chore: update lance dependency to v4.0.0-beta.11

.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.27.2-beta.1"
+current_version = "0.27.0-beta.5"
 parse = """(?x)
     (?P<major>0|[1-9]\\d*)\\.
     (?P<minor>0|[1-9]\\d*)\\.

.github/workflows/build_linux_wheel/action.yml

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

.github/workflows/nodejs.yml

@@ -7,7 +7,6 @@ on:
   pull_request:
     paths:
       - Cargo.toml
-      - Cargo.lock
       - nodejs/**
       - rust/**
       - docs/src/js/**

.github/workflows/npm-publish.yml

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

.github/workflows/pypi-publish.yml

@@ -9,7 +9,6 @@ on:
     paths:
       - .github/workflows/pypi-publish.yml
       - Cargo.toml # Change in dependency frequently breaks builds
-      - Cargo.lock
 
 env:
   PIP_EXTRA_INDEX_URL: "https://pypi.fury.io/lance-format/ https://pypi.fury.io/lancedb/"

.github/workflows/python.yml

@@ -7,7 +7,6 @@ on:
   pull_request:
     paths:
       - Cargo.toml
-      - Cargo.lock
       - python/**
       - rust/**
       - .github/workflows/python.yml

.github/workflows/rust.yml

@@ -7,7 +7,6 @@ on:
   pull_request:
     paths:
       - Cargo.toml
-      - Cargo.lock
       - rust/**
       - .github/workflows/rust.yml
 
@@ -207,14 +206,14 @@ jobs:
       - name: Downgrade  dependencies
         # These packages have newer requirements for MSRV
         run: |
-          cargo update -p aws-sdk-bedrockruntime --precise 1.77.0
-          cargo update -p aws-sdk-dynamodb --precise 1.68.0
-          cargo update -p aws-config --precise 1.6.0
-          cargo update -p aws-sdk-kms --precise 1.63.0
-          cargo update -p aws-sdk-s3 --precise 1.79.0
-          cargo update -p aws-sdk-sso --precise 1.62.0
-          cargo update -p aws-sdk-ssooidc --precise 1.63.0
-          cargo update -p aws-sdk-sts --precise 1.63.0
+          cargo update -p aws-sdk-bedrockruntime --precise 1.64.0
+          cargo update -p aws-sdk-dynamodb --precise 1.55.0
+          cargo update -p aws-config --precise 1.5.10
+          cargo update -p aws-sdk-kms --precise 1.51.0
+          cargo update -p aws-sdk-s3 --precise 1.65.0
+          cargo update -p aws-sdk-sso --precise 1.50.0
+          cargo update -p aws-sdk-ssooidc --precise 1.51.0
+          cargo update -p aws-sdk-sts --precise 1.51.0
           cargo update -p home --precise 0.5.9
       - name: cargo +${{ matrix.msrv }} check
         env:

Cargo.toml

@@ -15,20 +15,20 @@ categories = ["database-implementations"]
 rust-version = "1.91.0"
 
 [workspace.dependencies]
-lance = { version = "=4.0.0", default-features = false }
-lance-core = { version = "=4.0.0" }
-lance-datagen = { version = "=4.0.0" }
-lance-file = { version = "=4.0.0" }
-lance-io = { version = "=4.0.0", default-features = false }
-lance-index = { version = "=4.0.0" }
-lance-linalg = { version = "=4.0.0" }
-lance-namespace = { version = "=4.0.0" }
-lance-namespace-impls = { version = "=4.0.0", default-features = false }
-lance-table = { version = "=4.0.0" }
-lance-testing = { version = "=4.0.0" }
-lance-datafusion = { version = "=4.0.0" }
-lance-encoding = { version = "=4.0.0" }
-lance-arrow = { version = "=4.0.0" }
+lance = { "version" = "=4.0.0-beta.11", default-features = false, "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-core = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-datagen = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-file = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-io = { "version" = "=4.0.0-beta.11", default-features = false, "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-index = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-linalg = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace-impls = { "version" = "=4.0.0-beta.11", default-features = false, "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-table = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-testing = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-datafusion = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-encoding = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
+lance-arrow = { "version" = "=4.0.0-beta.11", "tag" = "v4.0.0-beta.11", "git" = "https://github.com/lance-format/lance.git" }
 ahash = "0.8"
 # Note that this one does not include pyarrow
 arrow = { version = "57.2", optional = false }

ci/check_lance_release.py

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

docker-compose.yml

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

docs/requirements.txt

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

docs/src/java/java.md

@@ -14,7 +14,7 @@ Add the following dependency to your `pom.xml`:
 <dependency>
     <groupId>com.lancedb</groupId>
     <artifactId>lancedb-core</artifactId>
-    <version>0.27.2-beta.1</version>
+    <version>0.27.0-beta.5</version>
 </dependency>
 ```
 

docs/src/js/classes/Table.md

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

docs/src/js/interfaces/OptimizeOptions.md

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

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

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

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

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

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

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

java/lancedb-core/pom.xml

@@ -8,7 +8,7 @@
     <parent>
       <groupId>com.lancedb</groupId>
       <artifactId>lancedb-parent</artifactId>
-      <version>0.27.2-beta.1</version>
+      <version>0.27.0-beta.5</version>
       <relativePath>../pom.xml</relativePath>
     </parent>
 

java/pom.xml

@@ -6,7 +6,7 @@
 
     <groupId>com.lancedb</groupId>
     <artifactId>lancedb-parent</artifactId>
-    <version>0.27.2-beta.1</version>
+    <version>0.27.0-beta.5</version>
     <packaging>pom</packaging>
     <name>${project.artifactId}</name>
     <description>LanceDB Java SDK Parent POM</description>
@@ -28,7 +28,7 @@
     <properties>
         <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
         <arrow.version>15.0.0</arrow.version>
-        <lance-core.version>3.0.1</lance-core.version>
+        <lance-core.version>4.0.0-beta.11</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 = "0.27.2-beta.1"
+version = "0.27.0-beta.5"
 license.workspace = true
 description.workspace = true
 repository.workspace = true
@@ -15,8 +15,6 @@ crate-type = ["cdylib"]
 async-trait.workspace = true
 arrow-ipc.workspace = true
 arrow-array.workspace = true
-arrow-buffer = "57.2"
-half.workspace = true
 arrow-schema.workspace = true
 env_logger.workspace = true
 futures.workspace = true

nodejs/__test__/vector_types.test.ts

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

nodejs/lancedb/arrow.ts

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

nodejs/lancedb/query.ts

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

nodejs/npm/darwin-arm64/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-darwin-arm64",
-	"version": "0.27.2-beta.1",
+	"version": "0.27.0-beta.5",
 	"os": ["darwin"],
 	"cpu": ["arm64"],
 	"main": "lancedb.darwin-arm64.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-gnu",
-	"version": "0.27.2-beta.1",
+	"version": "0.27.0-beta.5",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-gnu.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-musl",
-	"version": "0.27.2-beta.1",
+	"version": "0.27.0-beta.5",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-musl.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-gnu",
-	"version": "0.27.2-beta.1",
+	"version": "0.27.0-beta.5",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-gnu.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-musl",
-	"version": "0.27.2-beta.1",
+	"version": "0.27.0-beta.5",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-musl.node",

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

@@ -1,6 +1,6 @@
 {
   "name": "@lancedb/lancedb-win32-arm64-msvc",
-  "version": "0.27.2-beta.1",
+  "version": "0.27.0-beta.5",
   "os": [
     "win32"
   ],

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-win32-x64-msvc",
-	"version": "0.27.2-beta.1",
+	"version": "0.27.0-beta.5",
 	"os": ["win32"],
 	"cpu": ["x64"],
 	"main": "lancedb.win32-x64-msvc.node",

nodejs/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@lancedb/lancedb",
-  "version": "0.27.2-beta.1",
+  "version": "0.27.0-beta.5",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "@lancedb/lancedb",
-      "version": "0.27.2-beta.1",
+      "version": "0.27.0-beta.5",
       "cpu": [
         "x64",
         "arm64"

nodejs/package.json

@@ -11,7 +11,7 @@
     "ann"
   ],
   "private": false,
-  "version": "0.27.2-beta.1",
+  "version": "0.27.0-beta.5",
   "main": "dist/index.js",
   "exports": {
     ".": "./dist/index.js",

nodejs/src/query.rs

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

python/.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.30.2-beta.1"
+current_version = "0.30.0-beta.5"
 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 = "0.30.2-beta.1"
+version = "0.30.0-beta.5"
 edition.workspace = true
 description = "Python bindings for LanceDB"
 license.workspace = true
@@ -23,7 +23,6 @@ lance-namespace.workspace = true
 lance-namespace-impls.workspace = true
 lance-io.workspace = true
 env_logger.workspace = true
-log.workspace = true
 pyo3 = { version = "0.26", features = ["extension-module", "abi3-py39"] }
 pyo3-async-runtimes = { version = "0.26", features = [
     "attributes",

python/python/lancedb/_lancedb.pyi

@@ -135,10 +135,7 @@ class Table:
     def close(self) -> None: ...
     async def schema(self) -> pa.Schema: ...
     async def add(
-        self,
-        data: pa.RecordBatchReader,
-        mode: Literal["append", "overwrite"],
-        progress: Optional[Any] = None,
+        self, data: pa.RecordBatchReader, mode: Literal["append", "overwrite"]
     ) -> AddResult: ...
     async def update(
         self, updates: Dict[str, str], where: Optional[str]

python/python/lancedb/embeddings/utils.py

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

python/python/lancedb/query.py

@@ -70,7 +70,7 @@ def ensure_vector_query(
 ) -> Union[List[float], List[List[float]], pa.Array, List[pa.Array]]:
     if isinstance(val, list):
         if len(val) == 0:
-            raise ValueError("Vector query must be a non-empty list")
+            return ValueError("Vector query must be a non-empty list")
         sample = val[0]
     else:
         if isinstance(val, float):
@@ -83,7 +83,7 @@ def ensure_vector_query(
         return val
     if isinstance(sample, list):
         if len(sample) == 0:
-            raise ValueError("Vector query must be a non-empty list")
+            return ValueError("Vector query must be a non-empty list")
         if isinstance(sample[0], float):
             # val is list of list of floats
             return val
@@ -2205,8 +2205,8 @@ class LanceHybridQueryBuilder(LanceQueryBuilder):
             self._vector_query.select(self._columns)
             self._fts_query.select(self._columns)
         if self._where:
-            self._vector_query.where(self._where, not self._postfilter)
-            self._fts_query.where(self._where, not self._postfilter)
+            self._vector_query.where(self._where, self._postfilter)
+            self._fts_query.where(self._where, self._postfilter)
         if self._with_row_id:
             self._vector_query.with_row_id(True)
             self._fts_query.with_row_id(True)

python/python/lancedb/remote/table.py

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

python/python/lancedb/table.py

@@ -14,7 +14,6 @@ from functools import cached_property
 from typing import (
     TYPE_CHECKING,
     Any,
-    Callable,
     Dict,
     Iterable,
     List,
@@ -278,7 +277,7 @@ def _sanitize_data(
 
     if metadata:
         new_metadata = target_schema.metadata or {}
-        new_metadata.update(metadata)
+        new_metadata = new_metadata.update(metadata)
         target_schema = target_schema.with_metadata(new_metadata)
 
     _validate_schema(target_schema)
@@ -557,21 +556,6 @@ def _table_uri(base: str, table_name: str) -> str:
     return join_uri(base, f"{table_name}.lance")
 
 
-def _normalize_progress(progress):
-    """Normalize a ``progress`` parameter for :meth:`Table.add`.
-
-    Returns ``(progress_obj, owns)`` where *owns* is True when we created a
-    tqdm bar that the caller must close.
-    """
-    if progress is True:
-        from tqdm.auto import tqdm
-
-        return tqdm(unit=" rows"), True
-    if progress is False or progress is None:
-        return None, False
-    return progress, False
-
-
 class Table(ABC):
     """
     A Table is a collection of Records in a LanceDB Database.
@@ -990,7 +974,6 @@ class Table(ABC):
         mode: AddMode = "append",
         on_bad_vectors: OnBadVectorsType = "error",
         fill_value: float = 0.0,
-        progress: Optional[Union[bool, Callable, Any]] = None,
     ) -> AddResult:
         """Add more data to the [Table](Table).
 
@@ -1012,29 +995,6 @@ class Table(ABC):
             One of "error", "drop", "fill".
         fill_value: float, default 0.
             The value to use when filling vectors. Only used if on_bad_vectors="fill".
-        progress: bool, callable, or tqdm-like, optional
-            Progress reporting during the add operation. Can be:
-
-            - ``True`` to automatically create and display a tqdm progress
-              bar (requires ``tqdm`` to be installed)::
-
-                table.add(data, progress=True)
-
-            - A **callable** that receives a dict with keys ``output_rows``,
-              ``output_bytes``, ``total_rows``, ``elapsed_seconds``,
-              ``active_tasks``, ``total_tasks``, and ``done``::
-
-                def on_progress(p):
-                    print(f"{p['output_rows']}/{p['total_rows']} rows, "
-                          f"{p['active_tasks']}/{p['total_tasks']} workers")
-                table.add(data, progress=on_progress)
-
-            - A **tqdm-compatible** progress bar whose ``total`` and
-              ``update()`` will be called automatically. The postfix shows
-              write throughput (MB/s) and active worker count::
-
-                with tqdm() as pbar:
-                    table.add(data, progress=pbar)
 
         Returns
         -------
@@ -2532,7 +2492,6 @@ class LanceTable(Table):
         mode: AddMode = "append",
         on_bad_vectors: OnBadVectorsType = "error",
         fill_value: float = 0.0,
-        progress: Optional[Union[bool, Callable, Any]] = None,
     ) -> AddResult:
         """Add data to the table.
         If vector columns are missing and the table
@@ -2551,29 +2510,17 @@ class LanceTable(Table):
             One of "error", "drop", "fill", "null".
         fill_value: float, default 0.
             The value to use when filling vectors. Only used if on_bad_vectors="fill".
-        progress: bool, callable, or tqdm-like, optional
-            A callback or tqdm-compatible progress bar. See
-            :meth:`Table.add` for details.
 
         Returns
         -------
         int
             The number of vectors in the table.
         """
-        progress, owns = _normalize_progress(progress)
-        try:
-            return LOOP.run(
-                self._table.add(
-                    data,
-                    mode=mode,
-                    on_bad_vectors=on_bad_vectors,
-                    fill_value=fill_value,
-                    progress=progress,
-                )
+        return LOOP.run(
+            self._table.add(
+                data, mode=mode, on_bad_vectors=on_bad_vectors, fill_value=fill_value
             )
-        finally:
-            if owns:
-                progress.close()
+        )
 
     def merge(
         self,
@@ -3822,7 +3769,6 @@ class AsyncTable:
         mode: Optional[Literal["append", "overwrite"]] = "append",
         on_bad_vectors: Optional[OnBadVectorsType] = None,
         fill_value: Optional[float] = None,
-        progress: Optional[Union[bool, Callable, Any]] = None,
     ) -> AddResult:
         """Add more data to the [Table](Table).
 
@@ -3844,9 +3790,6 @@ class AsyncTable:
             One of "error", "drop", "fill", "null".
         fill_value: float, default 0.
             The value to use when filling vectors. Only used if on_bad_vectors="fill".
-        progress: callable or tqdm-like, optional
-            A callback or tqdm-compatible progress bar. See
-            :meth:`Table.add` for details.
 
         """
         schema = await self.schema()
@@ -3857,13 +3800,7 @@ class AsyncTable:
 
         # _santitize_data is an old code path, but we will use it until the
         # new code path is ready.
-        if mode == "overwrite":
-            # For overwrite, apply the same preprocessing as create_table
-            # so vector columns are inferred as FixedSizeList.
-            data, _ = sanitize_create_table(
-                data, None, on_bad_vectors=on_bad_vectors, fill_value=fill_value
-            )
-        elif on_bad_vectors != "error" or (
+        if on_bad_vectors != "error" or (
             schema.metadata is not None and b"embedding_functions" in schema.metadata
         ):
             data = _sanitize_data(
@@ -3876,9 +3813,8 @@ class AsyncTable:
             )
         _register_optional_converters()
         data = to_scannable(data)
-        progress, owns = _normalize_progress(progress)
         try:
-            return await self._inner.add(data, mode or "append", progress=progress)
+            return await self._inner.add(data, mode or "append")
         except RuntimeError as e:
             if "Cast error" in str(e):
                 raise ValueError(e)
@@ -3886,9 +3822,6 @@ class AsyncTable:
                 raise ValueError(e)
             else:
                 raise
-        finally:
-            if owns:
-                progress.close()
 
     def merge_insert(self, on: Union[str, Iterable[str]]) -> LanceMergeInsertBuilder:
         """
@@ -4818,16 +4751,7 @@ class IndexStatistics:
     num_indexed_rows: int
     num_unindexed_rows: int
     index_type: Literal[
-        "IVF_FLAT",
-        "IVF_SQ",
-        "IVF_PQ",
-        "IVF_RQ",
-        "IVF_HNSW_SQ",
-        "IVF_HNSW_PQ",
-        "FTS",
-        "BTREE",
-        "BITMAP",
-        "LABEL_LIST",
+        "IVF_PQ", "IVF_HNSW_PQ", "IVF_HNSW_SQ", "FTS", "BTREE", "BITMAP", "LABEL_LIST"
     ]
     distance_type: Optional[Literal["l2", "cosine", "dot"]] = None
     num_indices: Optional[int] = None

python/python/tests/test_embeddings.py

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

python/python/tests/test_hybrid_query.py

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

python/python/tests/test_index.py

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

python/python/tests/test_namespace.py

@@ -8,7 +8,6 @@ import shutil
 import pytest
 import pyarrow as pa
 import lancedb
-from lance_namespace.errors import NamespaceNotEmptyError, TableNotFoundError
 
 
 class TestNamespaceConnection:
@@ -131,7 +130,7 @@ class TestNamespaceConnection:
         assert len(list(db.table_names(namespace=["test_ns"]))) == 0
 
         # Should not be able to open dropped table
-        with pytest.raises(TableNotFoundError):
+        with pytest.raises(RuntimeError):
             db.open_table("table1", namespace=["test_ns"])
 
     def test_create_table_with_schema(self):
@@ -341,7 +340,7 @@ class TestNamespaceConnection:
         db.create_table("test_table", schema=schema, namespace=["test_namespace"])
 
         # Try to drop namespace with tables - should fail
-        with pytest.raises(NamespaceNotEmptyError):
+        with pytest.raises(RuntimeError, match="is not empty"):
             db.drop_namespace(["test_namespace"])
 
         # Drop table first

python/python/tests/test_namespace_integration.py

@@ -147,12 +147,7 @@ class TrackingNamespace(LanceNamespace):
         This simulates a credential rotation system where each call returns
         new credentials that expire after credential_expires_in_seconds.
         """
-        # Start from base storage options (endpoint, region, allow_http, etc.)
-        # because DirectoryNamespace returns None for storage_options from
-        # describe_table/declare_table when no credential vendor is configured.
-        modified = copy.deepcopy(self.base_storage_options)
-        if storage_options:
-            modified.update(storage_options)
+        modified = copy.deepcopy(storage_options) if storage_options else {}
 
         # Increment credentials to simulate rotation
         modified["aws_access_key_id"] = f"AKID_{count}"

python/python/tests/test_query.py

@@ -30,7 +30,6 @@ from lancedb.query import (
     PhraseQuery,
     Query,
     FullTextSearchQuery,
-    ensure_vector_query,
 )
 from lancedb.rerankers.cross_encoder import CrossEncoderReranker
 from lancedb.table import AsyncTable, LanceTable
@@ -1502,18 +1501,6 @@ def test_search_empty_table(mem_db):
     assert results == []
 
 
-def test_ensure_vector_query_empty_list():
-    """Regression: ensure_vector_query used to return instead of raise ValueError."""
-    with pytest.raises(ValueError, match="non-empty"):
-        ensure_vector_query([])
-
-
-def test_ensure_vector_query_nested_empty_list():
-    """Regression: ensure_vector_query used to return instead of raise ValueError."""
-    with pytest.raises(ValueError, match="non-empty"):
-        ensure_vector_query([[]])
-
-
 def test_fast_search(tmp_path):
     db = lancedb.connect(tmp_path)
 

python/python/tests/test_table.py

@@ -527,132 +527,6 @@ async def test_add_async(mem_db_async: AsyncConnection):
     assert await table.count_rows() == 3
 
 
-def test_add_overwrite_infers_vector_schema(mem_db: DBConnection):
-    """Overwrite should infer vector columns the same way create_table does.
-
-    Regression test for https://github.com/lancedb/lancedb/issues/3183
-    """
-    table = mem_db.create_table(
-        "test_overwrite_vec",
-        data=[
-            {"vector": [1.0, 2.0, 3.0, 4.0], "item": "foo"},
-            {"vector": [5.0, 6.0, 7.0, 8.0], "item": "bar"},
-        ],
-    )
-    # create_table infers vector as fixed_size_list<float32, 4>
-    original_type = table.schema.field("vector").type
-    assert pa.types.is_fixed_size_list(original_type)
-
-    # overwrite with plain Python lists (PyArrow infers list<double>)
-    table.add(
-        [
-            {"vector": [10.0, 20.0, 30.0, 40.0], "item": "baz"},
-        ],
-        mode="overwrite",
-    )
-    # overwrite should infer vector column the same way as create_table
-    new_type = table.schema.field("vector").type
-    assert pa.types.is_fixed_size_list(new_type), (
-        f"Expected fixed_size_list after overwrite, got {new_type}"
-    )
-
-
-def test_add_progress_callback(mem_db: DBConnection):
-    table = mem_db.create_table(
-        "test",
-        data=[{"id": 1}, {"id": 2}],
-    )
-
-    updates = []
-    table.add([{"id": 3}, {"id": 4}], progress=lambda p: updates.append(dict(p)))
-
-    assert len(table) == 4
-    # The done callback always fires, so we should always get at least one.
-    assert len(updates) >= 1, "expected at least one progress callback"
-    for p in updates:
-        assert "output_rows" in p
-        assert "output_bytes" in p
-        assert "total_rows" in p
-        assert "elapsed_seconds" in p
-        assert "active_tasks" in p
-        assert "total_tasks" in p
-        assert "done" in p
-    # The last callback should have done=True.
-    assert updates[-1]["done"] is True
-
-
-def test_add_progress_tqdm_like(mem_db: DBConnection):
-    """Test that a tqdm-like object gets total set and update() called."""
-
-    class FakeBar:
-        def __init__(self):
-            self.total = None
-            self.n = 0
-            self.postfix = None
-
-        def update(self, n):
-            self.n += n
-
-        def set_postfix_str(self, s):
-            self.postfix = s
-
-        def refresh(self):
-            pass
-
-    table = mem_db.create_table(
-        "test",
-        data=[{"id": 1}, {"id": 2}],
-    )
-
-    bar = FakeBar()
-    table.add([{"id": 3}, {"id": 4}], progress=bar)
-
-    assert len(table) == 4
-    # Postfix should contain throughput and worker count
-    if bar.postfix is not None:
-        assert "MB/s" in bar.postfix
-        assert "workers" in bar.postfix
-
-
-def test_add_progress_bool(mem_db: DBConnection):
-    """Test that progress=True creates and closes a tqdm bar automatically."""
-    table = mem_db.create_table(
-        "test",
-        data=[{"id": 1}, {"id": 2}],
-    )
-
-    table.add([{"id": 3}, {"id": 4}], progress=True)
-    assert len(table) == 4
-
-    # progress=False should be the same as None
-    table.add([{"id": 5}], progress=False)
-    assert len(table) == 5
-
-
-@pytest.mark.asyncio
-async def test_add_progress_callback_async(mem_db_async: AsyncConnection):
-    """Progress callbacks work through the async path too."""
-    table = await mem_db_async.create_table("test", data=[{"id": 1}, {"id": 2}])
-
-    updates = []
-    await table.add([{"id": 3}, {"id": 4}], progress=lambda p: updates.append(dict(p)))
-
-    assert await table.count_rows() == 4
-    assert len(updates) >= 1
-    assert updates[-1]["done"] is True
-
-
-def test_add_progress_callback_error(mem_db: DBConnection):
-    """A failing callback must not prevent the write from succeeding."""
-    table = mem_db.create_table("test", data=[{"id": 1}, {"id": 2}])
-
-    def bad_callback(p):
-        raise RuntimeError("boom")
-
-    table.add([{"id": 3}, {"id": 4}], progress=bad_callback)
-    assert len(table) == 4
-
-
 def test_polars(mem_db: DBConnection):
     data = {
         "vector": [[3.1, 4.1], [5.9, 26.5]],
@@ -2173,33 +2047,3 @@ def test_table_uri(tmp_path):
     db = lancedb.connect(tmp_path)
     table = db.create_table("my_table", data=[{"x": 0}])
     assert table.uri == str(tmp_path / "my_table.lance")
-
-
-def test_sanitize_data_metadata_not_stripped():
-    """Regression test: dict.update() returns None, so assigning its result
-    would silently replace metadata with None, causing with_metadata(None)
-    to strip all schema metadata from the target schema."""
-    from lancedb.table import _sanitize_data
-
-    schema = pa.schema(
-        [pa.field("x", pa.int64())],
-        metadata={b"existing_key": b"existing_value"},
-    )
-    batch = pa.record_batch([pa.array([1, 2, 3])], schema=schema)
-
-    # Use a different field type so the reader and target schemas differ,
-    # forcing _cast_to_target_schema to rebuild the schema with the
-    # target's metadata (instead of taking the fast-path).
-    target_schema = pa.schema(
-        [pa.field("x", pa.int32())],
-        metadata={b"existing_key": b"existing_value"},
-    )
-
-    reader = pa.RecordBatchReader.from_batches(schema, [batch])
-    metadata = {b"new_key": b"new_value"}
-    result = _sanitize_data(reader, target_schema=target_schema, metadata=metadata)
-
-    result_schema = result.schema
-    assert result_schema.metadata is not None
-    assert result_schema.metadata[b"existing_key"] == b"existing_value"
-    assert result_schema.metadata[b"new_key"] == b"new_value"

python/src/table.rs

@@ -19,7 +19,7 @@ use lancedb::table::{
     Table as LanceDbTable,
 };
 use pyo3::{
-    Bound, FromPyObject, Py, PyAny, PyRef, PyResult, Python,
+    Bound, FromPyObject, PyAny, PyRef, PyResult, Python,
     exceptions::{PyKeyError, PyRuntimeError, PyValueError},
     pyclass, pymethods,
     types::{IntoPyDict, PyAnyMethods, PyDict, PyDictMethods},
@@ -299,12 +299,10 @@ impl Table {
         })
     }
 
-    #[pyo3(signature = (data, mode, progress=None))]
     pub fn add<'a>(
         self_: PyRef<'a, Self>,
         data: PyScannable,
         mode: String,
-        progress: Option<Py<PyAny>>,
     ) -> PyResult<Bound<'a, PyAny>> {
         let mut op = self_.inner_ref()?.add(data);
         if mode == "append" {
@@ -314,81 +312,6 @@ impl Table {
         } else {
             return Err(PyValueError::new_err(format!("Invalid mode: {}", mode)));
         }
-        if let Some(progress_obj) = progress {
-            let is_callable = Python::attach(|py| progress_obj.bind(py).is_callable());
-            if is_callable {
-                // Callback: call with a dict of progress info.
-                op = op.progress(move |p| {
-                    Python::attach(|py| {
-                        let dict = PyDict::new(py);
-                        if let Err(e) = dict
-                            .set_item("output_rows", p.output_rows())
-                            .and_then(|_| dict.set_item("output_bytes", p.output_bytes()))
-                            .and_then(|_| dict.set_item("total_rows", p.total_rows()))
-                            .and_then(|_| {
-                                dict.set_item("elapsed_seconds", p.elapsed().as_secs_f64())
-                            })
-                            .and_then(|_| dict.set_item("active_tasks", p.active_tasks()))
-                            .and_then(|_| dict.set_item("total_tasks", p.total_tasks()))
-                            .and_then(|_| dict.set_item("done", p.done()))
-                        {
-                            log::warn!("progress dict error: {e}");
-                            return;
-                        }
-                        if let Err(e) = progress_obj.call1(py, (dict,)) {
-                            log::warn!("progress callback error: {e}");
-                        }
-                    });
-                });
-            } else {
-                // tqdm-like: has update() method.
-                let mut last_rows: usize = 0;
-                let mut total_set = false;
-                op = op.progress(move |p| {
-                    let current = p.output_rows();
-                    let prev = last_rows;
-                    last_rows = current;
-                    Python::attach(|py| {
-                        if let Some(total) = p.total_rows()
-                            && !total_set
-                        {
-                            if let Err(e) = progress_obj.setattr(py, "total", total) {
-                                log::warn!("progress setattr error: {e}");
-                            }
-                            total_set = true;
-                        }
-                        let delta = current.saturating_sub(prev);
-                        if delta > 0 {
-                            if let Err(e) = progress_obj.call_method1(py, "update", (delta,)) {
-                                log::warn!("progress update error: {e}");
-                            }
-                            // Show throughput and active workers in tqdm postfix.
-                            let elapsed = p.elapsed().as_secs_f64();
-                            if elapsed > 0.0 {
-                                let mb_per_sec = p.output_bytes() as f64 / elapsed / 1_000_000.0;
-                                let postfix = format!(
-                                    "{:.1} MB/s | {}/{} workers",
-                                    mb_per_sec,
-                                    p.active_tasks(),
-                                    p.total_tasks()
-                                );
-                                if let Err(e) =
-                                    progress_obj.call_method1(py, "set_postfix_str", (postfix,))
-                                {
-                                    log::warn!("progress set_postfix_str error: {e}");
-                                }
-                            }
-                        }
-                        if p.done() {
-                            // Force a final refresh so the bar shows completion.
-                            if let Err(e) = progress_obj.call_method0(py, "refresh") {
-                                log::warn!("progress refresh error: {e}");
-                            }
-                        }
-                    });
-                });
-            }
-        }
 
         future_into_py(self_.py(), async move {
             let result = op.execute().await.infer_error()?;

rust/lancedb/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "lancedb"
-version = "0.27.2-beta.1"
+version = "0.27.0-beta.5"
 edition.workspace = true
 description = "LanceDB: A serverless, low-latency vector database for AI applications"
 license.workspace = true

rust/lancedb/src/connection.rs

@@ -596,8 +596,11 @@ pub struct ConnectBuilder {
 }
 
 #[cfg(feature = "remote")]
-const ENV_VARS_TO_STORAGE_OPTS: [(&str, &str); 1] =
-    [("AZURE_STORAGE_ACCOUNT_NAME", "azure_storage_account_name")];
+const ENV_VARS_TO_STORAGE_OPTS: [(&str, &str); 3] = [
+    ("AZURE_STORAGE_ACCOUNT_NAME", "azure_storage_account_name"),
+    ("AZURE_CLIENT_ID", "azure_client_id"),
+    ("AZURE_TENANT_ID", "azure_tenant_id"),
+];
 
 impl ConnectBuilder {
     /// Create a new [`ConnectOptions`] with the given database URI.

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

@@ -240,7 +240,7 @@ impl Shuffler {
                     .await?;
                     // Need to read the entire file in a single batch for in-memory shuffling
                     let batch = reader.read_record_batch(0, reader.num_rows()).await?;
-                    let mut rng = rng.lock().unwrap_or_else(|e| e.into_inner());
+                    let mut rng = rng.lock().unwrap();
                     Self::shuffle_batch(&batch, &mut rng, clump_size)
                 }
             })

rust/lancedb/src/io/object_store/io_tracking.rs

@@ -66,13 +66,13 @@ impl IoTrackingStore {
     }
 
     fn record_read(&self, num_bytes: u64) {
-        let mut stats = self.stats.lock().unwrap_or_else(|e| e.into_inner());
+        let mut stats = self.stats.lock().unwrap();
         stats.read_iops += 1;
         stats.read_bytes += num_bytes;
     }
 
     fn record_write(&self, num_bytes: u64) {
-        let mut stats = self.stats.lock().unwrap_or_else(|e| e.into_inner());
+        let mut stats = self.stats.lock().unwrap();
         stats.write_iops += 1;
         stats.write_bytes += num_bytes;
     }
@@ -229,63 +229,10 @@ impl MultipartUpload for IoTrackingMultipartUpload {
 
     fn put_part(&mut self, payload: PutPayload) -> UploadPart {
         {
-            let mut stats = self.stats.lock().unwrap_or_else(|e| e.into_inner());
+            let mut stats = self.stats.lock().unwrap();
             stats.write_iops += 1;
             stats.write_bytes += payload.content_length() as u64;
         }
         self.target.put_part(payload)
     }
 }
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    /// Helper: poison a Mutex<IoStats> by panicking while holding the lock.
-    fn poison_stats(stats: &Arc<Mutex<IoStats>>) {
-        let stats_clone = stats.clone();
-        let handle = std::thread::spawn(move || {
-            let _guard = stats_clone.lock().unwrap();
-            panic!("intentional panic to poison stats mutex");
-        });
-        let _ = handle.join();
-        assert!(stats.lock().is_err(), "mutex should be poisoned");
-    }
-
-    #[test]
-    fn test_record_read_recovers_from_poisoned_lock() {
-        let stats = Arc::new(Mutex::new(IoStats::default()));
-        let store = IoTrackingStore {
-            target: Arc::new(object_store::memory::InMemory::new()),
-            stats: stats.clone(),
-        };
-
-        poison_stats(&stats);
-
-        // record_read should not panic
-        store.record_read(1024);
-
-        // Verify the stats were updated despite poisoning
-        let s = stats.lock().unwrap_or_else(|e| e.into_inner());
-        assert_eq!(s.read_iops, 1);
-        assert_eq!(s.read_bytes, 1024);
-    }
-
-    #[test]
-    fn test_record_write_recovers_from_poisoned_lock() {
-        let stats = Arc::new(Mutex::new(IoStats::default()));
-        let store = IoTrackingStore {
-            target: Arc::new(object_store::memory::InMemory::new()),
-            stats: stats.clone(),
-        };
-
-        poison_stats(&stats);
-
-        // record_write should not panic
-        store.record_write(2048);
-
-        let s = stats.lock().unwrap_or_else(|e| e.into_inner());
-        assert_eq!(s.write_iops, 1);
-        assert_eq!(s.write_bytes, 2048);
-    }
-}

rust/lancedb/src/query.rs

@@ -5,7 +5,7 @@ use std::sync::Arc;
 use std::{future::Future, time::Duration};
 
 use arrow::compute::concat_batches;
-use arrow_array::{Array, Float16Array, Float32Array, Float64Array, RecordBatch, make_array};
+use arrow_array::{Array, Float16Array, Float32Array, Float64Array, make_array};
 use arrow_schema::{DataType, SchemaRef};
 use datafusion_expr::Expr;
 use datafusion_physical_plan::ExecutionPlan;
@@ -17,17 +17,15 @@ use lance_datafusion::exec::execute_plan;
 use lance_index::scalar::FullTextSearchQuery;
 use lance_index::scalar::inverted::SCORE_COL;
 use lance_index::vector::DIST_COL;
+use lance_io::stream::RecordBatchStreamAdapter;
 
 use crate::DistanceType;
 use crate::error::{Error, Result};
 use crate::rerankers::rrf::RRFReranker;
 use crate::rerankers::{NormalizeMethod, Reranker, check_reranker_result};
 use crate::table::BaseTable;
-use crate::utils::{MaxBatchLengthStream, TimeoutStream};
-use crate::{
-    arrow::{SendableRecordBatchStream, SimpleRecordBatchStream},
-    table::AnyQuery,
-};
+use crate::utils::TimeoutStream;
+use crate::{arrow::SendableRecordBatchStream, table::AnyQuery};
 
 mod hybrid;
 
@@ -606,14 +604,6 @@ impl Default for QueryExecutionOptions {
     }
 }
 
-impl QueryExecutionOptions {
-    fn without_output_batch_length_limit(&self) -> Self {
-        let mut options = self.clone();
-        options.max_batch_length = 0;
-        options
-    }
-}
-
 /// A trait for a query object that can be executed to get results
 ///
 /// There are various kinds of queries but they all return results
@@ -1190,8 +1180,6 @@ impl VectorQuery {
         &self,
         options: QueryExecutionOptions,
     ) -> Result<SendableRecordBatchStream> {
-        let max_batch_length = options.max_batch_length as usize;
-        let internal_options = options.without_output_batch_length_limit();
         // clone query and specify we want to include row IDs, which can be needed for reranking
         let mut fts_query = Query::new(self.parent.clone());
         fts_query.request = self.request.base.clone();
@@ -1201,8 +1189,8 @@ impl VectorQuery {
 
         vector_query.request.base.full_text_search = None;
         let (fts_results, vec_results) = try_join!(
-            fts_query.execute_with_options(internal_options.clone()),
-            vector_query.inner_execute_with_options(internal_options)
+            fts_query.execute_with_options(options.clone()),
+            vector_query.inner_execute_with_options(options)
         )?;
 
         let (fts_results, vec_results) = try_join!(
@@ -1257,7 +1245,9 @@ impl VectorQuery {
             results = results.drop_column(ROW_ID)?;
         }
 
-        Ok(single_batch_stream(results, max_batch_length))
+        Ok(SendableRecordBatchStream::from(
+            RecordBatchStreamAdapter::new(results.schema(), stream::iter([Ok(results)])),
+        ))
     }
 
     async fn inner_execute_with_options(
@@ -1266,7 +1256,6 @@ impl VectorQuery {
     ) -> Result<SendableRecordBatchStream> {
         let plan = self.create_plan(options.clone()).await?;
         let inner = execute_plan(plan, Default::default())?;
-        let inner = MaxBatchLengthStream::new_boxed(inner, options.max_batch_length as usize);
         let inner = if let Some(timeout) = options.timeout {
             TimeoutStream::new_boxed(inner, timeout)
         } else {
@@ -1276,25 +1265,6 @@ impl VectorQuery {
     }
 }
 
-fn single_batch_stream(batch: RecordBatch, max_batch_length: usize) -> SendableRecordBatchStream {
-    let schema = batch.schema();
-    if max_batch_length == 0 || batch.num_rows() <= max_batch_length {
-        return Box::pin(SimpleRecordBatchStream::new(
-            stream::iter([Ok(batch)]),
-            schema,
-        ));
-    }
-
-    let mut batches = Vec::with_capacity(batch.num_rows().div_ceil(max_batch_length));
-    let mut offset = 0;
-    while offset < batch.num_rows() {
-        let length = (batch.num_rows() - offset).min(max_batch_length);
-        batches.push(Ok(batch.slice(offset, length)));
-        offset += length;
-    }
-    Box::pin(SimpleRecordBatchStream::new(stream::iter(batches), schema))
-}
-
 impl ExecutableQuery for VectorQuery {
     async fn create_plan(&self, options: QueryExecutionOptions) -> Result<Arc<dyn ExecutionPlan>> {
         let query = AnyQuery::VectorQuery(self.request.clone());
@@ -1783,50 +1753,6 @@ mod tests {
             .unwrap()
     }
 
-    async fn make_large_vector_table(tmp_dir: &tempfile::TempDir, rows: usize) -> Table {
-        let dataset_path = tmp_dir.path().join("large_test.lance");
-        let uri = dataset_path.to_str().unwrap();
-
-        let schema = Arc::new(ArrowSchema::new(vec![
-            ArrowField::new("id", DataType::Utf8, false),
-            ArrowField::new(
-                "vector",
-                DataType::FixedSizeList(
-                    Arc::new(ArrowField::new("item", DataType::Float32, true)),
-                    4,
-                ),
-                false,
-            ),
-        ]));
-
-        let ids = StringArray::from_iter_values((0..rows).map(|i| format!("row-{i}")));
-        let vectors = FixedSizeListArray::from_iter_primitive::<Float32Type, _, _>(
-            (0..rows).map(|i| Some(vec![Some(i as f32), Some(1.0), Some(2.0), Some(3.0)])),
-            4,
-        );
-        let batch =
-            RecordBatch::try_new(schema.clone(), vec![Arc::new(ids), Arc::new(vectors)]).unwrap();
-
-        let conn = connect(uri).execute().await.unwrap();
-        conn.create_table("my_table", vec![batch])
-            .execute()
-            .await
-            .unwrap()
-    }
-
-    async fn assert_stream_batches_at_most(
-        mut results: SendableRecordBatchStream,
-        max_batch_length: usize,
-    ) {
-        let mut saw_batch = false;
-        while let Some(batch) = results.next().await {
-            let batch = batch.unwrap();
-            saw_batch = true;
-            assert!(batch.num_rows() <= max_batch_length);
-        }
-        assert!(saw_batch);
-    }
-
     #[tokio::test]
     async fn test_execute_with_options() {
         let tmp_dir = tempdir().unwrap();
@@ -1846,83 +1772,6 @@ mod tests {
         }
     }
 
-    #[tokio::test]
-    async fn test_vector_query_execute_with_options_respects_max_batch_length() {
-        let tmp_dir = tempdir().unwrap();
-        let table = make_large_vector_table(&tmp_dir, 10_000).await;
-
-        let results = table
-            .query()
-            .nearest_to(vec![0.0, 1.0, 2.0, 3.0])
-            .unwrap()
-            .limit(10_000)
-            .execute_with_options(QueryExecutionOptions {
-                max_batch_length: 100,
-                ..Default::default()
-            })
-            .await
-            .unwrap();
-        assert_stream_batches_at_most(results, 100).await;
-    }
-
-    #[tokio::test]
-    async fn test_hybrid_query_execute_with_options_respects_max_batch_length() {
-        let tmp_dir = tempdir().unwrap();
-        let dataset_path = tmp_dir.path();
-        let conn = connect(dataset_path.to_str().unwrap())
-            .execute()
-            .await
-            .unwrap();
-
-        let dims = 2;
-        let rows = 512;
-        let schema = Arc::new(ArrowSchema::new(vec![
-            ArrowField::new("text", DataType::Utf8, false),
-            ArrowField::new(
-                "vector",
-                DataType::FixedSizeList(
-                    Arc::new(ArrowField::new("item", DataType::Float32, true)),
-                    dims,
-                ),
-                false,
-            ),
-        ]));
-
-        let text = StringArray::from_iter_values((0..rows).map(|_| "match"));
-        let vectors = FixedSizeListArray::from_iter_primitive::<Float32Type, _, _>(
-            (0..rows).map(|i| Some(vec![Some(i as f32), Some(0.0)])),
-            dims,
-        );
-        let record_batch =
-            RecordBatch::try_new(schema.clone(), vec![Arc::new(text), Arc::new(vectors)]).unwrap();
-        let table = conn
-            .create_table("my_table", record_batch)
-            .execute()
-            .await
-            .unwrap();
-
-        table
-            .create_index(&["text"], crate::index::Index::FTS(Default::default()))
-            .replace(true)
-            .execute()
-            .await
-            .unwrap();
-
-        let results = table
-            .query()
-            .full_text_search(FullTextSearchQuery::new("match".to_string()))
-            .limit(rows)
-            .nearest_to(&[0.0, 0.0])
-            .unwrap()
-            .execute_with_options(QueryExecutionOptions {
-                max_batch_length: 100,
-                ..Default::default()
-            })
-            .await
-            .unwrap();
-        assert_stream_batches_at_most(results, 100).await;
-    }
-
     #[tokio::test]
     async fn test_analyze_plan() {
         let tmp_dir = tempdir().unwrap();

rust/lancedb/src/remote/client.rs

@@ -443,13 +443,23 @@ impl<S: HttpSend> RestfulLanceDbClient<S> {
                 })?,
             );
         }
-        if let Some(v) = options.0.get("azure_storage_account_name") {
-            headers.insert(
-                HeaderName::from_static("x-azure-storage-account-name"),
-                HeaderValue::from_str(v).map_err(|_| Error::InvalidInput {
-                    message: format!("non-ascii storage account name '{}' provided", db_name),
-                })?,
-            );
+        // Map azure storage options to x-azure-* headers.
+        // The option key uses underscores (e.g. "azure_client_id") while the
+        // header uses hyphens (e.g. "x-azure-client-id").
+        let azure_opts: [(&str, &str); 3] = [
+            ("azure_storage_account_name", "x-azure-storage-account-name"),
+            ("azure_client_id", "x-azure-client-id"),
+            ("azure_tenant_id", "x-azure-tenant-id"),
+        ];
+        for (opt_key, header_name) in azure_opts {
+            if let Some(v) = options.0.get(opt_key) {
+                headers.insert(
+                    HeaderName::from_static(header_name),
+                    HeaderValue::from_str(v).map_err(|_| Error::InvalidInput {
+                        message: format!("non-ascii value '{}' for option '{}'", v, opt_key),
+                    })?,
+                );
+            }
         }
 
         for (key, value) in &config.extra_headers {
@@ -1072,4 +1082,34 @@ mod tests {
             _ => panic!("Expected Runtime error"),
         }
     }
+
+    #[test]
+    fn test_default_headers_azure_opts() {
+        let mut opts = HashMap::new();
+        opts.insert(
+            "azure_storage_account_name".to_string(),
+            "myaccount".to_string(),
+        );
+        opts.insert("azure_client_id".to_string(), "my-client-id".to_string());
+        opts.insert("azure_tenant_id".to_string(), "my-tenant-id".to_string());
+        let remote_opts = RemoteOptions::new(opts);
+
+        let headers = RestfulLanceDbClient::<Sender>::default_headers(
+            "test-key",
+            "us-east-1",
+            "testdb",
+            false,
+            &remote_opts,
+            None,
+            &ClientConfig::default(),
+        )
+        .unwrap();
+
+        assert_eq!(
+            headers.get("x-azure-storage-account-name").unwrap(),
+            "myaccount"
+        );
+        assert_eq!(headers.get("x-azure-client-id").unwrap(), "my-client-id");
+        assert_eq!(headers.get("x-azure-tenant-id").unwrap(), "my-tenant-id");
+    }
 }

rust/lancedb/src/remote/db.rs

@@ -72,10 +72,6 @@ impl ServerVersion {
     pub fn support_structural_fts(&self) -> bool {
         self.0 >= semver::Version::new(0, 3, 0)
     }
-
-    pub fn support_multipart_write(&self) -> bool {
-        self.0 >= semver::Version::new(0, 4, 0)
-    }
 }
 
 pub const OPT_REMOTE_PREFIX: &str = "remote_database_";
@@ -782,7 +778,12 @@ impl RemoteOptions {
 
 impl From<StorageOptions> for RemoteOptions {
     fn from(options: StorageOptions) -> Self {
-        let supported_opts = vec!["account_name", "azure_storage_account_name"];
+        let supported_opts = vec![
+            "account_name",
+            "azure_storage_account_name",
+            "azure_client_id",
+            "azure_tenant_id",
+        ];
         let mut filtered = HashMap::new();
         for opt in supported_opts {
             if let Some(v) = options.0.get(opt) {

rust/lancedb/src/remote/table/insert.rs

@@ -11,14 +11,10 @@ use arrow_ipc::CompressionType;
 use datafusion_common::{DataFusionError, Result as DataFusionResult};
 use datafusion_execution::{SendableRecordBatchStream, TaskContext};
 use datafusion_physical_expr::EquivalenceProperties;
-use datafusion_physical_plan::metrics::{ExecutionPlanMetricsSet, MetricsSet};
 use datafusion_physical_plan::stream::RecordBatchStreamAdapter;
-use datafusion_physical_plan::{
-    DisplayAs, DisplayFormatType, ExecutionPlan, ExecutionPlanProperties, PlanProperties,
-};
+use datafusion_physical_plan::{DisplayAs, DisplayFormatType, ExecutionPlan, PlanProperties};
 use futures::StreamExt;
 use http::header::CONTENT_TYPE;
-use lance::io::exec::utils::InstrumentedRecordBatchStreamAdapter;
 
 use crate::Error;
 use crate::remote::ARROW_STREAM_CONTENT_TYPE;
@@ -26,16 +22,13 @@ use crate::remote::client::{HttpSend, RestfulLanceDbClient, Sender};
 use crate::remote::table::RemoteTable;
 use crate::table::AddResult;
 use crate::table::datafusion::insert::COUNT_SCHEMA;
-use crate::table::write_progress::WriteProgressTracker;
 
 /// ExecutionPlan for inserting data into a remote LanceDB table.
 ///
-/// Streams data as Arrow IPC to `/v1/table/{id}/insert/` endpoint.
-///
-/// When `upload_id` is set, inserts are staged as part of a multipart write
-/// session and the plan supports multiple partitions for parallel uploads.
-/// Without `upload_id`, the plan requires a single partition and commits
-/// immediately.
+/// This plan:
+/// 1. Requires single partition (no parallel remote inserts yet)
+/// 2. Streams data as Arrow IPC to `/v1/table/{id}/insert/` endpoint
+/// 3. Stores AddResult for retrieval after execution
 #[derive(Debug)]
 pub struct RemoteInsertExec<S: HttpSend = Sender> {
     table_name: String,
@@ -45,69 +38,21 @@ pub struct RemoteInsertExec<S: HttpSend = Sender> {
     overwrite: bool,
     properties: PlanProperties,
     add_result: Arc<Mutex<Option<AddResult>>>,
-    metrics: ExecutionPlanMetricsSet,
-    upload_id: Option<String>,
-    tracker: Option<Arc<WriteProgressTracker>>,
 }
 
 impl<S: HttpSend + 'static> RemoteInsertExec<S> {
-    /// Create a new single-partition RemoteInsertExec.
+    /// Create a new RemoteInsertExec.
     pub fn new(
         table_name: String,
         identifier: String,
         client: RestfulLanceDbClient<S>,
         input: Arc<dyn ExecutionPlan>,
         overwrite: bool,
-        tracker: Option<Arc<WriteProgressTracker>>,
     ) -> Self {
-        Self::new_inner(
-            table_name, identifier, client, input, overwrite, None, tracker,
-        )
-    }
-
-    /// Create a multi-partition RemoteInsertExec for use with multipart writes.
-    ///
-    /// Each partition's insert is staged under the given `upload_id` without
-    /// committing. The caller is responsible for calling the complete (or abort)
-    /// endpoint after all partitions finish.
-    pub fn new_multipart(
-        table_name: String,
-        identifier: String,
-        client: RestfulLanceDbClient<S>,
-        input: Arc<dyn ExecutionPlan>,
-        overwrite: bool,
-        upload_id: String,
-        tracker: Option<Arc<WriteProgressTracker>>,
-    ) -> Self {
-        Self::new_inner(
-            table_name,
-            identifier,
-            client,
-            input,
-            overwrite,
-            Some(upload_id),
-            tracker,
-        )
-    }
-
-    fn new_inner(
-        table_name: String,
-        identifier: String,
-        client: RestfulLanceDbClient<S>,
-        input: Arc<dyn ExecutionPlan>,
-        overwrite: bool,
-        upload_id: Option<String>,
-        tracker: Option<Arc<WriteProgressTracker>>,
-    ) -> Self {
-        let num_partitions = if upload_id.is_some() {
-            input.output_partitioning().partition_count()
-        } else {
-            1
-        };
         let schema = COUNT_SCHEMA.clone();
         let properties = PlanProperties::new(
             EquivalenceProperties::new(schema),
-            datafusion_physical_plan::Partitioning::UnknownPartitioning(num_partitions),
+            datafusion_physical_plan::Partitioning::UnknownPartitioning(1),
             datafusion_physical_plan::execution_plan::EmissionType::Final,
             datafusion_physical_plan::execution_plan::Boundedness::Bounded,
         );
@@ -120,9 +65,6 @@ impl<S: HttpSend + 'static> RemoteInsertExec<S> {
             overwrite,
             properties,
             add_result: Arc::new(Mutex::new(None)),
-            metrics: ExecutionPlanMetricsSet::new(),
-            upload_id,
-            tracker,
         }
     }
 
@@ -130,10 +72,7 @@ impl<S: HttpSend + 'static> RemoteInsertExec<S> {
     // TODO: this will be used when we wire this up to Table::add().
     #[allow(dead_code)]
     pub fn add_result(&self) -> Option<AddResult> {
-        self.add_result
-            .lock()
-            .unwrap_or_else(|e| e.into_inner())
-            .clone()
+        self.add_result.lock().unwrap().clone()
     }
 
     /// Stream the input into an HTTP body as an Arrow IPC stream, capturing any
@@ -144,7 +83,6 @@ impl<S: HttpSend + 'static> RemoteInsertExec<S> {
     fn stream_as_http_body(
         data: SendableRecordBatchStream,
         error_tx: tokio::sync::oneshot::Sender<DataFusionError>,
-        tracker: Option<Arc<WriteProgressTracker>>,
     ) -> DataFusionResult<reqwest::Body> {
         let options = arrow_ipc::writer::IpcWriteOptions::default()
             .try_with_compression(Some(CompressionType::LZ4_FRAME))?;
@@ -156,46 +94,37 @@ impl<S: HttpSend + 'static> RemoteInsertExec<S> {
 
         let stream = futures::stream::try_unfold(
             (data, writer, Some(error_tx), false),
-            move |(mut data, mut writer, error_tx, finished)| {
-                let tracker = tracker.clone();
-                async move {
-                    if finished {
-                        return Ok(None);
+            move |(mut data, mut writer, error_tx, finished)| async move {
+                if finished {
+                    return Ok(None);
+                }
+                match data.next().await {
+                    Some(Ok(batch)) => {
+                        writer
+                            .write(&batch)
+                            .map_err(|e| std::io::Error::other(e.to_string()))?;
+                        let buffer = std::mem::take(writer.get_mut());
+                        Ok(Some((buffer, (data, writer, error_tx, false))))
                     }
-                    match data.next().await {
-                        Some(Ok(batch)) => {
-                            writer
-                                .write(&batch)
-                                .map_err(|e| std::io::Error::other(e.to_string()))?;
-                            let buffer = std::mem::take(writer.get_mut());
-                            if let Some(ref t) = tracker {
-                                t.record_bytes(buffer.len());
-                            }
-                            Ok(Some((buffer, (data, writer, error_tx, false))))
+                    Some(Err(e)) => {
+                        // Send the original error through the channel before
+                        // returning a generic error to reqwest.
+                        if let Some(tx) = error_tx {
+                            let _ = tx.send(e);
                         }
-                        Some(Err(e)) => {
-                            // Send the original error through the channel before
-                            // returning a generic error to reqwest.
-                            if let Some(tx) = error_tx {
-                                let _ = tx.send(e);
-                            }
-                            Err(std::io::Error::other(
-                                "input stream error (see error channel)",
-                            ))
-                        }
-                        None => {
-                            writer
-                                .finish()
-                                .map_err(|e| std::io::Error::other(e.to_string()))?;
-                            let buffer = std::mem::take(writer.get_mut());
-                            if buffer.is_empty() {
-                                Ok(None)
-                            } else {
-                                if let Some(ref t) = tracker {
-                                    t.record_bytes(buffer.len());
-                                }
-                                Ok(Some((buffer, (data, writer, None, true))))
-                            }
+                        Err(std::io::Error::other(
+                            "input stream error (see error channel)",
+                        ))
+                    }
+                    None => {
+                        writer
+                            .finish()
+                            .map_err(|e| std::io::Error::other(e.to_string()))?;
+                        let buffer = std::mem::take(writer.get_mut());
+                        if buffer.is_empty() {
+                            Ok(None)
+                        } else {
+                            Ok(Some((buffer, (data, writer, None, true))))
                         }
                     }
                 }
@@ -245,11 +174,8 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
     }
 
     fn required_input_distribution(&self) -> Vec<datafusion_physical_plan::Distribution> {
-        if self.upload_id.is_some() {
-            vec![datafusion_physical_plan::Distribution::UnspecifiedDistribution]
-        } else {
-            vec![datafusion_physical_plan::Distribution::SinglePartition]
-        }
+        // Until we have a separate commit endpoint, we need to do all inserts in a single partition
+        vec![datafusion_physical_plan::Distribution::SinglePartition]
     }
 
     fn benefits_from_input_partitioning(&self) -> Vec<bool> {
@@ -265,14 +191,12 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
                 "RemoteInsertExec requires exactly one child".to_string(),
             ));
         }
-        Ok(Arc::new(Self::new_inner(
+        Ok(Arc::new(Self::new(
             self.table_name.clone(),
             self.identifier.clone(),
             self.client.clone(),
             children[0].clone(),
             self.overwrite,
-            self.upload_id.clone(),
-            self.tracker.clone(),
         )))
     }
 
@@ -281,29 +205,18 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
         partition: usize,
         context: Arc<TaskContext>,
     ) -> DataFusionResult<SendableRecordBatchStream> {
-        if self.upload_id.is_none() && partition != 0 {
+        if partition != 0 {
             return Err(DataFusionError::Internal(
-                "RemoteInsertExec only supports single partition execution without upload_id"
-                    .to_string(),
+                "RemoteInsertExec only supports single partition execution".to_string(),
             ));
         }
 
-        let input_stream = self.input.execute(partition, context)?;
-        let input_schema = input_stream.schema();
-        let input_stream: SendableRecordBatchStream =
-            Box::pin(InstrumentedRecordBatchStreamAdapter::new(
-                input_schema,
-                input_stream,
-                partition,
-                &self.metrics,
-            ));
+        let input_stream = self.input.execute(0, context)?;
         let client = self.client.clone();
         let identifier = self.identifier.clone();
         let overwrite = self.overwrite;
         let add_result = self.add_result.clone();
         let table_name = self.table_name.clone();
-        let upload_id = self.upload_id.clone();
-        let tracker = self.tracker.clone();
 
         let stream = futures::stream::once(async move {
             let mut request = client
@@ -313,12 +226,9 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
             if overwrite {
                 request = request.query(&[("mode", "overwrite")]);
             }
-            if let Some(ref uid) = upload_id {
-                request = request.query(&[("upload_id", uid.as_str())]);
-            }
 
             let (error_tx, mut error_rx) = tokio::sync::oneshot::channel();
-            let body = Self::stream_as_http_body(input_stream, error_tx, tracker)?;
+            let body = Self::stream_as_http_body(input_stream, error_tx)?;
             let request = request.body(body);
 
             let result: DataFusionResult<(String, _)> = async {
@@ -352,43 +262,32 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
 
             let (request_id, response) = result?;
 
-            // For multipart writes, the staging response is not the final
-            // version. Only parse AddResult for non-multipart inserts.
-            if upload_id.is_none() {
-                let body_text = response.text().await.map_err(|e| {
+            let body_text = response.text().await.map_err(|e| {
+                DataFusionError::External(Box::new(Error::Http {
+                    source: Box::new(e),
+                    request_id: request_id.clone(),
+                    status_code: None,
+                }))
+            })?;
+
+            let parsed_result = if body_text.trim().is_empty() {
+                // Backward compatible with old servers
+                AddResult { version: 0 }
+            } else {
+                serde_json::from_str(&body_text).map_err(|e| {
                     DataFusionError::External(Box::new(Error::Http {
-                        source: Box::new(e),
+                        source: format!("Failed to parse add response: {}", e).into(),
                         request_id: request_id.clone(),
                         status_code: None,
                     }))
-                })?;
-
-                let parsed_result = if body_text.trim().is_empty() {
-                    // Backward compatible with old servers
-                    AddResult { version: 0 }
-                } else {
-                    serde_json::from_str(&body_text).map_err(|e| {
-                        DataFusionError::External(Box::new(Error::Http {
-                            source: format!("Failed to parse add response: {}", e).into(),
-                            request_id: request_id.clone(),
-                            status_code: None,
-                        }))
-                    })?
-                };
+                })?
+            };
 
+            {
                 let mut res_lock = add_result.lock().map_err(|_| {
                     DataFusionError::Execution("Failed to acquire lock for add_result".to_string())
                 })?;
                 *res_lock = Some(parsed_result);
-            } else {
-                // We don't use the body in this case, but we should still consume it.
-                let _ = response.bytes().await.map_err(|e| {
-                    DataFusionError::External(Box::new(Error::Http {
-                        source: Box::new(e),
-                        request_id: request_id.clone(),
-                        status_code: None,
-                    }))
-                })?;
             }
 
             // Return a single batch with count 0 (actual count is tracked in add_result)
@@ -402,10 +301,6 @@ impl<S: HttpSend + 'static> ExecutionPlan for RemoteInsertExec<S> {
             stream,
         )))
     }
-
-    fn metrics(&self) -> Option<MetricsSet> {
-        Some(self.metrics.clone_inner())
-    }
 }
 
 #[cfg(test)]

rust/lancedb/src/table.rs

@@ -74,10 +74,7 @@ pub mod optimize;
 pub mod query;
 pub mod schema_evolution;
 pub mod update;
-pub mod write_progress;
 use crate::index::waiter::wait_for_index;
-#[cfg(feature = "remote")]
-pub(crate) use add_data::PreprocessingOutput;
 pub use add_data::{AddDataBuilder, AddDataMode, AddResult, NaNVectorBehavior};
 pub use chrono::Duration;
 pub use delete::DeleteResult;
@@ -443,34 +440,6 @@ mod test_utils {
                 embedding_registry: Arc::new(MemoryRegistry::new()),
             }
         }
-
-        pub fn new_with_handler_version_and_config<T>(
-            name: impl Into<String>,
-            version: semver::Version,
-            handler: impl Fn(reqwest::Request) -> http::Response<T> + Clone + Send + Sync + 'static,
-            config: crate::remote::ClientConfig,
-        ) -> Self
-        where
-            T: Into<reqwest::Body>,
-        {
-            let inner = Arc::new(
-                crate::remote::table::RemoteTable::new_mock_with_version_and_config(
-                    name.into(),
-                    handler.clone(),
-                    Some(version),
-                    config.clone(),
-                ),
-            );
-            let database = Arc::new(crate::remote::db::RemoteDatabase::new_mock_with_config(
-                handler, config,
-            ));
-            Self {
-                inner,
-                database: Some(database),
-                // Registry is unused.
-                embedding_registry: Arc::new(MemoryRegistry::new()),
-            }
-        }
     }
 }
 
@@ -2229,26 +2198,21 @@ impl BaseTable for NativeTable {
 
         let table_schema = Schema::from(&ds.schema().clone());
 
-        let num_partitions = if let Some(parallelism) = add.write_parallelism {
-            parallelism
+        // Peek at the first batch to estimate a good partition count for
+        // write parallelism.
+        let mut peeked = PeekedScannable::new(add.data);
+        let num_partitions = if let Some(first_batch) = peeked.peek().await {
+            let max_partitions = lance_core::utils::tokio::get_num_compute_intensive_cpus();
+            estimate_write_partitions(
+                first_batch.get_array_memory_size(),
+                first_batch.num_rows(),
+                peeked.num_rows(),
+                max_partitions,
+            )
         } else {
-            // Peek at the first batch to estimate a good partition count for
-            // write parallelism.
-            let mut peeked = PeekedScannable::new(add.data);
-            let n = if let Some(first_batch) = peeked.peek().await {
-                let max_partitions = lance_core::utils::tokio::get_num_compute_intensive_cpus();
-                estimate_write_partitions(
-                    first_batch.get_array_memory_size(),
-                    first_batch.num_rows(),
-                    peeked.num_rows(),
-                    max_partitions,
-                )
-            } else {
-                1
-            };
-            add.data = Box::new(peeked);
-            n
+            1
         };
+        add.data = Box::new(peeked);
 
         let output = add.into_plan(&table_schema, &table_def)?;
 
@@ -2277,21 +2241,13 @@ impl BaseTable for NativeTable {
 
         let insert_exec = Arc::new(InsertExec::new(ds_wrapper.clone(), ds, plan, lance_params));
 
-        let tracker_for_tasks = output.tracker.clone();
-        if let Some(ref t) = tracker_for_tasks {
-            t.set_total_tasks(num_partitions);
-        }
-        let _finish = write_progress::FinishOnDrop(output.tracker);
-
         // Execute all partitions in parallel.
         let task_ctx = Arc::new(TaskContext::default());
         let handles = FuturesUnordered::new();
         for partition in 0..num_partitions {
             let exec = insert_exec.clone();
             let ctx = task_ctx.clone();
-            let tracker = tracker_for_tasks.clone();
             handles.push(tokio::spawn(async move {
-                let _guard = tracker.as_ref().map(|t| t.track_task());
                 let mut stream = exec
                     .execute(partition, ctx)
                     .map_err(|e| -> Error { e.into() })?;

rust/lancedb/src/table/add_data.rs

@@ -13,9 +13,6 @@ use crate::embeddings::EmbeddingRegistry;
 use crate::table::datafusion::cast::cast_to_table_schema;
 use crate::table::datafusion::reject_nan::reject_nan_vectors;
 use crate::table::datafusion::scannable_exec::ScannableExec;
-use crate::table::write_progress::ProgressCallback;
-use crate::table::write_progress::WriteProgress;
-use crate::table::write_progress::WriteProgressTracker;
 use crate::{Error, Result};
 
 use super::{BaseTable, TableDefinition, WriteOptions};
@@ -55,8 +52,6 @@ pub struct AddDataBuilder {
     pub(crate) write_options: WriteOptions,
     pub(crate) on_nan_vectors: NaNVectorBehavior,
     pub(crate) embedding_registry: Option<Arc<dyn EmbeddingRegistry>>,
-    pub(crate) progress_callback: Option<ProgressCallback>,
-    pub(crate) write_parallelism: Option<usize>,
 }
 
 impl std::fmt::Debug for AddDataBuilder {
@@ -82,8 +77,6 @@ impl AddDataBuilder {
             write_options: WriteOptions::default(),
             on_nan_vectors: NaNVectorBehavior::default(),
             embedding_registry,
-            progress_callback: None,
-            write_parallelism: None,
         }
     }
 
@@ -108,43 +101,7 @@ impl AddDataBuilder {
         self
     }
 
-    /// Set a callback to receive progress updates during the add operation.
-    ///
-    /// The callback is invoked once per batch written, and once more with
-    /// [`WriteProgress::done`] set to `true` when the write completes.
-    ///
-    /// ```
-    /// # use lancedb::Table;
-    /// # async fn example(table: &Table) -> Result<(), Box<dyn std::error::Error>> {
-    /// let batch = arrow_array::record_batch!(("id", Int32, [1, 2, 3])).unwrap();
-    /// table.add(batch)
-    ///     .progress(|p| println!("{}/{:?} rows", p.output_rows(), p.total_rows()))
-    ///     .execute()
-    ///     .await?;
-    /// # Ok(())
-    /// # }
-    /// ```
-    pub fn progress(mut self, callback: impl FnMut(&WriteProgress) + Send + 'static) -> Self {
-        self.progress_callback = Some(Arc::new(std::sync::Mutex::new(callback)));
-        self
-    }
-
-    /// Set the number of parallel write streams.
-    ///
-    /// By default, the number of streams is estimated from the data size.
-    /// Setting this to `1` disables parallel writes.
-    pub fn write_parallelism(mut self, parallelism: usize) -> Self {
-        self.write_parallelism = Some(parallelism);
-        self
-    }
-
     pub async fn execute(self) -> Result<AddResult> {
-        if self.write_parallelism.map(|p| p == 0).unwrap_or(false) {
-            return Err(Error::InvalidInput {
-                message: "write_parallelism must be greater than 0".to_string(),
-            });
-        }
-
         self.parent.clone().add(self).await
     }
 
@@ -173,11 +130,8 @@ impl AddDataBuilder {
             scannable_with_embeddings(self.data, table_def, self.embedding_registry.as_ref())?;
 
         let rescannable = self.data.rescannable();
-        let tracker = self
-            .progress_callback
-            .map(|cb| Arc::new(WriteProgressTracker::new(cb, self.data.num_rows())));
         let plan: Arc<dyn datafusion_physical_plan::ExecutionPlan> =
-            Arc::new(ScannableExec::new(self.data, tracker.clone()));
+            Arc::new(ScannableExec::new(self.data));
         // Skip casting when overwriting — the input schema replaces the table schema.
         let plan = if overwrite {
             plan
@@ -195,7 +149,6 @@ impl AddDataBuilder {
             rescannable,
             write_options: self.write_options,
             mode: self.mode,
-            tracker,
         })
     }
 }
@@ -208,7 +161,6 @@ pub struct PreprocessingOutput {
     pub rescannable: bool,
     pub write_options: WriteOptions,
     pub mode: AddDataMode,
-    pub tracker: Option<Arc<WriteProgressTracker>>,
 }
 
 /// Check that the input schema is valid for insert.

rust/lancedb/src/table/datafusion/insert.rs

@@ -12,16 +12,13 @@ use datafusion_common::{DataFusionError, Result as DataFusionResult};
 use datafusion_execution::{SendableRecordBatchStream, TaskContext};
 use datafusion_physical_expr::{EquivalenceProperties, Partitioning};
 use datafusion_physical_plan::execution_plan::{Boundedness, EmissionType};
-use datafusion_physical_plan::metrics::{ExecutionPlanMetricsSet, MetricBuilder, MetricsSet};
 use datafusion_physical_plan::stream::RecordBatchStreamAdapter;
 use datafusion_physical_plan::{
     DisplayAs, DisplayFormatType, ExecutionPlan, ExecutionPlanProperties, PlanProperties,
 };
-use futures::TryStreamExt;
 use lance::Dataset;
 use lance::dataset::transaction::{Operation, Transaction};
 use lance::dataset::{CommitBuilder, InsertBuilder, WriteParams};
-use lance::io::exec::utils::InstrumentedRecordBatchStreamAdapter;
 use lance_table::format::Fragment;
 
 use crate::table::dataset::DatasetConsistencyWrapper;
@@ -83,7 +80,6 @@ pub struct InsertExec {
     write_params: WriteParams,
     properties: PlanProperties,
     partial_transactions: Arc<Mutex<Vec<Transaction>>>,
-    metrics: ExecutionPlanMetricsSet,
 }
 
 impl InsertExec {
@@ -109,7 +105,6 @@ impl InsertExec {
             write_params,
             properties,
             partial_transactions: Arc::new(Mutex::new(Vec::with_capacity(num_partitions))),
-            metrics: ExecutionPlanMetricsSet::new(),
         }
     }
 }
@@ -181,19 +176,6 @@ impl ExecutionPlan for InsertExec {
         let total_partitions = self.input.output_partitioning().partition_count();
         let ds_wrapper = self.ds_wrapper.clone();
 
-        let output_bytes = MetricBuilder::new(&self.metrics).output_bytes(partition);
-        let input_schema = input_stream.schema();
-        let input_stream: SendableRecordBatchStream =
-            Box::pin(InstrumentedRecordBatchStreamAdapter::new(
-                input_schema,
-                input_stream.map_ok(move |batch| {
-                    output_bytes.add(batch.get_array_memory_size());
-                    batch
-                }),
-                partition,
-                &self.metrics,
-            ));
-
         let stream = futures::stream::once(async move {
             let transaction = InsertBuilder::new(dataset.clone())
                 .with_params(&write_params)
@@ -204,9 +186,7 @@ impl ExecutionPlan for InsertExec {
 
             let to_commit = {
                 // Don't hold the lock over an await point.
-                let mut txns = partial_transactions
-                    .lock()
-                    .unwrap_or_else(|e| e.into_inner());
+                let mut txns = partial_transactions.lock().unwrap();
                 txns.push(transaction);
                 if txns.len() == total_partitions {
                     Some(std::mem::take(&mut *txns))
@@ -235,10 +215,6 @@ impl ExecutionPlan for InsertExec {
             stream,
         )))
     }
-
-    fn metrics(&self) -> Option<MetricsSet> {
-        Some(self.metrics.clone_inner())
-    }
 }
 
 #[cfg(test)]

rust/lancedb/src/table/datafusion/scannable_exec.rs

@@ -7,21 +7,17 @@ use std::sync::{Arc, Mutex};
 use datafusion_common::{DataFusionError, Result as DFResult, Statistics, stats::Precision};
 use datafusion_execution::{SendableRecordBatchStream, TaskContext};
 use datafusion_physical_expr::{EquivalenceProperties, Partitioning};
-use datafusion_physical_plan::stream::RecordBatchStreamAdapter;
 use datafusion_physical_plan::{
     DisplayAs, DisplayFormatType, ExecutionPlan, PlanProperties, execution_plan::EmissionType,
 };
-use futures::TryStreamExt;
 
-use crate::table::write_progress::WriteProgressTracker;
 use crate::{arrow::SendableRecordBatchStreamExt, data::scannable::Scannable};
 
-pub(crate) struct ScannableExec {
-    // We don't require Scannable to be Sync, so we wrap it in a Mutex to allow safe concurrent access.
+pub struct ScannableExec {
+    // We don't require Scannable to by Sync, so we wrap it in a Mutex to allow safe concurrent access.
     source: Mutex<Box<dyn Scannable>>,
     num_rows: Option<usize>,
     properties: PlanProperties,
-    tracker: Option<Arc<WriteProgressTracker>>,
 }
 
 impl std::fmt::Debug for ScannableExec {
@@ -34,7 +30,7 @@ impl std::fmt::Debug for ScannableExec {
 }
 
 impl ScannableExec {
-    pub fn new(source: Box<dyn Scannable>, tracker: Option<Arc<WriteProgressTracker>>) -> Self {
+    pub fn new(source: Box<dyn Scannable>) -> Self {
         let schema = source.schema();
         let eq_properties = EquivalenceProperties::new(schema);
         let properties = PlanProperties::new(
@@ -50,7 +46,6 @@ impl ScannableExec {
             source,
             num_rows,
             properties,
-            tracker,
         }
     }
 }
@@ -107,18 +102,7 @@ impl ExecutionPlan for ScannableExec {
             Err(poison) => poison.into_inner().scan_as_stream(),
         };
 
-        let tracker = self.tracker.clone();
-        let stream = stream.into_df_stream().map_ok(move |batch| {
-            if let Some(ref t) = tracker {
-                t.record_batch(batch.num_rows(), batch.get_array_memory_size());
-            }
-            batch
-        });
-
-        Ok(Box::pin(RecordBatchStreamAdapter::new(
-            self.schema(),
-            stream,
-        )))
+        Ok(stream.into_df_stream())
     }
 
     fn partition_statistics(&self, _partition: Option<usize>) -> DFResult<Statistics> {

rust/lancedb/src/table/dataset.rs

@@ -82,7 +82,7 @@ impl DatasetConsistencyWrapper {
     /// pinned dataset regardless of consistency mode.
     pub async fn get(&self) -> Result<Arc<Dataset>> {
         {
-            let state = self.state.lock()?;
+            let state = self.state.lock().unwrap();
             if state.pinned_version.is_some() {
                 return Ok(state.dataset.clone());
             }
@@ -101,7 +101,7 @@ impl DatasetConsistencyWrapper {
             }
             ConsistencyMode::Strong => refresh_latest(self.state.clone()).await,
             ConsistencyMode::Lazy => {
-                let state = self.state.lock()?;
+                let state = self.state.lock().unwrap();
                 Ok(state.dataset.clone())
             }
         }
@@ -116,7 +116,7 @@ impl DatasetConsistencyWrapper {
     /// concurrent [`as_time_travel`](Self::as_time_travel) call), the update
     /// is silently ignored — the write already committed to storage.
     pub fn update(&self, dataset: Dataset) {
-        let mut state = self.state.lock().unwrap_or_else(|e| e.into_inner());
+        let mut state = self.state.lock().unwrap();
         if state.pinned_version.is_some() {
             // A concurrent as_time_travel() beat us here. The write succeeded
             // in storage, but since we're now pinned we don't advance the
@@ -139,7 +139,7 @@ impl DatasetConsistencyWrapper {
 
     /// Check that the dataset is in a mutable mode (Latest).
     pub fn ensure_mutable(&self) -> Result<()> {
-        let state = self.state.lock()?;
+        let state = self.state.lock().unwrap();
         if state.pinned_version.is_some() {
             Err(crate::Error::InvalidInput {
                 message: "table cannot be modified when a specific version is checked out"
@@ -152,16 +152,13 @@ impl DatasetConsistencyWrapper {
 
     /// Returns the version, if in time travel mode, or None otherwise.
     pub fn time_travel_version(&self) -> Option<u64> {
-        self.state
-            .lock()
-            .unwrap_or_else(|e| e.into_inner())
-            .pinned_version
+        self.state.lock().unwrap().pinned_version
     }
 
     /// Convert into a wrapper in latest version mode.
     pub async fn as_latest(&self) -> Result<()> {
         let dataset = {
-            let state = self.state.lock()?;
+            let state = self.state.lock().unwrap();
             if state.pinned_version.is_none() {
                 return Ok(());
             }
@@ -171,7 +168,7 @@ impl DatasetConsistencyWrapper {
         let latest_version = dataset.latest_version_id().await?;
         let new_dataset = dataset.checkout_version(latest_version).await?;
 
-        let mut state = self.state.lock()?;
+        let mut state = self.state.lock().unwrap();
         if state.pinned_version.is_some() {
             state.dataset = Arc::new(new_dataset);
             state.pinned_version = None;
@@ -187,7 +184,7 @@ impl DatasetConsistencyWrapper {
         let target_ref = target_version.into();
 
         let (should_checkout, dataset) = {
-            let state = self.state.lock()?;
+            let state = self.state.lock().unwrap();
             let should = match state.pinned_version {
                 None => true,
                 Some(version) => match &target_ref {
@@ -207,7 +204,7 @@ impl DatasetConsistencyWrapper {
         let new_dataset = dataset.checkout_version(target_ref).await?;
         let version_value = new_dataset.version().version;
 
-        let mut state = self.state.lock()?;
+        let mut state = self.state.lock().unwrap();
         state.dataset = Arc::new(new_dataset);
         state.pinned_version = Some(version_value);
         Ok(())
@@ -215,7 +212,7 @@ impl DatasetConsistencyWrapper {
 
     pub async fn reload(&self) -> Result<()> {
         let (dataset, pinned_version) = {
-            let state = self.state.lock()?;
+            let state = self.state.lock().unwrap();
             (state.dataset.clone(), state.pinned_version)
         };
 
@@ -233,7 +230,7 @@ impl DatasetConsistencyWrapper {
 
                 let new_dataset = dataset.checkout_version(version).await?;
 
-                let mut state = self.state.lock()?;
+                let mut state = self.state.lock().unwrap();
                 if state.pinned_version == Some(version) {
                     state.dataset = Arc::new(new_dataset);
                 }
@@ -245,14 +242,14 @@ impl DatasetConsistencyWrapper {
 }
 
 async fn refresh_latest(state: Arc<Mutex<DatasetState>>) -> Result<Arc<Dataset>> {
-    let dataset = { state.lock()?.dataset.clone() };
+    let dataset = { state.lock().unwrap().dataset.clone() };
 
     let mut ds = (*dataset).clone();
     ds.checkout_latest().await?;
     let new_arc = Arc::new(ds);
 
     {
-        let mut state = state.lock()?;
+        let mut state = state.lock().unwrap();
         if state.pinned_version.is_none()
             && new_arc.manifest().version >= state.dataset.manifest().version
         {
@@ -615,108 +612,4 @@ mod tests {
         let s = io_stats.incremental_stats();
         assert_eq!(s.read_iops, 0, "step 5, elapsed={:?}", start.elapsed());
     }
-
-    /// Helper: poison the mutex inside a DatasetConsistencyWrapper.
-    fn poison_state(wrapper: &DatasetConsistencyWrapper) {
-        let state = wrapper.state.clone();
-        let handle = std::thread::spawn(move || {
-            let _guard = state.lock().unwrap();
-            panic!("intentional panic to poison mutex");
-        });
-        let _ = handle.join(); // join collects the panic
-        assert!(wrapper.state.lock().is_err(), "mutex should be poisoned");
-    }
-
-    #[tokio::test]
-    async fn test_get_returns_error_on_poisoned_lock() {
-        let dir = tempfile::tempdir().unwrap();
-        let uri = dir.path().to_str().unwrap();
-        let ds = create_test_dataset(uri).await;
-
-        let wrapper = DatasetConsistencyWrapper::new_latest(ds, None);
-        poison_state(&wrapper);
-
-        // get() should return Err, not panic
-        let result = wrapper.get().await;
-        assert!(result.is_err());
-    }
-
-    #[tokio::test]
-    async fn test_ensure_mutable_returns_error_on_poisoned_lock() {
-        let dir = tempfile::tempdir().unwrap();
-        let uri = dir.path().to_str().unwrap();
-        let ds = create_test_dataset(uri).await;
-
-        let wrapper = DatasetConsistencyWrapper::new_latest(ds, None);
-        poison_state(&wrapper);
-
-        let result = wrapper.ensure_mutable();
-        assert!(result.is_err());
-    }
-
-    #[tokio::test]
-    async fn test_update_recovers_from_poisoned_lock() {
-        let dir = tempfile::tempdir().unwrap();
-        let uri = dir.path().to_str().unwrap();
-        let ds = create_test_dataset(uri).await;
-        let ds_v2 = append_to_dataset(uri).await;
-
-        let wrapper = DatasetConsistencyWrapper::new_latest(ds, None);
-        poison_state(&wrapper);
-
-        // update() returns (), should not panic
-        wrapper.update(ds_v2);
-    }
-
-    #[tokio::test]
-    async fn test_time_travel_version_recovers_from_poisoned_lock() {
-        let dir = tempfile::tempdir().unwrap();
-        let uri = dir.path().to_str().unwrap();
-        let ds = create_test_dataset(uri).await;
-
-        let wrapper = DatasetConsistencyWrapper::new_latest(ds, None);
-        poison_state(&wrapper);
-
-        // Should not panic, returns whatever was in the mutex
-        let _version = wrapper.time_travel_version();
-    }
-
-    #[tokio::test]
-    async fn test_as_latest_returns_error_on_poisoned_lock() {
-        let dir = tempfile::tempdir().unwrap();
-        let uri = dir.path().to_str().unwrap();
-        let ds = create_test_dataset(uri).await;
-
-        let wrapper = DatasetConsistencyWrapper::new_latest(ds, None);
-        poison_state(&wrapper);
-
-        let result = wrapper.as_latest().await;
-        assert!(result.is_err());
-    }
-
-    #[tokio::test]
-    async fn test_as_time_travel_returns_error_on_poisoned_lock() {
-        let dir = tempfile::tempdir().unwrap();
-        let uri = dir.path().to_str().unwrap();
-        let ds = create_test_dataset(uri).await;
-
-        let wrapper = DatasetConsistencyWrapper::new_latest(ds, None);
-        poison_state(&wrapper);
-
-        let result = wrapper.as_time_travel(1u64).await;
-        assert!(result.is_err());
-    }
-
-    #[tokio::test]
-    async fn test_reload_returns_error_on_poisoned_lock() {
-        let dir = tempfile::tempdir().unwrap();
-        let uri = dir.path().to_str().unwrap();
-        let ds = create_test_dataset(uri).await;
-
-        let wrapper = DatasetConsistencyWrapper::new_latest(ds, None);
-        poison_state(&wrapper);
-
-        let result = wrapper.reload().await;
-        assert!(result.is_err());
-    }
 }

rust/lancedb/src/table/query.rs

@@ -9,7 +9,7 @@ use crate::expr::expr_to_sql_string;
 use crate::query::{
     DEFAULT_TOP_K, QueryExecutionOptions, QueryFilter, QueryRequest, Select, VectorQueryRequest,
 };
-use crate::utils::{MaxBatchLengthStream, TimeoutStream, default_vector_column};
+use crate::utils::{TimeoutStream, default_vector_column};
 use arrow::array::{AsArray, FixedSizeListBuilder, Float32Builder};
 use arrow::datatypes::{Float32Type, UInt8Type};
 use arrow_array::Array;
@@ -66,7 +66,6 @@ async fn execute_generic_query(
 ) -> Result<DatasetRecordBatchStream> {
     let plan = create_plan(table, query, options.clone()).await?;
     let inner = execute_plan(plan, Default::default())?;
-    let inner = MaxBatchLengthStream::new_boxed(inner, options.max_batch_length as usize);
     let inner = if let Some(timeout) = options.timeout {
         TimeoutStream::new_boxed(inner, timeout)
     } else {
@@ -201,9 +200,7 @@ pub async fn create_plan(
         scanner.with_row_id();
     }
 
-    if options.max_batch_length > 0 {
-        scanner.batch_size(options.max_batch_length as usize);
-    }
+    scanner.batch_size(options.max_batch_length as usize);
 
     if query.base.fast_search {
         scanner.fast_search();

rust/lancedb/src/table/write_progress.rs

@@ -1,431 +0,0 @@
-// SPDX-License-Identifier: Apache-2.0
-// SPDX-FileCopyrightText: Copyright The LanceDB Authors
-
-//! Progress monitoring for write operations.
-//!
-//! You can add a callback to process progress in [`crate::table::AddDataBuilder::progress`].
-//! [`WriteProgress`] is the struct passed to the callback.
-
-use std::sync::atomic::{AtomicUsize, Ordering};
-use std::sync::{Arc, Mutex};
-use std::time::{Duration, Instant};
-
-/// Progress snapshot for a write operation.
-#[derive(Debug, Clone)]
-pub struct WriteProgress {
-    // These are private and only accessible via getters, to make it easy to add
-    // new fields without breaking existing callbacks.
-    elapsed: Duration,
-    output_rows: usize,
-    output_bytes: usize,
-    total_rows: Option<usize>,
-    active_tasks: usize,
-    total_tasks: usize,
-    done: bool,
-}
-
-impl WriteProgress {
-    /// Wall-clock time since monitoring started.
-    pub fn elapsed(&self) -> Duration {
-        self.elapsed
-    }
-
-    /// Number of rows written so far.
-    pub fn output_rows(&self) -> usize {
-        self.output_rows
-    }
-
-    /// Number of bytes written so far.
-    pub fn output_bytes(&self) -> usize {
-        self.output_bytes
-    }
-
-    /// Total rows expected.
-    ///
-    /// Populated when the input source reports a row count (e.g. a
-    /// [`arrow_array::RecordBatch`]).  Always `Some` when [`WriteProgress::done`]
-    /// is `true` — falling back to the actual number of rows written.
-    pub fn total_rows(&self) -> Option<usize> {
-        self.total_rows
-    }
-
-    /// Number of parallel write tasks currently in flight.
-    pub fn active_tasks(&self) -> usize {
-        self.active_tasks
-    }
-
-    /// Total number of parallel write tasks (i.e. the write parallelism).
-    pub fn total_tasks(&self) -> usize {
-        self.total_tasks
-    }
-
-    /// Whether the write operation has completed.
-    ///
-    /// The final callback always has `done = true`.  Callers can use this to
-    /// finalize progress bars or perform cleanup.
-    pub fn done(&self) -> bool {
-        self.done
-    }
-}
-
-/// Callback type for progress updates.
-///
-/// Callbacks are serialized by the tracker and are never invoked reentrantly,
-/// so `FnMut` is safe to use here.
-pub type ProgressCallback = Arc<Mutex<dyn FnMut(&WriteProgress) + Send>>;
-
-/// Tracks progress of a write operation and invokes a [`ProgressCallback`].
-///
-/// Call [`WriteProgressTracker::record_batch`] for each batch written.
-/// Call [`WriteProgressTracker::finish`] once after all data is written.
-///
-/// The callback is never invoked reentrantly: all state updates and callback
-/// invocations are serialized behind a single lock.
-impl std::fmt::Debug for WriteProgressTracker {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        f.debug_struct("WriteProgressTracker")
-            .field("total_rows", &self.total_rows)
-            .finish()
-    }
-}
-
-pub(crate) struct WriteProgressTracker {
-    rows_and_bytes: std::sync::Mutex<(usize, usize)>,
-    /// Wire bytes tracked separately by the insert layer. When set (> 0),
-    /// this takes precedence over the in-memory bytes from `rows_and_bytes`.
-    wire_bytes: AtomicUsize,
-    active_tasks: Arc<AtomicUsize>,
-    total_tasks: AtomicUsize,
-    start: Instant,
-    /// Known total rows from the input source, if available.
-    total_rows: Option<usize>,
-    callback: ProgressCallback,
-}
-
-impl WriteProgressTracker {
-    pub fn new(callback: ProgressCallback, total_rows: Option<usize>) -> Self {
-        Self {
-            rows_and_bytes: std::sync::Mutex::new((0, 0)),
-            wire_bytes: AtomicUsize::new(0),
-            active_tasks: Arc::new(AtomicUsize::new(0)),
-            total_tasks: AtomicUsize::new(1),
-            start: Instant::now(),
-            total_rows,
-            callback,
-        }
-    }
-
-    /// Set the total number of parallel write tasks (the write parallelism).
-    pub fn set_total_tasks(&self, n: usize) {
-        self.total_tasks.store(n, Ordering::Relaxed);
-    }
-
-    /// Increment the active task count. Returns a guard that decrements on drop.
-    pub fn track_task(&self) -> ActiveTaskGuard {
-        self.active_tasks.fetch_add(1, Ordering::Relaxed);
-        ActiveTaskGuard(self.active_tasks.clone())
-    }
-
-    /// Record a batch of rows passing through the scan node.
-    pub fn record_batch(&self, rows: usize, bytes: usize) {
-        // Lock order: callback first, then rows_and_bytes. This is the only
-        // order used anywhere, so deadlocks cannot occur.
-        let mut cb = self.callback.lock().unwrap_or_else(|e| e.into_inner());
-        let mut guard = self
-            .rows_and_bytes
-            .lock()
-            .unwrap_or_else(|e| e.into_inner());
-        guard.0 += rows;
-        guard.1 += bytes;
-        let progress = self.snapshot(guard.0, guard.1, false);
-        drop(guard);
-        cb(&progress);
-    }
-
-    /// Record wire bytes from the insert layer (e.g. IPC-encoded bytes for
-    /// remote writes). When wire bytes are recorded, they take precedence over
-    /// the in-memory Arrow bytes tracked by [`record_batch`].
-    pub fn record_bytes(&self, bytes: usize) {
-        self.wire_bytes.fetch_add(bytes, Ordering::Relaxed);
-    }
-
-    /// Emit the final progress callback indicating the write is complete.
-    ///
-    /// `total_rows` is always `Some` on the final callback: it uses the known
-    /// total if available, or falls back to the number of rows actually written.
-    pub fn finish(&self) {
-        let mut cb = self.callback.lock().unwrap_or_else(|e| e.into_inner());
-        let guard = self
-            .rows_and_bytes
-            .lock()
-            .unwrap_or_else(|e| e.into_inner());
-        let mut snap = self.snapshot(guard.0, guard.1, true);
-        snap.total_rows = Some(self.total_rows.unwrap_or(guard.0));
-        drop(guard);
-        cb(&snap);
-    }
-
-    fn snapshot(&self, rows: usize, in_memory_bytes: usize, done: bool) -> WriteProgress {
-        let wire = self.wire_bytes.load(Ordering::Relaxed);
-        // Prefer wire bytes (actual I/O size) when the insert layer is
-        // tracking them; fall back to in-memory Arrow size otherwise.
-        // TODO: for local writes, track actual bytes written by Lance
-        // instead of using in-memory Arrow size as a proxy.
-        let output_bytes = if wire > 0 { wire } else { in_memory_bytes };
-        WriteProgress {
-            elapsed: self.start.elapsed(),
-            output_rows: rows,
-            output_bytes,
-            total_rows: self.total_rows,
-            active_tasks: self.active_tasks.load(Ordering::Relaxed),
-            total_tasks: self.total_tasks.load(Ordering::Relaxed),
-            done,
-        }
-    }
-}
-
-/// RAII guard that decrements the active task count when dropped.
-pub(crate) struct ActiveTaskGuard(Arc<AtomicUsize>);
-
-impl Drop for ActiveTaskGuard {
-    fn drop(&mut self) {
-        self.0.fetch_sub(1, Ordering::Relaxed);
-    }
-}
-
-/// RAII guard that calls [`WriteProgressTracker::finish`] on drop.
-///
-/// This ensures the final `done=true` callback fires even if the write
-/// errors or the future is cancelled.
-pub(crate) struct FinishOnDrop(pub Option<Arc<WriteProgressTracker>>);
-
-impl Drop for FinishOnDrop {
-    fn drop(&mut self) {
-        if let Some(t) = self.0.take() {
-            t.finish();
-        }
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use std::sync::Arc;
-    use std::sync::atomic::{AtomicUsize, Ordering};
-
-    use arrow_array::record_batch;
-
-    use crate::connect;
-
-    #[tokio::test]
-    async fn test_progress_monitor_fires_callback() {
-        let db = connect("memory://").execute().await.unwrap();
-
-        let batch = record_batch!(("id", Int32, [1, 2, 3])).unwrap();
-        let table = db
-            .create_table("progress_test", batch)
-            .execute()
-            .await
-            .unwrap();
-
-        let callback_count = Arc::new(AtomicUsize::new(0));
-        let last_rows = Arc::new(AtomicUsize::new(0));
-        let max_active = Arc::new(AtomicUsize::new(0));
-        let last_total_tasks = Arc::new(AtomicUsize::new(0));
-        let cb_count = callback_count.clone();
-        let cb_rows = last_rows.clone();
-        let cb_active = max_active.clone();
-        let cb_total_tasks = last_total_tasks.clone();
-
-        let new_data = record_batch!(("id", Int32, [4, 5, 6])).unwrap();
-        table
-            .add(new_data)
-            .progress(move |p| {
-                cb_count.fetch_add(1, Ordering::SeqCst);
-                cb_rows.store(p.output_rows(), Ordering::SeqCst);
-                cb_active.fetch_max(p.active_tasks(), Ordering::SeqCst);
-                cb_total_tasks.store(p.total_tasks(), Ordering::SeqCst);
-            })
-            .execute()
-            .await
-            .unwrap();
-
-        assert_eq!(table.count_rows(None).await.unwrap(), 6);
-        assert!(callback_count.load(Ordering::SeqCst) >= 1);
-        // Progress tracks the newly inserted rows, not the total table size.
-        assert_eq!(last_rows.load(Ordering::SeqCst), 3);
-        // At least one callback should have seen an active task.
-        assert!(max_active.load(Ordering::SeqCst) >= 1);
-        // total_tasks should reflect the write parallelism.
-        assert!(last_total_tasks.load(Ordering::SeqCst) >= 1);
-    }
-
-    #[tokio::test]
-    async fn test_progress_done_fires_at_end() {
-        let db = connect("memory://").execute().await.unwrap();
-        let batch = record_batch!(("id", Int32, [1, 2, 3])).unwrap();
-        let table = db
-            .create_table("progress_done", batch)
-            .execute()
-            .await
-            .unwrap();
-
-        let seen_done = Arc::new(std::sync::Mutex::new(Vec::<bool>::new()));
-        let seen = seen_done.clone();
-
-        let new_data = record_batch!(("id", Int32, [4, 5, 6])).unwrap();
-        table
-            .add(new_data)
-            .progress(move |p| {
-                seen.lock().unwrap().push(p.done());
-            })
-            .execute()
-            .await
-            .unwrap();
-
-        let done_flags = seen_done.lock().unwrap();
-        assert!(!done_flags.is_empty(), "at least one callback must fire");
-        // Only the last callback should have done=true.
-        let last = *done_flags.last().unwrap();
-        assert!(last, "last callback must have done=true");
-        // All earlier callbacks should have done=false.
-        for &d in done_flags.iter().rev().skip(1) {
-            assert!(!d, "non-final callbacks must have done=false");
-        }
-    }
-
-    #[tokio::test]
-    async fn test_progress_total_rows_known() {
-        let db = connect("memory://").execute().await.unwrap();
-
-        let batch = record_batch!(("id", Int32, [1, 2, 3])).unwrap();
-        let table = db
-            .create_table("total_known", batch)
-            .execute()
-            .await
-            .unwrap();
-
-        let seen_total = Arc::new(std::sync::Mutex::new(Vec::new()));
-        let seen = seen_total.clone();
-
-        // RecordBatch implements Scannable with num_rows() -> Some(3)
-        let new_data = record_batch!(("id", Int32, [4, 5, 6])).unwrap();
-        table
-            .add(new_data)
-            .progress(move |p| {
-                seen.lock().unwrap().push(p.total_rows());
-            })
-            .execute()
-            .await
-            .unwrap();
-
-        let totals = seen_total.lock().unwrap();
-        // All callbacks (including done) should have total_rows = Some(3)
-        assert!(
-            totals.contains(&Some(3)),
-            "expected total_rows=Some(3) in at least one callback, got: {:?}",
-            *totals
-        );
-    }
-
-    #[tokio::test]
-    async fn test_progress_total_rows_unknown() {
-        use arrow_array::RecordBatchIterator;
-
-        let db = connect("memory://").execute().await.unwrap();
-
-        let batch = record_batch!(("id", Int32, [1, 2, 3])).unwrap();
-        let table = db
-            .create_table("total_unknown", batch)
-            .execute()
-            .await
-            .unwrap();
-
-        let seen_total = Arc::new(std::sync::Mutex::new(Vec::new()));
-        let seen = seen_total.clone();
-
-        // RecordBatchReader does not provide num_rows, so total_rows should be
-        // None in intermediate callbacks but always Some on the done callback.
-        let schema = arrow_schema::Schema::new(vec![arrow_schema::Field::new(
-            "id",
-            arrow_schema::DataType::Int32,
-            false,
-        )]);
-        let new_data: Box<dyn arrow_array::RecordBatchReader + Send> =
-            Box::new(RecordBatchIterator::new(
-                vec![Ok(record_batch!(("id", Int32, [4, 5, 6])).unwrap())],
-                Arc::new(schema),
-            ));
-        table
-            .add(new_data)
-            .progress(move |p| {
-                seen.lock().unwrap().push((p.total_rows(), p.done()));
-            })
-            .execute()
-            .await
-            .unwrap();
-
-        let entries = seen_total.lock().unwrap();
-        assert!(!entries.is_empty(), "at least one callback must fire");
-        for (total, done) in entries.iter() {
-            if *done {
-                assert!(
-                    total.is_some(),
-                    "done callback must have total_rows set, got: {:?}",
-                    total
-                );
-            } else {
-                assert_eq!(
-                    *total, None,
-                    "intermediate callback must have total_rows=None, got: {:?}",
-                    total
-                );
-            }
-        }
-    }
-
-    #[test]
-    fn test_record_batch_recovers_from_poisoned_callback_lock() {
-        use super::{ProgressCallback, WriteProgressTracker};
-        use std::sync::Mutex;
-
-        let callback: ProgressCallback = Arc::new(Mutex::new(|_: &super::WriteProgress| {}));
-
-        // Poison the callback mutex
-        let cb_clone = callback.clone();
-        let handle = std::thread::spawn(move || {
-            let _guard = cb_clone.lock().unwrap();
-            panic!("intentional panic to poison callback mutex");
-        });
-        let _ = handle.join();
-        assert!(
-            callback.lock().is_err(),
-            "callback mutex should be poisoned"
-        );
-
-        let tracker = WriteProgressTracker::new(callback, Some(100));
-
-        // record_batch should not panic
-        tracker.record_batch(10, 1024);
-    }
-
-    #[test]
-    fn test_finish_recovers_from_poisoned_callback_lock() {
-        use super::{ProgressCallback, WriteProgressTracker};
-        use std::sync::Mutex;
-
-        let callback: ProgressCallback = Arc::new(Mutex::new(|_: &super::WriteProgress| {}));
-
-        // Poison the callback mutex
-        let cb_clone = callback.clone();
-        let handle = std::thread::spawn(move || {
-            let _guard = cb_clone.lock().unwrap();
-            panic!("intentional panic to poison callback mutex");
-        });
-        let _ = handle.join();
-
-        let tracker = WriteProgressTracker::new(callback, Some(100));
-
-        // finish should not panic
-        tracker.finish();
-    }
-}

rust/lancedb/src/utils/background_cache.rs

@@ -122,7 +122,7 @@ where
     /// This is a cheap synchronous check useful as a fast path before
     /// constructing a fetch closure for [`get()`](Self::get).
     pub fn try_get(&self) -> Option<V> {
-        let cache = self.inner.lock().unwrap_or_else(|e| e.into_inner());
+        let cache = self.inner.lock().unwrap();
         cache.state.fresh_value(self.ttl, self.refresh_window)
     }
 
@@ -138,7 +138,7 @@ where
     {
         // Fast path: check if cache is fresh
         {
-            let cache = self.inner.lock().unwrap_or_else(|e| e.into_inner());
+            let cache = self.inner.lock().unwrap();
             if let Some(value) = cache.state.fresh_value(self.ttl, self.refresh_window) {
                 return Ok(value);
             }
@@ -147,7 +147,7 @@ where
         // Slow path
         let mut fetch = Some(fetch);
         let action = {
-            let mut cache = self.inner.lock().unwrap_or_else(|e| e.into_inner());
+            let mut cache = self.inner.lock().unwrap();
             self.determine_action(&mut cache, &mut fetch)
         };
 
@@ -161,7 +161,7 @@ where
     ///
     /// This avoids a blocking fetch on the first [`get()`](Self::get) call.
     pub fn seed(&self, value: V) {
-        let mut cache = self.inner.lock().unwrap_or_else(|e| e.into_inner());
+        let mut cache = self.inner.lock().unwrap();
         cache.state = State::Current(value, clock::now());
     }
 
@@ -170,7 +170,7 @@ where
     /// Any in-flight background fetch from before this call will not update the
     /// cache (the generation counter prevents stale writes).
     pub fn invalidate(&self) {
-        let mut cache = self.inner.lock().unwrap_or_else(|e| e.into_inner());
+        let mut cache = self.inner.lock().unwrap();
         cache.state = State::Empty;
         cache.generation += 1;
     }
@@ -267,7 +267,7 @@ where
         let fut_for_spawn = shared.clone();
         tokio::spawn(async move {
             let result = fut_for_spawn.await;
-            let mut cache = inner.lock().unwrap_or_else(|e| e.into_inner());
+            let mut cache = inner.lock().unwrap();
             // Only update if no invalidation has happened since we started
             if cache.generation != generation {
                 return;
@@ -590,67 +590,4 @@ mod tests {
         let v = cache.get(ok_fetcher(count.clone(), "fresh")).await.unwrap();
         assert_eq!(v, "fresh");
     }
-
-    /// Helper: poison the inner mutex of a BackgroundCache.
-    fn poison_cache(cache: &BackgroundCache<String, TestError>) {
-        let inner = cache.inner.clone();
-        let handle = std::thread::spawn(move || {
-            let _guard = inner.lock().unwrap();
-            panic!("intentional panic to poison mutex");
-        });
-        let _ = handle.join();
-        assert!(cache.inner.lock().is_err(), "mutex should be poisoned");
-    }
-
-    #[tokio::test]
-    async fn test_try_get_recovers_from_poisoned_lock() {
-        let cache = new_cache();
-        let count = Arc::new(AtomicUsize::new(0));
-
-        // Seed a value first
-        cache.get(ok_fetcher(count.clone(), "hello")).await.unwrap();
-        cache.get(ok_fetcher(count.clone(), "hello")).await.unwrap(); // peek
-
-        poison_cache(&cache);
-
-        // try_get() should not panic — it recovers via unwrap_or_else
-        let result = cache.try_get();
-        // The value may or may not be fresh depending on timing, but it must not panic
-        let _ = result;
-    }
-
-    #[tokio::test]
-    async fn test_get_recovers_from_poisoned_lock() {
-        let cache = new_cache();
-        let count = Arc::new(AtomicUsize::new(0));
-
-        poison_cache(&cache);
-
-        // get() should not panic — it recovers and can still fetch
-        let result = cache.get(ok_fetcher(count.clone(), "recovered")).await;
-        assert!(result.is_ok());
-        assert_eq!(result.unwrap(), "recovered");
-    }
-
-    #[tokio::test]
-    async fn test_seed_recovers_from_poisoned_lock() {
-        let cache = new_cache();
-        poison_cache(&cache);
-
-        // seed() should not panic
-        cache.seed("seeded".to_string());
-    }
-
-    #[tokio::test]
-    async fn test_invalidate_recovers_from_poisoned_lock() {
-        let cache = new_cache();
-        let count = Arc::new(AtomicUsize::new(0));
-
-        cache.get(ok_fetcher(count.clone(), "hello")).await.unwrap();
-
-        poison_cache(&cache);
-
-        // invalidate() should not panic
-        cache.invalidate();
-    }
 }

rust/lancedb/src/utils/mod.rs

@@ -335,85 +335,6 @@ impl Stream for TimeoutStream {
     }
 }
 
-/// A `Stream` wrapper that slices oversized batches to enforce a maximum batch length.
-pub struct MaxBatchLengthStream {
-    inner: SendableRecordBatchStream,
-    max_batch_length: Option<usize>,
-    buffered_batch: Option<RecordBatch>,
-    buffered_offset: usize,
-}
-
-impl MaxBatchLengthStream {
-    pub fn new(inner: SendableRecordBatchStream, max_batch_length: usize) -> Self {
-        Self {
-            inner,
-            max_batch_length: (max_batch_length > 0).then_some(max_batch_length),
-            buffered_batch: None,
-            buffered_offset: 0,
-        }
-    }
-
-    pub fn new_boxed(
-        inner: SendableRecordBatchStream,
-        max_batch_length: usize,
-    ) -> SendableRecordBatchStream {
-        if max_batch_length == 0 {
-            inner
-        } else {
-            Box::pin(Self::new(inner, max_batch_length))
-        }
-    }
-}
-
-impl RecordBatchStream for MaxBatchLengthStream {
-    fn schema(&self) -> SchemaRef {
-        self.inner.schema()
-    }
-}
-
-impl Stream for MaxBatchLengthStream {
-    type Item = DataFusionResult<RecordBatch>;
-
-    fn poll_next(
-        mut self: Pin<&mut Self>,
-        cx: &mut std::task::Context<'_>,
-    ) -> std::task::Poll<Option<Self::Item>> {
-        loop {
-            let Some(max_batch_length) = self.max_batch_length else {
-                return Pin::new(&mut self.inner).poll_next(cx);
-            };
-
-            if let Some(batch) = self.buffered_batch.clone() {
-                if self.buffered_offset < batch.num_rows() {
-                    let remaining = batch.num_rows() - self.buffered_offset;
-                    let length = remaining.min(max_batch_length);
-                    let sliced = batch.slice(self.buffered_offset, length);
-                    self.buffered_offset += length;
-                    if self.buffered_offset >= batch.num_rows() {
-                        self.buffered_batch = None;
-                        self.buffered_offset = 0;
-                    }
-                    return std::task::Poll::Ready(Some(Ok(sliced)));
-                }
-
-                self.buffered_batch = None;
-                self.buffered_offset = 0;
-            }
-
-            match Pin::new(&mut self.inner).poll_next(cx) {
-                std::task::Poll::Ready(Some(Ok(batch))) => {
-                    if batch.num_rows() <= max_batch_length {
-                        return std::task::Poll::Ready(Some(Ok(batch)));
-                    }
-                    self.buffered_batch = Some(batch);
-                    self.buffered_offset = 0;
-                }
-                other => return other,
-            }
-        }
-    }
-}
-
 #[cfg(test)]
 mod tests {
     use arrow_array::Int32Array;
@@ -549,7 +470,7 @@ mod tests {
         assert_eq!(string_to_datatype(string), Some(expected));
     }
 
-    fn sample_batch(num_rows: i32) -> RecordBatch {
+    fn sample_batch() -> RecordBatch {
         let schema = Arc::new(Schema::new(vec![Field::new(
             "col1",
             DataType::Int32,
@@ -557,14 +478,14 @@ mod tests {
         )]));
         RecordBatch::try_new(
             schema.clone(),
-            vec![Arc::new(Int32Array::from_iter_values(0..num_rows))],
+            vec![Arc::new(Int32Array::from(vec![1, 2, 3]))],
         )
         .unwrap()
     }
 
     #[tokio::test]
     async fn test_timeout_stream() {
-        let batch = sample_batch(3);
+        let batch = sample_batch();
         let schema = batch.schema();
         let mock_stream = stream::iter(vec![Ok(batch.clone()), Ok(batch.clone())]);
 
@@ -594,7 +515,7 @@ mod tests {
 
     #[tokio::test]
     async fn test_timeout_stream_zero_duration() {
-        let batch = sample_batch(3);
+        let batch = sample_batch();
         let schema = batch.schema();
         let mock_stream = stream::iter(vec![Ok(batch.clone()), Ok(batch.clone())]);
 
@@ -613,7 +534,7 @@ mod tests {
 
     #[tokio::test]
     async fn test_timeout_stream_completes_normally() {
-        let batch = sample_batch(3);
+        let batch = sample_batch();
         let schema = batch.schema();
         let mock_stream = stream::iter(vec![Ok(batch.clone()), Ok(batch.clone())]);
 
@@ -631,35 +552,4 @@ mod tests {
         // Stream should be empty now
         assert!(timeout_stream.next().await.is_none());
     }
-
-    async fn collect_batch_sizes(
-        stream: SendableRecordBatchStream,
-        max_batch_length: usize,
-    ) -> Vec<usize> {
-        let mut sliced_stream = MaxBatchLengthStream::new(stream, max_batch_length);
-        sliced_stream
-            .by_ref()
-            .map(|batch| batch.unwrap().num_rows())
-            .collect::<Vec<_>>()
-            .await
-    }
-
-    #[tokio::test]
-    async fn test_max_batch_length_stream_behaviors() {
-        let schema = sample_batch(7).schema();
-        let mock_stream = stream::iter(vec![Ok(sample_batch(2)), Ok(sample_batch(7))]);
-
-        let sendable_stream: SendableRecordBatchStream =
-            Box::pin(RecordBatchStreamAdapter::new(schema.clone(), mock_stream));
-        assert_eq!(
-            collect_batch_sizes(sendable_stream, 3).await,
-            vec![2, 3, 3, 1]
-        );
-
-        let sendable_stream: SendableRecordBatchStream = Box::pin(RecordBatchStreamAdapter::new(
-            schema,
-            stream::iter(vec![Ok(sample_batch(2)), Ok(sample_batch(7))]),
-        ));
-        assert_eq!(collect_batch_sizes(sendable_stream, 0).await, vec![2, 7]);
-    }
 }