can often be improved by creating a scalar index
 on the filter column(s).
+
+Calling this multiple times combines the filters with a logical AND rather
+than replacing the previous filter.
 ```
 
 #### Inherited from

docs/src/js/enumerations/OAuthFlowType.md

@@ -0,0 +1,29 @@
+[**@lancedb/lancedb**](../README.md) • **Docs**
+
+***
+
+[@lancedb/lancedb](../globals.md) / OAuthFlowType
+
+# Enumeration: OAuthFlowType
+
+OAuth authentication flow types.
+
+## Enumeration Members
+
+### AzureManagedIdentity
+
+```ts
+AzureManagedIdentity: "azure_managed_identity";
+```
+
+Azure Managed Identity via IMDS.
+
+***
+
+### ClientCredentials
+
+```ts
+ClientCredentials: "client_credentials";
+```
+
+Client Credentials grant (service-to-service / M2M).

docs/src/js/globals.md

@@ -12,6 +12,7 @@
 ## Enumerations
 
 - [FullTextQueryType](enumerations/FullTextQueryType.md)
+- [OAuthFlowType](enumerations/OAuthFlowType.md)
 - [Occur](enumerations/Occur.md)
 - [Operator](enumerations/Operator.md)
 
@@ -85,6 +86,8 @@
 - [ListNamespacesResponse](interfaces/ListNamespacesResponse.md)
 - [LsmWriteSpec](interfaces/LsmWriteSpec.md)
 - [MergeResult](interfaces/MergeResult.md)
+- [NativeOAuthConfig](interfaces/NativeOAuthConfig.md)
+- [OAuthConfig](interfaces/OAuthConfig.md)
 - [OpenTableOptions](interfaces/OpenTableOptions.md)
 - [OptimizeOptions](interfaces/OptimizeOptions.md)
 - [OptimizeStats](interfaces/OptimizeStats.md)

docs/src/js/interfaces/ConnectionOptions.md

@@ -64,6 +64,19 @@ client used by manifest-enabled native connections.
 
 ***
 
+### oauthConfig?
+
+```ts
+optional oauthConfig: NativeOAuthConfig;
+```
+
+(For LanceDB cloud only): OAuth configuration for IdP-based
+authentication (e.g., Azure Entra ID). When set, token acquisition
+and refresh are handled entirely in Rust. TypeScript users should pass
+the public `OAuthConfig` type exported from `@lancedb/lancedb`.
+
+***
+
 ### readConsistencyInterval?
 
 ```ts

docs/src/js/interfaces/NativeOAuthConfig.md

@@ -0,0 +1,88 @@
+[**@lancedb/lancedb**](../README.md) • **Docs**
+
+***
+
+[@lancedb/lancedb](../globals.md) / NativeOAuthConfig
+
+# Interface: NativeOAuthConfig
+
+OAuth configuration for LanceDB authentication.
+
+This is the generated napi-rs binding shape. TypeScript users should prefer
+the public `OAuthConfig` type exported from `@lancedb/lancedb`.
+
+All token acquisition and refresh is handled in the Rust layer.
+
+## Properties
+
+### clientId
+
+```ts
+clientId: string;
+```
+
+Application / Client ID.
+
+***
+
+### clientSecret?
+
+```ts
+optional clientSecret: string;
+```
+
+Client secret (required for client_credentials).
+
+***
+
+### flow?
+
+```ts
+optional flow: string;
+```
+
+Authentication flow: "client_credentials" or "azure_managed_identity"
+
+***
+
+### issuerUrl
+
+```ts
+issuerUrl: string;
+```
+
+OIDC issuer URL or OAuth authority URL.
+For Azure: `https://login.microsoftonline.com/{tenant_id}/v2.0`
+
+***
+
+### managedIdentityClientId?
+
+```ts
+optional managedIdentityClientId: string;
+```
+
+Client ID for user-assigned managed identity (azure_managed_identity).
+
+***
+
+### refreshBufferSecs?
+
+```ts
+optional refreshBufferSecs: number;
+```
+
+Seconds before expiry to trigger proactive refresh (default: 300).
+Keep this well below the token TTL; if it is greater than or equal to
+the TTL, each request refreshes the token.
+
+***
+
+### scopes
+
+```ts
+scopes: string[];
+```
+
+OAuth scopes to request. For Azure managed identity, exactly one scope
+or resource is required. For example: `["api://{app_id}/.default"]`

docs/src/js/interfaces/OAuthConfig.md

@@ -0,0 +1,111 @@
+[**@lancedb/lancedb**](../README.md) • **Docs**
+
+***
+
+[@lancedb/lancedb](../globals.md) / OAuthConfig
+
+# Interface: OAuthConfig
+
+OAuth configuration for LanceDB authentication.
+
+This is the public TypeScript OAuth configuration type. The generated
+`NativeOAuthConfig` type has the same runtime shape but is an implementation
+detail of the napi-rs binding.
+
+All token acquisition and refresh is handled in the Rust layer.
+This config is passed through to Rust via napi-rs.
+
+## Examples
+
+```typescript
+const config: OAuthConfig = {
+  issuerUrl: "https://login.microsoftonline.com/{tenant}/v2.0",
+  clientId: "app-id",
+  clientSecret: "secret",
+  scopes: ["api://lancedb-api/.default"],
+};
+```
+
+```typescript
+const config: OAuthConfig = {
+  issuerUrl: "https://login.microsoftonline.com/{tenant}/v2.0",
+  clientId: "app-id",
+  scopes: ["api://lancedb-api/.default"],
+  flow: OAuthFlowType.AzureManagedIdentity,
+};
+```
+
+## Properties
+
+### clientId
+
+```ts
+clientId: string;
+```
+
+Application / Client ID.
+
+***
+
+### clientSecret?
+
+```ts
+optional clientSecret: string;
+```
+
+Client secret (required for ClientCredentials).
+
+***
+
+### flow?
+
+```ts
+optional flow: OAuthFlowType;
+```
+
+Authentication flow (default: ClientCredentials).
+
+***
+
+### issuerUrl
+
+```ts
+issuerUrl: string;
+```
+
+OIDC issuer URL or OAuth authority URL.
+For Azure: `https://login.microsoftonline.com/{tenant_id}/v2.0`
+
+***
+
+### managedIdentityClientId?
+
+```ts
+optional managedIdentityClientId: string;
+```
+
+Client ID for user-assigned managed identity (AzureManagedIdentity).
+
+***
+
+### refreshBufferSecs?
+
+```ts
+optional refreshBufferSecs: number;
+```
+
+Seconds before expiry to trigger proactive refresh (default: 300).
+Keep this well below the token TTL; if it is greater than or equal to
+the TTL, each request refreshes the token.
+
+***
+
+### scopes
+
+```ts
+scopes: string[];
+```
+
+OAuth scopes to request.
+For Azure managed identity, exactly one scope or resource is required.
+For example: `["api://{app_id}/.default"]`

docs/src/js/interfaces/SplitRandomOptions.md

@@ -8,6 +8,14 @@
 
 ## Properties
 
+### clumpSize?
+
+```ts
+optional clumpSize: number;
+```
+
+***
+
 ### counts?
 
 ```ts

java/lancedb-core/pom.xml

@@ -8,7 +8,7 @@
     <parent>
       <groupId>com.lancedb</groupId>
       <artifactId>lancedb-parent</artifactId>
-      <version>0.31.0-beta.2</version>
+      <version>0.31.0-beta.6</version>
       <relativePath>../pom.xml</relativePath>
     </parent>
 

java/pom.xml

@@ -6,7 +6,7 @@
 
     <groupId>com.lancedb</groupId>
     <artifactId>lancedb-parent</artifactId>
-    <version>0.31.0-beta.2</version>
+    <version>0.31.0-beta.6</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>9.0.0-beta.8</lance-core.version>
+        <lance-core.version>9.0.0-beta.16</lance-core.version>
         <spotless.skip>false</spotless.skip>
         <spotless.version>2.30.0</spotless.version>
         <spotless.java.googlejavaformat.version>1.7</spotless.java.googlejavaformat.version>

nodejs/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "lancedb-nodejs"
 edition.workspace = true
-version = "0.31.0-beta.2"
+version = "0.31.0-beta.6"
 publish = false
 license.workspace = true
 description.workspace = true

nodejs/__test__/query.test.ts

@@ -215,6 +215,20 @@ describe("Query orderBy", () => {
     expect(results[2].score).toBeCloseTo(4.1, 0.001);
   });
 
+  it("should combine repeated where clauses with AND", async () => {
+    const results = await table
+      .query()
+      .where("score > 1.0")
+      .where("score < 3.0")
+      .orderBy({ columnName: "score" })
+      .toArray();
+    // Only rows matching both predicates should be returned, rather than the
+    // second where() silently replacing the first.
+    expect(results.length).toBe(2);
+    expect(results[0].score).toBeCloseTo(1.2, 0.001);
+    expect(results[1].score).toBeCloseTo(2.8, 0.001);
+  });
+
   it("should support method chaining with limit", async () => {
     const results = await table
       .query()

nodejs/lancedb/index.ts

@@ -52,6 +52,7 @@ export {
   SplitHashOptions,
   SplitSequentialOptions,
   ShuffleOptions,
+  OAuthConfig as NativeOAuthConfig,
 } from "./native.js";
 
 export {
@@ -130,6 +131,8 @@ export {
   TokenResponse,
 } from "./header";
 
+export { OAuthConfig, OAuthFlowType } from "./oauth";
+
 export { MergeInsertBuilder, WriteExecutionOptions } from "./merge";
 
 export * as embedding from "./embedding";

nodejs/lancedb/oauth.ts

@@ -0,0 +1,76 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+/**
+ * OAuth authentication flow types.
+ */
+export enum OAuthFlowType {
+  /** Client Credentials grant (service-to-service / M2M). */
+  ClientCredentials = "client_credentials",
+  /** Azure Managed Identity via IMDS. */
+  AzureManagedIdentity = "azure_managed_identity",
+}
+
+/**
+ * OAuth configuration for LanceDB authentication.
+ *
+ * This is the public TypeScript OAuth configuration type. The generated
+ * `NativeOAuthConfig` type has the same runtime shape but is an implementation
+ * detail of the napi-rs binding.
+ *
+ * All token acquisition and refresh is handled in the Rust layer.
+ * This config is passed through to Rust via napi-rs.
+ *
+ * @example Client Credentials (service-to-service):
+ * ```typescript
+ * const config: OAuthConfig = {
+ *   issuerUrl: "https://login.microsoftonline.com/{tenant}/v2.0",
+ *   clientId: "app-id",
+ *   clientSecret: "secret",
+ *   scopes: ["api://lancedb-api/.default"],
+ * };
+ * ```
+ *
+ * @example Azure Managed Identity:
+ * ```typescript
+ * const config: OAuthConfig = {
+ *   issuerUrl: "https://login.microsoftonline.com/{tenant}/v2.0",
+ *   clientId: "app-id",
+ *   scopes: ["api://lancedb-api/.default"],
+ *   flow: OAuthFlowType.AzureManagedIdentity,
+ * };
+ * ```
+ */
+export interface OAuthConfig {
+  /**
+   * OIDC issuer URL or OAuth authority URL.
+   * For Azure: `https://login.microsoftonline.com/{tenant_id}/v2.0`
+   */
+  issuerUrl: string;
+
+  /** Application / Client ID. */
+  clientId: string;
+
+  /**
+   * OAuth scopes to request.
+   * For Azure managed identity, exactly one scope or resource is required.
+   * For example: `["api://{app_id}/.default"]`
+   */
+  scopes: string[];
+
+  /** Authentication flow (default: ClientCredentials). */
+  flow?: OAuthFlowType;
+
+  /** Client secret (required for ClientCredentials). */
+  clientSecret?: string;
+
+  /** Client ID for user-assigned managed identity (AzureManagedIdentity). */
+  managedIdentityClientId?: string;
+
+  /**
+   * Seconds before expiry to trigger proactive refresh (default: 300).
+   * Keep this well below the token TTL; if it is greater than or equal to
+   * the TTL, each request refreshes the token.
+   */
+  refreshBufferSecs?: number;
+}

nodejs/lancedb/query.ts

@@ -362,6 +362,9 @@ export class StandardQueryBase<
    *
    * Filtering performance can often be improved by creating a scalar index
    * on the filter column(s).
+   *
+   * Calling this multiple times combines the filters with a logical AND rather
+   * than replacing the previous filter.
    */
   where(predicate: string): this {
     this.doCall((inner: NativeQueryType) => inner.onlyIf(predicate));

nodejs/npm/darwin-arm64/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-darwin-arm64",
-	"version": "0.31.0-beta.2",
+	"version": "0.31.0-beta.6",
 	"os": ["darwin"],
 	"cpu": ["arm64"],
 	"main": "lancedb.darwin-arm64.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-gnu",
-	"version": "0.31.0-beta.2",
+	"version": "0.31.0-beta.6",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-gnu.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-musl",
-	"version": "0.31.0-beta.2",
+	"version": "0.31.0-beta.6",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-musl.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-gnu",
-	"version": "0.31.0-beta.2",
+	"version": "0.31.0-beta.6",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-gnu.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-musl",
-	"version": "0.31.0-beta.2",
+	"version": "0.31.0-beta.6",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-musl.node",

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

@@ -1,6 +1,6 @@
 {
   "name": "@lancedb/lancedb-win32-arm64-msvc",
-  "version": "0.31.0-beta.2",
+  "version": "0.31.0-beta.6",
   "os": [
     "win32"
   ],

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-win32-x64-msvc",
-	"version": "0.31.0-beta.2",
+	"version": "0.31.0-beta.6",
 	"os": ["win32"],
 	"cpu": ["x64"],
 	"main": "lancedb.win32-x64-msvc.node",

nodejs/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@lancedb/lancedb",
-  "version": "0.31.0-beta.2",
+  "version": "0.31.0-beta.6",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "@lancedb/lancedb",
-      "version": "0.31.0-beta.2",
+      "version": "0.31.0-beta.6",
       "cpu": [
         "x64",
         "arm64"

nodejs/package.json

@@ -11,7 +11,7 @@
     "ann"
   ],
   "private": false,
-  "version": "0.31.0-beta.2",
+  "version": "0.31.0-beta.6",
   "main": "dist/index.js",
   "exports": {
     ".": "./dist/index.js",

nodejs/src/connection.rs

@@ -112,6 +112,12 @@ impl Connection {
 
         builder = builder.client_config(rust_config);
 
+        if let Some(oauth_config) = options.oauth_config {
+            let config: lancedb::remote::oauth::OAuthConfig =
+                oauth_config.try_into().default_error()?;
+            builder = builder.oauth_config(config);
+        }
+
         if let Some(api_key) = options.api_key {
             builder = builder.api_key(&api_key);
         }

nodejs/src/lib.rs

@@ -65,6 +65,11 @@ pub struct ConnectionOptions {
     /// (For LanceDB cloud only): the host to use for LanceDB cloud. Used
     /// for testing purposes.
     pub host_override: Option<String>,
+    /// (For LanceDB cloud only): OAuth configuration for IdP-based
+    /// authentication (e.g., Azure Entra ID). When set, token acquisition
+    /// and refresh are handled entirely in Rust. TypeScript users should pass
+    /// the public `OAuthConfig` type exported from `@lancedb/lancedb`.
+    pub oauth_config: Option<remote::OAuthConfig>,
 }
 
 #[napi(object)]

nodejs/src/merge.rs

@@ -3,7 +3,7 @@
 
 use std::time::Duration;
 
-use lancedb::{arrow::IntoArrow, ipc::ipc_file_to_batches, table::merge::MergeInsertBuilder};
+use lancedb::{ipc::ipc_file_to_batches, table::merge::MergeInsertBuilder};
 use napi::bindgen_prelude::*;
 use napi_derive::napi;
 
@@ -66,11 +66,9 @@ impl NativeMergeInsertBuilder {
 
     #[napi(catch_unwind)]
     pub async fn execute(&self, buf: Buffer) -> napi::Result<MergeResult> {
-        let data = ipc_file_to_batches(buf.to_vec())
-            .and_then(IntoArrow::into_arrow)
-            .map_err(|e| {
-                napi::Error::from_reason(format!("Failed to read IPC file: {}", convert_error(&e)))
-            })?;
+        let data = ipc_file_to_batches(buf.to_vec()).map_err(|e| {
+            napi::Error::from_reason(format!("Failed to read IPC file: {}", convert_error(&e)))
+        })?;
 
         let this = self.clone();
 

nodejs/src/permutation.rs

@@ -16,6 +16,7 @@ pub struct SplitRandomOptions {
     pub counts: Option<Vec<i64>>,
     pub fixed: Option<i64>,
     pub seed: Option<i64>,
+    pub clump_size: Option<i64>,
     pub split_names: Option<Vec<String>>,
 }
 
@@ -125,10 +126,15 @@ impl PermutationBuilder {
         };
 
         let seed = options.seed.map(|s| s as u64);
+        let clump_size = options.clump_size.map(|c| c as u64);
 
         self.modify(|builder| {
             builder.with_split_strategy(
-                SplitStrategy::Random { seed, sizes },
+                SplitStrategy::Random {
+                    seed,
+                    sizes,
+                    clump_size,
+                },
                 options.split_names.clone(),
             )
         })

nodejs/src/remote.rs

@@ -3,6 +3,7 @@
 
 use std::collections::HashMap;
 
+use lancedb::error::Error;
 use napi_derive::*;
 
 /// Timeout configuration for remote HTTP client.
@@ -140,6 +141,84 @@ impl From<TlsConfig> for lancedb::remote::TlsConfig {
     }
 }
 
+/// OAuth configuration for LanceDB authentication.
+///
+/// This is the generated napi-rs binding shape. TypeScript users should prefer
+/// the public `OAuthConfig` type exported from `@lancedb/lancedb`.
+///
+/// All token acquisition and refresh is handled in the Rust layer.
+#[napi(object)]
+#[derive(Clone)]
+pub struct OAuthConfig {
+    /// OIDC issuer URL or OAuth authority URL.
+    /// For Azure: `https://login.microsoftonline.com/{tenant_id}/v2.0`
+    pub issuer_url: String,
+    /// Application / Client ID.
+    pub client_id: String,
+    /// OAuth scopes to request. For Azure managed identity, exactly one scope
+    /// or resource is required. For example: `["api://{app_id}/.default"]`
+    pub scopes: Vec<String>,
+    /// Authentication flow: "client_credentials" or "azure_managed_identity"
+    pub flow: Option<String>,
+    /// Client secret (required for client_credentials).
+    pub client_secret: Option<String>,
+    /// Client ID for user-assigned managed identity (azure_managed_identity).
+    pub managed_identity_client_id: Option<String>,
+    /// Seconds before expiry to trigger proactive refresh (default: 300).
+    /// Keep this well below the token TTL; if it is greater than or equal to
+    /// the TTL, each request refreshes the token.
+    pub refresh_buffer_secs: Option<u32>,
+}
+
+impl std::fmt::Debug for OAuthConfig {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("OAuthConfig")
+            .field("issuer_url", &self.issuer_url)
+            .field("client_id", &self.client_id)
+            .field("scopes", &self.scopes)
+            .field("flow", &self.flow)
+            .field(
+                "client_secret",
+                &self.client_secret.as_deref().map(|_| "<redacted>"),
+            )
+            .field(
+                "managed_identity_client_id",
+                &self.managed_identity_client_id,
+            )
+            .field("refresh_buffer_secs", &self.refresh_buffer_secs)
+            .finish()
+    }
+}
+
+impl TryFrom<OAuthConfig> for lancedb::remote::oauth::OAuthConfig {
+    type Error = Error;
+
+    fn try_from(config: OAuthConfig) -> Result<Self, Self::Error> {
+        use lancedb::remote::oauth::OAuthFlow;
+
+        let flow = match config.flow.as_deref().unwrap_or("client_credentials") {
+            "client_credentials" => OAuthFlow::ClientCredentials,
+            "azure_managed_identity" => OAuthFlow::AzureManagedIdentity {
+                client_id: config.managed_identity_client_id,
+            },
+            other => {
+                return Err(Error::InvalidInput {
+                    message: format!("Unknown OAuth flow type: {other}"),
+                });
+            }
+        };
+
+        Ok(Self {
+            issuer_url: config.issuer_url,
+            client_id: config.client_id,
+            client_secret: config.client_secret,
+            scopes: config.scopes,
+            flow,
+            refresh_buffer_secs: config.refresh_buffer_secs.map(|v| v as u64),
+        })
+    }
+}
+
 impl From<ClientConfig> for lancedb::remote::ClientConfig {
     fn from(config: ClientConfig) -> Self {
         Self {
@@ -156,3 +235,45 @@ impl From<ClientConfig> for lancedb::remote::ClientConfig {
         }
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_unknown_oauth_flow_returns_invalid_input() {
+        let config = OAuthConfig {
+            issuer_url: "https://issuer.example.com".to_string(),
+            client_id: "client-id".to_string(),
+            scopes: vec!["scope".to_string()],
+            flow: Some("typo".to_string()),
+            client_secret: None,
+            managed_identity_client_id: None,
+            refresh_buffer_secs: None,
+        };
+
+        let err = lancedb::remote::oauth::OAuthConfig::try_from(config).unwrap_err();
+        assert!(matches!(
+            err,
+            Error::InvalidInput { message }
+                if message == "Unknown OAuth flow type: typo"
+        ));
+    }
+
+    #[test]
+    fn test_oauth_config_debug_redacts_client_secret() {
+        let config = OAuthConfig {
+            issuer_url: "https://issuer.example.com".to_string(),
+            client_id: "client-id".to_string(),
+            scopes: vec!["scope".to_string()],
+            flow: Some("client_credentials".to_string()),
+            client_secret: Some("super-secret".to_string()),
+            managed_identity_client_id: None,
+            refresh_buffer_secs: None,
+        };
+
+        let debug = format!("{config:?}");
+        assert!(!debug.contains("super-secret"));
+        assert!(debug.contains("client_secret: Some(\"<redacted>\")"));
+    }
+}

python/.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.34.0-beta.3"
+current_version = "0.34.0-beta.6"
 parse = """(?x)
     (?P<major>0|[1-9]\\d*)\\.
     (?P<minor>0|[1-9]\\d*)\\.

python/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "lancedb-python"
-version = "0.34.0-beta.3"
+version = "0.34.0-beta.6"
 publish = false
 edition.workspace = true
 description = "Python bindings for LanceDB"

python/benchmarks/bench_streaming_dataloader.py

@@ -0,0 +1,135 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+"""Benchmark for StreamingDataset throughput.
+
+Sweeps read_batch_size from 1 to 16384 to show how amortising the per-request
+overhead scales.  Each row at each chunk size is timed via the real
+StreamingDataset so the numbers reflect production code.
+
+Run with:
+    cd python
+    uv run --extra tests benchmarks/bench_streaming_dataloader.py
+
+Optional env vars:
+    BENCH_NUM_ROWS   — total rows in the table (default 49152 = 24 × 2048)
+    BENCH_NUM_SPLITS — number of splits (default 24)
+    BENCH_STEPS      — round-robin cycles to time per chunk size (default 100)
+    BENCH_ROW_BYTES  — approximate bytes per row padded with a binary column
+                       (default 4096, mimics a small embedding/image patch)
+"""
+
+import os
+import time
+import tempfile
+
+import pyarrow as pa
+import lancedb
+
+from lancedb.streaming import StreamingDataset
+
+# ---------------------------------------------------------------------------
+# Configuration
+# ---------------------------------------------------------------------------
+
+NUM_SPLITS = int(os.environ.get("BENCH_NUM_SPLITS", 24))
+# Default: 2048 rows per split so every chunk size up to 16Ki has ≥1 full
+# chunk (except 16Ki itself which gets a single full-split fetch — still valid).
+NUM_ROWS = int(os.environ.get("BENCH_NUM_ROWS", NUM_SPLITS * 2048))
+STEPS = int(os.environ.get("BENCH_STEPS", 100))
+ROW_BYTES = int(os.environ.get("BENCH_ROW_BYTES", 4096))
+
+assert NUM_ROWS % NUM_SPLITS == 0, "NUM_ROWS must be divisible by NUM_SPLITS"
+
+CHUNK_SIZES = [1, 4, 16, 64, 256, 1024, 4096, 16384]
+
+
+# ---------------------------------------------------------------------------
+# Table helpers
+# ---------------------------------------------------------------------------
+
+
+def make_table(db_path: str) -> lancedb.table.Table:
+    db = lancedb.connect(db_path)
+    payload = b"x" * ROW_BYTES
+    data = pa.table(
+        {
+            "id": pa.array(range(NUM_ROWS), type=pa.int32()),
+            "payload": pa.array([payload] * NUM_ROWS, type=pa.large_binary()),
+        }
+    )
+    return db.create_table("bench", data, mode="overwrite")
+
+
+# ---------------------------------------------------------------------------
+# Timing
+# ---------------------------------------------------------------------------
+
+
+def bench_chunk(table, chunk_size: int, steps: int) -> tuple[int, float]:
+    """Return (rows_drained, elapsed_seconds) for one timed run."""
+    total_rows = steps * NUM_SPLITS
+    ds = StreamingDataset(
+        table, num_splits=NUM_SPLITS, shuffle_seed=42, read_batch_size=chunk_size
+    )
+    count = 0
+    t0 = time.perf_counter()
+    for _ in ds:
+        count += 1
+        if count >= total_rows:
+            break
+    return count, time.perf_counter() - t0
+
+
+# ---------------------------------------------------------------------------
+# Main
+# ---------------------------------------------------------------------------
+
+
+def main() -> None:
+    rows_per_split = NUM_ROWS // NUM_SPLITS
+    print("Benchmark config:")
+    print(
+        f"  NUM_ROWS={NUM_ROWS}  NUM_SPLITS={NUM_SPLITS}  "
+        f"rows/split={rows_per_split}  STEPS={STEPS}  ROW_BYTES={ROW_BYTES}"
+    )
+    print(f"  ~{NUM_ROWS * ROW_BYTES / 1024 / 1024:.1f} MB total table size")
+    print()
+
+    with tempfile.TemporaryDirectory() as tmp:
+        print("Creating table...", flush=True)
+        table = make_table(tmp)
+
+        cols = (
+            f"{'chunk':>6}  {'rows':>6}  {'elapsed':>8}  {'rows/s':>10}  {'ms/step':>9}"
+        )
+        print(f"\n{cols}")
+        print("-" * 52)
+
+        for chunk in CHUNK_SIZES:
+            # Warm-up pass (one step's worth of rows)
+            warmup_ds = StreamingDataset(
+                table, num_splits=NUM_SPLITS, shuffle_seed=42, read_batch_size=chunk
+            )
+            warmup_count = 0
+            for _ in warmup_ds:
+                warmup_count += 1
+                if warmup_count >= NUM_SPLITS:
+                    break
+
+            drained, elapsed = bench_chunk(table, chunk, STEPS)
+            rows_per_sec = drained / elapsed if elapsed > 0 else float("inf")
+            ms_per_step = elapsed / STEPS * 1000
+
+            print(
+                f"{chunk:>6}  {drained:>6}  {elapsed:>7.3f}s  "
+                f"{rows_per_sec:>10.0f}  {ms_per_step:>8.1f}ms"
+            )
+
+    print()
+    print("Done.")
+
+
+if __name__ == "__main__":
+    main()

python/python/lancedb/__init__.py

@@ -89,6 +89,8 @@ def connect(
         If presented, connect to LanceDB cloud.
         Otherwise, connect to a database on file system or cloud storage.
         Can be set via environment variable `LANCEDB_API_KEY`.
+        OAuth configuration is currently supported only by ``connect_async``;
+        synchronous LanceDB Cloud connections require an API key.
     region: str, default "us-east-1"
         The region to use for LanceDB Cloud.
     host_override: str, optional
@@ -340,6 +342,7 @@ async def connect_async(
     session: Optional[Session] = None,
     manifest_enabled: bool = False,
     namespace_client_properties: Optional[Dict[str, str]] = None,
+    oauth_config=None,
 ) -> AsyncConnection:
     """Connect to a LanceDB database.
 
@@ -389,6 +392,10 @@ async def connect_async(
     namespace_client_properties : dict, optional
         Additional directory namespace client properties to use with
         ``manifest_enabled=True``.
+    oauth_config : OAuthConfig, optional
+        OAuth configuration for LanceDB Cloud/Enterprise. This is supported by
+        ``connect_async`` only; synchronous ``connect`` uses API key
+        authentication for ``db://`` URIs.
 
     Examples
     --------
@@ -435,6 +442,7 @@ async def connect_async(
             session,
             manifest_enabled,
             namespace_client_properties,
+            oauth_config,
         )
     )
 

python/python/lancedb/_lancedb.pyi

@@ -1,4 +1,5 @@
-from datetime import datetime, timedelta
+from datetime import date, datetime, timedelta
+from decimal import Decimal
 from typing import Dict, List, Optional, Tuple, Any, TypedDict, Union, Literal
 
 import pyarrow as pa
@@ -53,7 +54,9 @@ class PyExpr:
     def to_sql(self) -> str: ...
 
 def expr_col(name: str) -> PyExpr: ...
-def expr_lit(value: Union[bool, int, float, str, bytes]) -> PyExpr: ...
+def expr_lit(
+    value: Union[bool, int, float, str, bytes, date, datetime, Decimal],
+) -> PyExpr: ...
 def expr_func(name: str, args: List[PyExpr]) -> PyExpr: ...
 
 class Session:
@@ -280,6 +283,24 @@ async def connect(
     session: Optional[Session],
     manifest_enabled: bool = False,
     namespace_client_properties: Optional[Dict[str, str]] = None,
+    oauth_config: Optional[Any] = None,
+) -> Connection: ...
+def connect_namespace(
+    namespace_client_impl: str,
+    namespace_client_properties: Dict[str, str],
+    read_consistency_interval: Optional[float] = None,
+    storage_options: Optional[Dict[str, str]] = None,
+    session: Optional[Session] = None,
+    namespace_client_pushdown_operations: Optional[List[str]] = None,
+) -> Connection: ...
+def connect_namespace_client(
+    namespace_client: Any,
+    read_consistency_interval: Optional[float] = None,
+    storage_options: Optional[Dict[str, str]] = None,
+    session: Optional[Session] = None,
+    namespace_client_pushdown_operations: Optional[List[str]] = None,
+    namespace_client_impl: Optional[str] = None,
+    namespace_client_properties: Optional[Dict[str, str]] = None,
 ) -> Connection: ...
 
 class RecordBatchStream:

python/python/lancedb/expr.py

@@ -19,6 +19,8 @@ operators::
 
 from __future__ import annotations
 
+from datetime import date, datetime
+from decimal import Decimal
 from typing import Iterable, Union
 
 import pyarrow as pa
@@ -63,7 +65,7 @@ def _coerce(value: "ExprLike") -> "Expr":
 
 
 # Type alias used in annotations.
-ExprLike = Union["Expr", bool, int, float, str, bytes]
+ExprLike = Union["Expr", bool, int, float, str, bytes, date, datetime, Decimal]
 
 
 class Expr:
@@ -118,10 +120,18 @@ class Expr:
         """Logical AND (``expr_a & expr_b``)."""
         return Expr(self._inner.and_(_coerce(other)._inner))
 
+    def __rand__(self, other: ExprLike) -> "Expr":
+        """Right-hand logical AND (``True & expr``)."""
+        return Expr(_coerce(other)._inner.and_(self._inner))
+
     def __or__(self, other: "Expr") -> "Expr":
         """Logical OR (``expr_a | expr_b``)."""
         return Expr(self._inner.or_(_coerce(other)._inner))
 
+    def __ror__(self, other: ExprLike) -> "Expr":
+        """Right-hand logical OR (``False | expr``)."""
+        return Expr(_coerce(other)._inner.or_(self._inner))
+
     def __invert__(self) -> "Expr":
         """Logical NOT (``~expr``)."""
         return Expr(self._inner.not_())
@@ -266,13 +276,14 @@ def col(name: str) -> Expr:
     return Expr(expr_col(name))
 
 
-def lit(value: Union[bool, int, float, str, bytes]) -> Expr:
+def lit(value: Union[bool, int, float, str, bytes, date, datetime, Decimal]) -> Expr:
     """Create a literal (constant) value expression.
 
     Parameters
     ----------
     value:
-        A Python ``bool``, ``int``, ``float``, ``str``, or ``bytes``.
+        A Python ``bool``, ``int``, ``float``, ``str``, ``bytes``, ``date``,
+        ``datetime``, or ``Decimal``.
 
     Examples
     --------
@@ -280,6 +291,9 @@ def lit(value: Union[bool, int, float, str, bytes]) -> Expr:
     >>> col("price") * lit(1.1)
     Expr((price * 1.1))
     """
+    if not isinstance(value, (bool, int, float, str, bytes, date, datetime, Decimal)):
+        raise TypeError(f"Unsupported literal type: {type(value).__name__}")
+
     return Expr(expr_lit(value))
 
 

python/python/lancedb/merge.py

@@ -51,6 +51,15 @@ class LanceMergeInsertBuilder(object):
         If there are multiple matches then the behavior is undefined.
         Currently this causes multiple copies of the row to be created
         but that behavior is subject to change.
+
+        Parameters
+        ----------
+        where: Optional[str], default None
+            An optional filter to limit which rows are updated. Column
+            references in this expression must be prefixed with "target."
+            to refer to the existing table data. For example, to only
+            update rows where the existing color is red, use:
+            ``where="target.color = 'red'"``
         """
         self._when_matched_update_all = True
         self._when_matched_update_all_condition = where

python/python/lancedb/namespace.py

@@ -38,15 +38,13 @@ 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 lance_namespace.errors import NamespaceNotEmptyError, TableNotFoundError
+from lancedb._lancedb import (
+    connect_namespace as _connect_namespace,
+    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,
-    _normalize_drop_namespace_behavior,
-)
 from lance_namespace import (
     LanceNamespace,
     connect as namespace_connect,
@@ -55,13 +53,6 @@ from lance_namespace import (
     DropNamespaceResponse,
     ListNamespacesResponse,
     ListTablesResponse,
-    ListTablesRequest,
-    DescribeNamespaceRequest,
-    DropTableRequest,
-    RenameTableRequest,
-    ListNamespacesRequest,
-    CreateNamespaceRequest,
-    DropNamespaceRequest,
 )
 from lancedb.table import AsyncTable, LanceTable, Table
 from lancedb.util import validate_table_name
@@ -373,6 +364,23 @@ def _convert_pyarrow_schema_to_json(schema: pa.Schema) -> JsonArrowSchema:
     return JsonArrowSchema(fields=fields, metadata=meta)
 
 
+def _builds_namespace_natively(
+    namespace_client_impl: Optional[str],
+    namespace_client_properties: Optional[Dict[str, str]],
+) -> bool:
+    """Whether ``connect_namespace_client`` builds the namespace client natively
+    in Rust (installing the read-freshness context provider) rather than wrapping
+    the pre-built Python client.
+
+    Must mirror Rust ``build_namespace_natively`` in ``python/src/connection.rs``.
+    """
+    return namespace_client_impl == "rest" and bool(namespace_client_properties)
+
+
+def _supports_native_namespace(namespace_client_impl: str) -> bool:
+    return namespace_client_impl in {"dir", "rest"}
+
+
 class LanceNamespaceDBConnection(DBConnection):
     """
     A LanceDB connection that uses a namespace for table management.
@@ -383,7 +391,7 @@ class LanceNamespaceDBConnection(DBConnection):
 
     def __init__(
         self,
-        namespace_client: LanceNamespace,
+        namespace_client: Optional[LanceNamespace] = None,
         *,
         read_consistency_interval: Optional[timedelta] = None,
         storage_options: Optional[Dict[str, str]] = None,
@@ -391,6 +399,7 @@ class LanceNamespaceDBConnection(DBConnection):
         namespace_client_pushdown_operations: Optional[List[str]] = None,
         namespace_client_impl: Optional[str] = None,
         namespace_client_properties: Optional[Dict[str, str]] = None,
+        _inner: Optional[AsyncConnection] = None,
     ):
         """
         Initialize a namespace-based LanceDB connection.
@@ -432,23 +441,36 @@ 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,
-            )
+        # When the namespace connection or client is built natively in Rust, the
+        # underlying Rust table performs QueryTable pushdown through the
+        # read-freshness context provider, which the pure-Python ``query_table``
+        # path bypasses.
+        self._route_pushdown_to_rust = _inner is not None or _builds_namespace_natively(
+            namespace_client_impl, namespace_client_properties
         )
+        if _inner is not None:
+            self._inner = _inner
+        else:
+            if namespace_client is None:
+                raise ValueError("namespace_client is required without a native _inner")
+            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,
+                )
+            )
+        self._uri = self._inner.uri
 
     @override
     def serialize(self) -> str:
@@ -494,11 +516,11 @@ class LanceNamespaceDBConnection(DBConnection):
         )
         if namespace_path is None:
             namespace_path = []
-        request = ListTablesRequest(
-            id=namespace_path, page_token=page_token, limit=limit
+        return LOOP.run(
+            self._inner.table_names(
+                namespace_path=namespace_path, start_after=page_token, limit=limit
+            )
         )
-        response = self._namespace_client.list_tables(request)
-        return response.tables if response.tables else []
 
     @override
     def create_table(
@@ -543,6 +565,7 @@ class LanceNamespaceDBConnection(DBConnection):
             namespace_path=namespace_path,
             namespace_client=self._namespace_client,
             pushdown_operations=self._namespace_client_pushdown_operations,
+            route_pushdown_to_rust=self._route_pushdown_to_rust,
             _async=async_table,
         )
 
@@ -568,8 +591,8 @@ class LanceNamespaceDBConnection(DBConnection):
                     index_cache_size=index_cache_size,
                 )
             )
-        except RuntimeError as e:
-            if "Table not found" in str(e):
+        except (RuntimeError, ValueError) as e:
+            if "Table not found" in str(e) or "was not found" in str(e):
                 table_id = namespace_path + [name]
                 raise TableNotFoundError(f"Table not found: {'$'.join(table_id)}")
             raise
@@ -580,6 +603,7 @@ class LanceNamespaceDBConnection(DBConnection):
             namespace_path=namespace_path,
             namespace_client=self._namespace_client,
             pushdown_operations=self._namespace_client_pushdown_operations,
+            route_pushdown_to_rust=self._route_pushdown_to_rust,
             _async=async_table,
         )
         if branch is not None:
@@ -590,12 +614,9 @@ class LanceNamespaceDBConnection(DBConnection):
 
     @override
     def drop_table(self, name: str, namespace_path: Optional[List[str]] = None):
-        # Use namespace drop_table directly
         if namespace_path is None:
             namespace_path = []
-        table_id = namespace_path + [name]
-        request = DropTableRequest(id=table_id)
-        self._namespace_client.drop_table(request)
+        LOOP.run(self._inner.drop_table(name, namespace_path=namespace_path))
 
     @override
     def rename_table(
@@ -609,14 +630,19 @@ class LanceNamespaceDBConnection(DBConnection):
             cur_namespace_path = []
         if new_namespace_path is None:
             new_namespace_path = []
-        cur_table_id = cur_namespace_path + [cur_name]
-        new_namespace_id = new_namespace_path if new_namespace_path else None
-        request = RenameTableRequest(
-            id=cur_table_id,
-            new_table_name=new_name,
-            new_namespace_id=new_namespace_id,
-        )
-        self._namespace_client.rename_table(request)
+        try:
+            LOOP.run(
+                self._inner.rename_table(
+                    cur_name,
+                    new_name,
+                    cur_namespace_path=cur_namespace_path,
+                    new_namespace_path=new_namespace_path,
+                )
+            )
+        except RuntimeError as e:
+            if "rename_table not implemented" in str(e):
+                raise NotImplementedError("rename_table not implemented") from e
+            raise
 
     @override
     def drop_database(self):
@@ -628,8 +654,7 @@ class LanceNamespaceDBConnection(DBConnection):
     def drop_all_tables(self, namespace_path: Optional[List[str]] = None):
         if namespace_path is None:
             namespace_path = []
-        for table_name in self.table_names(namespace_path=namespace_path):
-            self.drop_table(table_name, namespace_path=namespace_path)
+        LOOP.run(self._inner.drop_all_tables(namespace_path=namespace_path))
 
     @override
     def list_namespaces(
@@ -659,13 +684,10 @@ class LanceNamespaceDBConnection(DBConnection):
         """
         if namespace_path is None:
             namespace_path = []
-        request = ListNamespacesRequest(
-            id=namespace_path, page_token=page_token, limit=limit
-        )
-        response = self._namespace_client.list_namespaces(request)
-        return ListNamespacesResponse(
-            namespaces=response.namespaces if response.namespaces else [],
-            page_token=response.page_token,
+        return LOOP.run(
+            self._inner.list_namespaces(
+                namespace_path=namespace_path, page_token=page_token, limit=limit
+            )
         )
 
     @override
@@ -693,14 +715,12 @@ class LanceNamespaceDBConnection(DBConnection):
         CreateNamespaceResponse
             Response containing the properties of the created namespace.
         """
-        request = CreateNamespaceRequest(
-            id=namespace_path,
-            mode=_normalize_create_namespace_mode(mode),
-            properties=properties,
-        )
-        response = self._namespace_client.create_namespace(request)
-        return CreateNamespaceResponse(
-            properties=response.properties if hasattr(response, "properties") else None
+        return LOOP.run(
+            self._inner.create_namespace(
+                namespace_path=namespace_path,
+                mode=mode,
+                properties=properties,
+            )
         )
 
     @override
@@ -728,20 +748,18 @@ class LanceNamespaceDBConnection(DBConnection):
         DropNamespaceResponse
             Response containing properties and transaction_id if applicable.
         """
-        request = DropNamespaceRequest(
-            id=namespace_path,
-            mode=_normalize_drop_namespace_mode(mode),
-            behavior=_normalize_drop_namespace_behavior(behavior),
-        )
-        response = self._namespace_client.drop_namespace(request)
-        return DropNamespaceResponse(
-            properties=(
-                response.properties if hasattr(response, "properties") else None
-            ),
-            transaction_id=(
-                response.transaction_id if hasattr(response, "transaction_id") else None
-            ),
-        )
+        try:
+            return LOOP.run(
+                self._inner.drop_namespace(
+                    namespace_path=namespace_path,
+                    mode=mode,
+                    behavior=behavior,
+                )
+            )
+        except RuntimeError as e:
+            if "Namespace not empty" in str(e):
+                raise NamespaceNotEmptyError(str(e)) from e
+            raise
 
     @override
     def describe_namespace(
@@ -760,11 +778,7 @@ class LanceNamespaceDBConnection(DBConnection):
         DescribeNamespaceResponse
             Response containing the namespace properties.
         """
-        request = DescribeNamespaceRequest(id=namespace_path)
-        response = self._namespace_client.describe_namespace(request)
-        return DescribeNamespaceResponse(
-            properties=response.properties if hasattr(response, "properties") else None
-        )
+        return LOOP.run(self._inner.describe_namespace(namespace_path))
 
     @override
     def list_tables(
@@ -794,13 +808,10 @@ class LanceNamespaceDBConnection(DBConnection):
         """
         if namespace_path is None:
             namespace_path = []
-        request = ListTablesRequest(
-            id=namespace_path, page_token=page_token, limit=limit
-        )
-        response = self._namespace_client.list_tables(request)
-        return ListTablesResponse(
-            tables=response.tables if response.tables else [],
-            page_token=response.page_token,
+        return LOOP.run(
+            self._inner.list_tables(
+                namespace_path=namespace_path, page_token=page_token, limit=limit
+            )
         )
 
     def _lance_table_from_uri(
@@ -856,6 +867,18 @@ class LanceNamespaceDBConnection(DBConnection):
         LanceNamespace
             The namespace client for this connection.
         """
+        if self._namespace_client is None:
+            if (
+                self._namespace_client_impl is None
+                or self._namespace_client_properties is None
+            ):
+                raise ValueError(
+                    "Cannot construct a Python namespace client without "
+                    "namespace implementation properties"
+                )
+            self._namespace_client = namespace_connect(
+                self._namespace_client_impl, self._namespace_client_properties
+            )
         return self._namespace_client
 
 
@@ -869,12 +892,15 @@ class AsyncLanceNamespaceDBConnection:
 
     def __init__(
         self,
-        namespace_client: LanceNamespace,
+        namespace_client: Optional[LanceNamespace] = None,
         *,
         read_consistency_interval: Optional[timedelta] = None,
         storage_options: Optional[Dict[str, str]] = None,
         session: Optional[Session] = None,
         namespace_client_pushdown_operations: Optional[List[str]] = None,
+        namespace_client_impl: Optional[str] = None,
+        namespace_client_properties: Optional[Dict[str, str]] = None,
+        _inner: Optional[AsyncConnection] = None,
     ):
         """
         Initialize an async namespace-based LanceDB connection.
@@ -900,6 +926,12 @@ class AsyncLanceNamespaceDBConnection:
               namespace.create_table() instead of using declare_table + local write.
 
             Default is None (no pushdown, all operations run locally).
+        namespace_client_impl : Optional[str]
+            The namespace implementation name used to create this connection.
+            Required (with ``namespace_client_properties``) for the Rust client to
+            be built natively and install the read-freshness provider.
+        namespace_client_properties : Optional[Dict[str, str]]
+            The namespace properties used to create this connection.
         """
         self._namespace_client = namespace_client
         self.read_consistency_interval = read_consistency_interval
@@ -908,23 +940,37 @@ 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,
-            )
+        self._namespace_client_impl = namespace_client_impl
+        self._namespace_client_properties = namespace_client_properties
+        # See LanceNamespaceDBConnection: when Rust owns the namespace
+        # connection/client, its table performs QueryTable pushdown through the
+        # read-freshness provider, so defer to it rather than the urllib3 client
+        # path (which omits x-lancedb-min-timestamp).
+        self._route_pushdown_to_rust = _inner is not None or _builds_namespace_natively(
+            namespace_client_impl, namespace_client_properties
         )
+        if _inner is not None:
+            self._inner = _inner
+        else:
+            if namespace_client is None:
+                raise ValueError("namespace_client is required without a native _inner")
+            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,
+                )
+            )
 
     async def table_names(
         self,
@@ -948,11 +994,9 @@ class AsyncLanceNamespaceDBConnection:
         )
         if namespace_path is None:
             namespace_path = []
-        request = ListTablesRequest(
-            id=namespace_path, page_token=page_token, limit=limit
+        return await self._inner.table_names(
+            namespace_path=namespace_path, start_after=page_token, limit=limit
         )
-        response = self._namespace_client.list_tables(request)
-        return response.tables if response.tables else []
 
     async def create_table(
         self,
@@ -992,6 +1036,7 @@ class AsyncLanceNamespaceDBConnection:
             namespace_path=namespace_path,
             namespace_client=self._namespace_client,
             pushdown_operations=self._namespace_client_pushdown_operations,
+            route_pushdown_to_rust=self._route_pushdown_to_rust,
         )
 
     async def open_table(
@@ -1014,8 +1059,8 @@ class AsyncLanceNamespaceDBConnection:
                 storage_options=storage_options,
                 index_cache_size=index_cache_size,
             )
-        except RuntimeError as e:
-            if "Table not found" in str(e):
+        except (RuntimeError, ValueError) as e:
+            if "Table not found" in str(e) or "was not found" in str(e):
                 table_id = namespace_path + [name]
                 raise TableNotFoundError(f"Table not found: {'$'.join(table_id)}")
             raise
@@ -1029,15 +1074,14 @@ class AsyncLanceNamespaceDBConnection:
             namespace_path=namespace_path,
             namespace_client=self._namespace_client,
             pushdown_operations=self._namespace_client_pushdown_operations,
+            route_pushdown_to_rust=self._route_pushdown_to_rust,
         )
 
     async def drop_table(self, name: str, namespace_path: Optional[List[str]] = None):
         """Drop a table from the namespace."""
         if namespace_path is None:
             namespace_path = []
-        table_id = namespace_path + [name]
-        request = DropTableRequest(id=table_id)
-        self._namespace_client.drop_table(request)
+        await self._inner.drop_table(name, namespace_path=namespace_path)
 
     async def rename_table(
         self,
@@ -1051,14 +1095,17 @@ class AsyncLanceNamespaceDBConnection:
             cur_namespace_path = []
         if new_namespace_path is None:
             new_namespace_path = []
-        cur_table_id = cur_namespace_path + [cur_name]
-        new_namespace_id = new_namespace_path if new_namespace_path else None
-        request = RenameTableRequest(
-            id=cur_table_id,
-            new_table_name=new_name,
-            new_namespace_id=new_namespace_id,
-        )
-        self._namespace_client.rename_table(request)
+        try:
+            await self._inner.rename_table(
+                cur_name,
+                new_name,
+                cur_namespace_path=cur_namespace_path,
+                new_namespace_path=new_namespace_path,
+            )
+        except RuntimeError as e:
+            if "rename_table not implemented" in str(e):
+                raise NotImplementedError("rename_table not implemented") from e
+            raise
 
     async def drop_database(self):
         """Deprecated method."""
@@ -1070,9 +1117,7 @@ class AsyncLanceNamespaceDBConnection:
         """Drop all tables in the namespace."""
         if namespace_path is None:
             namespace_path = []
-        table_names = await self.table_names(namespace_path=namespace_path)
-        for table_name in table_names:
-            await self.drop_table(table_name, namespace_path=namespace_path)
+        await self._inner.drop_all_tables(namespace_path=namespace_path)
 
     async def list_namespaces(
         self,
@@ -1101,13 +1146,8 @@ class AsyncLanceNamespaceDBConnection:
         """
         if namespace_path is None:
             namespace_path = []
-        request = ListNamespacesRequest(
-            id=namespace_path, page_token=page_token, limit=limit
-        )
-        response = self._namespace_client.list_namespaces(request)
-        return ListNamespacesResponse(
-            namespaces=response.namespaces if response.namespaces else [],
-            page_token=response.page_token,
+        return await self._inner.list_namespaces(
+            namespace_path=namespace_path, page_token=page_token, limit=limit
         )
 
     async def create_namespace(
@@ -1134,15 +1174,11 @@ class AsyncLanceNamespaceDBConnection:
         CreateNamespaceResponse
             Response containing the properties of the created namespace.
         """
-        request = CreateNamespaceRequest(
-            id=namespace_path,
-            mode=_normalize_create_namespace_mode(mode),
+        return await self._inner.create_namespace(
+            namespace_path=namespace_path,
+            mode=mode,
             properties=properties,
         )
-        response = self._namespace_client.create_namespace(request)
-        return CreateNamespaceResponse(
-            properties=response.properties if hasattr(response, "properties") else None
-        )
 
     async def drop_namespace(
         self,
@@ -1168,20 +1204,16 @@ class AsyncLanceNamespaceDBConnection:
         DropNamespaceResponse
             Response containing properties and transaction_id if applicable.
         """
-        request = DropNamespaceRequest(
-            id=namespace_path,
-            mode=_normalize_drop_namespace_mode(mode),
-            behavior=_normalize_drop_namespace_behavior(behavior),
-        )
-        response = self._namespace_client.drop_namespace(request)
-        return DropNamespaceResponse(
-            properties=(
-                response.properties if hasattr(response, "properties") else None
-            ),
-            transaction_id=(
-                response.transaction_id if hasattr(response, "transaction_id") else None
-            ),
-        )
+        try:
+            return await self._inner.drop_namespace(
+                namespace_path=namespace_path,
+                mode=mode,
+                behavior=behavior,
+            )
+        except RuntimeError as e:
+            if "Namespace not empty" in str(e):
+                raise NamespaceNotEmptyError(str(e)) from e
+            raise
 
     async def describe_namespace(
         self, namespace_path: List[str]
@@ -1199,11 +1231,7 @@ class AsyncLanceNamespaceDBConnection:
         DescribeNamespaceResponse
             Response containing the namespace properties.
         """
-        request = DescribeNamespaceRequest(id=namespace_path)
-        response = self._namespace_client.describe_namespace(request)
-        return DescribeNamespaceResponse(
-            properties=response.properties if hasattr(response, "properties") else None
-        )
+        return await self._inner.describe_namespace(namespace_path)
 
     async def list_tables(
         self,
@@ -1232,13 +1260,8 @@ class AsyncLanceNamespaceDBConnection:
         """
         if namespace_path is None:
             namespace_path = []
-        request = ListTablesRequest(
-            id=namespace_path, page_token=page_token, limit=limit
-        )
-        response = self._namespace_client.list_tables(request)
-        return ListTablesResponse(
-            tables=response.tables if response.tables else [],
-            page_token=response.page_token,
+        return await self._inner.list_tables(
+            namespace_path=namespace_path, page_token=page_token, limit=limit
         )
 
     async def namespace_client(self) -> LanceNamespace:
@@ -1252,6 +1275,18 @@ class AsyncLanceNamespaceDBConnection:
         LanceNamespace
             The namespace client for this connection.
         """
+        if self._namespace_client is None:
+            if (
+                self._namespace_client_impl is None
+                or self._namespace_client_properties is None
+            ):
+                raise ValueError(
+                    "Cannot construct a Python namespace client without "
+                    "namespace implementation properties"
+                )
+            self._namespace_client = namespace_connect(
+                self._namespace_client_impl, self._namespace_client_properties
+            )
         return self._namespace_client
 
 
@@ -1302,6 +1337,32 @@ def connect_namespace(
     LanceNamespaceDBConnection
         A namespace-based connection to LanceDB
     """
+    if _supports_native_namespace(namespace_client_impl):
+        inner = AsyncConnection(
+            _connect_namespace(
+                namespace_client_impl,
+                namespace_client_properties,
+                read_consistency_interval=(
+                    read_consistency_interval.total_seconds()
+                    if read_consistency_interval is not None
+                    else None
+                ),
+                storage_options=storage_options,
+                session=session,
+                namespace_client_pushdown_operations=namespace_client_pushdown_operations,
+            )
+        )
+        return LanceNamespaceDBConnection(
+            namespace_client=None,
+            read_consistency_interval=read_consistency_interval,
+            storage_options=storage_options,
+            session=session,
+            namespace_client_pushdown_operations=namespace_client_pushdown_operations,
+            namespace_client_impl=namespace_client_impl,
+            namespace_client_properties=namespace_client_properties,
+            _inner=inner,
+        )
+
     namespace_client = namespace_connect(
         namespace_client_impl, namespace_client_properties
     )
@@ -1377,6 +1438,32 @@ def connect_namespace_async(
     ...     tables = await db.table_names()
     ...     table = await db.create_table("my_table", schema=schema)
     """
+    if _supports_native_namespace(namespace_client_impl):
+        inner = AsyncConnection(
+            _connect_namespace(
+                namespace_client_impl,
+                namespace_client_properties,
+                read_consistency_interval=(
+                    read_consistency_interval.total_seconds()
+                    if read_consistency_interval is not None
+                    else None
+                ),
+                storage_options=storage_options,
+                session=session,
+                namespace_client_pushdown_operations=namespace_client_pushdown_operations,
+            )
+        )
+        return AsyncLanceNamespaceDBConnection(
+            namespace_client=None,
+            read_consistency_interval=read_consistency_interval,
+            storage_options=storage_options,
+            session=session,
+            namespace_client_pushdown_operations=namespace_client_pushdown_operations,
+            namespace_client_impl=namespace_client_impl,
+            namespace_client_properties=namespace_client_properties,
+            _inner=inner,
+        )
+
     namespace_client = namespace_connect(
         namespace_client_impl, namespace_client_properties
     )
@@ -1387,4 +1474,6 @@ def connect_namespace_async(
         storage_options=storage_options,
         session=session,
         namespace_client_pushdown_operations=namespace_client_pushdown_operations,
+        namespace_client_impl=namespace_client_impl,
+        namespace_client_properties=namespace_client_properties,
     )

python/python/lancedb/permutation.py

@@ -48,6 +48,14 @@ class PermutationBuilder:
         By default, the permutation builder will create a single split that contains all
         rows in the same order as the base table.
         """
+        if not hasattr(table, "_inner"):
+            raise TypeError(
+                f"PermutationBuilder requires a local LanceTable, "
+                f"got {type(table).__name__}. "
+                "The permutation API is not supported on remote tables. "
+                "Remote tables connect to LanceDB Cloud or Enterprise and do not have "
+                "direct access to the underlying Lance dataset needed for permutations."
+            )
         self._async = async_permutation_builder(table)
 
     def split_random(
@@ -57,6 +65,7 @@ class PermutationBuilder:
         counts: Optional[list[int]] = None,
         fixed: Optional[int] = None,
         seed: Optional[int] = None,
+        clump_size: Optional[int] = None,
         split_names: Optional[list[str]] = None,
     ) -> "PermutationBuilder":
         """
@@ -79,6 +88,9 @@ class PermutationBuilder:
         Rows will be randomly assigned to splits.  The optional seed can be provided to
         make the assignment deterministic.
 
+        If clump_size is provided, rows are shuffled as contiguous groups of that size,
+        preserving I/O locality while still randomising the split assignment.
+
         The optional split_names can be provided to name the splits.  If not provided,
         the splits can only be referenced by their index.
         """
@@ -87,6 +99,7 @@ class PermutationBuilder:
             counts=counts,
             fixed=fixed,
             seed=seed,
+            clump_size=clump_size,
             split_names=split_names,
         )
         return self

python/python/lancedb/query.py

@@ -119,6 +119,27 @@ def _filter_to_sql(filter: Optional[Union[str, Expr]]) -> Optional[str]:
     return filter
 
 
+def _combine_where(
+    existing: Optional[Union[str, Expr]], new: Union[str, Expr]
+) -> Union[str, Expr]:
+    """Combine a new filter with an existing one using a logical AND.
+
+    Calling ``where`` more than once composes the filters with AND instead of
+    replacing the previous filter. Two :class:`~lancedb.expr.Expr` filters are
+    combined as an expression; otherwise both filters are lowered to SQL strings
+    and combined as SQL.
+    """
+    if existing is None:
+        return new
+    existing_is_expr = isinstance(existing, Expr)
+    new_is_expr = isinstance(new, Expr)
+    if existing_is_expr and new_is_expr:
+        return existing & new
+    existing_sql = existing.to_sql() if existing_is_expr else existing
+    new_sql = new.to_sql() if new_is_expr else new
+    return f"({existing_sql}) AND ({new_sql})"
+
+
 def _projection_to_scanner_kwargs(
     columns: Optional[
         Union[
@@ -1148,8 +1169,13 @@ class LanceQueryBuilder(ABC):
         -------
         LanceQueryBuilder
             The LanceQueryBuilder object.
+
+        Notes
+        -----
+        Calling this multiple times combines the filters with a logical AND
+        rather than replacing the previous filter.
         """
-        self._where = where
+        self._where = _combine_where(self._where, where)
         self._postfilter = not prefilter
         return self
 
@@ -1693,8 +1719,13 @@ class LanceVectorQueryBuilder(LanceQueryBuilder):
         -------
         LanceQueryBuilder
             The LanceQueryBuilder object.
+
+        Notes
+        -----
+        Calling this multiple times combines the filters with a logical AND
+        rather than replacing the previous filter.
         """
-        self._where = where
+        self._where = _combine_where(self._where, where)
         if prefilter is not None:
             self._postfilter = not prefilter
         return self
@@ -2894,6 +2925,9 @@ class AsyncStandardQuery(AsyncQueryBase):
 
         Filtering performance can often be improved by creating a scalar index
         on the filter column(s).
+
+        Calling this multiple times combines the filters with a logical AND
+        rather than replacing the previous filter.
         """
         if isinstance(predicate, Expr):
             self._inner.where_expr(predicate._inner)

python/python/lancedb/remote/__init__.py

@@ -9,6 +9,7 @@ from typing import List, Optional
 from lancedb import __version__
 
 from .header import HeaderProvider
+from .oauth import OAuthConfig, OAuthFlowType
 
 __all__ = [
     "TimeoutConfig",
@@ -16,6 +17,8 @@ __all__ = [
     "TlsConfig",
     "ClientConfig",
     "HeaderProvider",
+    "OAuthConfig",
+    "OAuthFlowType",
 ]
 
 

python/python/lancedb/remote/oauth.py

@@ -0,0 +1,75 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+from dataclasses import dataclass, field
+from enum import Enum
+from typing import List, Optional
+
+
+class OAuthFlowType(str, Enum):
+    """OAuth authentication flow types."""
+
+    CLIENT_CREDENTIALS = "client_credentials"
+    """Client Credentials grant (service-to-service / M2M)."""
+
+    AZURE_MANAGED_IDENTITY = "azure_managed_identity"
+    """Azure Managed Identity via IMDS."""
+
+
+@dataclass
+class OAuthConfig:
+    """OAuth configuration for LanceDB authentication.
+
+    All token acquisition and refresh is handled in the Rust layer.
+    This config is passed through to Rust via PyO3.
+
+    Parameters
+    ----------
+    issuer_url : str
+        OIDC issuer URL or OAuth authority URL.
+        For Azure: ``https://login.microsoftonline.com/{tenant_id}/v2.0``
+    client_id : str
+        Application / Client ID.
+    scopes : List[str]
+        OAuth scopes to request.
+        For Azure managed identity, exactly one scope or resource is required.
+        For example: ``["api://{app_id}/.default"]``
+    flow : OAuthFlowType
+        Authentication flow to use. Default: CLIENT_CREDENTIALS.
+    client_secret : Optional[str]
+        Client secret (required for CLIENT_CREDENTIALS).
+    managed_identity_client_id : Optional[str]
+        Client ID for user-assigned managed identity (AZURE_MANAGED_IDENTITY).
+    refresh_buffer_secs : Optional[int]
+        Seconds before expiry to trigger proactive refresh (default: 300).
+        Keep this well below the token TTL; if it is greater than or equal to
+        the TTL, each request refreshes the token.
+
+    Examples
+    --------
+    Client Credentials (service-to-service):
+
+    >>> config = OAuthConfig(
+    ...     issuer_url="https://login.microsoftonline.com/{tenant}/v2.0",
+    ...     client_id="app-id",
+    ...     client_secret="secret",
+    ...     scopes=["api://lancedb-api/.default"],
+    ... )
+
+    Azure Managed Identity:
+
+    >>> config = OAuthConfig(
+    ...     issuer_url="https://login.microsoftonline.com/{tenant}/v2.0",
+    ...     client_id="app-id",
+    ...     scopes=["api://lancedb-api/.default"],
+    ...     flow=OAuthFlowType.AZURE_MANAGED_IDENTITY,
+    ... )
+    """
+
+    issuer_url: str
+    client_id: str
+    scopes: List[str]
+    flow: OAuthFlowType = OAuthFlowType.CLIENT_CREDENTIALS
+    client_secret: Optional[str] = field(default=None, repr=False)
+    managed_identity_client_id: Optional[str] = None
+    refresh_buffer_secs: Optional[int] = None

python/python/lancedb/rerankers/mrr.py

@@ -156,9 +156,16 @@ class MRRReranker(Reranker):
                 reciprocal_rank = 1.0 / rank
                 mrr_score_map[result_id].append(reciprocal_rank)
 
+        # MRR averages the reciprocal rank across *all* ranking systems, treating
+        # a system in which a document does not appear as a reciprocal rank of 0.
+        # We therefore divide by the total number of systems, not by the number of
+        # systems the document happens to appear in -- otherwise a document found
+        # by a single ranking would outrank one ranked highly by every system,
+        # defeating the purpose of fusing the rankings.
+        num_systems = len(vector_results)
         final_mrr_scores = {}
         for result_id, reciprocal_ranks in mrr_score_map.items():
-            mean_rr = np.mean(reciprocal_ranks)
+            mean_rr = float(np.sum(reciprocal_ranks)) / num_systems
             final_mrr_scores[result_id] = mean_rr
 
         combined = pa.concat_tables(vector_results, **self._concat_tables_args)

python/python/lancedb/streaming.py

@@ -0,0 +1,607 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+"""Elastic streaming dataloader for PyTorch.
+
+Provides StreamingDataset, a PyTorch IterableDataset that guarantees:
+
+- **Elastic determinism**: for a fixed (num_splits, shuffle_seed, epoch) the set
+  of samples that forms each global training step is identical regardless of
+  world_size or num_workers.
+- **Resumability**: state_dict / load_state_dict capture per-split consumption
+  counts so training can resume from an exact mid-epoch position even when the
+  distributed topology changes between runs.
+"""
+
+import ctypes
+import logging
+import os
+import random
+import threading
+import time
+from collections import deque
+from concurrent.futures import ThreadPoolExecutor
+from multiprocessing import RawArray
+from typing import Any, Callable, Iterator, Optional
+
+from torch.utils.data import IterableDataset, get_worker_info
+
+from .permutation import (
+    Permutation,
+    Transforms,
+    permutation_builder,
+    _table_from_pickle_state,
+    _table_to_pickle_state,
+)
+
+logger = logging.getLogger(__name__)
+
+# Multiplier used to combine shuffle_seed and epoch into a single permutation
+# seed.  Chosen to be a large prime so different (seed, epoch) pairs produce
+# distinct seeds for any practically encountered epoch count.
+_EPOCH_PRIME = 100003
+
+DEFAULT_READ_BATCH_SIZE = 64
+DEFAULT_PREFETCH_BATCHES = 4
+
+
+class StreamingDataset(IterableDataset):
+    """An elastic, resumable PyTorch IterableDataset backed by a LanceDB table.
+
+    The table is partitioned into ``num_splits`` fixed splits using a
+    deterministic random shuffle controlled by ``shuffle_seed`` and ``epoch``.
+    Each rank is assigned a contiguous block of splits, and within a rank each
+    DataLoader worker is assigned a contiguous sub-block.  Samples are yielded
+    by round-robining over the assigned splits, one sample per split per cycle.
+
+    Internally ``__iter__`` runs a two-stage pipeline:
+
+    - **Stage 1 (I/O)**: one thread pool with ``num_splits * prefetch_batches``
+      workers fetches raw ``RecordBatch`` objects from LanceDB in parallel
+      across all splits and places them in a per-split raw-batch queue.
+    - **Stage 2 (transform)**: a second thread pool with ``os.cpu_count()``
+      workers picks up raw batches, applies the transform, and places the
+      results in a per-split cooked-row queue.
+
+    The main thread round-robins over the cooked queues, yielding one row per
+    split per cycle.
+
+    Parameters
+    ----------
+    table:
+        LanceDB table to stream from.
+    num_splits:
+        Number of fixed splits to partition the table into.  Must be divisible
+        by ``world_size``.  When used with DataLoader workers it must also be
+        divisible by ``world_size * num_workers``.  Defaults to ``world_size``.
+        If the row count (after any ``filter``) is not evenly divisible by
+        ``num_splits``, the surplus rows — at most ``num_splits - 1`` per epoch
+        — are silently dropped to keep all splits the same length.
+    shuffle:
+        Whether to randomly assign rows to splits.  When ``True`` (the
+        default) rows are shuffled using ``shuffle_seed`` and ``epoch``.
+        When ``False`` rows are divided into splits sequentially in storage
+        order, which can be useful for deterministic debugging or evaluation.
+    shuffle_seed:
+        Base seed for the random permutation.  Combined with ``epoch`` so
+        each epoch produces a different ordering.  Pass ``None`` to generate
+        a random seed at construction time.
+    epoch:
+        Current training epoch.  Combined with ``shuffle_seed`` so that each
+        epoch produces a different sample ordering.
+    rank:
+        This process's rank in the distributed training group.
+    world_size:
+        Total number of processes in the distributed training group.
+    read_batch_size:
+        Number of rows fetched from each split in a single ``take_offsets``
+        call.  Larger values amortise per-request overhead (critical on object
+        storage) at the cost of higher memory usage per split buffer.  Defaults
+        to ``DEFAULT_READ_BATCH_SIZE`` (64).
+    prefetch_batches:
+        Number of I/O batches to keep in flight per split.  Higher values
+        overlap storage latency with transform and training compute at the cost
+        of more memory and threads.  Defaults to ``DEFAULT_PREFETCH_BATCHES``
+        (4).
+    columns:
+        Optional list of column names to read.  When set, only those columns
+        are fetched from storage; all others are omitted.  ``None`` (the
+        default) reads every column.
+    shuffle_clump_size:
+        When set, rows are shuffled in contiguous groups of this size rather
+        than individually.  Larger clumps improve I/O locality (important on
+        object storage) at the cost of reduced randomness.  ``None`` (the
+        default) shuffles rows individually.
+    filter:
+        Optional SQL filter expression (e.g. ``"label = 'dog'"``).  Only rows
+        that satisfy the predicate are included in the permutation.  The filter
+        is applied during permutation construction so split sizes reflect the
+        filtered row count.
+    transform:
+        Optional callable applied to each ``pyarrow.RecordBatch`` before rows
+        are yielded.  Receives one batch at a time and must return an iterable
+        whose length equals the number of rows in the batch.  When ``None``
+        (the default) rows are returned as plain Python dicts.
+    worker_info_override:
+        If set, used in place of ``torch.utils.data.get_worker_info()`` to
+        determine the DataLoader worker assignment.  Intended for unit tests
+        that need to simulate multiple workers without spawning real processes.
+        If both this and the real worker info are non-None a warning is logged
+        and the override takes precedence.
+    """
+
+    def __init__(
+        self,
+        table,
+        *,
+        num_splits: Optional[int] = None,
+        shuffle: bool = True,
+        shuffle_seed: Optional[int] = 0,
+        epoch: int = 0,
+        rank: int = 0,
+        world_size: int = 1,
+        read_batch_size: int = DEFAULT_READ_BATCH_SIZE,
+        prefetch_batches: int = DEFAULT_PREFETCH_BATCHES,
+        columns: Optional[list[str]] = None,
+        shuffle_clump_size: Optional[int] = None,
+        filter: Optional[str] = None,
+        transform: Optional[Callable] = None,
+        connection_factory: Optional[Callable[[str], Any]] = None,
+        worker_info_override=None,
+    ):
+        super().__init__()
+        if num_splits is None:
+            num_splits = world_size
+        if shuffle_seed is None:
+            shuffle_seed = random.randrange(2**32)
+        if num_splits % world_size != 0:
+            raise ValueError(
+                f"num_splits ({num_splits}) must be divisible by "
+                f"world_size ({world_size})"
+            )
+
+        self._table = table
+        self._num_splits = num_splits
+        self._shuffle = shuffle
+        self._shuffle_seed = shuffle_seed
+        self._epoch = epoch
+        self._rank = rank
+        self._world_size = world_size
+        self._read_batch_size = read_batch_size
+        self._prefetch_batches = prefetch_batches
+        self._columns = columns
+        self._shuffle_clump_size = shuffle_clump_size
+        self._filter = filter
+        self._transform = transform
+        self._connection_factory = connection_factory
+        self._worker_info_override = worker_info_override
+
+        # Live references to pipeline state, set only while __iter__ is running
+        # in the same process.  Used by the observability properties when the
+        # DataLoader runs with num_workers=0.
+        self._raw_batches_ref: Optional[list[deque]] = None
+        self._cooked_ref: Optional[list[deque]] = None
+        self._fetch_head_ref: Optional[list[int]] = None
+        self._split_sizes_ref: Optional[list[int]] = None
+        self._local_consumed_ref: Optional[list[int]] = None
+
+        # Shared-memory counters written by __iter__ (which may run in a
+        # DataLoader worker process) and read by the observability properties
+        # in the main process.  RawArray is picklable via the forkserver
+        # reduction protocol so it survives the dataset pickle round-trip.
+        # Layout: [unscanned_rows, raw_rows, cooked_rows, consumed_rows,
+        #          bytes_loaded, fetch_time_us, transform_time_us]
+        self._worker_stats: RawArray = RawArray(ctypes.c_int64, 7)
+
+        # Cumulative bytes of Arrow buffer data fetched across all iterations.
+        self._bytes_loaded: int = 0
+        # Cumulative seconds spent in LanceDB I/O and in transform functions.
+        self._fetch_time: float = 0.0
+        self._transform_time: float = 0.0
+
+        # Number of samples each split has already been consumed.  At global
+        # step boundaries all splits have consumed this many samples, so a
+        # single scalar captures the topology-independent checkpoint state.
+        self._resume_offset: int = 0
+
+        # Build the permutation table once, deterministically.
+        builder = permutation_builder(table)
+        if filter is not None:
+            builder = builder.filter(filter)
+        if shuffle:
+            perm_seed = shuffle_seed + epoch * _EPOCH_PRIME
+            self._perm_table = builder.split_random(
+                fixed=num_splits, seed=perm_seed, clump_size=shuffle_clump_size
+            ).execute()
+        else:
+            self._perm_table = builder.split_sequential(fixed=num_splits).execute()
+
+        # Contiguous block of global split indices assigned to this rank.
+        splits_per_rank = num_splits // world_size
+        rank_start = rank * splits_per_rank
+        self._rank_splits: list[int] = list(
+            range(rank_start, rank_start + splits_per_rank)
+        )
+
+    def _resolve_my_splits(self) -> list[int]:
+        """Return the split indices this instance should read in __iter__."""
+        torch_worker_info = get_worker_info()
+        if self._worker_info_override is not None:
+            if torch_worker_info is not None:
+                logger.warning(
+                    "worker_info_override is set but get_worker_info() also returned a "
+                    "non-None value; ignoring the real torch worker info and using the "
+                    "override instead.  This may lead to duplicated or incorrect data "
+                    "from the dataset."
+                )
+            worker_info = self._worker_info_override
+        else:
+            worker_info = torch_worker_info
+
+        if worker_info is None:
+            return self._rank_splits
+
+        num_workers: int = worker_info.num_workers
+        worker_id: int = worker_info.id
+        n_rank_splits = len(self._rank_splits)
+        if n_rank_splits % num_workers != 0:
+            raise ValueError(
+                f"Number of rank splits ({n_rank_splits}) must be divisible by "
+                f"num_workers ({num_workers})"
+            )
+        splits_per_worker = n_rank_splits // num_workers
+        start = worker_id * splits_per_worker
+        return self._rank_splits[start : start + splits_per_worker]
+
+    def __iter__(self) -> Iterator[dict[str, Any]]:
+        if self._raw_batches_ref is not None:
+            raise RuntimeError(
+                "StreamingDataset does not support concurrent iteration. "
+                "Only one active iterator per dataset instance is allowed."
+            )
+        my_splits = self._resolve_my_splits()
+        if not my_splits:
+            return
+
+        # Set identity transform on each Permutation so __getitems__ returns
+        # the raw RecordBatch.  Stage 2 applies the real transform.
+        permutations: list[Permutation] = []
+        for split_idx in my_splits:
+            perm = Permutation.from_tables(
+                self._table, self._perm_table, split=split_idx
+            )
+            if self._columns is not None:
+                perm = perm.select_columns(self._columns)
+            perm = perm.with_transform(lambda batch: batch)
+            if self._resume_offset > 0:
+                perm = perm.with_skip(self._resume_offset)
+            permutations.append(perm)
+
+        n = len(permutations)
+        split_sizes = [perm.num_rows for perm in permutations]
+        initial_offset = self._resume_offset
+        local_consumed = [0] * n
+
+        batch_size = self._read_batch_size
+        max_prefetch = self._prefetch_batches
+        cpu_workers = os.cpu_count() or 1
+        final_transform = (
+            self._transform if self._transform is not None else Transforms.arrow2python
+        )
+
+        # Per-split pipeline state.
+        fetch_head = [0] * n
+        io_pending = [deque() for _ in range(n)]  # Future[RecordBatch]
+        raw_batches = [deque() for _ in range(n)]  # RecordBatch — fetched, awaiting tx
+        tx_pending = [deque() for _ in range(n)]  # Future[list[Any]]
+        cooked = [deque() for _ in range(n)]  # rows ready to yield
+
+        # Limit simultaneous transforms to cpu_workers across all splits.
+        tx_semaphore = threading.Semaphore(cpu_workers)
+
+        # ── Stage 1 helpers ───────────────────────────────────────────────────
+
+        def _io_call(perm, indices):
+            t0 = time.perf_counter()
+            batch = perm.__getitems__(indices)
+            self._bytes_loaded += batch.nbytes
+            self._fetch_time += time.perf_counter() - t0
+            return batch
+
+        def _submit_io(i: int) -> None:
+            remaining = split_sizes[i] - fetch_head[i]
+            if remaining <= 0:
+                return
+            fetch = min(batch_size, remaining)
+            start = fetch_head[i]
+            fetch_head[i] += fetch
+            perm_i = permutations[i]
+            indices = list(range(start, start + fetch))
+            io_pending[i].append(io_pool.submit(_io_call, perm_i, indices))
+
+        def _fill_io(i: int) -> None:
+            while len(io_pending[i]) < max_prefetch and fetch_head[i] < split_sizes[i]:
+                _submit_io(i)
+
+        def _drain_io(i: int) -> None:
+            """Move completed I/O futures into raw_batches non-blockingly."""
+            while io_pending[i] and io_pending[i][0].done():
+                raw_batches[i].append(io_pending[i].popleft().result())
+
+        # ── Stage 2 helpers ───────────────────────────────────────────────────
+
+        def _tx_call_guarded(batch):
+            try:
+                t0 = time.perf_counter()
+                result = final_transform(batch)
+                self._transform_time += time.perf_counter() - t0
+                return result
+            finally:
+                tx_semaphore.release()
+
+        def _try_submit_tx(i: int) -> None:
+            """Submit transforms for raw_batches[i] up to available capacity."""
+            while raw_batches[i] and tx_semaphore.acquire(blocking=False):
+                batch = raw_batches[i].popleft()
+                tx_pending[i].append(tx_pool.submit(_tx_call_guarded, batch))
+
+        def _drain_tx(i: int) -> None:
+            """Move completed transform futures into cooked non-blockingly."""
+            while tx_pending[i] and tx_pending[i][0].done():
+                cooked[i].extend(tx_pending[i].popleft().result())
+
+        # ── Combined advance ──────────────────────────────────────────────────
+
+        def _advance(i: int) -> None:
+            """Non-blocking pipeline pump for split i."""
+            _drain_io(i)
+            _drain_tx(i)
+            _try_submit_tx(i)
+            _fill_io(i)
+
+        def _ensure_cooked(i: int) -> None:
+            """Ensure cooked[i] has at least one row, blocking if necessary."""
+            _advance(i)
+            while not cooked[i]:
+                if tx_pending[i]:
+                    # Wait for the oldest in-flight transform.
+                    cooked[i].extend(tx_pending[i].popleft().result())
+                    _advance(i)
+                elif raw_batches[i]:
+                    # Acquire a transform slot (may block briefly if all
+                    # cpu_workers are busy with other splits).
+                    tx_semaphore.acquire()
+                    batch = raw_batches[i].popleft()
+                    tx_pending[i].append(tx_pool.submit(_tx_call_guarded, batch))
+                elif io_pending[i]:
+                    # Block on the oldest in-flight I/O fetch.
+                    raw_batches[i].append(io_pending[i].popleft().result())
+                    _advance(i)
+                else:
+                    break  # split exhausted
+
+        # ── Main loop ─────────────────────────────────────────────────────────
+
+        with ThreadPoolExecutor(max_workers=n * max_prefetch) as io_pool:
+            with ThreadPoolExecutor(max_workers=cpu_workers) as tx_pool:
+                self._raw_batches_ref = raw_batches
+                self._cooked_ref = cooked
+                self._fetch_head_ref = fetch_head
+                self._split_sizes_ref = split_sizes
+                self._local_consumed_ref = local_consumed
+                try:
+                    for i in range(n):
+                        _fill_io(i)
+
+                    while True:
+                        # Stop when any split is exhausted (all exhaust
+                        # simultaneously: equal split sizes + round-robin).
+                        if any(local_consumed[i] >= split_sizes[i] for i in range(n)):
+                            break
+
+                        for i in range(n):
+                            _ensure_cooked(i)
+                            row = cooked[i].popleft()
+                            local_consumed[i] += 1
+                            _advance(i)
+
+                            # After the last split in each cycle: update the
+                            # global offset and refresh the shared-memory stats
+                            # so the main process can observe pipeline depth
+                            # even when __iter__ runs in a worker process.
+                            if i == n - 1:
+                                self._resume_offset = initial_offset + local_consumed[i]
+                                ws = self._worker_stats
+                                ws[0] = sum(
+                                    split_sizes[j] - fetch_head[j] for j in range(n)
+                                )
+                                ws[1] = sum(
+                                    batch.num_rows for q in raw_batches for batch in q
+                                )
+                                ws[2] = sum(len(q) for q in cooked)
+                                ws[3] = sum(local_consumed)
+                                ws[4] = self._bytes_loaded
+                                ws[5] = int(self._fetch_time * 1_000_000)
+                                ws[6] = int(self._transform_time * 1_000_000)
+
+                            yield row
+                finally:
+                    self._raw_batches_ref = None
+                    self._cooked_ref = None
+                    self._fetch_head_ref = None
+                    self._split_sizes_ref = None
+                    self._local_consumed_ref = None
+
+    @property
+    def bytes_loaded(self) -> int:
+        """Cumulative bytes of raw Arrow buffer data fetched from storage.
+
+        Measured on the ``RecordBatch`` before any transform is applied, so
+        the value reflects actual I/O rather than the size of transformed
+        output.  Accumulates across multiple iterations of the same dataset
+        instance and is never reset automatically.
+        """
+        if self._raw_batches_ref is not None:
+            return self._bytes_loaded
+        return int(self._worker_stats[4])
+
+    @property
+    def fetch_time(self) -> float:
+        """Cumulative seconds spent waiting for data from LanceDB.
+
+        Measured per batch in the Stage 1 I/O threads as the total elapsed
+        time of the ``take_offsets`` call.  Accumulates across all splits and
+        all iterations.
+        """
+        if self._raw_batches_ref is not None:
+            return self._fetch_time
+        return self._worker_stats[5] / 1_000_000
+
+    @property
+    def transform_time(self) -> float:
+        """Cumulative seconds spent applying the transform.
+
+        Measured per batch in the Stage 2 transform threads as the elapsed
+        time inside the transform callable (or the default ``arrow2python``
+        conversion when no transform is set).  Accumulates across all splits
+        and all iterations.
+        """
+        if self._raw_batches_ref is not None:
+            return self._transform_time
+        return self._worker_stats[6] / 1_000_000
+
+    @property
+    def raw_queue_depth(self) -> int:
+        """Number of raw rows waiting for a transform thread across all splits.
+
+        A persistently non-zero value means Stage 2 (transform) is the
+        bottleneck: I/O is completing faster than transforms can consume
+        batches.  Returns 0 when not iterating.
+        """
+        if self._raw_batches_ref is not None:
+            return sum(batch.num_rows for q in self._raw_batches_ref for batch in q)
+        return int(self._worker_stats[1])
+
+    @property
+    def prefetch_queue_depth(self) -> int:
+        """Number of rows transformed and ready to yield across all splits.
+
+        Counts rows whose transform has completed and are sitting in memory
+        waiting for the main thread — rows that can be handed off with no
+        I/O or CPU wait.  Returns 0 when not iterating.
+        """
+        if self._cooked_ref is not None:
+            return sum(len(q) for q in self._cooked_ref)
+        return int(self._worker_stats[2])
+
+    @property
+    def unscanned_rows(self) -> int:
+        """Number of rows not yet submitted to the I/O stage across all splits.
+
+        Decreases as the I/O stage submits fetch requests.  When this reaches
+        zero all data has been requested from storage (though it may not have
+        arrived yet).  Returns 0 when not iterating.
+        """
+        if self._fetch_head_ref is not None:
+            return sum(
+                size - head
+                for size, head in zip(self._split_sizes_ref, self._fetch_head_ref)
+            )
+        return int(self._worker_stats[0])
+
+    @property
+    def consumed_rows(self) -> int:
+        """Number of rows already yielded to the caller across all splits.
+
+        Monotonically increases throughout iteration.  Returns 0 when not
+        iterating.
+        """
+        if self._local_consumed_ref is not None:
+            return sum(self._local_consumed_ref)
+        return int(self._worker_stats[3])
+
+    def __getstate__(self):
+        """Support pickling for multi-worker DataLoader (forkserver / spawn).
+
+        The live LanceDB table object contains non-picklable connection state
+        (sockets, Rust-backed PyO3 objects).  If a ``connection_factory`` was
+        supplied only the table name is serialised; the factory is called in
+        the worker to reopen the connection without embedding any credentials.
+        Without a factory the table's own picklable reopen state is captured
+        via ``_table_to_pickle_state`` (mirrors the ``Permutation`` approach).
+        """
+        state = self.__dict__.copy()
+        # _table: replace with reconnect info (credentials must not be embedded).
+        state["_table_name"] = self._table.name
+        if self._connection_factory is not None:
+            state["_table"] = None
+        else:
+            state["_table"] = _table_to_pickle_state(self._table)
+        # _perm_table: always in-memory; serialise as Arrow data (mirrors
+        # how Permutation.__getstate__ handles its permutation_table).
+        state["_perm_table"] = (
+            self._perm_table.name,
+            self._perm_table.to_arrow(),
+        )
+        for key in (
+            "_raw_batches_ref",
+            "_cooked_ref",
+            "_fetch_head_ref",
+            "_split_sizes_ref",
+            "_local_consumed_ref",
+        ):
+            state[key] = None
+        return state
+
+    def __setstate__(self, state):
+        """Reconnect to LanceDB after unpickling in a worker process."""
+        from . import connect as _connect
+
+        table_name = state.pop("_table_name")
+        table_state = state.pop("_table")
+        perm_name, perm_data = state.pop("_perm_table")
+        self.__dict__.update(state)
+        if self._connection_factory is not None:
+            self._table = self._connection_factory(table_name)
+        else:
+            self._table = _table_from_pickle_state(table_state)
+        self._perm_table = _connect("memory://").create_table(perm_name, perm_data)
+
+    def state_dict(self) -> dict:
+        """Snapshot the dataset's consumption state.
+
+        The returned dict is topology-independent: at global step boundaries
+        every split has been consumed the same number of times (by the
+        round-robin design), so the per-split count is a single uniform value
+        that is identical across all ranks and DataLoader workers.
+        """
+        return {
+            "shuffle_seed": self._shuffle_seed,
+            "num_splits": self._num_splits,
+            "epoch": self._epoch,
+            "samples_consumed_per_split": [self._resume_offset] * self._num_splits,
+        }
+
+    def load_state_dict(self, state: dict) -> None:
+        """Resume from a previously snapshotted state.
+
+        Raises ``ValueError`` if ``num_splits`` or ``shuffle_seed`` differ
+        from the checkpoint, since a different split structure or shuffle order
+        makes mid-epoch resumption meaningless.
+        """
+        if state["num_splits"] != self._num_splits:
+            raise ValueError(
+                f"num_splits mismatch: checkpoint has {state['num_splits']}, "
+                f"current dataset has {self._num_splits}"
+            )
+        if state["shuffle_seed"] != self._shuffle_seed:
+            raise ValueError(
+                f"shuffle_seed mismatch: checkpoint has {state['shuffle_seed']}, "
+                f"current dataset has {self._shuffle_seed}"
+            )
+        consumed = state["samples_consumed_per_split"]
+        # All entries are equal at step boundaries; use the first.
+        if isinstance(consumed, list):
+            self._resume_offset = consumed[0] if consumed else 0
+        else:
+            self._resume_offset = int(consumed)

python/python/lancedb/table.py

@@ -651,6 +651,16 @@ def _append_vector_columns(
                     col_data = func.compute_source_embeddings_with_retry(
                         batch[conf.source_column]
                     )
+                    # Replace vectors with wrong length (including empty lists
+                    # returned for inputs like empty strings) with None so that
+                    # _handle_bad_vectors can process them according to the
+                    # on_bad_vectors policy instead of crashing when PyArrow
+                    # tries to cast them into a fixed-size list array.
+                    expected_ndims = conf.function.ndims()
+                    col_data = [
+                        v if v is not None and len(v) == expected_ndims else None
+                        for v in col_data
+                    ]
                     if no_vector_column:
                         batch = batch.append_column(
                             schema.field(vector_column),
@@ -2022,6 +2032,7 @@ class LanceTable(Table):
         namespace_client: Optional[Any] = None,
         managed_versioning: Optional[bool] = None,
         pushdown_operations: Optional[set] = None,
+        route_pushdown_to_rust: bool = False,
         _async: AsyncTable = None,
     ):
         if namespace_path is None:
@@ -2031,6 +2042,14 @@ class LanceTable(Table):
         self._location = location  # Store location for use in _dataset_path
         self._namespace_client = namespace_client
         self._pushdown_operations = pushdown_operations or set()
+        # When the connection built the namespace client natively (e.g. an
+        # enterprise "rest" connection), the underlying Rust table already
+        # executes QueryTable pushdown itself -- and, unlike this Python urllib3
+        # path, it routes through the read-freshness context provider that emits
+        # the ``x-lancedb-min-timestamp`` header. So we must defer pushdown to
+        # Rust instead of calling the Python ``namespace_client.query_table``
+        # directly, or reads silently bypass read-freshness (stale results).
+        self._route_pushdown_to_rust = route_pushdown_to_rust
         if _async is not None:
             self._table = _async
         else:
@@ -2133,12 +2152,19 @@ class LanceTable(Table):
 
         branch = self.current_branch()
         version = None if branch is not None else self.version
-        if self._namespace_client is not None:
+        namespace_client = self._namespace_client
+        if namespace_client is None:
+            conn_uri = getattr(self._conn, "uri", "")
+            if get_uri_scheme(conn_uri) == "namespace":
+                namespace_client = self._conn.namespace_client()
+                self._namespace_client = namespace_client
+
+        if namespace_client is not None:
             table_id = self._namespace_path + [self.name]
             ds = lance.dataset(
                 version=version,
                 storage_options=self._conn.storage_options,
-                namespace_client=self._namespace_client,
+                namespace_client=namespace_client,
                 table_id=table_id,
                 **kwargs,
             )
@@ -2241,6 +2267,7 @@ class LanceTable(Table):
             namespace_path=self._namespace_path,
             namespace_client=self._namespace_client,
             pushdown_operations=self._pushdown_operations,
+            route_pushdown_to_rust=self._route_pushdown_to_rust,
             location=self._location,
             _async=async_table,
         )
@@ -2391,8 +2418,11 @@ class LanceTable(Table):
         Returns
         -------
         pa.Table"""
-        if _should_push_down_query_table(
-            self._namespace_client, self._pushdown_operations
+        if (
+            _should_push_down_query_table(
+                self._namespace_client, self._pushdown_operations
+            )
+            and not self._route_pushdown_to_rust
         ):
             return self._execute_query(Query()).read_all()
 
@@ -3344,6 +3374,7 @@ class LanceTable(Table):
         location: Optional[str] = None,
         namespace_client: Optional[Any] = None,
         pushdown_operations: Optional[set] = None,
+        route_pushdown_to_rust: bool = False,
     ):
         """
         Create a new table.
@@ -3406,6 +3437,7 @@ class LanceTable(Table):
         self._location = location
         self._namespace_client = namespace_client
         self._pushdown_operations = pushdown_operations or set()
+        self._route_pushdown_to_rust = route_pushdown_to_rust
 
         if data_storage_version is not None:
             warnings.warn(
@@ -3519,6 +3551,7 @@ class LanceTable(Table):
             _should_push_down_query_table(
                 self._namespace_client, self._pushdown_operations
             )
+            and not self._route_pushdown_to_rust
             and self.current_branch() is None
         ):
             from lancedb.namespace import _execute_server_side_query
@@ -3997,7 +4030,16 @@ def _handle_bad_vector_column(
         dim = _infer_vector_dim(vec_arr)
         if dim is None:
             return data
-    has_wrong_dim = pc.not_equal(pc.list_value_length(vec_arr), dim)
+
+    is_null = pc.is_null(vec_arr)
+    # pc.list_value_length returns null for null list entries, so
+    # pc.not_equal(null, dim) also returns null. Use or_kleene so that
+    # True OR null = True (Kleene three-valued logic), ensuring null vectors
+    # are counted as wrong-dim.
+    has_wrong_dim = pc.or_kleene(
+        is_null,
+        pc.not_equal(pc.list_value_length(vec_arr), dim),
+    )
 
     has_bad_vectors = pc.any(has_nan).as_py() or pc.any(has_wrong_dim).as_py()
 
@@ -4260,6 +4302,7 @@ class AsyncTable:
         namespace_path: Optional[List[str]] = None,
         namespace_client: Optional[Any] = None,
         pushdown_operations: Optional[set] = None,
+        route_pushdown_to_rust: bool = False,
     ):
         """Create a new AsyncTable object.
 
@@ -4272,6 +4315,9 @@ class AsyncTable:
         self._namespace_path = namespace_path or []
         self._namespace_client = namespace_client
         self._pushdown_operations = pushdown_operations or set()
+        # See LanceTable.__init__: defer QueryTable pushdown to Rust (which emits
+        # the read-freshness header) for natively-built namespace clients.
+        self._route_pushdown_to_rust = route_pushdown_to_rust
 
     def _set_namespace_context(
         self,
@@ -4279,10 +4325,12 @@ class AsyncTable:
         namespace_path: Optional[List[str]] = None,
         namespace_client: Optional[Any] = None,
         pushdown_operations: Optional[set] = None,
+        route_pushdown_to_rust: bool = False,
     ) -> "AsyncTable":
         self._namespace_path = namespace_path or []
         self._namespace_client = namespace_client
         self._pushdown_operations = pushdown_operations or set()
+        self._route_pushdown_to_rust = route_pushdown_to_rust
         return self
 
     def __repr__(self):
@@ -4492,8 +4540,11 @@ class AsyncTable:
         -------
         pa.Table
         """
-        if _should_push_down_query_table(
-            self._namespace_client, self._pushdown_operations
+        if (
+            _should_push_down_query_table(
+                self._namespace_client, self._pushdown_operations
+            )
+            and not self._route_pushdown_to_rust
         ):
             return (await self._execute_query(Query())).read_all()
 
@@ -5177,8 +5228,11 @@ class AsyncTable:
         batch_size: Optional[int] = None,
         timeout: Optional[timedelta] = None,
     ) -> pa.RecordBatchReader:
-        if _should_push_down_query_table(
-            self._namespace_client, self._pushdown_operations
+        if (
+            _should_push_down_query_table(
+                self._namespace_client, self._pushdown_operations
+            )
+            and not self._route_pushdown_to_rust
         ):
             from lancedb.namespace import _execute_server_side_query
 

python/python/tests/test_embeddings.py

@@ -242,6 +242,49 @@ def test_embedding_with_bad_results(tmp_path):
     assert tbl["vector"].null_count == 1
 
 
+def test_embedding_with_empty_output_vectors(tmp_path):
+    """Regression test for issue #1672.
+
+    When an embedding function returns an empty list (e.g. for empty-string
+    inputs), _append_vector_columns used to crash because PyArrow cannot cast
+    [] into a fixed-size list element.  The fix replaces wrong-length vectors
+    with None before building the Arrow array so that _handle_bad_vectors can
+    process them normally.
+    """
+
+    @register("empty-vec-embedding")
+    class EmptyVecEmbeddingFunction(TextEmbeddingFunction):
+        def ndims(self):
+            return 128
+
+        def generate_embeddings(self, texts: Union[List[str], np.ndarray]) -> list:
+            # Simulate a model that returns an empty list for blank inputs
+            return [
+                [] if text.strip() == "" else np.random.randn(self.ndims()).tolist()
+                for text in texts
+            ]
+
+    db = lancedb.connect(tmp_path)
+    registry = EmbeddingFunctionRegistry.get_instance()
+    model = registry.get("empty-vec-embedding").create()
+
+    class Schema(LanceModel):
+        text: str = model.SourceField()
+        vector: Vector(model.ndims()) = model.VectorField()
+
+    table = db.create_table("test_empty_vec", schema=Schema, mode="overwrite")
+
+    # Should not crash; the row with the empty string should be dropped
+    table.add(
+        [{"text": "hello world"}, {"text": ""}, {"text": "foo"}],
+        on_bad_vectors="drop",
+    )
+
+    assert len(table) == 2
+    texts = table.to_arrow()["text"].to_pylist()
+    assert "" not in texts
+
+
 def test_with_existing_vectors(tmp_path):
     @register("mock-embedding")
     class MockEmbeddingFunction(TextEmbeddingFunction):

python/python/tests/test_expr.py

@@ -3,10 +3,14 @@
 
 """Tests for the type-safe expression builder API."""
 
-import pytest
+from datetime import date, datetime, timedelta, timezone
+from decimal import Decimal
+
 import pyarrow as pa
+import pytest
+
 import lancedb
-from lancedb.expr import Expr, col, lit, func
+from lancedb.expr import Expr, col, func, lit
 
 
 # ── unit tests for Expr construction ─────────────────────────────────────────
@@ -54,6 +58,28 @@ class TestExprConstruction:
         with pytest.raises(Exception):
             func("not_a_real_function", col("x"))
 
+    def test_lit_date(self):
+        e = lit(date(2024, 1, 1))
+        assert isinstance(e, Expr)
+
+    def test_lit_datetime(self):
+        # Naive datetime
+        e = lit(datetime(2024, 1, 1, 10, 0))
+        assert isinstance(e, Expr)
+
+    def test_lit_datetime_tz(self):
+        # Timezone-aware datetime
+        tz = timezone(timedelta(hours=5))
+        dt = datetime(2024, 1, 1, 10, 0, tzinfo=tz)
+        e = lit(dt)
+        assert isinstance(e, Expr)
+
+    def test_lit_decimal_precision(self):
+        # High precision Decimal that would be rounded if converted to float
+        d = Decimal("1.234567890123456789")
+        e = lit(d)
+        assert isinstance(e, Expr)
+
 
 class TestExprOperators:
     def test_eq_operator(self):
@@ -142,6 +168,20 @@ class TestExprOperators:
         assert isinstance(e, Expr)
         assert e.to_sql() == "(name = 'alice')"
 
+    def test_reflexive_comparisons(self):
+        # 10 < col("age") swaps to col("age") > 10
+        assert (10 < col("age")).to_sql() == "(age > 10)"
+        assert (10 <= col("age")).to_sql() == "(age >= 10)"
+        assert (10 > col("age")).to_sql() == "(age < 10)"
+        assert (10 >= col("age")).to_sql() == "(age <= 10)"
+        assert (10 == col("age")).to_sql() == "(age = 10)"
+        assert (10 != col("age")).to_sql() == "(age <> 10)"
+
+    def test_reflexive_logical(self):
+        # True & Expr calls Expr.__rand__(True)
+        assert (True & (col("age") > 18)).to_sql() == "(true AND (age > 18))"
+        assert (False | (col("age") > 18)).to_sql() == "(false OR (age > 18))"
+
 
 class TestExprBytesLiteral:
     def test_bytes_to_sql(self):
@@ -282,6 +322,40 @@ class TestExprRepr:
             {e: 1}
 
 
+class TestExprReflexive:
+    def test_reflexive_eq(self):
+        e = 1 == col("x")
+        assert isinstance(e, Expr)
+        assert e.to_sql() == "(x = 1)"
+
+    def test_reflexive_ne(self):
+        e = 1 != col("x")
+        assert isinstance(e, Expr)
+        assert e.to_sql() == "(x <> 1)"
+
+    def test_reflexive_lt(self):
+        # 1 < x  =>  (x > 1)
+        e = 1 < col("x")
+        assert isinstance(e, Expr)
+        assert e.to_sql() == "(x > 1)"
+
+    def test_reflexive_gt(self):
+        # 1 > x  =>  (x < 1)
+        e = 1 > col("x")
+        assert isinstance(e, Expr)
+        assert e.to_sql() == "(x < 1)"
+
+    def test_reflexive_and(self):
+        e = True & col("active")
+        assert isinstance(e, Expr)
+        assert e.to_sql() == "(true AND active)"
+
+    def test_reflexive_or(self):
+        e = False | col("inactive")
+        assert isinstance(e, Expr)
+        assert e.to_sql() == "(false OR inactive)"
+
+
 # ── integration tests: end-to-end query against a real table ─────────────────
 
 
@@ -432,6 +506,72 @@ class TestColNamingIntegration:
         assert sorted(result["upper_name"].to_pylist()) == ["ALICE", "BOB", "CHARLIE"]
 
 
+@pytest.fixture
+def type_check_table(tmp_path):
+    """Fixture that creates a table with Date32 and Decimal128 columns."""
+    db = lancedb.connect(str(tmp_path))
+    schema = pa.schema(
+        [
+            ("date", pa.date32()),
+            ("decimal", pa.decimal128(10, 2)),
+            ("binary", pa.binary()),
+        ]
+    )
+    data = pa.table(
+        {
+            "date": [date(2024, 1, 1), date(2024, 1, 2)],
+            "decimal": [Decimal("10.50"), Decimal("20.75")],
+            "binary": [b"\x01", b"\x02"],
+        },
+        schema=schema,
+    )
+    return db.create_table("extended_types", data)
+
+
+class TestExtendedTypeIntegration:
+    """Integration tests verifying that typed literals work correctly in filters."""
+
+    def test_date_integration(self, type_check_table):
+        """Verify that Date32 literals are correctly parsed and filtered."""
+        result = (
+            type_check_table.search()
+            .where(col("date") == lit(date(2024, 1, 1)))
+            .to_arrow()
+        )
+        assert result.num_rows == 1
+        assert result["date"][0].as_py() == date(2024, 1, 1)
+
+    def test_decimal_integration(self, tmp_path):
+        """A Decimal literal must retain full 128-bit precision in a filter.
+
+        1.234567890123456789 and 1.234567890123456790 differ only in the last
+        digit and are indistinguishable once truncated to f64.  The filter
+        therefore returns the single expected row only if ``lit(Decimal)``
+        produces a true ``Decimal128`` scalar rather than being coerced to f64.
+        """
+        low = Decimal("1.234567890123456789")
+        high = Decimal("1.234567890123456790")
+
+        db = lancedb.connect(str(tmp_path / "decimal_precision"))
+        schema = pa.schema([("val", pa.decimal128(19, 18))])
+        table = db.create_table(
+            "decimal_precision",
+            pa.table({"val": [low, high]}, schema=schema),
+        )
+
+        result = table.search().where(col("val") < lit(high)).to_arrow()
+        assert result.num_rows == 1
+        assert result["val"][0].as_py() == low
+
+    def test_binary_integration(self, type_check_table):
+        """Verify that Binary literals are correctly filtered."""
+        result = (
+            type_check_table.search().where(col("binary") == lit(b"\x01")).to_arrow()
+        )
+        assert result.num_rows == 1
+        assert result["binary"][0].as_py() == b"\x01"
+
+
 # ── bytes / binary column integration tests ───────────────────────────────────
 
 

python/python/tests/test_namespace.py

@@ -5,6 +5,7 @@
 
 import tempfile
 import shutil
+import importlib
 import pytest
 import pyarrow as pa
 import lancedb
@@ -65,6 +66,9 @@ def _namespace_lance_table(namespace_client: _NamespaceClient) -> LanceTable:
     table._namespace_path = ["geneva"]
     table._namespace_client = namespace_client
     table._pushdown_operations = {"QueryTable"}
+    # This test exercises the Python-side pushdown path (non-native client), so
+    # pushdown is not routed to Rust.
+    table._route_pushdown_to_rust = False
     return table
 
 
@@ -100,6 +104,40 @@ class TestNamespaceConnection:
         assert isinstance(db, lancedb.LanceNamespaceDBConnection)
         assert len(list(db.table_names())) == 0
 
+    def test_sync_builtin_namespace_uses_rust_without_python_client(self, monkeypatch):
+        """Built-in sync namespace connections should not construct or call the
+        Python namespace client for normal namespace/table management."""
+        namespace_module = importlib.import_module("lancedb.namespace")
+
+        def fail_namespace_connect(*args, **kwargs):
+            raise AssertionError("Python namespace client should not be constructed")
+
+        monkeypatch.setattr(
+            namespace_module, "namespace_connect", fail_namespace_connect
+        )
+
+        db = lancedb.connect_namespace("dir", {"root": self.temp_dir})
+        assert isinstance(db, lancedb.LanceNamespaceDBConnection)
+        assert db._namespace_client is None
+        assert db._route_pushdown_to_rust is True
+
+        db.create_namespace(["test_ns"])
+        assert "test_ns" in db.list_namespaces().namespaces
+
+        schema = pa.schema([pa.field("id", pa.int64())])
+        table = db.create_table("test_table", schema=schema, namespace_path=["test_ns"])
+        assert table.namespace == ["test_ns"]
+        assert "test_table" in db.table_names(namespace_path=["test_ns"])
+        assert "test_table" in db.list_tables(namespace_path=["test_ns"]).tables
+
+        opened = db.open_table("test_table", namespace_path=["test_ns"])
+        assert opened.namespace == ["test_ns"]
+
+        db.drop_table("test_table", namespace_path=["test_ns"])
+        assert db.list_tables(namespace_path=["test_ns"]).tables == []
+        db.drop_namespace(["test_ns"])
+        assert "test_ns" not in db.list_namespaces().namespaces
+
     def test_create_table_through_namespace(self):
         """Test creating a table through namespace."""
         db = lancedb.connect_namespace("dir", {"root": self.temp_dir})
@@ -561,6 +599,61 @@ class TestAsyncNamespaceConnection:
         table_names = await db.table_names()
         assert len(list(table_names)) == 0
 
+    async def test_async_builtin_namespace_uses_rust_without_python_client(
+        self, monkeypatch
+    ):
+        """Built-in async namespace connections should not construct or call the
+        Python namespace client for normal namespace/table management."""
+        namespace_module = importlib.import_module("lancedb.namespace")
+
+        def fail_namespace_connect(*args, **kwargs):
+            raise AssertionError("Python namespace client should not be constructed")
+
+        monkeypatch.setattr(
+            namespace_module, "namespace_connect", fail_namespace_connect
+        )
+
+        db = lancedb.connect_namespace_async("dir", {"root": self.temp_dir})
+        assert isinstance(db, lancedb.AsyncLanceNamespaceDBConnection)
+        assert db._namespace_client is None
+        assert db._route_pushdown_to_rust is True
+
+        await db.create_namespace(["test_ns"])
+        assert "test_ns" in (await db.list_namespaces()).namespaces
+
+        schema = pa.schema([pa.field("id", pa.int64())])
+        table = await db.create_table(
+            "test_table", schema=schema, namespace_path=["test_ns"]
+        )
+        assert table._namespace_path == ["test_ns"]
+        assert table._namespace_client is None
+        assert table._route_pushdown_to_rust is True
+        assert "test_table" in await db.table_names(namespace_path=["test_ns"])
+        assert "test_table" in (await db.list_tables(namespace_path=["test_ns"])).tables
+
+        opened = await db.open_table("test_table", namespace_path=["test_ns"])
+        assert opened._namespace_path == ["test_ns"]
+
+        await db.drop_table("test_table", namespace_path=["test_ns"])
+        assert (await db.list_tables(namespace_path=["test_ns"])).tables == []
+        await db.drop_namespace(["test_ns"])
+        assert "test_ns" not in (await db.list_namespaces()).namespaces
+
+    async def test_async_namespace_client_is_lazy(self):
+        """namespace_client() should still return the backing client on demand."""
+        pytest.importorskip("lance")
+        from lance.namespace import DirectoryNamespace
+
+        db = lancedb.connect_namespace_async("dir", {"root": self.temp_dir})
+        assert db._namespace_client is None
+
+        ns_client = await db.namespace_client()
+
+        assert isinstance(ns_client, DirectoryNamespace)
+        namespace_id = ns_client.namespace_id().replace("\\\\", "\\")
+        assert str(self.temp_dir) in namespace_id
+        assert db._namespace_client is ns_client
+
     # Async connect via namespace helper is not enabled yet.
 
     async def test_create_table_async(self):
@@ -805,6 +898,39 @@ class TestPushdownOperations:
         db = lancedb.connect_namespace("dir", {"root": self.temp_dir})
         assert len(db._namespace_client_pushdown_operations) == 0
 
+    def test_route_pushdown_to_rust_for_native_rest(self):
+        """A natively-built rest connection must defer QueryTable pushdown to
+        Rust so reads carry the x-lancedb-min-timestamp read-freshness header."""
+        db = lancedb.connect_namespace(
+            "rest",
+            {"uri": "http://localhost:12345"},
+            namespace_client_pushdown_operations=["QueryTable"],
+        )
+        assert db._route_pushdown_to_rust is True
+
+    def test_route_pushdown_to_rust_for_native_dir(self):
+        """The sync dir connection is natively built and defers QueryTable
+        pushdown to Rust."""
+        db = lancedb.connect_namespace("dir", {"root": self.temp_dir})
+        assert db._route_pushdown_to_rust is True
+
+    def test_async_route_pushdown_to_rust_for_native_rest(self):
+        """The async connection must not silently bypass the read-freshness fix:
+        a natively-built rest connection defers pushdown to Rust (regression test
+        for the async path omitting the freshness header)."""
+        db = lancedb.connect_namespace_async(
+            "rest",
+            {"uri": "http://localhost:12345"},
+            namespace_client_pushdown_operations=["QueryTable"],
+        )
+        assert db._route_pushdown_to_rust is True
+
+    def test_async_route_pushdown_to_rust_for_native_dir(self):
+        """The async dir connection is natively built and defers QueryTable
+        pushdown to Rust."""
+        db = lancedb.connect_namespace_async("dir", {"root": self.temp_dir})
+        assert db._route_pushdown_to_rust is True
+
     def test_lance_table_to_arrow_uses_query_pushdown(self):
         namespace_client = _NamespaceClient()
         table = _namespace_lance_table(namespace_client)

python/python/tests/test_query.py

@@ -502,6 +502,61 @@ def test_with_row_id(table: lancedb.table.Table):
     assert rs["_rowid"].to_pylist() == [0, 1]
 
 
+def test_where_repeated_combines_with_and(table: lancedb.table.Table):
+    # Calling where() more than once should AND the filters together instead of
+    # silently replacing the previous one (regression test for #2649).
+    builder = table.search().where("id >= 1").where("id < 2")
+    assert builder._where == "(id >= 1) AND (id < 2)"
+
+    ids = [row["id"] for row in builder.limit(10).to_list()]
+    assert ids == [1]
+
+
+def test_where_repeated_combines_expr(table: lancedb.table.Table):
+    from lancedb.expr import col, lit
+
+    builder = table.search().where(col("id") >= lit(1)).where(col("id") < lit(2))
+    ids = [row["id"] for row in builder.limit(10).to_list()]
+    assert ids == [1]
+
+
+def test_where_mixed_filter_kinds_combines(table: lancedb.table.Table):
+    # Mixing a SQL string filter with an expression filter lowers the
+    # expression to SQL and combines them as SQL strings.
+    from lancedb.expr import col, lit
+
+    builder = table.search().where("id >= 1").where(col("id") < lit(2))
+    ids = [row["id"] for row in builder.limit(10).to_list()]
+    assert ids == [1]
+
+
+@pytest.mark.asyncio
+async def test_where_repeated_combines_with_and_async(table_async: AsyncTable):
+    ids = [
+        row["id"]
+        for row in (
+            await table_async.query().where("id >= 1").where("id < 2").to_list()
+        )
+    ]
+    assert ids == [1]
+
+
+@pytest.mark.asyncio
+async def test_where_mixed_filter_kinds_combines_async(table_async: AsyncTable):
+    from lancedb.expr import col, lit
+
+    ids = [
+        row["id"]
+        for row in (
+            await table_async.query()
+            .where("id >= 1")
+            .where(col("id") < lit(2))
+            .to_list()
+        )
+    ]
+    assert ids == [1]
+
+
 def test_distance_range(table: lancedb.table.Table):
     q = [0, 0]
     rs = table.search(q).to_arrow()

python/python/tests/test_remote_db.py

@@ -412,10 +412,12 @@ def test_remote_permutation_is_picklable():
             content_len = int(request.headers.get("Content-Length"))
             body = json.loads(request.rfile.read(content_len))
             if "filter" in body:
-                match = re.search(r"_rowoffset in \((.*?)\)", body["filter"])
-                offsets = [int(offset.strip()) for offset in match.group(1).split(",")]
+                match = re.search(
+                    r"_rowoffset\s+in\s+\((.*?)\)", body["filter"], re.IGNORECASE
+                )
+                offsets = [int(o.strip()) for o in match.group(1).split(",")]
             else:
-                offsets = rows
+                offsets = list(range(len(rows)))
             table = pa.table({"a": [rows[offset] for offset in offsets]})
 
             request.send_response(200)

python/python/tests/test_rerankers.py

@@ -350,6 +350,38 @@ def test_mrr_reranker_empty_input():
         reranker.rerank_multivector([])
 
 
+def test_mrr_multivector_rewards_consensus():
+    # Reciprocal ranks must be averaged across *all* ranking systems, treating a
+    # missing system as 0. A document ranked first by every system must outrank a
+    # document ranked first by only one of them.
+    reranker = MRRReranker()
+
+    def ranking(row_ids):
+        return pa.table({"_rowid": pa.array(row_ids, type=pa.int64())})
+
+    # Doc 1 is rank 1 in only the first system; doc 2 is rank 1 in two systems
+    # and rank 2 in the third (strong cross-system consensus).
+    rs1 = ranking([1, 2, 3])
+    rs2 = ranking([2, 3, 4])
+    rs3 = ranking([2, 5, 6])
+
+    result = reranker.rerank_multivector([rs1, rs2, rs3])
+    scores = {
+        row_id: score
+        for row_id, score in zip(
+            result["_rowid"].to_pylist(),
+            result["_relevance_score"].to_pylist(),
+        )
+    }
+
+    # sum of reciprocal ranks / number of systems
+    assert scores[1] == pytest.approx(1.0 / 3)
+    assert scores[2] == pytest.approx((0.5 + 1.0 + 1.0) / 3)
+    assert scores[2] > scores[1]
+    # The consensus document ranks first overall.
+    assert result["_rowid"].to_pylist()[0] == 2
+
+
 def test_rrf_reranker_distance():
     data = pa.table(
         {

python/python/tests/test_table.py

@@ -1137,6 +1137,16 @@ def test_namespace_open_table_with_branch_version(tmp_path):
     assert db.open_table("t", namespace_path=["ns1"], branch="exp").count_rows() == 3
 
 
+def test_namespace_root_table_to_lance_uses_namespace_client(tmp_path):
+    pytest.importorskip("lance")  # "dir" impl is lance.namespace.DirectoryNamespace
+    db = lancedb.connect_namespace("dir", {"root": str(tmp_path)})
+    table = db.create_table("t", [{"i": 0}])
+
+    assert table._namespace_client is None
+    assert table.to_lance().count_rows() == 1
+    assert table._namespace_client is not None
+
+
 @pytest.mark.asyncio
 async def test_async_namespace_open_table_with_branch_version(tmp_path):
     pytest.importorskip("lance")  # "dir" impl is lance.namespace.DirectoryNamespace

python/src/connection.rs

@@ -539,7 +539,7 @@ 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, manifest_enabled=false, namespace_client_properties=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, oauth_config=None))]
 #[allow(clippy::too_many_arguments)]
 pub fn connect(
     py: Python<'_>,
@@ -553,6 +553,7 @@ pub fn connect(
     session: Option<crate::session::Session>,
     manifest_enabled: bool,
     namespace_client_properties: Option<HashMap<String, String>>,
+    oauth_config: Option<crate::oauth::PyOAuthConfig>,
 ) -> PyResult<Bound<'_, PyAny>> {
     future_into_py(py, async move {
         let mut builder = lancedb::connect(&uri);
@@ -582,6 +583,11 @@ pub fn connect(
         if let Some(client_config) = client_config {
             builder = builder.client_config(client_config.into());
         }
+        if let Some(oauth_config) = oauth_config {
+            let config: lancedb::remote::oauth::OAuthConfig =
+                oauth_config.try_into().infer_error()?;
+            builder = builder.oauth_config(config);
+        }
         if let Some(session) = session {
             builder = builder.session(session.inner.clone());
         }
@@ -610,24 +616,38 @@ pub fn connect_namespace_client(
     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,
-    );
+    // Prefer building the namespace natively from (impl, properties) so the
+    // read-freshness provider installed
+    let database = if build_namespace_natively(namespace_client_impl.as_deref(), &ns_properties) {
+        let ns_impl = namespace_client_impl.expect("impl present per build_namespace_natively");
+        crate::runtime::block_on(LanceNamespaceDatabase::connect(
+            &ns_impl,
+            ns_properties,
+            storage_options,
+            read_consistency_interval,
+            session,
+            namespace_client_pushdown_operations,
+        ))
+        .infer_error()?
+    } else {
+        let namespace_client = extract_namespace_arc(py, namespace_client)?;
+        LanceNamespaceDatabase::from_namespace_client(
+            namespace_client,
+            namespace_client_impl.unwrap_or_else(|| "python".to_string()),
+            ns_properties,
+            storage_options,
+            read_consistency_interval,
+            session,
+            namespace_client_pushdown_operations,
+        )
+    };
 
     Ok(Connection::new(LanceConnection::new(
         Arc::new(database),
@@ -635,6 +655,56 @@ pub fn connect_namespace_client(
     )))
 }
 
+#[pyfunction]
+#[pyo3(signature = (
+    namespace_client_impl,
+    namespace_client_properties,
+    read_consistency_interval=None,
+    storage_options=None,
+    session=None,
+    namespace_client_pushdown_operations=None,
+))]
+#[allow(clippy::too_many_arguments)]
+pub fn connect_namespace(
+    namespace_client_impl: String,
+    namespace_client_properties: HashMap<String, String>,
+    read_consistency_interval: Option<f64>,
+    storage_options: Option<HashMap<String, String>>,
+    session: Option<crate::session::Session>,
+    namespace_client_pushdown_operations: Option<Vec<String>>,
+) -> PyResult<Connection> {
+    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 mut builder =
+        lancedb::connect_namespace(&namespace_client_impl, namespace_client_properties)
+            .pushdown_operations(namespace_client_pushdown_operations);
+    if let Some(storage_options) = storage_options {
+        builder = builder.storage_options(storage_options);
+    }
+    if let Some(read_consistency_interval) = read_consistency_interval {
+        builder = builder.read_consistency_interval(read_consistency_interval);
+    }
+    if let Some(session) = session {
+        builder = builder.session(session.inner.clone());
+    }
+
+    Ok(Connection::new(
+        crate::runtime::block_on(builder.execute()).infer_error()?,
+    ))
+}
+
+/// Whether to build the namespace natively (from impl + properties) instead of
+/// wrapping a pre-built client. Native construction is required for the
+/// read-freshness provider to be installed
+fn build_namespace_natively(
+    namespace_client_impl: Option<&str>,
+    namespace_client_properties: &HashMap<String, String>,
+) -> bool {
+    matches!(namespace_client_impl, Some("rest")) && !namespace_client_properties.is_empty()
+}
+
 #[derive(FromPyObject)]
 pub struct PyClientConfig {
     user_agent: String,
@@ -733,3 +803,36 @@ impl From<PyClientConfig> for lancedb::remote::ClientConfig {
         }
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn props(pairs: &[(&str, &str)]) -> HashMap<String, String> {
+        pairs
+            .iter()
+            .map(|(k, v)| (k.to_string(), v.to_string()))
+            .collect()
+    }
+
+    #[test]
+    fn native_build_only_for_rest_with_properties() {
+        let rest = props(&[("uri", "http://localhost:10024")]);
+
+        // rest + non-empty properties -> build natively (installs the
+        // read-freshness provider so checkout_latest() busts the server cache).
+        assert!(build_namespace_natively(Some("rest"), &rest));
+
+        // dir is local (no server cache) -> wrap the pre-built client unchanged.
+        assert!(!build_namespace_natively(
+            Some("dir"),
+            &props(&[("root", "/tmp")])
+        ));
+
+        // No impl: only a pre-built client was handed in -> wrap it as-is.
+        assert!(!build_namespace_natively(None, &rest));
+
+        // rest but no properties: nothing to build a connection from -> wrap.
+        assert!(!build_namespace_natively(Some("rest"), &HashMap::new()));
+    }
+}

python/src/expr.rs

@@ -7,12 +7,14 @@
 //! build type-safe filter / projection expressions that map directly to
 //! DataFusion [`Expr`] nodes, bypassing SQL string parsing.
 
+use std::ops::{Add, Div, Mul, Not, Sub};
+
 use arrow::{datatypes::DataType, pyarrow::PyArrowType};
 use datafusion_common::ScalarValue;
 use lancedb::expr::{
     DfExpr, col as ldb_col, contains, expr_cast, is_in, lit as df_lit, lower, upper,
 };
-use pyo3::types::PyBytes;
+use pyo3::types::{PyBytes, PyDate, PyDateTime};
 use pyo3::{Bound, PyAny, PyResult, exceptions::PyValueError, prelude::*, pyfunction};
 
 /// A type-safe DataFusion expression.
@@ -63,30 +65,30 @@ impl PyExpr {
         Self(self.0.clone().or(other.0.clone()))
     }
 
+    /// Logical NOT.
     fn not_(&self) -> Self {
-        use std::ops::Not;
         Self(self.0.clone().not())
     }
 
     // ── arithmetic ───────────────────────────────────────────────────────────
 
+    /// Add expressions.
     fn add(&self, other: &Self) -> Self {
-        use std::ops::Add;
         Self(self.0.clone().add(other.0.clone()))
     }
 
+    /// Subtract expressions.
     fn sub(&self, other: &Self) -> Self {
-        use std::ops::Sub;
         Self(self.0.clone().sub(other.0.clone()))
     }
 
+    /// Multiply expressions.
     fn mul(&self, other: &Self) -> Self {
-        use std::ops::Mul;
         Self(self.0.clone().mul(other.0.clone()))
     }
 
+    /// Divide expressions.
     fn div(&self, other: &Self) -> Self {
-        use std::ops::Div;
         Self(self.0.clone().div(other.0.clone()))
     }
 
@@ -153,7 +155,8 @@ pub fn expr_col(name: &str) -> PyExpr {
 
 /// Create a literal value expression.
 ///
-/// Supported Python types: `bool`, `int`, `float`, `str`, `bytes`.
+/// Supported Python types: `bool`, `int`, `float`, `str`, `bytes`, `date`,
+/// `datetime`, `Decimal`.
 #[pyfunction]
 pub fn expr_lit(value: Bound<'_, PyAny>) -> PyResult<PyExpr> {
     // bool must be checked before int because bool is a subclass of int in Python
@@ -163,6 +166,19 @@ pub fn expr_lit(value: Bound<'_, PyAny>) -> PyResult<PyExpr> {
     if let Ok(i) = value.extract::<i64>() {
         return Ok(PyExpr(df_lit(i)));
     }
+    // Decimal must be checked before f64: Python's Decimal implements __float__,
+    // so value.extract::<f64>() would succeed and silently truncate the value to
+    // f64, losing precision. Build a Decimal128 scalar to preserve it instead.
+    if value.get_type().name()? == "Decimal" {
+        let s = value.call_method0("__str__")?.extract::<String>()?;
+        // Parse the decimal string into an i128 value, precision, and scale.
+        let (val, precision, scale) = parse_decimal(&s)?;
+        return Ok(PyExpr(df_lit(ScalarValue::Decimal128(
+            Some(val),
+            precision,
+            scale,
+        ))));
+    }
     if let Ok(f) = value.extract::<f64>() {
         return Ok(PyExpr(df_lit(f)));
     }
@@ -173,12 +189,48 @@ pub fn expr_lit(value: Bound<'_, PyAny>) -> PyResult<PyExpr> {
         let bytes = value.extract::<Vec<u8>>()?;
         return Ok(PyExpr(df_lit(ScalarValue::Binary(Some(bytes)))));
     }
+
+    // datetime.datetime is a subclass of datetime.date, so it must be checked first.
+    if let Ok(dt) = value.cast::<PyDateTime>() {
+        let ts: f64 = dt.call_method0("timestamp")?.extract()?;
+        let micros = (ts * 1_000_000.0).round() as i64;
+        return Ok(PyExpr(df_lit(ScalarValue::TimestampMicrosecond(
+            Some(micros),
+            None,
+        ))));
+    }
+    if let Ok(d) = value.cast::<PyDate>() {
+        let ordinal: i32 = d.call_method0("toordinal")?.extract()?;
+        let days = ordinal - 719163; // Unix epoch is 1970-01-01
+        return Ok(PyExpr(df_lit(ScalarValue::Date32(Some(days)))));
+    }
+
     Err(PyValueError::new_err(format!(
-        "unsupported literal type: {}. Supported: bool, int, float, str, bytes",
+        "unsupported literal type: {}. Supported: bool, int, float, str, bytes, date, datetime, Decimal",
         value.get_type().name()?
     )))
 }
 
+fn parse_decimal(s: &str) -> PyResult<(i128, u8, i8)> {
+    let s = s.trim();
+    let dot_pos = s.find('.');
+    let scale = if let Some(pos) = dot_pos {
+        (s.len() - pos - 1) as i8
+    } else {
+        0
+    };
+
+    let digits = s.replace('.', "");
+    let val = digits
+        .parse::<i128>()
+        .map_err(|e| PyValueError::new_err(format!("failed to parse decimal digits: {}", e)))?;
+
+    // Precision is total number of digits
+    let precision = digits.trim_start_matches('-').len() as u8;
+
+    Ok((val, precision, scale))
+}
+
 /// Call an arbitrary registered SQL function by name.
 ///
 /// See `lancedb::expr::func` for the list of supported function names.

python/src/lib.rs

@@ -2,7 +2,7 @@
 // SPDX-FileCopyrightText: Copyright The LanceDB Authors
 
 use arrow::RecordBatchStream;
-use connection::{Connection, connect, connect_namespace_client};
+use connection::{Connection, connect, connect_namespace, connect_namespace_client};
 use env_logger::Env;
 use expr::{PyExpr, expr_col, expr_func, expr_lit};
 use index::IndexConfig;
@@ -26,6 +26,7 @@ pub mod expr;
 pub mod header;
 pub mod index;
 pub mod namespace;
+pub mod oauth;
 pub mod permutation;
 pub mod query;
 pub mod runtime;
@@ -61,6 +62,7 @@ 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, 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)?)?;

python/src/oauth.rs

@@ -0,0 +1,72 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+use pyo3::FromPyObject;
+
+use lancedb::error::Error;
+use lancedb::remote::oauth::{OAuthConfig, OAuthFlow};
+
+/// Python-side OAuth configuration, extracted via FromPyObject.
+/// Maps to `lancedb.remote.oauth.OAuthConfig` Python dataclass.
+#[derive(FromPyObject)]
+pub struct PyOAuthConfig {
+    pub issuer_url: String,
+    pub client_id: String,
+    pub scopes: Vec<String>,
+    pub flow: String,
+    pub client_secret: Option<String>,
+    pub managed_identity_client_id: Option<String>,
+    pub refresh_buffer_secs: Option<u64>,
+}
+
+impl TryFrom<PyOAuthConfig> for OAuthConfig {
+    type Error = Error;
+
+    fn try_from(py: PyOAuthConfig) -> Result<Self, Self::Error> {
+        let flow = match py.flow.as_str() {
+            "client_credentials" => OAuthFlow::ClientCredentials,
+            "azure_managed_identity" => OAuthFlow::AzureManagedIdentity {
+                client_id: py.managed_identity_client_id,
+            },
+            other => {
+                return Err(Error::InvalidInput {
+                    message: format!("Unknown OAuth flow type: {other}"),
+                });
+            }
+        };
+
+        Ok(Self {
+            issuer_url: py.issuer_url,
+            client_id: py.client_id,
+            client_secret: py.client_secret,
+            scopes: py.scopes,
+            flow,
+            refresh_buffer_secs: py.refresh_buffer_secs,
+        })
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_unknown_oauth_flow_returns_invalid_input() {
+        let config = PyOAuthConfig {
+            issuer_url: "https://issuer.example.com".to_string(),
+            client_id: "client-id".to_string(),
+            scopes: vec!["scope".to_string()],
+            flow: "typo".to_string(),
+            client_secret: None,
+            managed_identity_client_id: None,
+            refresh_buffer_secs: None,
+        };
+
+        let err = OAuthConfig::try_from(config).unwrap_err();
+        assert!(matches!(
+            err,
+            Error::InvalidInput { message }
+                if message == "Unknown OAuth flow type: typo"
+        ));
+    }
+}

python/src/permutation.rs

@@ -79,13 +79,14 @@ impl PyAsyncPermutationBuilder {
 
 #[pymethods]
 impl PyAsyncPermutationBuilder {
-    #[pyo3(signature = (*, ratios=None, counts=None, fixed=None, seed=None, split_names=None))]
+    #[pyo3(signature = (*, ratios=None, counts=None, fixed=None, seed=None, clump_size=None, split_names=None))]
     pub fn split_random(
         slf: PyRefMut<'_, Self>,
         ratios: Option<Vec<f64>>,
         counts: Option<Vec<u64>>,
         fixed: Option<u64>,
         seed: Option<u64>,
+        clump_size: Option<u64>,
         split_names: Option<Vec<String>>,
     ) -> PyResult<Self> {
         // Check that exactly one split type is provided
@@ -111,7 +112,14 @@ impl PyAsyncPermutationBuilder {
         };
 
         slf.modify(|builder| {
-            builder.with_split_strategy(SplitStrategy::Random { seed, sizes }, split_names)
+            builder.with_split_strategy(
+                SplitStrategy::Random {
+                    seed,
+                    sizes,
+                    clump_size,
+                },
+                split_names,
+            )
         })
     }
 

python/src/runtime.rs

@@ -56,6 +56,15 @@ fn get_runtime() -> &'static runtime::Runtime {
     unsafe { &*new_ptr }
 }
 
+/// Block the current thread on a future using the shared runtime.
+///
+/// For sync `#[pyfunction]`s that need to drive an async operation (e.g.
+/// building a namespace client). Must not be called from within the runtime's
+/// own worker threads.
+pub fn block_on<F: std::future::Future>(fut: F) -> F::Output {
+    get_runtime().block_on(fut)
+}
+
 /// Runs in async-signal context after `fork()` in the child.  We can only
 /// touch atomics here; we deliberately leak the previous runtime because
 /// dropping a tokio `Runtime` would try to join its (now-dead) worker

python/tests/test_oauth.py

@@ -0,0 +1,33 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+import importlib.util
+import sys
+from pathlib import Path
+
+
+def _load_oauth_module():
+    oauth_path = (
+        Path(__file__).parents[1] / "python" / "lancedb" / "remote" / "oauth.py"
+    )
+    spec = importlib.util.spec_from_file_location("lancedb_remote_oauth", oauth_path)
+    module = importlib.util.module_from_spec(spec)
+    assert spec.loader is not None
+    sys.modules[spec.name] = module
+    spec.loader.exec_module(module)
+    return module
+
+
+def test_oauth_config_repr_redacts_client_secret():
+    oauth = _load_oauth_module()
+
+    config = oauth.OAuthConfig(
+        issuer_url="https://issuer.example.com",
+        client_id="client-id",
+        scopes=["scope"],
+        client_secret="super-secret",
+    )
+
+    rendered = repr(config)
+    assert "super-secret" not in rendered
+    assert "client_secret" not in rendered

rust/lancedb/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "lancedb"
-version = "0.31.0-beta.2"
+version = "0.31.0-beta.6"
 edition.workspace = true
 description = "LanceDB: A serverless, low-latency vector database for AI applications"
 license.workspace = true
@@ -14,6 +14,7 @@ rust-version.workspace = true
 ahash = { workspace = true }
 arrow = { workspace = true }
 arrow-array = { workspace = true }
+arrow-buffer = { workspace = true }
 arrow-data = { workspace = true }
 arrow-schema = { workspace = true }
 arrow-select = { workspace = true }
@@ -50,7 +51,7 @@ lance-namespace = { workspace = true }
 lance-namespace-impls = { workspace = true }
 moka = { workspace = true }
 pin-project = { workspace = true }
-tokio = { version = "1.23", features = ["rt-multi-thread"] }
+tokio = { version = "1.23", features = ["rt-multi-thread", "sync"] }
 log.workspace = true
 async-trait = "0"
 bytes = "1"
@@ -75,6 +76,7 @@ reqwest = { version = "0.12.0", default-features = false, features = [
     "stream",
 ], optional = true }
 http = { version = "1", optional = true } # Matching what is in reqwest
+urlencoding = { version = "2", optional = true }
 uuid = { version = "1.7.0", features = ["v4", "v5"] }
 polars-arrow = { version = ">=0.37,<0.40.0", optional = true }
 polars = { version = ">=0.37,<0.40.0", optional = true }
@@ -93,6 +95,7 @@ semver = { workspace = true }
 anyhow = "1"
 tempfile = "3.5.0"
 random_word = { version = "0.4.3", features = ["en"] }
+tokio = { version = "1.23", features = ["io-util", "macros", "net", "rt-multi-thread", "sync"] }
 uuid = { version = "1.7.0", features = ["v4"] }
 walkdir = "2"
 aws-sdk-dynamodb = { version = "1.55.0" }
@@ -105,6 +108,8 @@ http-body = "1"                                        # Matching reqwest
 rstest = "0.23.0"
 test-log = "0.2"
 serial_test = "3"
+[target.'cfg(unix)'.dev-dependencies]
+pprof = { version = "0.14", features = ["flamegraph"] }
 
 
 [features]
@@ -129,7 +134,13 @@ huggingface = [
     "lance-namespace-impls/dir-huggingface",
 ]
 dynamodb = ["lance/dynamodb", "aws"]
-remote = ["dep:reqwest", "dep:http", "lance-namespace-impls/rest", "lance-namespace-impls/rest-adapter"]
+remote = [
+    "dep:reqwest",
+    "dep:http",
+    "dep:urlencoding",
+    "lance-namespace-impls/rest",
+    "lance-namespace-impls/rest-adapter",
+]
 fp16kernels = ["lance-linalg/fp16kernels"]
 s3-test = []
 bedrock = ["dep:aws-sdk-bedrockruntime"]
@@ -155,9 +166,16 @@ required-features = ["sentence-transformers"]
 name = "bedrock"
 required-features = ["bedrock"]
 
+[[example]]
+name = "bench_streaming_dataloader"
+
 [[example]]
 name = "simple"
 
+[[example]]
+name = "polars"
+required-features = ["polars"]
+
 [[example]]
 name = "full_text_search"
 

rust/lancedb/examples/bench_streaming_dataloader.rs

@@ -0,0 +1,272 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+//! Benchmark + CPU profiler for the PermutationReader used by the elastic
+//! streaming dataloader.
+//!
+//! Normal sweep:
+//!   cargo run --release --example bench_streaming_dataloader
+//!
+//! Flamegraph (self-contained, no perf/dtrace needed):
+//!   BENCH_PROFILE=1 BENCH_CHUNK=64 cargo run --release \
+//!       --example bench_streaming_dataloader
+//!   # writes flamegraph.svg in the current directory
+//!
+//! Environment variables:
+//!   BENCH_NUM_ROWS   – total rows (default 49152 = 24 × 2048)
+//!   BENCH_NUM_SPLITS – number of splits (default 24)
+//!   BENCH_STEPS      – round-robin cycles per chunk-size trial (default 200)
+//!   BENCH_ROW_BYTES  – bytes of payload per row (default 4096)
+//!   BENCH_CHUNK      – restrict sweep to this single chunk size
+//!   BENCH_PROFILE    – if set to "1", capture a pprof flamegraph SVG
+
+use std::{sync::Arc, time::Instant};
+
+use arrow_array::{Int32Array, LargeBinaryArray, RecordBatch};
+use arrow_schema::{DataType, Field, Schema};
+use lancedb::{
+    Result, Table,
+    arrow::{SendableRecordBatchStream, SimpleRecordBatchStream},
+    connect,
+    dataloader::permutation::{
+        builder::{PermutationBuilder, ShuffleStrategy},
+        reader::PermutationReader,
+        split::{SplitSizes, SplitStrategy},
+    },
+    query::Select,
+};
+
+fn env_usize(key: &str, default: usize) -> usize {
+    std::env::var(key)
+        .ok()
+        .and_then(|v| v.parse().ok())
+        .unwrap_or(default)
+}
+
+// ---------------------------------------------------------------------------
+// Table creation
+// ---------------------------------------------------------------------------
+
+async fn make_base_table(num_rows: usize, row_bytes: usize) -> Result<Table> {
+    let schema = Arc::new(Schema::new(vec![
+        Field::new("id", DataType::Int32, false),
+        Field::new("payload", DataType::LargeBinary, false),
+    ]));
+    let payload = vec![0u8; row_bytes];
+    let ids: Int32Array = (0..num_rows as i32).collect();
+    let payloads: LargeBinaryArray = (0..num_rows).map(|_| Some(payload.as_slice())).collect();
+    let batch = RecordBatch::try_new(schema.clone(), vec![Arc::new(ids), Arc::new(payloads)])?;
+    let stream: SendableRecordBatchStream = Box::pin(SimpleRecordBatchStream::new(
+        futures::stream::once(std::future::ready(Ok(batch))),
+        schema,
+    ));
+    let db = connect("memory:///").execute().await?;
+    db.create_table("base", stream).execute().await
+}
+
+async fn make_permutation_table(base: &Table, num_splits: usize) -> Result<Table> {
+    PermutationBuilder::new(base.clone())
+        .with_split_strategy(
+            SplitStrategy::Random {
+                seed: Some(42),
+                sizes: SplitSizes::Fixed(num_splits as u64),
+                clump_size: None,
+            },
+            None,
+        )
+        .with_shuffle_strategy(ShuffleStrategy::Random {
+            seed: Some(42),
+            clump_size: None,
+        })
+        .build()
+        .await
+}
+
+// ---------------------------------------------------------------------------
+// Round-robin hot loop (mirrors StreamingDataset.__iter__)
+// ---------------------------------------------------------------------------
+
+async fn run_hot_loop(
+    readers: &[PermutationReader],
+    chunk_size: usize,
+    steps: usize,
+) -> Result<(usize, f64)> {
+    let n = readers.len();
+    let split_sizes: Vec<usize> = readers.iter().map(|r| r.count_rows() as usize).collect();
+
+    struct SplitBuf {
+        batch: Option<RecordBatch>,
+        row_in_batch: usize,
+        consumed: usize,
+    }
+    let mut bufs: Vec<SplitBuf> = (0..n)
+        .map(|_| SplitBuf {
+            batch: None,
+            row_in_batch: 0,
+            consumed: 0,
+        })
+        .collect();
+
+    // Pre-fill
+    for i in 0..n {
+        let fetch = chunk_size.min(split_sizes[i]);
+        if fetch > 0 {
+            let offsets: Vec<u64> = (0..fetch as u64).collect();
+            bufs[i].batch = Some(readers[i].take_offsets(&offsets, Select::All).await?);
+        }
+    }
+
+    let mut total_rows = 0usize;
+    let t0 = Instant::now();
+
+    'outer: for _step in 0..steps {
+        for i in 0..n {
+            if bufs[i].consumed >= split_sizes[i] {
+                break 'outer;
+            }
+            let need_refill = bufs[i]
+                .batch
+                .as_ref()
+                .map(|b| bufs[i].row_in_batch >= b.num_rows())
+                .unwrap_or(true);
+            if need_refill {
+                let start = bufs[i].consumed as u64;
+                let remaining = (split_sizes[i] - bufs[i].consumed) as u64;
+                let fetch = chunk_size.min(remaining as usize);
+                let offsets: Vec<u64> = (start..start + fetch as u64).collect();
+                bufs[i].batch = Some(readers[i].take_offsets(&offsets, Select::All).await?);
+                bufs[i].row_in_batch = 0;
+            }
+            bufs[i].row_in_batch += 1;
+            bufs[i].consumed += 1;
+            total_rows += 1;
+        }
+    }
+
+    Ok((total_rows, t0.elapsed().as_secs_f64()))
+}
+
+// ---------------------------------------------------------------------------
+// Main
+// ---------------------------------------------------------------------------
+
+#[tokio::main]
+async fn main() -> Result<()> {
+    let num_splits = env_usize("BENCH_NUM_SPLITS", 24);
+    let num_rows = env_usize("BENCH_NUM_ROWS", num_splits * 2048);
+    let steps = env_usize("BENCH_STEPS", 200);
+    let row_bytes = env_usize("BENCH_ROW_BYTES", 4096);
+    let single_chunk: Option<usize> = std::env::var("BENCH_CHUNK")
+        .ok()
+        .and_then(|v| v.parse().ok());
+    let do_profile = std::env::var("BENCH_PROFILE")
+        .map(|v| v == "1")
+        .unwrap_or(false);
+
+    assert_eq!(
+        num_rows % num_splits,
+        0,
+        "NUM_ROWS must be divisible by NUM_SPLITS"
+    );
+
+    println!("Benchmark config:");
+    println!(
+        "  num_rows={}  num_splits={}  rows/split={}  steps={}  row_bytes={}",
+        num_rows,
+        num_splits,
+        num_rows / num_splits,
+        steps,
+        row_bytes,
+    );
+    println!(
+        "  ~{:.1} MB total",
+        (num_rows * row_bytes) as f64 / (1024.0 * 1024.0)
+    );
+    println!();
+
+    print!("Building base table... ");
+    let _ = std::io::Write::flush(&mut std::io::stdout());
+    let base = make_base_table(num_rows, row_bytes).await?;
+    println!("done");
+
+    print!("Building permutation table... ");
+    let _ = std::io::Write::flush(&mut std::io::stdout());
+    let perm = make_permutation_table(&base, num_splits).await?;
+    println!("done");
+
+    print!("Building {} PermutationReaders... ", num_splits);
+    let _ = std::io::Write::flush(&mut std::io::stdout());
+    let base_inner = base.base_table().clone();
+    let perm_inner = perm.base_table().clone();
+    let mut readers = Vec::with_capacity(num_splits);
+    for split in 0..num_splits {
+        readers.push(
+            PermutationReader::try_from_tables(
+                base_inner.clone(),
+                perm_inner.clone(),
+                split as u64,
+            )
+            .await?,
+        );
+    }
+    println!("done ({} rows/split)", readers[0].count_rows());
+    println!();
+
+    let chunk_sizes: Vec<usize> = if let Some(c) = single_chunk {
+        vec![c]
+    } else {
+        vec![1, 4, 16, 64, 256, 1024, 4096, 16384]
+    };
+
+    if do_profile {
+        #[cfg(unix)]
+        {
+            let chunk = chunk_sizes[0];
+            println!("Profiling chunk={chunk} for {steps} steps...");
+            // Warm-up outside the profiler window
+            let _ = run_hot_loop(&readers, chunk, 1).await?;
+
+            let guard = pprof::ProfilerGuardBuilder::default()
+                .frequency(1000)
+                .build()
+                .unwrap();
+
+            let (rows, elapsed) = run_hot_loop(&readers, chunk, steps).await?;
+
+            if let Ok(report) = guard.report().build() {
+                let svg_path = "flamegraph.svg";
+                let file = std::fs::File::create(svg_path).unwrap();
+                report.flamegraph(file).unwrap();
+                println!("Flamegraph written to {svg_path}");
+            }
+
+            let rows_per_sec = rows as f64 / elapsed;
+            println!("chunk={chunk}  {rows} rows  {elapsed:.3}s  {rows_per_sec:.0} rows/s");
+        }
+        #[cfg(not(unix))]
+        {
+            println!("Flamegraph profiling (BENCH_PROFILE=1) is not supported on this platform.");
+            println!("Run without BENCH_PROFILE to get throughput numbers.");
+        }
+    } else {
+        println!(
+            "{:>6}  {:>7}  {:>8}  {:>11}  {:>10}",
+            "chunk", "rows", "elapsed", "rows/s", "ms/step"
+        );
+        println!("{}", "-".repeat(52));
+
+        for &chunk in &chunk_sizes {
+            let _ = run_hot_loop(&readers, chunk, 1).await?;
+            let (rows, elapsed) = run_hot_loop(&readers, chunk, steps).await?;
+            let rows_per_sec = rows as f64 / elapsed;
+            let ms_per_step = elapsed / steps as f64 * 1000.0;
+            println!(
+                "{:>6}  {:>7}  {:>7.3}s  {:>11.0}  {:>9.1}ms",
+                chunk, rows, elapsed, rows_per_sec, ms_per_step,
+            );
+        }
+    }
+
+    println!("\nDone.");
+    Ok(())
+}

rust/lancedb/examples/polars.rs

@@ -0,0 +1,47 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+//! This example demonstrates ingesting a Polars DataFrame into LanceDB and
+//! reading it back out as a Polars DataFrame.
+
+use lancedb::arrow::IntoPolars;
+use lancedb::query::ExecutableQuery;
+use lancedb::{Result, connect};
+use polars::prelude::{DataFrame, NamedFrom, Series};
+
+fn make_dataframe() -> DataFrame {
+    let ids = Series::new("id", &[1i32, 2, 3, 4, 5]);
+    let names = Series::new("name", &["Alice", "Bob", "Carol", "Dave", "Eve"]);
+    let scores = Series::new("score", &[9.5f64, 8.1, 7.3, 9.0, 6.5]);
+    DataFrame::new(vec![ids, names, scores]).unwrap()
+}
+
+#[tokio::main]
+async fn main() -> Result<()> {
+    let tmp = tempfile::tempdir().unwrap();
+    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
+
+    // Ingest a Polars DataFrame directly — DataFrame now implements Scannable.
+    let df = make_dataframe();
+    println!("Input DataFrame:\n{df}");
+
+    let table = db.create_table("people", df).execute().await?;
+
+    // Append more rows.
+    let more = DataFrame::new(vec![
+        Series::new("id", &[6i32, 7]),
+        Series::new("name", &["Frank", "Grace"]),
+        Series::new("score", &[7.8f64, 8.9]),
+    ])
+    .unwrap();
+    table.add(more).execute().await?;
+
+    // Read back as a Polars DataFrame.
+    let result_df = table.query().execute().await?.into_polars().await?;
+
+    println!(
+        "\nRound-tripped DataFrame ({} rows):\n{result_df}",
+        result_df.height()
+    );
+    Ok(())
+}

rust/lancedb/src/arrow.rs

@@ -3,7 +3,19 @@
 
 use std::{pin::Pin, sync::Arc};
 
+// Re-export the arrow crates we depend on so downstream consumers can build
+// `RecordBatch`/arrays/builders against the exact same arrow line lancedb was
+// compiled against, instead of declaring their own (potentially mismatched)
+// direct arrow dependencies. See https://github.com/lancedb/lancedb/issues/3575.
+pub use arrow;
+pub use arrow_array;
+pub use arrow_buffer;
+pub use arrow_cast;
+pub use arrow_data;
+pub use arrow_ipc;
+pub use arrow_ord;
 pub use arrow_schema;
+pub use arrow_select;
 use datafusion_common::DataFusionError;
 use datafusion_physical_plan::stream::RecordBatchStreamAdapter;
 use futures::{Stream, StreamExt, TryStreamExt};
@@ -112,54 +124,14 @@ impl<S: Stream<Item = Result<arrow_array::RecordBatch>>> RecordBatchStream
 
 /// A trait for converting incoming data to Arrow
 ///
-/// Integrations should implement this trait to allow data to be
-/// imported directly from the integration.  For example, implementing
-/// this trait for `Vec<Vec<...>>` would allow the `Vec` to be directly
-/// used in methods like [`crate::connection::Connection::create_table`]
-/// or [`crate::table::Table::add`]
-pub trait IntoArrow {
-    /// Convert the data into an iterator of Arrow batches
-    fn into_arrow(self) -> Result<Box<dyn arrow_array::RecordBatchReader + Send>>;
-}
-
 pub type BoxedRecordBatchReader = Box<dyn arrow_array::RecordBatchReader + Send>;
 
-impl<T: arrow_array::RecordBatchReader + Send + 'static> IntoArrow for T {
-    fn into_arrow(self) -> Result<Box<dyn arrow_array::RecordBatchReader + Send>> {
-        Ok(Box::new(self))
-    }
-}
-
-/// A trait for converting incoming data to Arrow asynchronously
-///
-/// Serves the same purpose as [`IntoArrow`], but for asynchronous data.
-///
-/// Note: Arrow has no async equivalent to RecordBatchReader and so
-pub trait IntoArrowStream {
-    /// Convert the data into a stream of Arrow batches
-    fn into_arrow(self) -> Result<SendableRecordBatchStream>;
-}
-
 impl<S: Stream<Item = Result<arrow_array::RecordBatch>>> SimpleRecordBatchStream<S> {
     pub fn new(stream: S, schema: Arc<arrow_schema::Schema>) -> Self {
         Self { schema, stream }
     }
 }
 
-impl IntoArrowStream for SendableRecordBatchStream {
-    fn into_arrow(self) -> Result<SendableRecordBatchStream> {
-        Ok(self)
-    }
-}
-
-impl IntoArrowStream for datafusion_physical_plan::SendableRecordBatchStream {
-    fn into_arrow(self) -> Result<SendableRecordBatchStream> {
-        let schema = self.schema();
-        let stream = self.map_err(|df_err| df_err.into());
-        Ok(Box::pin(SimpleRecordBatchStream::new(stream, schema)))
-    }
-}
-
 pub trait LanceDbDatagenExt {
     fn into_ldb_stream(
         self,
@@ -264,9 +236,7 @@ impl IntoPolars for SendableRecordBatchStream {
 #[cfg(all(test, feature = "polars"))]
 mod tests {
     use super::SendableRecordBatchStream;
-    use crate::arrow::{
-        IntoArrow, IntoPolars, PolarsDataFrameRecordBatchReader, SimpleRecordBatchStream,
-    };
+    use crate::arrow::{IntoPolars, PolarsDataFrameRecordBatchReader, SimpleRecordBatchStream};
     use polars::prelude::{DataFrame, NamedFrom, Series};
 
     fn get_record_batch_reader_from_polars() -> Box<dyn arrow_array::RecordBatchReader + Send> {
@@ -280,10 +250,7 @@ mod tests {
         float_series = Series::new("float", &[2.0]);
         let df2 = DataFrame::new(vec![string_series, int_series, float_series]).unwrap();
 
-        PolarsDataFrameRecordBatchReader::new(df1.vstack(&df2).unwrap())
-            .unwrap()
-            .into_arrow()
-            .unwrap()
+        Box::new(PolarsDataFrameRecordBatchReader::new(df1.vstack(&df2).unwrap()).unwrap())
     }
 
     #[test]

rust/lancedb/src/connection.rs

@@ -667,6 +667,8 @@ pub struct ConnectRequest {
 pub struct ConnectBuilder {
     request: ConnectRequest,
     embedding_registry: Option<Arc<dyn EmbeddingRegistry>>,
+    #[cfg(feature = "remote")]
+    oauth_config: Option<crate::remote::OAuthConfig>,
 }
 
 #[cfg(feature = "remote")]
@@ -688,6 +690,8 @@ impl ConnectBuilder {
                 session: None,
             },
             embedding_registry: None,
+            #[cfg(feature = "remote")]
+            oauth_config: None,
         }
     }
 
@@ -776,6 +780,19 @@ impl ConnectBuilder {
         self
     }
 
+    /// Configure OAuth authentication for LanceDB Cloud/Enterprise.
+    ///
+    /// This creates an [`OAuthHeaderProvider`](crate::remote::OAuthHeaderProvider)
+    /// from the given config and sets it as the header provider. OAuth cannot
+    /// be combined with an API key or another header provider.
+    ///
+    /// Token acquisition and refresh are handled in Rust.
+    #[cfg(feature = "remote")]
+    pub fn oauth_config(mut self, config: crate::remote::OAuthConfig) -> Self {
+        self.oauth_config = Some(config);
+        self
+    }
+
     /// Provide a custom [`EmbeddingRegistry`] to use for this connection.
     pub fn embedding_registry(mut self, registry: Arc<dyn EmbeddingRegistry>) -> Self {
         self.embedding_registry = Some(registry);
@@ -921,9 +938,40 @@ impl ConnectBuilder {
         let region = options.region.ok_or_else(|| Error::InvalidInput {
             message: "A region is required when connecting to LanceDb Cloud".to_string(),
         })?;
-        let api_key = options.api_key.ok_or_else(|| Error::InvalidInput {
-            message: "An api_key is required when connecting to LanceDb Cloud".to_string(),
-        })?;
+        let api_key = match (&self.oauth_config, &options.api_key) {
+            (Some(_), None) => String::new(),
+            (Some(_), Some(_)) => {
+                return Err(Error::InvalidInput {
+                    message:
+                        "api_key and oauth_config cannot both be set when connecting to LanceDb Cloud"
+                            .to_string(),
+                });
+            }
+            (None, Some(key)) => key.clone(),
+            (None, None) => {
+                return Err(Error::InvalidInput {
+                    message:
+                        "An api_key or oauth_config is required when connecting to LanceDb Cloud"
+                            .to_string(),
+                });
+            }
+        };
+
+        if self.oauth_config.is_some() && self.request.client_config.header_provider.is_some() {
+            return Err(Error::InvalidInput {
+                message:
+                    "oauth_config and client_config.header_provider cannot both be set when connecting to LanceDb Cloud"
+                        .to_string(),
+            });
+        }
+
+        let mut client_config = self.request.client_config;
+
+        if let Some(oauth_config) = self.oauth_config {
+            let provider = crate::remote::OAuthHeaderProvider::new(oauth_config)?;
+            client_config.header_provider =
+                Some(Arc::new(provider) as Arc<dyn crate::remote::HeaderProvider>);
+        }
 
         let storage_options = StorageOptions(options.storage_options.clone());
         let internal = Arc::new(crate::remote::db::RemoteDatabase::try_new(
@@ -931,7 +979,7 @@ impl ConnectBuilder {
             &api_key,
             &region,
             options.host_override,
-            self.request.client_config,
+            client_config,
             storage_options.into(),
             self.request.read_consistency_interval,
         )?);
@@ -1240,6 +1288,83 @@ mod tests {
         assert_eq!(Some(&"EXPLICIT-VALUE".to_string()), options.get(opts_key));
     }
 
+    #[cfg(feature = "remote")]
+    #[tokio::test]
+    async fn test_connect_rejects_api_key_with_oauth_config() {
+        let oauth_config = crate::remote::OAuthConfig {
+            issuer_url: "https://issuer.example.com".to_string(),
+            client_id: "client-id".to_string(),
+            client_secret: Some("secret".to_string()),
+            scopes: vec!["scope".to_string()],
+            flow: crate::remote::OAuthFlow::ClientCredentials,
+            refresh_buffer_secs: None,
+        };
+
+        let result = ConnectBuilder::new("db://my-container/my-prefix")
+            .region("us-east-1")
+            .api_key("my-api-key")
+            .oauth_config(oauth_config)
+            .execute()
+            .await;
+
+        match result {
+            Err(Error::InvalidInput { message })
+                if message
+                    == "api_key and oauth_config cannot both be set when connecting to LanceDb Cloud" =>
+                {}
+            Err(err) => panic!("expected InvalidInput, got {err:?}"),
+            Ok(_) => panic!("expected api_key and oauth_config to be rejected"),
+        }
+    }
+
+    #[cfg(feature = "remote")]
+    #[tokio::test]
+    async fn test_connect_rejects_header_provider_with_oauth_config() {
+        #[derive(Debug)]
+        struct TestHeaderProvider;
+
+        #[async_trait::async_trait]
+        impl crate::remote::HeaderProvider for TestHeaderProvider {
+            async fn get_headers(&self) -> Result<HashMap<String, String>> {
+                Ok(HashMap::from([(
+                    "authorization".to_string(),
+                    "Bearer token".to_string(),
+                )]))
+            }
+        }
+
+        let oauth_config = crate::remote::OAuthConfig {
+            issuer_url: "https://issuer.example.com".to_string(),
+            client_id: "client-id".to_string(),
+            client_secret: Some("secret".to_string()),
+            scopes: vec!["scope".to_string()],
+            flow: crate::remote::OAuthFlow::ClientCredentials,
+            refresh_buffer_secs: None,
+        };
+        let client_config = crate::remote::ClientConfig {
+            header_provider: Some(
+                Arc::new(TestHeaderProvider) as Arc<dyn crate::remote::HeaderProvider>
+            ),
+            ..Default::default()
+        };
+
+        let result = ConnectBuilder::new("db://my-container/my-prefix")
+            .region("us-east-1")
+            .client_config(client_config)
+            .oauth_config(oauth_config)
+            .execute()
+            .await;
+
+        match result {
+            Err(Error::InvalidInput { message })
+                if message
+                    == "oauth_config and client_config.header_provider cannot both be set when connecting to LanceDb Cloud" =>
+                {}
+            Err(err) => panic!("expected InvalidInput, got {err:?}"),
+            Ok(_) => panic!("expected header_provider and oauth_config to be rejected"),
+        }
+    }
+
     #[cfg(not(windows))]
     #[tokio::test]
     async fn test_connect_relative() {

rust/lancedb/src/data/scannable.rs

@@ -185,6 +185,43 @@ impl Scannable for SendableRecordBatchStream {
     }
 }
 
+#[cfg(feature = "polars")]
+impl Scannable for polars::frame::DataFrame {
+    fn schema(&self) -> SchemaRef {
+        crate::polars_arrow_convertors::convert_polars_df_schema_to_arrow_rb_schema(
+            self.schema().clone(),
+        )
+        .expect("failed to convert Polars DataFrame schema to Arrow schema")
+    }
+
+    fn scan_as_stream(&mut self) -> SendableRecordBatchStream {
+        let schema = Scannable::schema(self);
+        let batches: crate::Result<Vec<RecordBatch>> =
+            match crate::arrow::PolarsDataFrameRecordBatchReader::new(self.clone()) {
+                Err(e) => Err(e),
+                Ok(reader) => reader.map(|b| b.map_err(Into::into)).collect(),
+            };
+        match batches {
+            Err(e) => Box::pin(SimpleRecordBatchStream {
+                schema,
+                stream: once(async move { Err(e) }),
+            }),
+            Ok(batches) => {
+                let stream = futures::stream::iter(batches.into_iter().map(Ok));
+                Box::pin(SimpleRecordBatchStream { schema, stream })
+            }
+        }
+    }
+
+    fn num_rows(&self) -> Option<usize> {
+        Some(self.height())
+    }
+
+    fn rescannable(&self) -> bool {
+        true
+    }
+}
+
 #[async_trait]
 impl StreamingWriteSource for Box<dyn Scannable> {
     fn arrow_schema(&self) -> SchemaRef {
@@ -1089,4 +1126,60 @@ mod tests {
             );
         }
     }
+
+    #[cfg(feature = "polars")]
+    mod polars_tests {
+        use super::*;
+        use crate::arrow::IntoPolars;
+        use crate::query::ExecutableQuery;
+        use polars::prelude::{DataFrame, NamedFrom, Series};
+
+        fn make_df() -> DataFrame {
+            DataFrame::new(vec![
+                Series::new("id", &[1i32, 2, 3]),
+                Series::new("val", &[1.1f64, 2.2, 3.3]),
+            ])
+            .unwrap()
+        }
+
+        #[tokio::test]
+        async fn test_dataframe_scannable_round_trip() {
+            let tmp = tempfile::tempdir().unwrap();
+            let db = crate::connect(tmp.path().to_str().unwrap())
+                .execute()
+                .await
+                .unwrap();
+
+            let df = make_df();
+            let table = db.create_table("t", df.clone()).execute().await.unwrap();
+
+            // Append the same rows again.
+            table.add(df.clone()).execute().await.unwrap();
+
+            let result = table
+                .query()
+                .execute()
+                .await
+                .unwrap()
+                .into_polars()
+                .await
+                .unwrap();
+
+            assert_eq!(result.height(), df.height() * 2);
+            assert_eq!(result.schema(), df.schema());
+        }
+
+        #[tokio::test]
+        async fn test_dataframe_scannable_rescannable() {
+            let mut df = make_df();
+            assert!(df.rescannable());
+
+            let batches1: Vec<RecordBatch> = df.scan_as_stream().try_collect().await.unwrap();
+            assert_eq!(batches1.iter().map(|b| b.num_rows()).sum::<usize>(), 3);
+
+            // Can be scanned again.
+            let batches2: Vec<RecordBatch> = df.scan_as_stream().try_collect().await.unwrap();
+            assert_eq!(batches2.iter().map(|b| b.num_rows()).sum::<usize>(), 3);
+        }
+    }
 }

rust/lancedb/src/dataloader/permutation/builder.rs

@@ -391,6 +391,7 @@ mod tests {
                 SplitStrategy::Random {
                     seed: Some(42),
                     sizes: SplitSizes::Percentages(vec![0.05, 0.30]),
+                    clump_size: None,
                 },
                 None,
             )

rust/lancedb/src/dataloader/permutation/reader.rs

@@ -21,6 +21,7 @@ use arrow::compute::concat_batches;
 use arrow::datatypes::UInt64Type;
 use arrow_array::{RecordBatch, UInt64Array};
 use arrow_schema::SchemaRef;
+use datafusion_expr::{Expr, col, lit};
 use futures::{StreamExt, TryStreamExt};
 use lance::dataset::scanner::DatasetRecordBatchStream;
 use lance::io::RecordBatchStream;
@@ -196,17 +197,10 @@ impl PermutationReader {
             .expect_ok()?
             .values();
 
-        let filter = format!(
-            "_rowid in ({})",
-            row_ids
-                .iter()
-                .map(|o| o.to_string())
-                .collect::<Vec<_>>()
-                .join(",")
-        );
+        let in_list: Vec<Expr> = row_ids.iter().map(|id| lit(*id)).collect();
 
         let base_query = QueryRequest {
-            filter: Some(QueryFilter::Sql(filter)),
+            filter: Some(QueryFilter::Datafusion(col(ROW_ID).in_list(in_list, false))),
             select: selection,
             with_row_id: true,
             ..Default::default()

rust/lancedb/src/dataloader/permutation/split.rs

@@ -35,9 +35,13 @@ pub enum SplitStrategy {
     /// Rows will be randomly assigned to splits
     ///
     /// A seed can be provided to make the assignment deterministic.
+    ///
+    /// A clump_size can be provided to shuffle contiguous groups of rows together,
+    /// preserving I/O locality while still randomising the split assignment.
     Random {
         seed: Option<u64>,
         sizes: SplitSizes,
+        clump_size: Option<u64>,
     },
     /// Rows will be assigned to splits based on the values in the specified columns.
     ///
@@ -323,13 +327,17 @@ impl Splitter {
                 self.apply_sequential(source, num_rows, &SplitSizes::Counts(vec![num_rows]))
                     .await
             }
-            SplitStrategy::Random { seed, sizes } => {
+            SplitStrategy::Random {
+                seed,
+                sizes,
+                clump_size,
+            } => {
                 let shuffler = Shuffler::new(ShufflerConfig {
                     seed: *seed,
                     // In this case we are only shuffling row ids so we can use a large max_rows_per_file
                     max_rows_per_file: 10 * 1024 * 1024,
                     temp_dir: self.temp_dir.clone(),
-                    clump_size: None,
+                    clump_size: *clump_size,
                 });
 
                 let shuffled = shuffler.shuffle(source, num_rows).await?;
@@ -692,6 +700,7 @@ mod tests {
             SplitStrategy::Random {
                 seed: Some(42),
                 sizes: SplitSizes::Fixed(3),
+                clump_size: None,
             },
         );
 
@@ -718,6 +727,7 @@ mod tests {
             SplitStrategy::Random {
                 seed: Some(42),
                 sizes: SplitSizes::Counts(vec![5, 15, 10]),
+                clump_size: None,
             },
         );
 
@@ -744,6 +754,7 @@ mod tests {
             SplitStrategy::Random {
                 seed: Some(42),
                 sizes: SplitSizes::Percentages(vec![0.217, 0.168, 0.17]),
+                clump_size: None,
             },
         );
 

rust/lancedb/src/lib.rs

@@ -342,3 +342,9 @@ pub use connection::connect_namespace;
 /// Re-export Lance Session and ObjectStoreRegistry for custom session creation
 pub use lance::session::Session;
 pub use lance_io::object_store::ObjectStoreRegistry;
+
+/// Re-export DataFusion so consumers can build the `Expr` values that public
+/// query/merge APIs (e.g. [`query::QueryBase::only_if_expr`]) accept without
+/// declaring their own (potentially mismatched) direct `datafusion` dependency.
+/// See <https://github.com/lancedb/lancedb/issues/3575>.
+pub use datafusion;

rust/lancedb/src/query.rs

@@ -7,7 +7,7 @@ use std::{future::Future, time::Duration};
 use arrow::compute::concat_batches;
 use arrow_array::{Array, Float16Array, Float32Array, Float64Array, RecordBatch, make_array};
 use arrow_schema::{DataType, SchemaRef};
-use datafusion_expr::Expr;
+use datafusion_expr::{Expr, col, lit};
 use datafusion_physical_plan::ExecutionPlan;
 use futures::{FutureExt, TryFutureExt, TryStreamExt, stream, try_join};
 use half::f16;
@@ -401,6 +401,9 @@ pub trait QueryBase {
     ///
     /// Filtering performance can often be improved by creating a scalar index
     /// on the filter column(s).
+    ///
+    /// Calling this multiple times combines the filters with a logical AND
+    /// (i.e. `(previous) AND (new)`) rather than replacing the previous filter.
     fn only_if(self, filter: impl AsRef<str>) -> Self;
 
     /// Only return rows which match the filter, using an expression builder.
@@ -423,6 +426,9 @@ pub trait QueryBase {
     ///
     /// Note: Expression filters are not supported for remote/server-side queries.
     /// Use [`QueryBase::only_if`] with SQL strings for remote tables.
+    ///
+    /// Calling this multiple times combines the expressions with a logical AND
+    /// rather than replacing the previous filter.
     fn only_if_expr(self, filter: datafusion_expr::Expr) -> Self;
 
     /// Perform a full text search on the table.
@@ -535,12 +541,13 @@ impl<T: HasQuery> QueryBase for T {
     }
 
     fn only_if(mut self, filter: impl AsRef<str>) -> Self {
-        self.mut_query().filter = Some(QueryFilter::Sql(filter.as_ref().to_string()));
+        self.mut_query()
+            .add_filter(QueryFilter::Sql(filter.as_ref().to_string()));
         self
     }
 
     fn only_if_expr(mut self, filter: datafusion_expr::Expr) -> Self {
-        self.mut_query().filter = Some(QueryFilter::Datafusion(filter));
+        self.mut_query().add_filter(QueryFilter::Datafusion(filter));
         self
     }
 
@@ -716,6 +723,39 @@ pub enum QueryFilter {
     Datafusion(Expr),
 }
 
+/// Combine two filters with a logical AND.
+///
+/// This is used when a query receives more than one filter (for example when
+/// `where`/`only_if` is called multiple times) so the filters are composed
+/// with AND rather than the later filter silently replacing the earlier one.
+///
+/// SQL string and expression filters are combined within their own
+/// representation. When the two representations are mixed, the expression is
+/// lowered to SQL (via [`crate::expr::expr_to_sql_string`]) and the filters are
+/// combined as SQL strings. Substrait filters cannot be combined and return an
+/// error.
+fn and_filters(existing: QueryFilter, new: QueryFilter) -> Result<QueryFilter> {
+    match (existing, new) {
+        (QueryFilter::Sql(lhs), QueryFilter::Sql(rhs)) => {
+            Ok(QueryFilter::Sql(format!("({lhs}) AND ({rhs})")))
+        }
+        (QueryFilter::Datafusion(lhs), QueryFilter::Datafusion(rhs)) => {
+            Ok(QueryFilter::Datafusion(lhs.and(rhs)))
+        }
+        (QueryFilter::Sql(lhs), QueryFilter::Datafusion(rhs)) => {
+            let rhs = crate::expr::expr_to_sql_string(&rhs)?;
+            Ok(QueryFilter::Sql(format!("({lhs}) AND ({rhs})")))
+        }
+        (QueryFilter::Datafusion(lhs), QueryFilter::Sql(rhs)) => {
+            let lhs = crate::expr::expr_to_sql_string(&lhs)?;
+            Ok(QueryFilter::Sql(format!("({lhs}) AND ({rhs})")))
+        }
+        _ => Err(Error::InvalidInput {
+            message: "cannot combine a Substrait filter with another filter".to_string(),
+        }),
+    }
+}
+
 /// A basic query into a table without any kind of search
 ///
 /// This will result in a (potentially filtered) scan if executed
@@ -730,6 +770,13 @@ pub struct QueryRequest {
     /// Apply filter to the returned rows.
     pub filter: Option<QueryFilter>,
 
+    /// An error recorded while combining repeated filters that could not be
+    /// composed (see [`QueryRequest::add_filter`]). It is surfaced when the
+    /// query is executed via [`QueryRequest::check_filter`]. We defer the error
+    /// because the builder methods that set filters return `Self` rather than a
+    /// `Result`.
+    pub(crate) filter_error: Option<String>,
+
     /// Perform a full text search on the table.
     pub full_text_search: Option<FullTextSearchQuery>,
 
@@ -775,6 +822,7 @@ impl Default for QueryRequest {
             limit: None,
             offset: None,
             filter: None,
+            filter_error: None,
             full_text_search: None,
             select: Select::All,
             fast_search: false,
@@ -788,6 +836,41 @@ impl Default for QueryRequest {
     }
 }
 
+impl QueryRequest {
+    /// Add a filter, combining it with any existing filter using a logical AND.
+    ///
+    /// If the new filter cannot be combined with the existing one (because they
+    /// use different representations) the error is recorded and surfaced later
+    /// by [`Self::check_filter`].
+    pub(crate) fn add_filter(&mut self, new: QueryFilter) {
+        self.filter = Some(match self.filter.take() {
+            None => new,
+            Some(existing) => match and_filters(existing, new) {
+                Ok(combined) => combined,
+                Err(err) => {
+                    // The filters were consumed while attempting to combine
+                    // them; the recorded error is surfaced by `check_filter`
+                    // before the query executes.
+                    self.filter_error = Some(err.to_string());
+                    return;
+                }
+            },
+        });
+    }
+
+    /// Return an error if combining filters failed (see [`Self::add_filter`]).
+    ///
+    /// This must be called by every backend before executing a query.
+    pub(crate) fn check_filter(&self) -> Result<()> {
+        if let Some(message) = &self.filter_error {
+            return Err(Error::InvalidInput {
+                message: message.clone(),
+            });
+        }
+        Ok(())
+    }
+}
+
 /// A builder for LanceDB queries.
 ///
 /// See [`crate::Table::query`] for more details on queries
@@ -1385,18 +1468,13 @@ impl TakeQuery {
     ///
     /// See [`crate::Table::take_offsets`] for more details.
     pub fn from_offsets(parent: Arc<dyn BaseTable>, offsets: Vec<u64>) -> Self {
-        let filter = format!(
-            "_rowoffset in ({})",
-            offsets
-                .iter()
-                .map(|o| o.to_string())
-                .collect::<Vec<_>>()
-                .join(",")
-        );
+        let in_list: Vec<Expr> = offsets.iter().map(|o| lit(*o)).collect();
         Self {
             parent,
             request: QueryRequest {
-                filter: Some(QueryFilter::Sql(filter)),
+                filter: Some(QueryFilter::Datafusion(
+                    col("_rowoffset").in_list(in_list, false),
+                )),
                 ..Default::default()
             },
         }
@@ -1406,18 +1484,11 @@ impl TakeQuery {
     ///
     /// See [`crate::Table::take_row_ids`] for more details.
     pub fn from_row_ids(parent: Arc<dyn BaseTable>, row_ids: Vec<u64>) -> Self {
-        let filter = format!(
-            "_rowid in ({})",
-            row_ids
-                .iter()
-                .map(|o| o.to_string())
-                .collect::<Vec<_>>()
-                .join(",")
-        );
+        let in_list: Vec<Expr> = row_ids.iter().map(|id| lit(*id)).collect();
         Self {
             parent,
             request: QueryRequest {
-                filter: Some(QueryFilter::Sql(filter)),
+                filter: Some(QueryFilter::Datafusion(col(ROW_ID).in_list(in_list, false))),
                 ..Default::default()
             },
         }
@@ -1682,6 +1753,70 @@ mod tests {
         }
     }
 
+    #[tokio::test]
+    async fn test_repeated_only_if_combines_with_and() {
+        use crate::expr::{col, lit};
+
+        let tmp_dir = tempdir().unwrap();
+        let dataset_path = tmp_dir.path().join("test.lance");
+        let uri = dataset_path.to_str().unwrap();
+
+        let conn = connect(uri).execute().await.unwrap();
+        let table = conn
+            .create_table("my_table", make_non_empty_batches())
+            .execute()
+            .await
+            .unwrap();
+
+        let query = table.query().only_if("id > 0").only_if("id < 100");
+        match &query.request.filter {
+            Some(QueryFilter::Sql(sql)) => assert_eq!(sql, "(id > 0) AND (id < 100)"),
+            other => panic!("expected combined SQL filter, got {other:?}"),
+        }
+
+        // A single filter is left untouched.
+        let query = table.query().only_if("id > 0");
+        match &query.request.filter {
+            Some(QueryFilter::Sql(sql)) => assert_eq!(sql, "id > 0"),
+            other => panic!("expected single SQL filter, got {other:?}"),
+        }
+
+        // Expression filters are combined with a logical AND as well.
+        let query = table
+            .query()
+            .only_if_expr(col("id").gt(lit(0i32)))
+            .only_if_expr(col("id").lt(lit(100i32)));
+        match &query.request.filter {
+            Some(QueryFilter::Datafusion(expr)) => {
+                assert_eq!(
+                    expr,
+                    &col("id").gt(lit(0i32)).and(col("id").lt(lit(100i32)))
+                );
+            }
+            other => panic!("expected combined Datafusion filter, got {other:?}"),
+        }
+
+        // Mixing an SQL string filter with an expression filter lowers the
+        // expression to SQL and combines them as SQL strings.
+        let query = table
+            .query()
+            .only_if("id > 0")
+            .only_if_expr(col("id").lt(lit(100i32)));
+        match &query.request.filter {
+            Some(QueryFilter::Sql(sql)) => {
+                let expected = format!(
+                    "(id > 0) AND ({})",
+                    crate::expr::expr_to_sql_string(&col("id").lt(lit(100i32))).unwrap()
+                );
+                assert_eq!(sql, &expected);
+            }
+            other => panic!("expected combined SQL filter, got {other:?}"),
+        }
+        assert!(query.request.check_filter().is_ok());
+        // The combined filter executes without error.
+        query.execute().await.unwrap();
+    }
+
     #[tokio::test]
     async fn test_select_with_transform() {
         // TODO: Switch back to memory://foo after https://github.com/lancedb/lancedb/issues/1051

rust/lancedb/src/remote.rs

@@ -8,6 +8,7 @@
 
 pub(crate) mod client;
 pub(crate) mod db;
+pub mod oauth;
 mod retry;
 pub(crate) mod table;
 pub(crate) mod util;
@@ -20,3 +21,4 @@ const JSON_CONTENT_TYPE: &str = "application/json";
 
 pub use client::{ClientConfig, HeaderProvider, RetryConfig, TimeoutConfig, TlsConfig};
 pub use db::{RemoteDatabaseOptions, RemoteDatabaseOptionsBuilder};
+pub use oauth::{OAuthConfig, OAuthFlow, OAuthHeaderProvider};

rust/lancedb/src/remote/oauth.rs

@@ -0,0 +1,907 @@
+// SPDX-License-Identifier: Apache-2.0
+// SPDX-FileCopyrightText: Copyright The LanceDB Authors
+
+use std::collections::HashMap;
+use std::net::IpAddr;
+use std::sync::Arc;
+use std::time::{Duration, Instant};
+
+use async_trait::async_trait;
+use log::debug;
+use reqwest::Client;
+use serde::Deserialize;
+use tokio::sync::RwLock;
+
+use crate::error::{Error, Result};
+use crate::remote::client::HeaderProvider;
+
+const DEFAULT_REFRESH_BUFFER_SECS: u64 = 300;
+const DEFAULT_TOKEN_TTL_SECS: u64 = 3600;
+const AZURE_IMDS_ENDPOINT: &str = "http://169.254.169.254/metadata/identity/oauth2/token";
+const AZURE_IMDS_API_VERSION: &str = "2018-02-01";
+
+/// OAuth authentication flow configuration.
+#[derive(Debug, Clone)]
+pub enum OAuthFlow {
+    /// Client Credentials grant (service-to-service / M2M).
+    /// Requires `client_secret` in [`OAuthConfig`].
+    ClientCredentials,
+
+    /// Azure Managed Identity via IMDS.
+    /// Works on Azure VMs, AKS, App Service, and Azure Functions.
+    /// IMDS requests bypass proxy settings because the endpoint is link-local.
+    AzureManagedIdentity {
+        /// Client ID for user-assigned managed identity.
+        /// Omit for system-assigned managed identity.
+        client_id: Option<String>,
+    },
+}
+
+/// OAuth configuration for LanceDB authentication.
+///
+/// All token acquisition and refresh is handled in the Rust layer.
+/// Python and TypeScript bindings expose this as a plain config object.
+#[derive(Clone)]
+pub struct OAuthConfig {
+    /// OIDC issuer URL or OAuth authority URL.
+    /// For Azure: `https://login.microsoftonline.com/{tenant_id}/v2.0`
+    pub issuer_url: String,
+
+    /// Application / Client ID.
+    pub client_id: String,
+
+    /// Client secret (required for `ClientCredentials`, optional for others).
+    pub client_secret: Option<String>,
+
+    /// OAuth scopes to request.
+    /// For Azure managed identity, exactly one scope or resource is required.
+    /// For example: `["api://{app_id}/.default"]`
+    pub scopes: Vec<String>,
+
+    /// Authentication flow to use.
+    pub flow: OAuthFlow,
+
+    /// Seconds before token expiry to trigger proactive refresh (default: 300).
+    /// Keep this well below the token TTL; if it is greater than or equal to
+    /// the TTL, each request refreshes the token.
+    pub refresh_buffer_secs: Option<u64>,
+}
+
+impl std::fmt::Debug for OAuthConfig {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("OAuthConfig")
+            .field("issuer_url", &self.issuer_url)
+            .field("client_id", &self.client_id)
+            .field(
+                "client_secret",
+                &self.client_secret.as_deref().map(|_| "<redacted>"),
+            )
+            .field("scopes", &self.scopes)
+            .field("flow", &self.flow)
+            .field("refresh_buffer_secs", &self.refresh_buffer_secs)
+            .finish()
+    }
+}
+
+// -- OIDC Discovery --
+
+#[derive(Clone, Debug, Deserialize)]
+struct OidcDiscovery {
+    token_endpoint: String,
+}
+
+// -- Token Response --
+
+#[derive(Deserialize)]
+struct TokenResponse {
+    access_token: String,
+    /// Token lifetime in seconds.
+    /// Some providers (Azure IMDS) return this as a string, so we accept both.
+    #[serde(default, deserialize_with = "deserialize_optional_u64_or_string")]
+    expires_in: Option<u64>,
+    #[serde(default)]
+    #[allow(dead_code)]
+    token_type: Option<String>,
+}
+
+impl std::fmt::Debug for TokenResponse {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("TokenResponse")
+            .field("access_token", &"<redacted>")
+            .field("expires_in", &self.expires_in)
+            .field("token_type", &self.token_type)
+            .finish()
+    }
+}
+
+fn deserialize_optional_u64_or_string<'de, D>(
+    deserializer: D,
+) -> std::result::Result<Option<u64>, D::Error>
+where
+    D: serde::Deserializer<'de>,
+{
+    use serde::de;
+
+    struct U64OrString;
+    impl<'de> de::Visitor<'de> for U64OrString {
+        type Value = Option<u64>;
+
+        fn expecting(&self, formatter: &mut std::fmt::Formatter) -> std::fmt::Result {
+            formatter.write_str("an integer, an integer-valued float, a numeric string, or null")
+        }
+
+        fn visit_u64<E: de::Error>(self, v: u64) -> std::result::Result<Self::Value, E> {
+            Ok(Some(v))
+        }
+
+        fn visit_i64<E: de::Error>(self, v: i64) -> std::result::Result<Self::Value, E> {
+            if v < 0 {
+                return Err(E::custom(format!("invalid expires_in value: {v}")));
+            }
+            Ok(Some(v as u64))
+        }
+
+        fn visit_f64<E: de::Error>(self, v: f64) -> std::result::Result<Self::Value, E> {
+            if !v.is_finite() || v < 0.0 || v.fract() != 0.0 || v > u64::MAX as f64 {
+                return Err(E::custom(format!("invalid expires_in value: {v}")));
+            }
+            Ok(Some(v as u64))
+        }
+
+        fn visit_str<E: de::Error>(self, v: &str) -> std::result::Result<Self::Value, E> {
+            v.parse::<u64>().map(Some).map_err(de::Error::custom)
+        }
+
+        fn visit_none<E: de::Error>(self) -> std::result::Result<Self::Value, E> {
+            Ok(None)
+        }
+
+        fn visit_unit<E: de::Error>(self) -> std::result::Result<Self::Value, E> {
+            Ok(None)
+        }
+    }
+
+    deserializer.deserialize_any(U64OrString)
+}
+
+// -- Internal Token State --
+
+struct TokenState {
+    access_token: Option<String>,
+    expires_at: Option<Instant>,
+}
+
+impl TokenState {
+    fn new() -> Self {
+        Self {
+            access_token: None,
+            expires_at: None,
+        }
+    }
+
+    fn is_expired(&self, buffer: Duration) -> bool {
+        match (self.access_token.as_ref(), self.expires_at) {
+            (Some(_), Some(expires_at)) => Instant::now() + buffer >= expires_at,
+            (None, _) => true,
+            (Some(_), None) => true,
+        }
+    }
+
+    fn update(&mut self, resp: &TokenResponse) {
+        self.access_token = Some(resp.access_token.clone());
+        let expires_in = resp.expires_in.unwrap_or(DEFAULT_TOKEN_TTL_SECS);
+        self.expires_at = Some(Instant::now() + Duration::from_secs(expires_in));
+    }
+}
+
+#[async_trait]
+trait TokenSource: Send + Sync + std::fmt::Debug {
+    async fn fetch_token(&self) -> Result<TokenResponse>;
+}
+
+struct ClientCredentialsSource {
+    issuer_url: String,
+    client_id: String,
+    client_secret: String,
+    scopes: Vec<String>,
+    http_client: Client,
+    discovery: RwLock<Option<OidcDiscovery>>,
+}
+
+impl std::fmt::Debug for ClientCredentialsSource {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("ClientCredentialsSource")
+            .field("issuer_url", &self.issuer_url)
+            .field("client_id", &self.client_id)
+            .field("client_secret", &"<redacted>")
+            .field("scopes", &self.scopes)
+            .finish()
+    }
+}
+
+impl ClientCredentialsSource {
+    fn new(
+        issuer_url: String,
+        client_id: String,
+        client_secret: Option<String>,
+        scopes: Vec<String>,
+    ) -> Result<Self> {
+        let client_secret = client_secret.ok_or(Error::InvalidInput {
+            message: "client_secret is required for ClientCredentials flow".to_string(),
+        })?;
+        Self::validate_issuer_transport(&issuer_url)?;
+
+        let http_client = Client::builder()
+            .timeout(Duration::from_secs(30))
+            .build()
+            .map_err(|e| Error::Runtime {
+                message: format!("Failed to create HTTP client for OAuth: {e}"),
+            })?;
+
+        Ok(Self {
+            issuer_url,
+            client_id,
+            client_secret,
+            scopes,
+            http_client,
+            discovery: RwLock::new(None),
+        })
+    }
+
+    fn validate_issuer_transport(issuer_url: &str) -> Result<()> {
+        let issuer = url::Url::parse(issuer_url).map_err(|e| Error::InvalidInput {
+            message: format!("Invalid OAuth issuer_url: {e}"),
+        })?;
+
+        match issuer.scheme() {
+            "https" => Ok(()),
+            "http" if Self::is_loopback_issuer(&issuer) => Ok(()),
+            _ => Err(Error::InvalidInput {
+                message:
+                    "ClientCredentials OAuth issuer_url must use https, except for loopback hosts"
+                        .to_string(),
+            }),
+        }
+    }
+
+    fn is_loopback_issuer(issuer: &url::Url) -> bool {
+        let Some(host) = issuer.host_str() else {
+            return false;
+        };
+
+        host.eq_ignore_ascii_case("localhost")
+            || host
+                .parse::<IpAddr>()
+                .map(|addr| addr.is_loopback())
+                .unwrap_or(false)
+    }
+
+    async fn get_discovery(&self) -> Result<OidcDiscovery> {
+        {
+            let cached = self.discovery.read().await;
+            if let Some(ref disc) = *cached {
+                return Ok(disc.clone());
+            }
+        }
+
+        let mut cache = self.discovery.write().await;
+        // Double-check
+        if let Some(ref disc) = *cache {
+            return Ok(disc.clone());
+        }
+
+        let discovery_url = format!(
+            "{}/.well-known/openid-configuration",
+            self.issuer_url.trim_end_matches('/')
+        );
+
+        debug!("Fetching OIDC discovery from {}", discovery_url);
+
+        let resp = self
+            .http_client
+            .get(&discovery_url)
+            .send()
+            .await
+            .map_err(|e| Error::Runtime {
+                message: format!("Failed to fetch OIDC discovery document: {e}"),
+            })?;
+
+        if !resp.status().is_success() {
+            return Err(Error::Runtime {
+                message: format!(
+                    "OIDC discovery failed with status {}: {}",
+                    resp.status(),
+                    resp.text().await.unwrap_or_default()
+                ),
+            });
+        }
+
+        let disc: OidcDiscovery = resp.json().await.map_err(|e| Error::Runtime {
+            message: format!("Failed to parse OIDC discovery document: {e}"),
+        })?;
+
+        let result = disc.clone();
+
+        *cache = Some(disc);
+        Ok(result)
+    }
+
+    async fn get_token_endpoint(&self) -> Result<String> {
+        self.get_discovery().await.map(|disc| disc.token_endpoint)
+    }
+
+    fn scopes_string(&self) -> String {
+        self.scopes.join(" ")
+    }
+
+    async fn post_token_request(
+        &self,
+        endpoint: &str,
+        params: &[(&str, &str)],
+    ) -> Result<TokenResponse> {
+        let resp = self
+            .http_client
+            .post(endpoint)
+            .form(params)
+            .send()
+            .await
+            .map_err(|e| Error::Runtime {
+                message: format!("Token request to {endpoint} failed: {e}"),
+            })?;
+
+        if !resp.status().is_success() {
+            return Err(Error::Runtime {
+                message: format!(
+                    "Token request failed with status {}: {}",
+                    resp.status(),
+                    resp.text().await.unwrap_or_default()
+                ),
+            });
+        }
+
+        resp.json().await.map_err(|e| Error::Runtime {
+            message: format!("Failed to parse token response: {e}"),
+        })
+    }
+}
+
+#[async_trait]
+impl TokenSource for ClientCredentialsSource {
+    async fn fetch_token(&self) -> Result<TokenResponse> {
+        let token_endpoint = self.get_token_endpoint().await?;
+        let scope = self.scopes_string();
+        let params = [
+            ("grant_type", "client_credentials"),
+            ("client_id", self.client_id.as_str()),
+            ("client_secret", self.client_secret.as_str()),
+            ("scope", scope.as_str()),
+        ];
+
+        self.post_token_request(&token_endpoint, &params).await
+    }
+}
+
+struct AzureImdsSource {
+    client_id: Option<String>,
+    resource: String,
+    http_client: Client,
+}
+
+impl std::fmt::Debug for AzureImdsSource {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("AzureImdsSource")
+            .field("client_id", &self.client_id)
+            .field("resource", &self.resource)
+            .finish()
+    }
+}
+
+impl AzureImdsSource {
+    fn new(scopes: Vec<String>, client_id: Option<String>) -> Result<Self> {
+        let resource = Self::resource_from_scopes(&scopes)?;
+        let http_client = Client::builder()
+            .timeout(Duration::from_secs(30))
+            .no_proxy()
+            .build()
+            .map_err(|e| Error::Runtime {
+                message: format!("Failed to create HTTP client for Azure IMDS OAuth: {e}"),
+            })?;
+
+        Ok(Self {
+            client_id,
+            resource,
+            http_client,
+        })
+    }
+
+    fn resource_from_scopes(scopes: &[String]) -> Result<String> {
+        let [scope] = scopes else {
+            return Err(Error::InvalidInput {
+                message: "AzureManagedIdentity flow requires exactly one OAuth scope or resource"
+                    .to_string(),
+            });
+        };
+
+        Ok(scope.strip_suffix("/.default").unwrap_or(scope).to_string())
+    }
+}
+
+#[async_trait]
+impl TokenSource for AzureImdsSource {
+    async fn fetch_token(&self) -> Result<TokenResponse> {
+        let mut url = format!(
+            "{AZURE_IMDS_ENDPOINT}?api-version={AZURE_IMDS_API_VERSION}&resource={}",
+            urlencoding::encode(&self.resource),
+        );
+        if let Some(cid) = self.client_id.as_deref() {
+            url.push_str(&format!("&client_id={}", urlencoding::encode(cid)));
+        }
+
+        let resp = self
+            .http_client
+            .get(&url)
+            .header("Metadata", "true")
+            .send()
+            .await
+            .map_err(|e| Error::Runtime {
+                message: format!("Azure IMDS request failed: {e}"),
+            })?;
+
+        if !resp.status().is_success() {
+            return Err(Error::Runtime {
+                message: format!(
+                    "Azure IMDS returned status {}: {}",
+                    resp.status(),
+                    resp.text().await.unwrap_or_default()
+                ),
+            });
+        }
+
+        resp.json().await.map_err(|e| Error::Runtime {
+            message: format!("Failed to parse IMDS token response: {e}"),
+        })
+    }
+}
+
+/// OAuth header provider that manages the full token lifecycle.
+///
+/// Implements [`HeaderProvider`] to inject `Authorization: Bearer <token>`
+/// headers into every LanceDB request, with automatic token refresh.
+pub struct OAuthHeaderProvider {
+    token_source: Box<dyn TokenSource>,
+    token_state: Arc<RwLock<TokenState>>,
+    refresh_buffer: Duration,
+}
+
+impl std::fmt::Debug for OAuthHeaderProvider {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("OAuthHeaderProvider")
+            .field("token_source", &self.token_source)
+            .finish()
+    }
+}
+
+impl OAuthHeaderProvider {
+    /// Create a new OAuth header provider from configuration.
+    pub fn new(config: OAuthConfig) -> Result<Self> {
+        let OAuthConfig {
+            issuer_url,
+            client_id,
+            client_secret,
+            scopes,
+            flow,
+            refresh_buffer_secs,
+        } = config;
+
+        if scopes.is_empty() {
+            return Err(Error::InvalidInput {
+                message: "At least one OAuth scope is required".to_string(),
+            });
+        }
+
+        let refresh_buffer =
+            Duration::from_secs(refresh_buffer_secs.unwrap_or(DEFAULT_REFRESH_BUFFER_SECS));
+        let token_source: Box<dyn TokenSource> = match flow {
+            OAuthFlow::ClientCredentials => Box::new(ClientCredentialsSource::new(
+                issuer_url,
+                client_id,
+                client_secret,
+                scopes,
+            )?),
+            OAuthFlow::AzureManagedIdentity { client_id } => {
+                Box::new(AzureImdsSource::new(scopes, client_id)?)
+            }
+        };
+
+        Ok(Self {
+            token_source,
+            token_state: Arc::new(RwLock::new(TokenState::new())),
+            refresh_buffer,
+        })
+    }
+
+    /// Get a valid access token, refreshing if necessary.
+    async fn get_valid_token(&self) -> Result<String> {
+        // Fast path: check if current token is still valid
+        {
+            let state = self.token_state.read().await;
+            if !state.is_expired(self.refresh_buffer)
+                && let Some(ref token) = state.access_token
+            {
+                return Ok(token.clone());
+            }
+        }
+
+        // Slow path: acquire or refresh token
+        let mut state = self.token_state.write().await;
+
+        // Double-check after acquiring write lock
+        if !state.is_expired(self.refresh_buffer)
+            && let Some(ref token) = state.access_token
+        {
+            return Ok(token.clone());
+        }
+
+        debug!("Acquiring new OAuth token via {:?}", self.token_source);
+        let resp = self.token_source.fetch_token().await?;
+
+        state.update(&resp);
+        Ok(resp.access_token)
+    }
+}
+
+#[async_trait]
+impl HeaderProvider for OAuthHeaderProvider {
+    async fn get_headers(&self) -> Result<HashMap<String, String>> {
+        let token = self.get_valid_token().await?;
+        Ok(HashMap::from([(
+            "authorization".to_string(),
+            format!("Bearer {token}"),
+        )]))
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use std::sync::atomic::{AtomicUsize, Ordering};
+
+    use tokio::io::{AsyncReadExt, AsyncWriteExt};
+    use tokio::net::{TcpListener, TcpStream};
+    use tokio::task::JoinHandle;
+
+    #[test]
+    fn test_token_state_expiry() {
+        let mut state = TokenState::new();
+        assert!(state.is_expired(Duration::from_secs(0)));
+
+        state.access_token = Some("tok".to_string());
+        state.expires_at = Some(Instant::now() + Duration::from_secs(600));
+        assert!(!state.is_expired(Duration::from_secs(300)));
+        assert!(state.is_expired(Duration::from_secs(601)));
+
+        state.expires_at = None;
+        assert!(state.is_expired(Duration::from_secs(0)));
+    }
+
+    #[test]
+    fn test_token_state_uses_default_expiry() {
+        let mut state = TokenState::new();
+        let response = TokenResponse {
+            access_token: "tok".to_string(),
+            expires_in: None,
+            token_type: None,
+        };
+
+        state.update(&response);
+
+        assert!(!state.is_expired(Duration::from_secs(DEFAULT_TOKEN_TTL_SECS - 1)));
+        assert!(state.is_expired(Duration::from_secs(DEFAULT_TOKEN_TTL_SECS + 1)));
+    }
+
+    #[test]
+    fn test_token_response_accepts_float_expires_in() {
+        let response: TokenResponse =
+            serde_json::from_str(r#"{"access_token":"tok","expires_in":3600.0}"#).unwrap();
+
+        assert_eq!(response.expires_in, Some(3600));
+    }
+
+    #[test]
+    fn test_token_response_rejects_negative_expires_in() {
+        let err =
+            serde_json::from_str::<TokenResponse>(r#"{"access_token":"tok","expires_in":-1}"#)
+                .unwrap_err();
+
+        assert!(err.to_string().contains("invalid expires_in value: -1"));
+    }
+
+    #[test]
+    fn test_token_response_debug_redacts_access_token() {
+        let response = TokenResponse {
+            access_token: "secret-token".to_string(),
+            expires_in: Some(3600),
+            token_type: Some("Bearer".to_string()),
+        };
+
+        let debug = format!("{response:?}");
+        assert!(!debug.contains("secret-token"));
+        assert!(debug.contains("access_token: \"<redacted>\""));
+    }
+
+    #[test]
+    fn test_scopes_string() {
+        let source = ClientCredentialsSource::new(
+            "https://login.microsoftonline.com/tenant/v2.0".to_string(),
+            "app-id".to_string(),
+            Some("secret".to_string()),
+            vec!["scope1".to_string(), "scope2".to_string()],
+        )
+        .unwrap();
+
+        assert_eq!(source.scopes_string(), "scope1 scope2");
+    }
+
+    #[test]
+    fn test_oauth_config_debug_redacts_client_secret() {
+        let config = OAuthConfig {
+            issuer_url: "https://issuer.example.com".to_string(),
+            client_id: "client-id".to_string(),
+            client_secret: Some("super-secret".to_string()),
+            scopes: vec!["scope".to_string()],
+            flow: OAuthFlow::ClientCredentials,
+            refresh_buffer_secs: None,
+        };
+
+        let debug = format!("{config:?}");
+        assert!(!debug.contains("super-secret"));
+        assert!(debug.contains("client_secret: Some(\"<redacted>\")"));
+    }
+
+    #[test]
+    fn test_oauth_header_provider_debug_redacts_client_secret() {
+        let config = OAuthConfig {
+            issuer_url: "https://issuer.example.com".to_string(),
+            client_id: "client-id".to_string(),
+            client_secret: Some("super-secret".to_string()),
+            scopes: vec!["scope".to_string()],
+            flow: OAuthFlow::ClientCredentials,
+            refresh_buffer_secs: None,
+        };
+
+        let provider = OAuthHeaderProvider::new(config).unwrap();
+        let debug = format!("{provider:?}");
+        assert!(!debug.contains("super-secret"));
+        assert!(debug.contains("client_secret: \"<redacted>\""));
+    }
+
+    #[test]
+    fn test_managed_identity_resource_from_default_scope() {
+        assert_eq!(
+            AzureImdsSource::resource_from_scopes(&["api://test/.default".to_string()]).unwrap(),
+            "api://test"
+        );
+    }
+
+    #[test]
+    fn test_managed_identity_resource_without_default_suffix() {
+        assert_eq!(
+            AzureImdsSource::resource_from_scopes(&["api://test".to_string()]).unwrap(),
+            "api://test"
+        );
+    }
+
+    #[test]
+    fn test_managed_identity_rejects_multiple_scopes() {
+        let config = OAuthConfig {
+            issuer_url: "https://login.microsoftonline.com/tenant/v2.0".to_string(),
+            client_id: "app-id".to_string(),
+            client_secret: None,
+            scopes: vec![
+                "api://test-a/.default".to_string(),
+                "api://test-b/.default".to_string(),
+            ],
+            flow: OAuthFlow::AzureManagedIdentity { client_id: None },
+            refresh_buffer_secs: None,
+        };
+        assert!(OAuthHeaderProvider::new(config).is_err());
+    }
+
+    #[tokio::test]
+    async fn test_token_endpoint_requires_discovery_success() {
+        let (issuer_url, server) = spawn_discovery_error_server().await;
+        let source = ClientCredentialsSource::new(
+            issuer_url,
+            "client-id".to_string(),
+            Some("secret".to_string()),
+            vec!["scope".to_string()],
+        )
+        .unwrap();
+
+        let err = source.get_token_endpoint().await.unwrap_err();
+        assert!(matches!(
+            err,
+            Error::Runtime { message }
+                if message.contains("OIDC discovery failed with status 503")
+        ));
+        server.await.unwrap();
+    }
+
+    #[test]
+    fn test_client_credentials_requires_secret() {
+        let config = OAuthConfig {
+            issuer_url: "https://login.microsoftonline.com/tenant/v2.0".to_string(),
+            client_id: "app-id".to_string(),
+            client_secret: None,
+            scopes: vec!["scope".to_string()],
+            flow: OAuthFlow::ClientCredentials,
+            refresh_buffer_secs: None,
+        };
+        assert!(OAuthHeaderProvider::new(config).is_err());
+    }
+
+    #[test]
+    fn test_client_credentials_rejects_insecure_non_loopback_issuer() {
+        let config = OAuthConfig {
+            issuer_url: "http://issuer.example.com".to_string(),
+            client_id: "app-id".to_string(),
+            client_secret: Some("secret".to_string()),
+            scopes: vec!["scope".to_string()],
+            flow: OAuthFlow::ClientCredentials,
+            refresh_buffer_secs: None,
+        };
+
+        let err = OAuthHeaderProvider::new(config).unwrap_err();
+        assert!(matches!(
+            err,
+            Error::InvalidInput { message }
+                if message == "ClientCredentials OAuth issuer_url must use https, except for loopback hosts"
+        ));
+    }
+
+    #[test]
+    fn test_empty_scopes_rejected() {
+        let config = OAuthConfig {
+            issuer_url: "https://login.microsoftonline.com/tenant/v2.0".to_string(),
+            client_id: "app-id".to_string(),
+            client_secret: None,
+            scopes: vec![],
+            flow: OAuthFlow::AzureManagedIdentity { client_id: None },
+            refresh_buffer_secs: None,
+        };
+        assert!(OAuthHeaderProvider::new(config).is_err());
+    }
+
+    #[tokio::test]
+    async fn test_client_credentials_token_lifecycle() {
+        let (issuer_url, token_requests, server) = spawn_oauth_server().await;
+        let config = OAuthConfig {
+            issuer_url,
+            client_id: "client-id".to_string(),
+            client_secret: Some("secret".to_string()),
+            scopes: vec!["scope".to_string()],
+            flow: OAuthFlow::ClientCredentials,
+            refresh_buffer_secs: Some(0),
+        };
+        let provider = OAuthHeaderProvider::new(config).unwrap();
+
+        let headers = provider.get_headers().await.unwrap();
+        assert_eq!(headers.get("authorization").unwrap(), "Bearer token-1");
+        assert_eq!(token_requests.load(Ordering::SeqCst), 1);
+
+        let headers = provider.get_headers().await.unwrap();
+        assert_eq!(headers.get("authorization").unwrap(), "Bearer token-1");
+        assert_eq!(token_requests.load(Ordering::SeqCst), 1);
+
+        provider.token_state.write().await.expires_at =
+            Some(Instant::now() - Duration::from_secs(1));
+
+        let headers = provider.get_headers().await.unwrap();
+        assert_eq!(headers.get("authorization").unwrap(), "Bearer token-2");
+        assert_eq!(token_requests.load(Ordering::SeqCst), 2);
+
+        server.await.unwrap();
+    }
+
+    async fn spawn_oauth_server() -> (String, Arc<AtomicUsize>, JoinHandle<()>) {
+        let listener = TcpListener::bind("127.0.0.1:0").await.unwrap();
+        let addr = listener.local_addr().unwrap();
+        let issuer_url = format!("http://{addr}");
+        let token_requests = Arc::new(AtomicUsize::new(0));
+        let server_token_requests = Arc::clone(&token_requests);
+
+        let server = tokio::spawn(async move {
+            for _ in 0..3 {
+                let (mut stream, _) = listener.accept().await.unwrap();
+                let (request_line, body) = read_http_request(&mut stream).await;
+
+                if request_line.starts_with("GET /.well-known/openid-configuration ") {
+                    let discovery = format!(r#"{{"token_endpoint":"http://{addr}/token"}}"#);
+                    write_json_response(&mut stream, "200 OK", &discovery).await;
+                } else if request_line.starts_with("POST /token ") {
+                    assert!(body.contains("grant_type=client_credentials"));
+                    assert!(body.contains("client_id=client-id"));
+                    assert!(body.contains("client_secret=secret"));
+                    assert!(body.contains("scope=scope"));
+
+                    let token_num = server_token_requests.fetch_add(1, Ordering::SeqCst) + 1;
+                    let token = format!(
+                        r#"{{"access_token":"token-{token_num}","expires_in":3600,"token_type":"Bearer"}}"#
+                    );
+                    write_json_response(&mut stream, "200 OK", &token).await;
+                } else {
+                    write_json_response(&mut stream, "404 Not Found", "{}").await;
+                }
+            }
+        });
+
+        (issuer_url, token_requests, server)
+    }
+
+    async fn spawn_discovery_error_server() -> (String, JoinHandle<()>) {
+        let listener = TcpListener::bind("127.0.0.1:0").await.unwrap();
+        let addr = listener.local_addr().unwrap();
+        let issuer_url = format!("http://{addr}");
+
+        let server = tokio::spawn(async move {
+            let (mut stream, _) = listener.accept().await.unwrap();
+            let (request_line, _) = read_http_request(&mut stream).await;
+            assert!(request_line.starts_with("GET /.well-known/openid-configuration "));
+            write_json_response(&mut stream, "503 Service Unavailable", "{}").await;
+        });
+
+        (issuer_url, server)
+    }
+
+    async fn read_http_request(stream: &mut TcpStream) -> (String, String) {
+        let mut buffer = Vec::new();
+        let mut header_end = None;
+
+        while header_end.is_none() {
+            let mut chunk = [0; 1024];
+            let read = stream.read(&mut chunk).await.unwrap();
+            assert_ne!(read, 0, "connection closed before request headers");
+            buffer.extend_from_slice(&chunk[..read]);
+            header_end = find_subsequence(&buffer, b"\r\n\r\n").map(|pos| pos + 4);
+        }
+
+        let header_end = header_end.unwrap();
+        let headers = String::from_utf8_lossy(&buffer[..header_end]).to_string();
+        let request_line = headers.lines().next().unwrap_or_default().to_string();
+        let content_length = headers
+            .lines()
+            .find_map(|line| {
+                let (name, value) = line.split_once(':')?;
+                name.eq_ignore_ascii_case("content-length")
+                    .then(|| value.trim().parse::<usize>().ok())
+                    .flatten()
+            })
+            .unwrap_or(0);
+
+        while buffer.len() < header_end + content_length {
+            let mut chunk = [0; 1024];
+            let read = stream.read(&mut chunk).await.unwrap();
+            assert_ne!(read, 0, "connection closed before request body");
+            buffer.extend_from_slice(&chunk[..read]);
+        }
+
+        let body =
+            String::from_utf8_lossy(&buffer[header_end..header_end + content_length]).to_string();
+
+        (request_line, body)
+    }
+
+    fn find_subsequence(haystack: &[u8], needle: &[u8]) -> Option<usize> {
+        haystack
+            .windows(needle.len())
+            .position(|window| window == needle)
+    }
+
+    async fn write_json_response(stream: &mut TcpStream, status: &str, body: &str) {
+        let response = format!(
+            "HTTP/1.1 {status}\r\ncontent-type: application/json\r\ncontent-length: {}\r\nconnection: close\r\n\r\n{body}",
+            body.len()
+        );
+        stream.write_all(response.as_bytes()).await.unwrap();
+    }
+}

rust/lancedb/src/remote/table.rs

@@ -70,18 +70,29 @@ use tokio::sync::RwLock;
 const REQUEST_TIMEOUT_HEADER: HeaderName = HeaderName::from_static("x-request-timeout-ms");
 const MIN_VERSION_HEADER: HeaderName = HeaderName::from_static("x-lancedb-min-version");
 const MIN_TIMESTAMP_HEADER: HeaderName = HeaderName::from_static("x-lancedb-min-timestamp");
+const MIN_READ_VERSION_HEADER: HeaderName = HeaderName::from_static("x-lancedb-min-read-version");
+const VERSION_HEADER: HeaderName = HeaderName::from_static("x-lancedb-version");
 const METRIC_TYPE_KEY: &str = "metric_type";
 const INDEX_TYPE_KEY: &str = "index_type";
 const SCHEMA_CACHE_TTL: Duration = Duration::from_secs(30);
 const SCHEMA_CACHE_REFRESH_WINDOW: Duration = Duration::from_secs(5);
 
-/// Per-table state driving the freshness headers (`x-lancedb-min-version` and
-/// `x-lancedb-min-timestamp`) sent on read requests.
+/// Per-table state driving the freshness headers (`x-lancedb-min-version`,
+/// `x-lancedb-min-timestamp`, and `x-lancedb-min-read-version`) sent on read
+/// requests.
 #[derive(Debug, Default, Clone, Copy)]
 struct FreshnessState {
     /// Provides read-your-write within a single handle: writes that return a
     /// version update this, and reads send it as `x-lancedb-min-version`.
     min_version: Option<u64>,
+    /// Highest dataset version observed in a *read* response on this handle.
+    /// Reads send it as `x-lancedb-min-read-version` so a load-balanced query
+    /// node whose cache is behind this version must refresh before serving,
+    /// giving monotonic reads across nodes regardless of which one the load
+    /// balancer routes to. Sourced only from reads (always committed dataset
+    /// versions), never from writes (which may return WAL entry ids), so it is
+    /// unaffected by the WAL/version mismatch that retired `min_version`.
+    min_read_version: Option<u64>,
     /// Wall-clock time captured at the last [`BaseTable::checkout_latest`]
     /// call. Subsequent reads send
     /// `max(baseline, now - read_consistency_interval)` as
@@ -102,6 +113,7 @@ struct FreshnessState {
 struct FreshnessHeaders {
     min_version: Option<u64>,
     min_timestamp: Option<SystemTime>,
+    min_read_version: Option<u64>,
 }
 
 impl FreshnessHeaders {
@@ -113,6 +125,9 @@ impl FreshnessHeaders {
             let dt: chrono::DateTime<chrono::Utc> = ts.into();
             request = request.header(MIN_TIMESTAMP_HEADER, dt.to_rfc3339());
         }
+        if let Some(v) = self.min_read_version {
+            request = request.header(MIN_READ_VERSION_HEADER, v.to_string());
+        }
         request
     }
 }
@@ -597,6 +612,7 @@ impl<S: HttpSend> RemoteTable<S> {
         body: &mut serde_json::Value,
         params: &QueryRequest,
     ) -> Result<()> {
+        params.check_filter()?;
         body["prefilter"] = params.prefilter.into();
         if let Some(offset) = params.offset {
             body["offset"] = serde_json::Value::Number(serde_json::Number::from(offset));
@@ -884,6 +900,7 @@ impl<S: HttpSend> RemoteTable<S> {
                 self.client.read_consistency_interval,
                 SystemTime::now(),
             ),
+            min_read_version: state.min_read_version,
         }
     }
 
@@ -905,6 +922,30 @@ impl<S: HttpSend> RemoteTable<S> {
         state.min_version = Some(state.min_version.map_or(version, |v| v.max(version)));
     }
 
+    /// Record a dataset version observed in a *read* response so subsequent
+    /// reads request at least this version via `x-lancedb-min-read-version`,
+    /// giving monotonic reads across load-balanced query nodes. A returned `0`
+    /// (or absent header from an old server) is ignored.
+    fn track_read_version(&self, version: u64) {
+        if version == 0 {
+            return;
+        }
+        let mut state = self.freshness.lock().unwrap();
+        state.min_read_version = Some(state.min_read_version.map_or(version, |v| v.max(version)));
+    }
+
+    /// Parse the `x-lancedb-version` response header (the dataset version a read
+    /// reflects) and fold it into the read-version watermark.
+    fn track_read_version_from_headers(&self, headers: &reqwest::header::HeaderMap) {
+        if let Some(version) = headers
+            .get(&VERSION_HEADER)
+            .and_then(|value| value.to_str().ok())
+            .and_then(|value| value.parse::<u64>().ok())
+        {
+            self.track_read_version(version);
+        }
+    }
+
     async fn execute_query(
         &self,
         query: &AnyQuery,
@@ -928,6 +969,7 @@ impl<S: HttpSend> RemoteTable<S> {
 
         let futures = requests.into_iter().map(|req| async move {
             let (request_id, response) = self.send(req, true).await?;
+            self.track_read_version_from_headers(response.headers());
             self.read_arrow_stream(&request_id, response).await
         });
         let streams = futures::future::try_join_all(futures);
@@ -1545,11 +1587,12 @@ impl<S: HttpSend> BaseTable for RemoteTable<S> {
         *write_guard = None;
         drop(write_guard);
 
-        // Drop any per-handle write tracking; subsequent reads use the
+        // Drop any per-handle read/write tracking; subsequent reads use the
         // baseline timestamp captured now to guarantee freshness.
         *self.freshness.lock().unwrap() = FreshnessState {
             min_version: None,
             checkout_baseline: Some(SystemTime::now()),
+            min_read_version: None,
         };
 
         // Invalidate schema cache since we're switching versions
@@ -1805,6 +1848,7 @@ impl<S: HttpSend> BaseTable for RemoteTable<S> {
             }
         };
 
+        self.track_read_version_from_headers(response.headers());
         let body = response.text().await.err_to_http(request_id.clone())?;
 
         serde_json::from_str(&body).map_err(|e| Error::Http {
@@ -7124,6 +7168,7 @@ mod tests {
         let state = FreshnessState {
             min_version: None,
             checkout_baseline: Some(baseline),
+            min_read_version: None,
         };
         assert_eq!(compute_min_timestamp(&state, None, now), Some(baseline));
 
@@ -7148,6 +7193,7 @@ mod tests {
         let state = FreshnessState {
             min_version: None,
             checkout_baseline: Some(baseline),
+            min_read_version: None,
         };
         assert_eq!(
             compute_min_timestamp(&state, Some(Duration::from_secs(10)), now),
@@ -7159,6 +7205,7 @@ mod tests {
         let state = FreshnessState {
             min_version: None,
             checkout_baseline: Some(recent_baseline),
+            min_read_version: None,
         };
         assert_eq!(
             compute_min_timestamp(&state, Some(Duration::from_secs(60)), now),
@@ -7303,6 +7350,106 @@ mod tests {
         );
     }
 
+    /// A handler that records every request's headers and answers each read with
+    /// an `x-lancedb-version` response header taken from `versions` (by call
+    /// index, saturating at the last entry). An empty string means "no header".
+    fn read_version_handler(
+        versions: &'static [&'static str],
+    ) -> (
+        impl Fn(reqwest::Request) -> http::Response<String> + Clone + Send + Sync + 'static,
+        Arc<std::sync::Mutex<Vec<http::HeaderMap>>>,
+    ) {
+        let requests = Arc::new(std::sync::Mutex::new(Vec::new()));
+        let requests_c = requests.clone();
+        let call = Arc::new(AtomicUsize::new(0));
+        let handler = move |request: reqwest::Request| {
+            requests_c.lock().unwrap().push(request.headers().clone());
+            let i = call.fetch_add(1, Ordering::SeqCst).min(versions.len() - 1);
+            let mut builder = http::Response::builder().status(200);
+            if !versions[i].is_empty() {
+                builder = builder.header("x-lancedb-version", versions[i]);
+            }
+            builder.body("42".to_string()).unwrap()
+        };
+        (handler, requests)
+    }
+
+    #[tokio::test]
+    async fn test_read_version_watermark_tracked_and_sent() {
+        let (handler, requests) = read_version_handler(&["100", "100"]);
+        let table = Table::new_with_handler("my_table", handler);
+
+        // First read has no watermark yet; the response advertises version 100,
+        // so the second read must floor the server at 100.
+        table.count_rows(None).await.unwrap();
+        table.count_rows(None).await.unwrap();
+
+        let reqs = requests.lock().unwrap();
+        assert!(!reqs[0].contains_key("x-lancedb-min-read-version"));
+        assert_eq!(
+            reqs[1]
+                .get("x-lancedb-min-read-version")
+                .unwrap()
+                .to_str()
+                .unwrap(),
+            "100"
+        );
+    }
+
+    #[tokio::test]
+    async fn test_read_version_watermark_keeps_max() {
+        // Server reports 100 then a stale 50; the watermark must not regress.
+        let (handler, requests) = read_version_handler(&["100", "50", "50"]);
+        let table = Table::new_with_handler("my_table", handler);
+
+        table.count_rows(None).await.unwrap();
+        table.count_rows(None).await.unwrap();
+        table.count_rows(None).await.unwrap();
+
+        let reqs = requests.lock().unwrap();
+        assert_eq!(
+            reqs[2]
+                .get("x-lancedb-min-read-version")
+                .unwrap()
+                .to_str()
+                .unwrap(),
+            "100"
+        );
+    }
+
+    #[tokio::test]
+    async fn test_read_version_absent_header_no_watermark() {
+        // An old server that doesn't return the version header leaves the
+        // watermark unset, preserving backward compatibility.
+        let (handler, requests) = read_version_handler(&[""]);
+        let table = Table::new_with_handler("my_table", handler);
+
+        table.count_rows(None).await.unwrap();
+        table.count_rows(None).await.unwrap();
+
+        let reqs = requests.lock().unwrap();
+        assert!(!reqs[1].contains_key("x-lancedb-min-read-version"));
+    }
+
+    #[tokio::test]
+    async fn test_read_version_watermark_reset_on_checkout_latest() {
+        let (handler, requests) = read_version_handler(&["100", "100"]);
+        let table = Table::new_with_handler("my_table", handler);
+
+        table.count_rows(None).await.unwrap();
+        table.checkout_latest().await.unwrap();
+        table.count_rows(None).await.unwrap();
+
+        // The read after checkout_latest starts from a clean slate.
+        let reqs = requests.lock().unwrap();
+        assert!(
+            !reqs
+                .last()
+                .unwrap()
+                .contains_key("x-lancedb-min-read-version")
+        );
+    }
+
     /// Like `capturing_handler`, but keeps a per-path snapshot of the headers
     /// from every request so tests can assert on a specific endpoint.
     #[allow(clippy::type_complexity)]

rust/lancedb/src/table/query.rs

@@ -35,6 +35,15 @@ pub enum AnyQuery {
     VectorQuery(VectorQueryRequest),
 }
 
+impl AnyQuery {
+    pub(crate) fn base(&self) -> &QueryRequest {
+        match self {
+            Self::Query(query) => query,
+            Self::VectorQuery(query) => &query.base,
+        }
+    }
+}
+
 //Decide between namespace or local
 pub async fn execute_query(
     table: &NativeTable,
@@ -108,6 +117,7 @@ pub async fn create_plan(
         AnyQuery::VectorQuery(query) => query.clone(),
         AnyQuery::Query(query) => VectorQueryRequest::from_plain_query(query.clone()),
     };
+    query.base.check_filter()?;
 
     let ds_ref = table.dataset.get().await?;
     let schema = ds_ref.schema();
@@ -357,6 +367,7 @@ async fn execute_namespace_query(
 
 /// Convert an AnyQuery to the namespace QueryTableRequest format.
 fn convert_to_namespace_query(query: &AnyQuery) -> Result<NsQueryTableRequest> {
+    query.base().check_filter()?;
     match query {
         AnyQuery::VectorQuery(vq) => {
             // Extract the query vector(s)
@@ -579,24 +590,45 @@ fn array_to_f32_vec(arr: &Arc<dyn arrow_array::Array>) -> Result<Vec<f32>> {
     })
 }
 
+/// Magic bytes that prefix (and suffix) the Arrow IPC *file* format.
+const ARROW_IPC_FILE_MAGIC: &[u8] = b"ARROW1";
+
 /// Parse Arrow IPC response from the namespace server.
+///
+/// The server may return either the Arrow IPC *file* format or the *stream*
+/// format. REST/phalanx returns the file format (it begins with the `ARROW1`
+/// magic); reading that with a `StreamReader` fails with "failed to fill whole
+/// buffer". Detect the magic and pick the matching reader so both are handled.
 async fn parse_arrow_ipc_response(bytes: bytes::Bytes) -> Result<DatasetRecordBatchStream> {
-    use arrow_ipc::reader::StreamReader;
+    use arrow_ipc::reader::{FileReader, StreamReader};
     use std::io::Cursor;
 
-    let cursor = Cursor::new(bytes);
-    let reader = StreamReader::try_new(cursor, None).map_err(|e| Error::Runtime {
-        message: format!("Failed to parse Arrow IPC response: {}", e),
-    })?;
-
-    // Collect all record batches
-    let schema = reader.schema();
-    let batches: Vec<_> = reader
-        .into_iter()
-        .collect::<std::result::Result<Vec<_>, _>>()
-        .map_err(|e| Error::Runtime {
-            message: format!("Failed to read Arrow IPC batches: {}", e),
+    let (schema, batches) = if bytes.starts_with(ARROW_IPC_FILE_MAGIC) {
+        let reader = FileReader::try_new(Cursor::new(bytes), None).map_err(|e| Error::Runtime {
+            message: format!("Failed to parse Arrow IPC file response: {}", e),
         })?;
+        let schema = reader.schema();
+        let batches = reader
+            .into_iter()
+            .collect::<std::result::Result<Vec<_>, _>>()
+            .map_err(|e| Error::Runtime {
+                message: format!("Failed to read Arrow IPC file batches: {}", e),
+            })?;
+        (schema, batches)
+    } else {
+        let reader =
+            StreamReader::try_new(Cursor::new(bytes), None).map_err(|e| Error::Runtime {
+                message: format!("Failed to parse Arrow IPC response: {}", e),
+            })?;
+        let schema = reader.schema();
+        let batches = reader
+            .into_iter()
+            .collect::<std::result::Result<Vec<_>, _>>()
+            .map_err(|e| Error::Runtime {
+                message: format!("Failed to read Arrow IPC batches: {}", e),
+            })?;
+        (schema, batches)
+    };
 
     // Create a stream from the batches
     let stream = futures::stream::iter(batches.into_iter().map(Ok));
@@ -624,6 +656,59 @@ mod tests {
         FixedSizeListArray::try_new_from_values(Float32Array::from(values), dimension).unwrap()
     }
 
+    #[tokio::test]
+    async fn test_parse_arrow_ipc_response_handles_file_and_stream() {
+        use arrow_array::{Int32Array, RecordBatch};
+        use arrow_ipc::writer::{FileWriter, StreamWriter};
+        use arrow_schema::{DataType, Field, Schema};
+
+        let schema = Arc::new(Schema::new(vec![Field::new("id", DataType::Int32, false)]));
+        let batch = RecordBatch::try_new(
+            schema.clone(),
+            vec![Arc::new(Int32Array::from(vec![1, 2, 3])) as ArrayRef],
+        )
+        .unwrap();
+
+        // Arrow IPC *file* format -- what REST/phalanx returns. Previously this
+        // failed with "failed to fill whole buffer" because we used a StreamReader.
+        let mut file_buf = Vec::new();
+        {
+            let mut writer = FileWriter::try_new(&mut file_buf, &schema).unwrap();
+            writer.write(&batch).unwrap();
+            writer.finish().unwrap();
+        }
+        assert!(file_buf.starts_with(ARROW_IPC_FILE_MAGIC));
+        let rows: usize = parse_arrow_ipc_response(bytes::Bytes::from(file_buf))
+            .await
+            .unwrap()
+            .try_collect::<Vec<_>>()
+            .await
+            .unwrap()
+            .iter()
+            .map(|b| b.num_rows())
+            .sum();
+        assert_eq!(rows, 3);
+
+        // Arrow IPC *stream* format must still parse.
+        let mut stream_buf = Vec::new();
+        {
+            let mut writer = StreamWriter::try_new(&mut stream_buf, &schema).unwrap();
+            writer.write(&batch).unwrap();
+            writer.finish().unwrap();
+        }
+        assert!(!stream_buf.starts_with(ARROW_IPC_FILE_MAGIC));
+        let rows: usize = parse_arrow_ipc_response(bytes::Bytes::from(stream_buf))
+            .await
+            .unwrap()
+            .try_collect::<Vec<_>>()
+            .await
+            .unwrap()
+            .iter()
+            .map(|b| b.num_rows())
+            .sum();
+        assert_eq!(rows, 3);
+    }
+
     #[test]
     fn test_convert_to_namespace_query_vector() {
         let query_vector = Arc::new(Float32Array::from(vec![1.0, 2.0, 3.0, 4.0]));