feat(query): add approx mode to vector queries (#3549)

## Feature

### What is the new feature?

Adds Rust core API support for configuring vector query approximation
mode with `ApproxMode::{Fast, Normal, Accurate}`.

### Why do we need this feature?

Lance already exposes `lance_index::vector::ApproxMode` and scanner
support for controlling the speed/accuracy tradeoff for approximate
vector search. LanceDB Rust queries need to expose and pass this setting
through for local/native and remote vector searches.

### How does it work?

- Adds public `ApproxMode` in `rust/lancedb`, with lowercase serde,
`Default::Normal`, parse/display, and conversions to/from Lance's
`ApproxMode`.
- Adds `approx_mode: Option<ApproxMode>` to `VectorQueryRequest` and a
`VectorQuery::approx_mode(...)` builder.
- Applies the mode to native/local Lance scanners after `nearest(...)`
when explicitly set.
- Sends `approx_mode` in remote query JSON only when explicitly set;
default requests omit it.

## Validation

- `cargo fmt --all`
- `cargo test --quiet --features remote approx_mode`
- `cargo test --quiet --features remote
test_query_vector_default_values`
- `cargo check --quiet --features remote --tests --examples`
- `git diff --check`

.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.31.0-beta.1"
+current_version = "0.30.1-beta.2"
 parse = """(?x)
     (?P<major>0|[1-9]\\d*)\\.
     (?P<minor>0|[1-9]\\d*)\\.
@@ -23,8 +23,6 @@ allow_dirty = true
 commit = true
 message = "Bump version: {current_version} → {new_version}"
 commit_args = ""
-# bump-my-version >=1.4.0 rejects pre_commit_hooks containing shell syntax unless opted in.
-allow_shell_hooks = true
 
 # Java maven files
 pre_commit_hooks = [

Cargo.lock

@@ -3432,8 +3432,8 @@ checksum = "42703706b716c37f96a77aea830392ad231f44c9e9a67872fa5548707e11b11c"
 
 [[package]]
 name = "fsst"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow-array",
  "rand 0.9.4",
@@ -4735,8 +4735,8 @@ checksum = "e037a2e1d8d5fdbd49b16a4ea09d5d6401c1f29eca5ff29d03d3824dba16256a"
 
 [[package]]
 name = "lance"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arc-swap",
  "arrow",
@@ -4810,8 +4810,8 @@ dependencies = [
 
 [[package]]
 name = "lance-arrow"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -4832,7 +4832,7 @@ dependencies = [
 [[package]]
 name = "lance-arrow-scalar"
 version = "58.0.0"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -4846,7 +4846,7 @@ dependencies = [
 [[package]]
 name = "lance-arrow-stats"
 version = "58.0.0"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow-array",
  "arrow-schema",
@@ -4855,8 +4855,8 @@ dependencies = [
 
 [[package]]
 name = "lance-bitpacking"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrayref",
  "paste",
@@ -4865,8 +4865,8 @@ dependencies = [
 
 [[package]]
 name = "lance-core"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -4904,8 +4904,8 @@ dependencies = [
 
 [[package]]
 name = "lance-datafusion"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow",
  "arrow-array",
@@ -4935,8 +4935,8 @@ dependencies = [
 
 [[package]]
 name = "lance-datagen"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow",
  "arrow-array",
@@ -4953,8 +4953,8 @@ dependencies = [
 
 [[package]]
 name = "lance-derive"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "proc-macro2",
  "quote",
@@ -4963,8 +4963,8 @@ dependencies = [
 
 [[package]]
 name = "lance-encoding"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow-arith",
  "arrow-array",
@@ -4999,8 +4999,8 @@ dependencies = [
 
 [[package]]
 name = "lance-file"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow-arith",
  "arrow-array",
@@ -5030,8 +5030,8 @@ dependencies = [
 
 [[package]]
 name = "lance-index"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arc-swap",
  "arrow",
@@ -5096,8 +5096,8 @@ dependencies = [
 
 [[package]]
 name = "lance-io"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow",
  "arrow-arith",
@@ -5138,8 +5138,8 @@ dependencies = [
 
 [[package]]
 name = "lance-linalg"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -5154,8 +5154,8 @@ dependencies = [
 
 [[package]]
 name = "lance-namespace"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow",
  "async-trait",
@@ -5167,8 +5167,8 @@ dependencies = [
 
 [[package]]
 name = "lance-namespace-impls"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow",
  "arrow-ipc",
@@ -5222,8 +5222,8 @@ dependencies = [
 
 [[package]]
 name = "lance-select"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -5238,8 +5238,8 @@ dependencies = [
 
 [[package]]
 name = "lance-table"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow",
  "arrow-array",
@@ -5278,8 +5278,8 @@ dependencies = [
 
 [[package]]
 name = "lance-testing"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "arrow-array",
  "arrow-schema",
@@ -5292,21 +5292,20 @@ dependencies = [
 
 [[package]]
 name = "lance-tokenizer"
-version = "8.0.0-rc.1"
-source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-rc.1#eea4095b188bf2ba2fa95d934a2f5d6c2c9e661c"
+version = "8.0.0-beta.16"
+source = "git+https://github.com/lance-format/lance.git?tag=v8.0.0-beta.16#6e734df607f2841fe3bba82f05a90f3174933bab"
 dependencies = [
  "icu_segmenter",
  "jieba-rs",
  "lindera",
  "rust-stemmers",
  "serde",
- "stop-words",
  "unicode-normalization",
 ]
 
 [[package]]
 name = "lancedb"
-version = "0.31.0-beta.1"
+version = "0.30.1-beta.2"
 dependencies = [
  "ahash",
  "anyhow",
@@ -5389,7 +5388,7 @@ dependencies = [
 
 [[package]]
 name = "lancedb-nodejs"
-version = "0.31.0-beta.1"
+version = "0.30.1-beta.2"
 dependencies = [
  "arrow-array",
  "arrow-buffer",
@@ -5414,7 +5413,7 @@ dependencies = [
 
 [[package]]
 name = "lancedb-python"
-version = "0.34.0-beta.1"
+version = "0.33.1-beta.2"
 dependencies = [
  "arrow",
  "async-trait",
@@ -9206,15 +9205,6 @@ version = "0.2.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "e51f1e89f093f99e7432c491c382b88a6860a5adbe6bf02574bf0a08efff1978"
 
-[[package]]
-name = "stop-words"
-version = "0.10.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d68df56303396bcfb639455b3c166804aeb7994005010aab5e9e8a1277b8871d"
-dependencies = [
- "serde_json",
-]
-
 [[package]]
 name = "str_stack"
 version = "0.1.1"

Cargo.toml

@@ -13,20 +13,20 @@ categories = ["database-implementations"]
 rust-version = "1.91.0"
 
 [workspace.dependencies]
-lance = { "version" = "=8.0.0-rc.1", default-features = false, "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-core = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-datagen = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-file = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-io = { "version" = "=8.0.0-rc.1", default-features = false, "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-index = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-linalg = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-namespace = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-namespace-impls = { "version" = "=8.0.0-rc.1", default-features = false, "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-table = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-testing = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-datafusion = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-encoding = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
-lance-arrow = { "version" = "=8.0.0-rc.1", "tag" = "v8.0.0-rc.1", "git" = "https://github.com/lance-format/lance.git" }
+lance = { "version" = "=8.0.0-beta.16", default-features = false, "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-core = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-datagen = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-file = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-io = { "version" = "=8.0.0-beta.16", default-features = false, "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-index = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-linalg = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-namespace-impls = { "version" = "=8.0.0-beta.16", default-features = false, "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-table = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-testing = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-datafusion = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-encoding = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
+lance-arrow = { "version" = "=8.0.0-beta.16", "tag" = "v8.0.0-beta.16", "git" = "https://github.com/lance-format/lance.git" }
 ahash = "0.8"
 # Note that this one does not include pyarrow
 arrow = { version = "58.0.0", optional = false }

docs/src/java/java.md

@@ -14,7 +14,7 @@ Add the following dependency to your `pom.xml`:
 <dependency>
     <groupId>com.lancedb</groupId>
     <artifactId>lancedb-core</artifactId>
-    <version>0.31.0-beta.1</version>
+    <version>0.30.1-beta.2</version>
 </dependency>
 ```
 

java/lancedb-core/pom.xml

@@ -8,7 +8,7 @@
     <parent>
       <groupId>com.lancedb</groupId>
       <artifactId>lancedb-parent</artifactId>
-      <version>0.31.0-beta.1</version>
+      <version>0.30.1-beta.2</version>
       <relativePath>../pom.xml</relativePath>
     </parent>
 

java/pom.xml

@@ -6,7 +6,7 @@
 
     <groupId>com.lancedb</groupId>
     <artifactId>lancedb-parent</artifactId>
-    <version>0.31.0-beta.1</version>
+    <version>0.30.1-beta.2</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>8.0.0-rc.1</lance-core.version>
+        <lance-core.version>8.0.0-beta.16</lance-core.version>
         <spotless.skip>false</spotless.skip>
         <spotless.version>2.30.0</spotless.version>
         <spotless.java.googlejavaformat.version>1.7</spotless.java.googlejavaformat.version>

nodejs/Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "lancedb-nodejs"
 edition.workspace = true
-version = "0.31.0-beta.1"
+version = "0.30.1-beta.2"
 publish = false
 license.workspace = true
 description.workspace = true

nodejs/npm/darwin-arm64/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-darwin-arm64",
-	"version": "0.31.0-beta.1",
+	"version": "0.30.1-beta.2",
 	"os": ["darwin"],
 	"cpu": ["arm64"],
 	"main": "lancedb.darwin-arm64.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-gnu",
-	"version": "0.31.0-beta.1",
+	"version": "0.30.1-beta.2",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-gnu.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-arm64-musl",
-	"version": "0.31.0-beta.1",
+	"version": "0.30.1-beta.2",
 	"os": ["linux"],
 	"cpu": ["arm64"],
 	"main": "lancedb.linux-arm64-musl.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-gnu",
-	"version": "0.31.0-beta.1",
+	"version": "0.30.1-beta.2",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-gnu.node",

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-linux-x64-musl",
-	"version": "0.31.0-beta.1",
+	"version": "0.30.1-beta.2",
 	"os": ["linux"],
 	"cpu": ["x64"],
 	"main": "lancedb.linux-x64-musl.node",

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

@@ -1,6 +1,6 @@
 {
   "name": "@lancedb/lancedb-win32-arm64-msvc",
-  "version": "0.31.0-beta.1",
+  "version": "0.30.1-beta.2",
   "os": [
     "win32"
   ],

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

@@ -1,6 +1,6 @@
 {
 	"name": "@lancedb/lancedb-win32-x64-msvc",
-	"version": "0.31.0-beta.1",
+	"version": "0.30.1-beta.2",
 	"os": ["win32"],
 	"cpu": ["x64"],
 	"main": "lancedb.win32-x64-msvc.node",

nodejs/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@lancedb/lancedb",
-  "version": "0.31.0-beta.1",
+  "version": "0.30.1-beta.2",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "@lancedb/lancedb",
-      "version": "0.31.0-beta.1",
+      "version": "0.30.1-beta.2",
       "cpu": [
         "x64",
         "arm64"

nodejs/package.json

@@ -11,7 +11,7 @@
     "ann"
   ],
   "private": false,
-  "version": "0.31.0-beta.1",
+  "version": "0.30.1-beta.2",
   "main": "dist/index.js",
   "exports": {
     ".": "./dist/index.js",

python/.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.34.0-beta.1"
+current_version = "0.33.1-beta.2"
 parse = """(?x)
     (?P<major>0|[1-9]\\d*)\\.
     (?P<minor>0|[1-9]\\d*)\\.
@@ -23,8 +23,6 @@ allow_dirty = true
 commit = true
 message = "Bump version: {current_version} → {new_version}"
 commit_args = ""
-# bump-my-version >=1.4.0 rejects pre_commit_hooks containing shell syntax unless opted in.
-allow_shell_hooks = true
 
 # Update Cargo.lock after version bump
 pre_commit_hooks = [

python/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "lancedb-python"
-version = "0.34.0-beta.1"
+version = "0.33.1-beta.2"
 publish = false
 edition.workspace = true
 description = "Python bindings for LanceDB"

python/python/lancedb/namespace.py

@@ -71,9 +71,6 @@ from lancedb.embeddings import EmbeddingFunctionConfig
 from ._lancedb import Session
 
 
-_MAX_QUERY_K = 2**31 - 1
-
-
 def _query_to_namespace_request(
     table_id: List[str],
     query: "Query",
@@ -151,8 +148,7 @@ def _query_to_namespace_request(
     if query.limit is not None:
         k = query.limit
     elif query.vector is None and query.full_text_query is None:
-        # limit k to max i32 value to avoid client overflows
-        k = _MAX_QUERY_K
+        k = sys.maxsize
     else:
         k = 10
 

python/python/lancedb/pydantic.py

@@ -275,18 +275,7 @@ def _py_type_to_arrow_type(py_type: Type[Any], field: FieldInfo) -> pa.DataType:
         tz = get_extras(field, "tz")
         return pa.timestamp("us", tz=tz)
     elif getattr(py_type, "__origin__", None) in (list, tuple):
-        # A bare, unparameterised ``typing.List`` / ``typing.Tuple`` matches this
-        # branch (its ``__origin__`` is ``list`` / ``tuple``) but has no
-        # ``__args__``, so we cannot infer the element type. Raise a clear
-        # ``TypeError`` instead of crashing with an opaque ``AttributeError``.
-        args = getattr(py_type, "__args__", None)
-        if not args:
-            raise TypeError(
-                "Converting Pydantic type to Arrow Type: unsupported type "
-                f"{py_type}. Specify the element type, e.g. List[int] instead "
-                "of a bare List."
-            )
-        child = args[0]
+        child = py_type.__args__[0]
         return _pydantic_list_child_to_arrow(child, field)
     raise TypeError(
         f"Converting Pydantic type to Arrow Type: unsupported type {py_type}."

python/python/lancedb/scannable.py

@@ -86,10 +86,7 @@ def _from_list(data: list) -> Scannable:
 
 @to_scannable.register(dict)
 def _from_dict(data: dict) -> Scannable:
-    raise ValueError(
-        "Cannot create or add rows from a single dictionary. "
-        "Use a list of dictionaries instead."
-    )
+    raise ValueError("Cannot add a single dictionary to a table. Use a list.")
 
 
 @to_scannable.register(LanceModel)

python/python/lancedb/table.py

@@ -243,10 +243,7 @@ def _into_pyarrow_reader(
         raise ValueError("Cannot add a single LanceModel to a table. Use a list.")
 
     if isinstance(data, dict):
-        raise ValueError(
-            "Cannot create or add rows from a single dictionary. "
-            "Use a list of dictionaries instead."
-        )
+        raise ValueError("Cannot add a single dictionary to a table. Use a list.")
 
     if isinstance(data, list):
         # Handle empty list case

python/python/lancedb/util.py

@@ -373,15 +373,9 @@ def _(value: list):
 @value_to_sql.register(dict)
 def _(value: dict):
     # https://datafusion.apache.org/user-guide/sql/scalar_functions.html#named-struct
-    # Render the field name through value_to_sql(str(...)) as well so that keys
-    # containing characters meaningful in SQL (e.g. a single quote) are escaped
-    # the same way string values are. A bare f"'{k}'" would emit invalid SQL for
-    # a key like "it's".
     return (
         "named_struct("
-        + ", ".join(
-            f"{value_to_sql(str(k))}, {value_to_sql(v)}" for k, v in value.items()
-        )
+        + ", ".join(f"'{k}', {value_to_sql(v)}" for k, v in value.items())
         + ")"
     )
 

python/python/tests/test_index.py

@@ -91,9 +91,7 @@ async def test_create_scalar_index(some_table: AsyncTable):
     # Can recreate if replace=True
     await some_table.create_index("id", replace=True)
     indices = await some_table.list_indices()
-    assert str(indices).startswith(
-        '[IndexConfig(name="id_idx", index_type="BTree", columns=["id"]'
-    )
+    assert str(indices) == '[Index(BTree, columns=["id"], name="id_idx")]'
     assert len(indices) == 1
     assert indices[0].index_type == "BTree"
     assert indices[0].columns == ["id"]
@@ -108,27 +106,6 @@ async def test_create_scalar_index(some_table: AsyncTable):
     assert len(indices) == 0
 
 
-@pytest.mark.asyncio
-async def test_index_config_repr(db_async):
-    # Use >= 1000 rows so the thousands separator in the repr is exercised.
-    nrows = 1500
-    table = await db_async.create_table(
-        "repr_table", pa.Table.from_pydict({"id": list(range(nrows))})
-    )
-    await table.create_index("id", config=BTree())
-    indices = await table.list_indices()
-    assert len(indices) == 1
-
-    r = repr(indices[0])
-    assert r.startswith('IndexConfig(name="id_idx", index_type="BTree", columns=["id"]')
-    # Integer counts use `_` thousands separators (valid Python int syntax).
-    assert "num_indexed_rows=1_500" in r
-    assert "num_unindexed_rows=0" in r
-    # created_at renders as a datetime so the value round-trips.
-    assert "created_at=datetime.datetime(" in r
-    assert r.endswith(")")
-
-
 @pytest.mark.asyncio
 async def test_create_nested_scalar_index_lists_canonical_paths(db_async):
     metadata_type = pa.struct(
@@ -221,9 +198,7 @@ async def test_create_nested_scalar_index_lists_canonical_paths(db_async):
 async def test_create_fixed_size_binary_index(some_table: AsyncTable):
     await some_table.create_index("fsb", config=BTree())
     indices = await some_table.list_indices()
-    assert str(indices).startswith(
-        '[IndexConfig(name="fsb_idx", index_type="BTree", columns=["fsb"]'
-    )
+    assert str(indices) == '[Index(BTree, columns=["fsb"], name="fsb_idx")]'
     assert len(indices) == 1
     assert indices[0].index_type == "BTree"
     assert indices[0].columns == ["fsb"]
@@ -272,9 +247,7 @@ async def test_create_bitmap_index(some_table: AsyncTable):
 async def test_create_label_list_index(some_table: AsyncTable):
     await some_table.create_index("tags", config=LabelList())
     indices = await some_table.list_indices()
-    assert str(indices).startswith(
-        '[IndexConfig(name="tags_idx", index_type="LabelList", columns=["tags"]'
-    )
+    assert str(indices) == '[Index(LabelList, columns=["tags"], name="tags_idx")]'
     plan = await some_table.query().where("array_has(tags, 'tag0')").explain_plan()
     assert "ScalarIndexQuery" in plan
 
@@ -289,9 +262,7 @@ async def test_create_large_list_label_list_index(db_async):
 
     await table.create_index("tags", config=LabelList())
     indices = await table.list_indices()
-    assert str(indices).startswith(
-        '[IndexConfig(name="tags_idx", index_type="LabelList", columns=["tags"]'
-    )
+    assert str(indices) == '[Index(LabelList, columns=["tags"], name="tags_idx")]'
     plan = await table.query().where("array_has(tags, 'shared')").explain_plan()
     assert "ScalarIndexQuery" in plan
 
@@ -328,9 +299,7 @@ async def test_create_label_list_index_rejects_list_struct(db_async):
 async def test_full_text_search_index(some_table: AsyncTable):
     await some_table.create_index("tags", config=FTS(with_position=False))
     indices = await some_table.list_indices()
-    assert str(indices).startswith(
-        '[IndexConfig(name="tags_idx", index_type="FTS", columns=["tags"]'
-    )
+    assert str(indices) == '[Index(FTS, columns=["tags"], name="tags_idx")]'
 
     await some_table.prewarm_index("tags_idx")
 

python/python/tests/test_namespace.py

@@ -5,11 +5,11 @@
 
 import tempfile
 import shutil
+import sys
 import pytest
 import pyarrow as pa
 import lancedb
 from lance_namespace.errors import NamespaceNotEmptyError, TableNotFoundError
-from lancedb.namespace import _MAX_QUERY_K
 from lancedb.table import AsyncTable, LanceTable
 
 
@@ -816,13 +816,10 @@ class TestPushdownOperations:
             ["geneva", "hist"],
             ["geneva", "hist"],
         ]
-        # Unlimited reads cap k at i32::MAX (the namespace query_table `k`
-        # field is i32); sys.maxsize would overflow the Rust binding.
         assert [request.k for request in namespace_client.requests] == [
-            _MAX_QUERY_K,
-            _MAX_QUERY_K,
+            sys.maxsize,
+            sys.maxsize,
         ]
-        assert all(r.k <= 2**31 - 1 for r in namespace_client.requests)
 
 
 @pytest.mark.asyncio
@@ -877,13 +874,10 @@ class TestAsyncPushdownOperations:
             ["geneva", "hist"],
             ["geneva", "hist"],
         ]
-        # Unlimited reads cap k at i32::MAX (the namespace query_table `k`
-        # field is i32); sys.maxsize would overflow the Rust binding.
         assert [request.k for request in namespace_client.requests] == [
-            _MAX_QUERY_K,
-            _MAX_QUERY_K,
+            sys.maxsize,
+            sys.maxsize,
         ]
-        assert all(r.k <= 2**31 - 1 for r in namespace_client.requests)
 
 
 def test_local_table_to_arrow_and_to_pandas_are_unchanged(tmp_path):

python/python/tests/test_pydantic.py

@@ -188,18 +188,6 @@ def test_nested_struct_list():
     assert schema == expect_schema
 
 
-def test_bare_generic_raises_type_error():
-    # A bare, unparameterised List/Tuple has no element type to map to Arrow.
-    # It should raise a clear TypeError, not crash with AttributeError: __args__.
-    for bare in (List, Tuple):
-
-        class TestModel(pydantic.BaseModel):
-            items: bare
-
-        with pytest.raises(TypeError, match="unsupported type"):
-            pydantic_to_schema(TestModel)
-
-
 def test_nested_struct_list_optional():
     class SplitInfo(pydantic.BaseModel):
         start_frame: int

python/python/tests/test_table.py

@@ -301,16 +301,6 @@ def test_create_table(mem_db: DBConnection):
         assert expected == tbl
 
 
-def test_create_table_rejects_single_dictionary(mem_db: DBConnection):
-    data = {"vector": [3.1, 4.1], "item": "foo", "price": 10.0}
-    with pytest.raises(ValueError) as excep_info:
-        mem_db.create_table("test", data=data)
-    assert (
-        str(excep_info.value) == "Cannot create or add rows from a single dictionary. "
-        "Use a list of dictionaries instead."
-    )
-
-
 def test_empty_table(mem_db: DBConnection):
     schema = pa.schema(
         [
@@ -340,8 +330,8 @@ def test_add_dictionary(mem_db: DBConnection):
     with pytest.raises(ValueError) as excep_info:
         tbl.add(data=data)
     assert (
-        str(excep_info.value) == "Cannot create or add rows from a single dictionary. "
-        "Use a list of dictionaries instead."
+        str(excep_info.value)
+        == "Cannot add a single dictionary to a table. Use a list."
     )
 
 

python/python/tests/test_util.py

@@ -149,21 +149,6 @@ def test_value_to_sql_dict():
     assert value_to_sql({}) == "named_struct()"
 
 
-def test_value_to_sql_dict_key_escaping():
-    # Struct field names that contain a single quote must be escaped (doubled)
-    # the same way string values are, otherwise value_to_sql emits invalid SQL
-    # such as named_struct('it's', 1).
-    assert value_to_sql({"it's": 1}) == "named_struct('it''s', 1)"
-    assert (
-        value_to_sql({"o'brien": "d'angelo"}) == "named_struct('o''brien', 'd''angelo')"
-    )
-    # Escaping also applies to keys of nested structs.
-    assert (
-        value_to_sql({"outer": {"in'r": 1}})
-        == "named_struct('outer', named_struct('in''r', 1))"
-    )
-
-
 def test_value_to_sql_numpy_scalars():
     # numpy scalars (e.g. pulled from an ndarray or a pandas column) must
     # convert the same way as their native Python counterparts. np.float64

python/src/index.rs

@@ -319,53 +319,11 @@ pub struct IndexConfig {
 
 #[pymethods]
 impl IndexConfig {
-    pub fn __repr__(&self, py: Python<'_>) -> String {
-        let mut fields = vec![
-            format!("name={:?}", self.name),
-            format!("index_type={:?}", self.index_type),
-            format!("columns={:?}", self.columns),
-        ];
-        if let Some(v) = &self.index_uuid {
-            fields.push(format!("index_uuid={:?}", v));
-        }
-        if let Some(v) = &self.type_url {
-            fields.push(format!("type_url={:?}", v));
-        }
-        if let Some(v) = self.created_at {
-            // Render the datetime's own Python repr so the value round-trips,
-            // falling back to RFC 3339 if the conversion ever fails.
-            let rendered = v
-                .into_pyobject(py)
-                .ok()
-                .and_then(|obj| obj.into_any().repr().ok())
-                .map(|r| r.to_string())
-                .unwrap_or_else(|| v.to_rfc3339());
-            fields.push(format!("created_at={}", rendered));
-        }
-        if let Some(v) = self.num_indexed_rows {
-            fields.push(format!("num_indexed_rows={}", fmt_thousands(v)));
-        }
-        if let Some(v) = self.num_unindexed_rows {
-            fields.push(format!("num_unindexed_rows={}", fmt_thousands(v)));
-        }
-        if let Some(v) = self.size_bytes {
-            fields.push(format!("size_bytes={}", fmt_thousands(v)));
-        }
-        if let Some(v) = self.num_segments {
-            fields.push(format!("num_segments={}", v));
-        }
-        if let Some(v) = self.index_version {
-            fields.push(format!("index_version={}", v));
-        }
-        if let Some(v) = &self.index_details {
-            let details = v
-                .bind(py)
-                .repr()
-                .map(|r| r.to_string())
-                .unwrap_or_else(|_| "<unavailable>".to_string());
-            fields.push(format!("index_details={}", details));
-        }
-        format!("IndexConfig({})", fields.join(", "))
+    pub fn __repr__(&self) -> String {
+        format!(
+            "Index({}, columns={:?}, name=\"{}\")",
+            self.index_type, self.columns, self.name
+        )
     }
 
     // For backwards-compatibility with the old sync SDK, we also support getting
@@ -394,23 +352,6 @@ impl IndexConfig {
     }
 }
 
-/// Format an integer with `_` thousands separators, e.g. `24_500_213`.
-///
-/// Underscores are valid Python int-literal syntax, so the repr stays
-/// copy-pasteable and machine-parseable while remaining readable.
-fn fmt_thousands(n: u64) -> String {
-    let digits = n.to_string();
-    let bytes = digits.as_bytes();
-    let mut out = String::with_capacity(digits.len() + digits.len() / 3);
-    for (i, b) in bytes.iter().enumerate() {
-        if i > 0 && (bytes.len() - i).is_multiple_of(3) {
-            out.push('_');
-        }
-        out.push(*b as char);
-    }
-    out
-}
-
 fn parse_index_details(py: Python<'_>, s: String) -> Py<PyAny> {
     let json = py.import("json").expect("json module is always available");
     match json.call_method1("loads", (s.as_str(),)) {

rust/lancedb/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "lancedb"
-version = "0.31.0-beta.1"
+version = "0.30.1-beta.2"
 edition.workspace = true
 description = "LanceDB: A serverless, low-latency vector database for AI applications"
 license.workspace = true

rust/lancedb/src/blob.rs

@@ -1,126 +0,0 @@
-// SPDX-License-Identifier: Apache-2.0
-// SPDX-FileCopyrightText: Copyright The LanceDB Authors
-
-//! Lance blob v2 columns store large binary payloads out of line.
-//!
-//! Declare a column with [`blob`]. On write, [`crate::table::Table::add`] coerces
-//! raw `Binary` / `LargeBinary` into the blob struct layout. Queries return
-//! small descriptors, not bytes.
-//!
-//! Blob tables require Lance file format >= 2.2 and stable row ids at create.
-
-use arrow_schema::{Field, Schema};
-use lance::dataset::WriteParams;
-use lance_arrow::FieldExt;
-use lance_encoding::version::LanceFileVersion;
-
-/// Creates an Arrow field for a Lance blob v2 column.
-///
-/// `Struct<data, uri>` with the `lance.blob.v2` marker. Same layout Lance
-/// expects on write.
-///
-/// ```
-/// use arrow_schema::{DataType, Field, Schema};
-///
-/// let schema = Schema::new(vec![
-///     Field::new("id", DataType::Int64, false),
-///     lancedb::blob("image", true),
-/// ]);
-/// ```
-///
-/// Blob tables use Lance file format >= 2.2 and stable row ids at create.
-pub fn blob(name: impl AsRef<str>, nullable: bool) -> Field {
-    lance::blob::blob_field(name.as_ref(), nullable)
-}
-
-/// Returns true if `schema` declares any blob v2 column.
-pub(crate) fn has_blob_columns(schema: &Schema) -> bool {
-    schema.fields().iter().any(|field| field.is_blob_v2())
-}
-
-/// Bumps storage format to at least [`LanceFileVersion::V2_2`] for blob schemas.
-pub(crate) fn ensure_blob_storage_version(schema: &Schema, params: &mut WriteParams) {
-    if !has_blob_columns(schema) {
-        return;
-    }
-
-    let resolved = params
-        .data_storage_version
-        .unwrap_or(LanceFileVersion::Stable)
-        .resolve();
-    if resolved < LanceFileVersion::V2_2 {
-        params.data_storage_version = Some(LanceFileVersion::V2_2);
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-    use arrow_schema::DataType;
-    use lance_arrow::ARROW_EXT_NAME_KEY;
-
-    fn blob_schema() -> Schema {
-        Schema::new(vec![
-            Field::new("id", DataType::Int64, false),
-            blob("image", true),
-        ])
-    }
-
-    #[test]
-    fn blob_field_carries_v2_extension_marker() {
-        let field = blob("image", true);
-        assert_eq!(
-            field.metadata().get(ARROW_EXT_NAME_KEY).map(String::as_str),
-            Some("lance.blob.v2")
-        );
-        assert!(matches!(field.data_type(), DataType::Struct(_)));
-    }
-
-    #[test]
-    fn has_blob_columns_detects_blob_fields() {
-        assert!(has_blob_columns(&blob_schema()));
-        let plain = Schema::new(vec![Field::new("id", DataType::Int64, false)]);
-        assert!(!has_blob_columns(&plain));
-    }
-
-    #[test]
-    fn storage_version_bumps_to_v2_2() {
-        let mut params = WriteParams::default();
-        ensure_blob_storage_version(&blob_schema(), &mut params);
-        assert_eq!(
-            params.data_storage_version.unwrap().resolve(),
-            LanceFileVersion::V2_2
-        );
-    }
-
-    #[test]
-    fn storage_version_overrides_lower_explicit_version() {
-        let mut params = WriteParams {
-            data_storage_version: Some(LanceFileVersion::V2_0),
-            ..Default::default()
-        };
-        ensure_blob_storage_version(&blob_schema(), &mut params);
-        assert_eq!(
-            params.data_storage_version.unwrap().resolve(),
-            LanceFileVersion::V2_2
-        );
-    }
-
-    #[test]
-    fn storage_version_keeps_higher_explicit_version() {
-        let mut params = WriteParams {
-            data_storage_version: Some(LanceFileVersion::V2_3),
-            ..Default::default()
-        };
-        ensure_blob_storage_version(&blob_schema(), &mut params);
-        assert_eq!(params.data_storage_version.unwrap(), LanceFileVersion::V2_3);
-    }
-
-    #[test]
-    fn storage_version_noop_without_blob_columns() {
-        let schema = Schema::new(vec![Field::new("id", DataType::Int64, false)]);
-        let mut params = WriteParams::default();
-        ensure_blob_storage_version(&schema, &mut params);
-        assert!(params.data_storage_version.is_none());
-    }
-}

rust/lancedb/src/database.rs

@@ -32,7 +32,6 @@ use crate::table::{BaseTable, WriteOptions};
 
 pub mod listing;
 pub mod namespace;
-pub(crate) mod read_freshness;
 
 pub trait DatabaseOptions {
     fn serialize_into_map(&self, map: &mut HashMap<String, String>);

rust/lancedb/src/database/listing.rs

@@ -18,7 +18,6 @@ use lance_table::io::commit::commit_handler_from_url;
 use object_store::local::LocalFileSystem;
 use snafu::ResultExt;
 
-use crate::blob::{ensure_blob_storage_version, has_blob_columns};
 use crate::connection::ConnectRequest;
 use crate::database::ReadConsistency;
 use crate::database::namespace::LanceNamespaceDatabase;
@@ -839,16 +838,13 @@ impl ListingDatabase {
             write_params.enable_v2_manifest_paths = enable_v2_manifest_paths;
         }
 
-        let data_schema = request.data.arrow_schema();
-        if let Some(enable_stable_row_ids) = stable_row_ids_override
-            .or(self.new_table_config.enable_stable_row_ids)
-            .or(has_blob_columns(&data_schema).then_some(true))
+        // Apply enable_stable_row_ids: table-level override takes precedence over connection config
+        if let Some(enable_stable_row_ids) =
+            stable_row_ids_override.or(self.new_table_config.enable_stable_row_ids)
         {
             write_params.enable_stable_row_ids = enable_stable_row_ids;
         }
 
-        ensure_blob_storage_version(&data_schema, &mut write_params);
-
         if matches!(&request.mode, CreateTableMode::Overwrite) {
             write_params.mode = WriteMode::Overwrite;
         }

rust/lancedb/src/database/namespace.rs

@@ -4,7 +4,7 @@
 //! Namespace-based database implementation that delegates table management to lance-namespace
 
 use std::collections::{HashMap, HashSet};
-use std::sync::{Arc, Mutex};
+use std::sync::Arc;
 
 use async_trait::async_trait;
 use lance::io::commit::namespace_manifest::LanceNamespaceExternalManifestStore;
@@ -23,16 +23,12 @@ use lance_namespace_impls::ConnectBuilder;
 use lance_table::io::commit::CommitHandler;
 use lance_table::io::commit::external_manifest::ExternalManifestCommitHandler;
 
-use crate::blob::{ensure_blob_storage_version, has_blob_columns};
 use crate::connection::NamespaceClientPushdownOperation;
 use crate::database::ReadConsistency;
 use crate::database::listing::{
     NewTableConfig, OPT_NEW_TABLE_ENABLE_STABLE_ROW_IDS, OPT_NEW_TABLE_STORAGE_VERSION,
     OPT_NEW_TABLE_V2_MANIFEST_PATHS,
 };
-use crate::database::read_freshness::{
-    FreshnessBaselines, ReadFreshnessContextProvider, TableFreshness,
-};
 use crate::error::{Error, Result};
 use crate::table::{NativeTable, map_namespace_lance_error};
 use lance::dataset::WriteMode;
@@ -55,10 +51,6 @@ fn is_table_already_exists_namespace_error(err: &lance::Error) -> bool {
     false
 }
 
-/// Object-id delimiter default (matches `RestNamespaceBuilder`'s); overridable
-/// via the `delimiter` property.
-const DEFAULT_NAMESPACE_DELIMITER: &str = "$";
-
 /// A database implementation that uses lance-namespace for table management
 pub struct LanceNamespaceDatabase {
     namespace: Arc<dyn LanceNamespace>,
@@ -78,17 +70,6 @@ pub struct LanceNamespaceDatabase {
     ns_properties: HashMap<String, String>,
     // Options for tables created by this connection
     new_table_config: NewTableConfig,
-    // Per-table read-freshness baselines, shared with the context provider.
-    freshness_baselines: FreshnessBaselines,
-    // Delimiter for building freshness keys; see `table_freshness`.
-    delimiter: String,
-}
-
-fn resolve_delimiter(ns_properties: &HashMap<String, String>) -> String {
-    ns_properties
-        .get("delimiter")
-        .cloned()
-        .unwrap_or_else(|| DEFAULT_NAMESPACE_DELIMITER.to_string())
 }
 
 impl LanceNamespaceDatabase {
@@ -101,9 +82,6 @@ impl LanceNamespaceDatabase {
         session: Option<Arc<lance::session::Session>>,
         namespace_client_pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
     ) -> Self {
-        // Client is pre-built, so we can't install the freshness provider here;
-        // baselines are still tracked for a uniform bump path.
-        let delimiter = resolve_delimiter(&namespace_client_properties);
         Self {
             namespace: namespace_client,
             storage_options,
@@ -114,8 +92,6 @@ impl LanceNamespaceDatabase {
             ns_impl: namespace_client_impl,
             ns_properties: namespace_client_properties,
             new_table_config: NewTableConfig::default(),
-            freshness_baselines: Arc::new(Mutex::new(HashMap::new())),
-            delimiter,
         }
     }
 
@@ -160,19 +136,10 @@ impl LanceNamespaceDatabase {
         if let Some(ref sess) = session {
             builder = builder.session(sess.clone());
         }
-
-        // Install the read-freshness provider before building the client.
-        let freshness_baselines: FreshnessBaselines = Arc::new(Mutex::new(HashMap::new()));
-        builder = builder.context_provider(Arc::new(ReadFreshnessContextProvider::new(
-            freshness_baselines.clone(),
-            read_consistency_interval,
-        )));
-
         let namespace = builder.connect().await.map_err(|e| Error::InvalidInput {
             message: format!("Failed to connect to namespace: {:?}", e),
         })?;
 
-        let delimiter = resolve_delimiter(&ns_properties);
         Ok(Self {
             namespace,
             storage_options,
@@ -183,20 +150,9 @@ impl LanceNamespaceDatabase {
             ns_impl: ns_impl.to_string(),
             ns_properties,
             new_table_config,
-            freshness_baselines,
-            delimiter,
         })
     }
 
-    /// Build a table's freshness handle, keyed to match the `object_id` the
-    /// namespace client sends on reads (table-id parts joined by the delimiter).
-    fn table_freshness(&self, namespace_path: &[String], name: &str) -> TableFreshness {
-        let mut parts = namespace_path.to_vec();
-        parts.push(name.to_string());
-        let key = parts.join(&self.delimiter);
-        TableFreshness::new(self.freshness_baselines.clone(), key)
-    }
-
     fn extract_storage_overrides(
         &self,
         request: &DbCreateTableRequest,
@@ -258,16 +214,12 @@ impl LanceNamespaceDatabase {
             params.enable_v2_manifest_paths = enable_v2_manifest_paths;
         }
 
-        let data_schema = request.data.schema();
-        if let Some(enable_stable_row_ids) = stable_row_ids_override
-            .or(self.new_table_config.enable_stable_row_ids)
-            .or(has_blob_columns(data_schema.as_ref()).then_some(true))
+        if let Some(enable_stable_row_ids) =
+            stable_row_ids_override.or(self.new_table_config.enable_stable_row_ids)
         {
             params.enable_stable_row_ids = enable_stable_row_ids;
         }
 
-        ensure_blob_storage_version(data_schema.as_ref(), params);
-
         Ok(())
     }
 }
@@ -379,8 +331,7 @@ impl Database for LanceNamespaceDatabase {
                         self.pushdown_operations.clone(),
                         self.session.clone(),
                     )
-                    .await?
-                    .with_freshness(self.table_freshness(&request.namespace_path, &request.name));
+                    .await?;
 
                     return Ok(Arc::new(native_table));
                 }
@@ -511,8 +462,7 @@ impl Database for LanceNamespaceDatabase {
             self.pushdown_operations.clone(),
             self.session.clone(),
         )
-        .await?
-        .with_freshness(self.table_freshness(&request.namespace_path, &request.name));
+        .await?;
 
         Ok(Arc::new(native_table))
     }
@@ -528,8 +478,7 @@ impl Database for LanceNamespaceDatabase {
             self.pushdown_operations.clone(),
             self.session.clone(),
         )
-        .await?
-        .with_freshness(self.table_freshness(&request.namespace_path, &request.name));
+        .await?;
 
         Ok(Arc::new(native_table))
     }

rust/lancedb/src/database/read_freshness.rs

@@ -1,312 +0,0 @@
-// SPDX-License-Identifier: Apache-2.0
-// SPDX-FileCopyrightText: Copyright The LanceDB Authors
-
-//! Read-freshness signaling for the lance-namespace path.
-//!
-//! Against a server that serves cached table metadata up to some staleness
-//! window, a handle that just wrote (or asked for the latest version via
-//! `checkout_latest`) can still read a stale snapshot. To prevent that, reads
-//! routed through the namespace client carry an `x-lancedb-min-timestamp`
-//! header naming the oldest snapshot the caller will accept.
-//!
-//! This mirrors `remote::table`: a per-table baseline is bumped to "now" on
-//! every write and on `checkout_latest()`, and reads send
-//! `max(baseline, now - read_consistency_interval)`. Since the namespace client
-//! takes no headers directly, a [`DynamicContextProvider`] injects it per request.
-
-use std::collections::HashMap;
-use std::sync::{Arc, Mutex};
-use std::time::{Duration, SystemTime};
-
-use lance_namespace_impls::{DynamicContextProvider, OperationInfo};
-
-/// Provider context keys prefixed with `headers.` become HTTP headers (prefix
-/// stripped), so this emits the `x-lancedb-min-timestamp` header.
-const MIN_TIMESTAMP_CONTEXT_KEY: &str = "headers.x-lancedb-min-timestamp";
-
-/// Per-table freshness baselines (keyed by namespace object id), shared between
-/// the provider that reads them and the table handles that bump them.
-pub type FreshnessBaselines = Arc<Mutex<HashMap<String, SystemTime>>>;
-
-/// `max(baseline, now - interval)`, or `None` when neither constraint applies.
-fn compute_min_timestamp(
-    baseline: Option<SystemTime>,
-    interval: Option<Duration>,
-    now: SystemTime,
-) -> Option<SystemTime> {
-    let interval_based = match interval {
-        None => None,
-        Some(d) if d.is_zero() => Some(now),
-        Some(d) => Some(now.checked_sub(d).unwrap_or(now)),
-    };
-    match (interval_based, baseline) {
-        (None, None) => None,
-        (Some(t), None) | (None, Some(t)) => Some(t),
-        (Some(a), Some(b)) => Some(a.max(b)),
-    }
-}
-
-/// Advance the baseline to `now`, never backwards, so a concurrent handle's
-/// write can't lower a floor another handle already set.
-fn next_freshness_baseline(prev: Option<SystemTime>, now: SystemTime) -> SystemTime {
-    match prev {
-        Some(p) => p.max(now),
-        None => now,
-    }
-}
-
-/// A handle's view of the shared baseline map for a single table.
-#[derive(Clone, Debug)]
-pub struct TableFreshness {
-    baselines: FreshnessBaselines,
-    /// Namespace object id for this table (matches the read's `object_id`).
-    key: String,
-}
-
-impl TableFreshness {
-    pub fn new(baselines: FreshnessBaselines, key: String) -> Self {
-        Self { baselines, key }
-    }
-
-    pub fn bump(&self) {
-        let now = SystemTime::now();
-        let mut baselines = self.baselines.lock().unwrap();
-        let prev = baselines.get(&self.key).copied();
-        baselines.insert(self.key.clone(), next_freshness_baseline(prev, now));
-    }
-}
-
-/// Read ops that can be served stale and so carry the freshness floor.
-/// `list_table_versions` resolves "latest" for managed-versioning tables, so it
-/// is what makes `checkout_latest()` observe a prior write.
-fn is_read_operation(operation: &str) -> bool {
-    matches!(
-        operation,
-        "describe_table" | "list_table_versions" | "query_table" | "list_tables"
-    )
-}
-
-/// Injects `x-lancedb-min-timestamp` on namespace reads, per addressed table.
-#[derive(Debug)]
-pub struct ReadFreshnessContextProvider {
-    baselines: FreshnessBaselines,
-    read_consistency_interval: Option<Duration>,
-}
-
-impl ReadFreshnessContextProvider {
-    pub fn new(baselines: FreshnessBaselines, read_consistency_interval: Option<Duration>) -> Self {
-        Self {
-            baselines,
-            read_consistency_interval,
-        }
-    }
-}
-
-impl DynamicContextProvider for ReadFreshnessContextProvider {
-    fn provide_context(&self, info: &OperationInfo) -> HashMap<String, String> {
-        if !is_read_operation(&info.operation) {
-            return HashMap::new();
-        }
-
-        let baseline = self.baselines.lock().unwrap().get(&info.object_id).copied();
-        match compute_min_timestamp(baseline, self.read_consistency_interval, SystemTime::now()) {
-            Some(ts) => {
-                let dt: chrono::DateTime<chrono::Utc> = ts.into();
-                HashMap::from([(MIN_TIMESTAMP_CONTEXT_KEY.to_string(), dt.to_rfc3339())])
-            }
-            None => HashMap::new(),
-        }
-    }
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    /// Allowed slop when comparing a header timestamp against a locally
-    /// captured wall-clock bound. Tests run fast enough that 1s is plenty.
-    const TOLERANCE: Duration = Duration::from_secs(1);
-
-    fn parse_header_ts(headers: &HashMap<String, String>) -> SystemTime {
-        let value = headers
-            .get(MIN_TIMESTAMP_CONTEXT_KEY)
-            .expect("expected min-timestamp context key");
-        chrono::DateTime::parse_from_rfc3339(value)
-            .unwrap()
-            .with_timezone(&chrono::Utc)
-            .into()
-    }
-
-    #[test]
-    fn test_compute_min_timestamp_combines_baseline_and_interval() {
-        let now = SystemTime::now();
-        let baseline = now - Duration::from_secs(60);
-
-        // No interval, no baseline -> no header.
-        assert_eq!(compute_min_timestamp(None, None, now), None);
-
-        // Baseline only -> baseline.
-        assert_eq!(
-            compute_min_timestamp(Some(baseline), None, now),
-            Some(baseline)
-        );
-
-        // ZERO interval, no baseline -> now (strong consistency).
-        assert_eq!(
-            compute_min_timestamp(None, Some(Duration::ZERO), now),
-            Some(now)
-        );
-
-        // Positive interval, no baseline -> now - interval.
-        assert_eq!(
-            compute_min_timestamp(None, Some(Duration::from_secs(10)), now),
-            Some(now - Duration::from_secs(10))
-        );
-
-        // Both: pick the more-recent (tighter) constraint.
-        // baseline = now-60, now-interval = now-10. now-10 is newer.
-        assert_eq!(
-            compute_min_timestamp(Some(baseline), Some(Duration::from_secs(10)), now),
-            Some(now - Duration::from_secs(10))
-        );
-
-        // Both, baseline newer: pick baseline.
-        let recent_baseline = now - Duration::from_secs(5);
-        assert_eq!(
-            compute_min_timestamp(Some(recent_baseline), Some(Duration::from_secs(60)), now),
-            Some(recent_baseline)
-        );
-    }
-
-    #[test]
-    fn test_next_freshness_baseline_is_monotonic() {
-        let now = SystemTime::now();
-        let earlier = now - Duration::from_secs(30);
-        let later = now + Duration::from_secs(30);
-
-        // No prior baseline -> now.
-        assert_eq!(next_freshness_baseline(None, now), now);
-        // Prior baseline older than now -> now.
-        assert_eq!(next_freshness_baseline(Some(earlier), now), now);
-        // Prior baseline newer than now -> keep the newer baseline.
-        assert_eq!(next_freshness_baseline(Some(later), now), later);
-    }
-
-    fn provider_with(
-        entries: &[(&str, SystemTime)],
-        interval: Option<Duration>,
-    ) -> ReadFreshnessContextProvider {
-        let map: HashMap<String, SystemTime> =
-            entries.iter().map(|(k, v)| (k.to_string(), *v)).collect();
-        ReadFreshnessContextProvider::new(Arc::new(Mutex::new(map)), interval)
-    }
-
-    #[test]
-    fn test_provider_emits_header_at_or_after_bumped_baseline() {
-        // A baseline set "now" with no interval: every read op must carry a
-        // floor at or after that baseline. `list_table_versions` is the hook
-        // that makes managed-versioning `checkout_latest()` observe a write.
-        let baseline = SystemTime::now();
-        let provider = provider_with(&[("ns$tbl", baseline)], None);
-
-        // These ops are keyed by the table id, so they pick up the per-table
-        // baseline. (`list_tables` is keyed by the namespace, so it is covered
-        // separately by the interval-floor test.)
-        for op in ["describe_table", "list_table_versions", "query_table"] {
-            let ctx = provider.provide_context(&OperationInfo::new(op, "ns$tbl"));
-            let sent = parse_header_ts(&ctx);
-            assert!(
-                sent >= baseline - TOLERANCE && sent <= baseline + TOLERANCE,
-                "operation {op} should carry a floor at the bumped baseline"
-            );
-        }
-    }
-
-    #[test]
-    fn test_provider_list_tables_uses_interval_floor_not_table_baseline() {
-        // `list_tables` is addressed by the namespace id, which never matches a
-        // per-table baseline key, so a bumped table baseline must not leak onto
-        // it. With no interval it sends nothing; with one it sends now-interval.
-        let provider = provider_with(&[("ns$tbl", SystemTime::now())], None);
-        let ctx = provider.provide_context(&OperationInfo::new("list_tables", "ns"));
-        assert!(
-            ctx.is_empty(),
-            "list_tables must not inherit a per-table baseline"
-        );
-
-        let interval = Duration::from_secs(30);
-        let provider = provider_with(&[("ns$tbl", SystemTime::now())], Some(interval));
-        let before = SystemTime::now();
-        let ctx = provider.provide_context(&OperationInfo::new("list_tables", "ns"));
-        let after = SystemTime::now();
-        let sent = parse_header_ts(&ctx);
-        assert!(
-            sent >= before - interval - TOLERANCE && sent <= after - interval + TOLERANCE,
-            "list_tables should carry the interval floor"
-        );
-    }
-
-    #[test]
-    fn test_provider_no_header_for_empty_baseline_and_no_interval() {
-        // Manual consistency (no interval) on a table that was never bumped:
-        // no floor, so the server may serve from cache.
-        let provider = provider_with(&[], None);
-        let ctx = provider.provide_context(&OperationInfo::new("describe_table", "ns$tbl"));
-        assert!(ctx.is_empty());
-    }
-
-    #[test]
-    fn test_provider_interval_floor_applies_without_baseline() {
-        // With a consistency interval and no baseline, the floor is now-interval.
-        let interval = Duration::from_secs(30);
-        let provider = provider_with(&[], Some(interval));
-
-        let before = SystemTime::now();
-        let ctx = provider.provide_context(&OperationInfo::new("query_table", "ns$tbl"));
-        let after = SystemTime::now();
-
-        let sent = parse_header_ts(&ctx);
-        assert!(
-            sent >= before - interval - TOLERANCE && sent <= after - interval + TOLERANCE,
-            "expected floor at roughly now - interval"
-        );
-    }
-
-    #[test]
-    fn test_provider_non_read_ops_emit_nothing() {
-        // Even with a fresh baseline and a zero interval, a non-read operation
-        // (which establishes rather than consumes a baseline) sends no header.
-        let provider = provider_with(&[("ns$tbl", SystemTime::now())], Some(Duration::ZERO));
-        for op in [
-            "create_table",
-            "register_table",
-            "drop_table",
-            "rename_table",
-            // Pinned to an immutable version, so it cannot be served stale.
-            "describe_table_version",
-        ] {
-            let ctx = provider.provide_context(&OperationInfo::new(op, "ns$tbl"));
-            assert!(
-                ctx.is_empty(),
-                "operation {op} must not send a freshness header"
-            );
-        }
-    }
-
-    #[test]
-    fn test_provider_uses_per_table_baseline() {
-        // The floor is looked up by object id, so an unrelated table's baseline
-        // does not leak onto another table's read.
-        let baseline = SystemTime::now();
-        let provider = provider_with(&[("ns$has_baseline", baseline)], None);
-
-        // The bumped table gets a header.
-        let hit =
-            provider.provide_context(&OperationInfo::new("describe_table", "ns$has_baseline"));
-        assert!(!hit.is_empty());
-
-        // A different table with no baseline (and no interval) gets nothing.
-        let miss = provider.provide_context(&OperationInfo::new("describe_table", "ns$other"));
-        assert!(miss.is_empty());
-    }
-}

rust/lancedb/src/embeddings/bedrock.rs

@@ -13,7 +13,7 @@ use serde_json::{Value, json};
 use super::EmbeddingFunction;
 use crate::{Error, Result};
 
-use tokio::runtime::{Handle, RuntimeFlavor};
+use tokio::runtime::Handle;
 use tokio::task::block_in_place;
 
 #[derive(Debug)]
@@ -148,12 +148,6 @@ impl BedrockEmbeddingFunction {
             _ => unreachable!(),
         };
 
-        // Bedrock's SDK is async but this trait method is synchronous, so we
-        // bridge with `block_in_place` + `block_on`. That requires a
-        // multi-threaded Tokio runtime; return a typed error instead of
-        // panicking when no compatible runtime is available.
-        let handle = current_multi_thread_handle()?;
-
         for text in texts {
             let request_body = match self.model {
                 BedrockEmbeddingModel::TitanEmbedding => {
@@ -169,28 +163,24 @@ impl BedrockEmbeddingFunction {
                 }
             };
 
-            // Serialize before entering the blocking section so a serialization
-            // failure surfaces as a typed error rather than an `unwrap` panic.
-            let body = serde_json::to_vec(&request_body).map_err(|e| Error::Runtime {
-                message: format!("Failed to serialize Bedrock request: {e}"),
-            })?;
-
             let client = self.client.clone();
             let model_id = self.model.model_id().to_string();
+            let request_body = request_body.clone();
 
-            let response = block_in_place(|| {
-                handle.block_on(async move {
+            let response = block_in_place(move || {
+                Handle::current().block_on(async move {
                     client
                         .invoke_model()
                         .model_id(model_id)
-                        .body(aws_sdk_bedrockruntime::primitives::Blob::new(body))
+                        .body(aws_sdk_bedrockruntime::primitives::Blob::new(
+                            serde_json::to_vec(&request_body).unwrap(),
+                        ))
                         .send()
                         .await
-                        .map_err(|e| Error::Runtime {
-                            message: format!("Bedrock invoke_model request failed: {e}"),
-                        })
+                        .map_err(Box::new)
                 })
-            })?;
+            })
+            .unwrap();
 
             let response_json: Value =
                 serde_json::from_slice(response.body.as_ref()).map_err(|e| Error::Runtime {
@@ -198,12 +188,22 @@ impl BedrockEmbeddingFunction {
                 })?;
 
             let embedding = match self.model {
-                BedrockEmbeddingModel::TitanEmbedding => {
-                    json_array_to_f32(&response_json["embedding"], "embedding")?
-                }
-                BedrockEmbeddingModel::CohereLarge => {
-                    json_array_to_f32(&response_json["embeddings"][0], "embeddings")?
-                }
+                BedrockEmbeddingModel::TitanEmbedding => response_json["embedding"]
+                    .as_array()
+                    .ok_or_else(|| Error::Runtime {
+                        message: "Missing embedding in response".to_string(),
+                    })?
+                    .iter()
+                    .map(|v| v.as_f64().unwrap() as f32)
+                    .collect::<Vec<f32>>(),
+                BedrockEmbeddingModel::CohereLarge => response_json["embeddings"][0]
+                    .as_array()
+                    .ok_or_else(|| Error::Runtime {
+                        message: "Missing embeddings in response".to_string(),
+                    })?
+                    .iter()
+                    .map(|v| v.as_f64().unwrap() as f32)
+                    .collect::<Vec<f32>>(),
             };
 
             builder.append_slice(&embedding);
@@ -212,86 +212,3 @@ impl BedrockEmbeddingFunction {
         Ok(builder.finish())
     }
 }
-
-/// Returns a handle to the current multi-threaded Tokio runtime, or a typed
-/// [`Error::Runtime`] when called outside a runtime or on the current-thread
-/// runtime. This keeps the synchronous-over-async bridge in
-/// [`BedrockEmbeddingFunction::compute_inner`] from panicking on runtime
-/// configurations that cannot support `block_in_place`.
-fn current_multi_thread_handle() -> Result<Handle> {
-    let handle = Handle::try_current().map_err(|e| Error::Runtime {
-        message: format!("Bedrock embedding must be called from within a Tokio runtime: {e}"),
-    })?;
-    if handle.runtime_flavor() == RuntimeFlavor::CurrentThread {
-        return Err(Error::Runtime {
-            message: "Bedrock embedding requires a multi-threaded Tokio runtime; the \
-                      current-thread runtime cannot use `block_in_place`"
-                .to_string(),
-        });
-    }
-    Ok(handle)
-}
-
-/// Converts a JSON value expected to be an array of numbers into `Vec<f32>`.
-///
-/// Returns a typed [`Error::Runtime`] (rather than panicking) when the value is
-/// not an array or contains a non-numeric element, so malformed provider
-/// responses degrade gracefully.
-fn json_array_to_f32(value: &Value, field: &str) -> Result<Vec<f32>> {
-    let arr = value.as_array().ok_or_else(|| Error::Runtime {
-        message: format!("Missing or non-array '{field}' field in Bedrock response"),
-    })?;
-    arr.iter()
-        .map(|v| {
-            v.as_f64().map(|f| f as f32).ok_or_else(|| Error::Runtime {
-                message: format!("Non-numeric value in Bedrock '{field}' embedding: {v}"),
-            })
-        })
-        .collect()
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn json_array_to_f32_parses_numbers() {
-        let v = json!([1.0, 2, -3.5]);
-        let out = json_array_to_f32(&v, "embedding").unwrap();
-        assert_eq!(out, vec![1.0_f32, 2.0, -3.5]);
-    }
-
-    #[test]
-    fn json_array_to_f32_rejects_non_array() {
-        // Missing field indexes to `Value::Null`; a malformed payload should be
-        // a typed error, not a panic.
-        let v = json!({"unexpected": "shape"});
-        let err = json_array_to_f32(&v["embedding"], "embedding").unwrap_err();
-        assert!(matches!(err, Error::Runtime { .. }), "got {err:?}");
-    }
-
-    #[test]
-    fn json_array_to_f32_rejects_non_numeric_element() {
-        let v = json!([1.0, "not-a-number", 3.0]);
-        let err = json_array_to_f32(&v, "embedding").unwrap_err();
-        assert!(matches!(err, Error::Runtime { .. }), "got {err:?}");
-    }
-
-    #[test]
-    fn handle_errors_without_runtime() {
-        // No Tokio runtime in scope -> typed error instead of a panic.
-        let err = current_multi_thread_handle().unwrap_err();
-        assert!(matches!(err, Error::Runtime { .. }), "got {err:?}");
-    }
-
-    #[tokio::test(flavor = "current_thread")]
-    async fn handle_errors_on_current_thread_runtime() {
-        let err = current_multi_thread_handle().unwrap_err();
-        assert!(matches!(err, Error::Runtime { .. }), "got {err:?}");
-    }
-
-    #[tokio::test(flavor = "multi_thread")]
-    async fn handle_ok_on_multi_thread_runtime() {
-        current_multi_thread_handle().expect("multi-threaded runtime should be accepted");
-    }
-}

rust/lancedb/src/lib.rs

@@ -163,7 +163,6 @@
 //! ```
 
 pub mod arrow;
-pub mod blob;
 pub mod connection;
 pub mod data;
 pub mod database;
@@ -189,7 +188,6 @@ use std::{fmt::Display, str::FromStr};
 
 use serde::{Deserialize, Serialize};
 
-pub use blob::blob;
 pub use connection::{ConnectNamespaceBuilder, Connection};
 pub use error::{Error, Result};
 use lance_index::vector::ApproxMode as LanceApproxMode;

rust/lancedb/src/remote/table.rs

@@ -1352,35 +1352,6 @@ impl<S: HttpSend + 'static> RemoteTable<S> {
     }
 }
 
-/// Deserialize an index's `created_at` field.
-///
-/// The server returns this as an RFC 3339 string (e.g. `"2026-06-18T21:37:36.637Z"`),
-/// but older deployments sent a unix timestamp in milliseconds. Accept both so the
-/// client works against any server version.
-fn deserialize_created_at<'de, D>(
-    deserializer: D,
-) -> std::result::Result<Option<DateTime<Utc>>, D::Error>
-where
-    D: serde::Deserializer<'de>,
-{
-    use serde::de::Error as _;
-
-    #[derive(Deserialize)]
-    #[serde(untagged)]
-    enum CreatedAt {
-        Rfc3339(String),
-        Millis(i64),
-    }
-
-    match Option::<CreatedAt>::deserialize(deserializer)? {
-        None => Ok(None),
-        Some(CreatedAt::Rfc3339(s)) => DateTime::parse_from_rfc3339(&s)
-            .map(|dt| Some(dt.with_timezone(&Utc)))
-            .map_err(D::Error::custom),
-        Some(CreatedAt::Millis(ms)) => Ok(DateTime::from_timestamp_millis(ms)),
-    }
-}
-
 impl<S: HttpSend + 'static> RemoteTable<S> {
     /// Parse the response from `/index/list/` into `IndexConfig` entries.
     ///
@@ -1409,7 +1380,7 @@ impl<S: HttpSend + 'static> RemoteTable<S> {
             // Used as the sentinel to decide whether to skip the stats call.
             index_type: Option<IndexType>,
             index_uuid: Option<String>,
-            #[serde(default, deserialize_with = "deserialize_created_at")]
+            #[serde(default, with = "chrono::serde::ts_milliseconds_option")]
             created_at: Option<DateTime<Utc>>,
             num_indexed_rows: Option<u64>,
             num_unindexed_rows: Option<u64>,
@@ -4707,7 +4678,7 @@ mod tests {
                                 "num_segments": 2,
                                 "index_version": 1,
                                 "index_details": "{\"num_partitions\":16}",
-                                "created_at": "2026-06-18T21:37:36.637Z",
+                                "created_at": 1700000000000i64,
                                 "type_url": "type.googleapis.com/lance.index.vector.IvfPq",
                             },
                             {
@@ -4757,10 +4728,7 @@ mod tests {
             vec_idx.type_url,
             Some("type.googleapis.com/lance.index.vector.IvfPq".to_string())
         );
-        assert_eq!(
-            vec_idx.created_at,
-            Some("2026-06-18T21:37:36.637Z".parse::<DateTime<Utc>>().unwrap())
-        );
+        assert!(vec_idx.created_at.is_some());
 
         let text_idx = &indices[1];
         assert_eq!(text_idx.name, "text_idx");
@@ -4781,36 +4749,6 @@ mod tests {
         assert_eq!(text_idx.created_at, None);
     }
 
-    #[test]
-    fn test_deserialize_created_at() {
-        #[derive(Deserialize)]
-        struct Wrapper {
-            #[serde(default, deserialize_with = "deserialize_created_at")]
-            created_at: Option<DateTime<Utc>>,
-        }
-
-        // RFC 3339 string (current server format).
-        let w: Wrapper =
-            serde_json::from_str(r#"{"created_at": "2026-06-18T21:37:36.637Z"}"#).unwrap();
-        assert_eq!(
-            w.created_at,
-            Some("2026-06-18T21:37:36.637Z".parse::<DateTime<Utc>>().unwrap())
-        );
-
-        // Unix milliseconds (legacy server format).
-        let w: Wrapper = serde_json::from_str(r#"{"created_at": 1700000000000}"#).unwrap();
-        assert_eq!(w.created_at, DateTime::from_timestamp_millis(1700000000000));
-
-        // Null and missing both yield None.
-        let w: Wrapper = serde_json::from_str(r#"{"created_at": null}"#).unwrap();
-        assert_eq!(w.created_at, None);
-        let w: Wrapper = serde_json::from_str(r#"{}"#).unwrap();
-        assert_eq!(w.created_at, None);
-
-        // A malformed string is rejected rather than silently dropped to None.
-        assert!(serde_json::from_str::<Wrapper>(r#"{"created_at": "not-a-date"}"#).is_err());
-    }
-
     #[tokio::test]
     async fn test_list_versions() {
         let table = Table::new_with_handler("my_table", |request| {

rust/lancedb/src/table.rs

@@ -43,7 +43,6 @@ use crate::connection::NamespaceClientPushdownOperation;
 
 use crate::data::scannable::{PeekedScannable, Scannable, estimate_write_partitions};
 use crate::database::Database;
-use crate::database::read_freshness::TableFreshness;
 use crate::embeddings::{EmbeddingDefinition, EmbeddingRegistry, MemoryRegistry};
 use crate::error::{Error, Result};
 use crate::index::IndexStatistics;
@@ -1764,8 +1763,6 @@ pub struct NativeTable {
     // Operations to push down to the namespace server.
     // pub(crate) so query.rs can access the field for server-side query execution.
     pub(crate) pushdown_operations: HashSet<NamespaceClientPushdownOperation>,
-    // Read-freshness baseline; `Some` only for namespace-backed tables.
-    freshness: Option<TableFreshness>,
 }
 
 impl std::fmt::Debug for NativeTable {
@@ -1926,7 +1923,6 @@ impl NativeTable {
             read_consistency_interval,
             namespace_client,
             pushdown_operations,
-            freshness: None,
         })
     }
 
@@ -1938,12 +1934,6 @@ impl NativeTable {
         self
     }
 
-    /// Attach the read-freshness baseline handle (namespace connections only).
-    pub(crate) fn with_freshness(mut self, freshness: TableFreshness) -> Self {
-        self.freshness = Some(freshness);
-        self
-    }
-
     /// Build a sibling `NativeTable` with the same identity but a different
     /// (independent) dataset wrapper — used to hand out branch-scoped handles.
     fn with_dataset(&self, dataset: dataset::DatasetConsistencyWrapper) -> Self {
@@ -1956,14 +1946,6 @@ impl NativeTable {
             read_consistency_interval: self.read_consistency_interval,
             namespace_client: self.namespace_client.clone(),
             pushdown_operations: self.pushdown_operations.clone(),
-            freshness: self.freshness.clone(),
-        }
-    }
-
-    /// Bump the read-freshness baseline; no-op for non-namespace tables.
-    fn bump_freshness(&self) {
-        if let Some(freshness) = &self.freshness {
-            freshness.bump();
         }
     }
 
@@ -2063,7 +2045,6 @@ impl NativeTable {
             read_consistency_interval,
             namespace_client: stored_namespace_client,
             pushdown_operations,
-            freshness: None,
         })
     }
 
@@ -2153,7 +2134,6 @@ impl NativeTable {
             read_consistency_interval,
             namespace_client,
             pushdown_operations,
-            freshness: None,
         })
     }
 
@@ -2285,7 +2265,6 @@ impl NativeTable {
             read_consistency_interval,
             namespace_client: stored_namespace_client,
             pushdown_operations,
-            freshness: None,
         })
     }
 
@@ -2445,8 +2424,6 @@ impl BaseTable for NativeTable {
     }
 
     async fn checkout_latest(&self) -> Result<()> {
-        // Bump before resolving "latest" so that request carries the floor.
-        self.bump_freshness();
         self.dataset.as_latest().await?;
         self.dataset.reload().await
     }
@@ -2534,8 +2511,6 @@ impl BaseTable for NativeTable {
             debug_assert_eq!(dataset.version().version, version);
             dataset.restore().await?;
         }
-        // Restore moves "latest", so bump before resolving it (as RemoteTable does).
-        self.bump_freshness();
         self.dataset.as_latest().await?;
         Ok(())
     }
@@ -2616,13 +2591,7 @@ impl BaseTable for NativeTable {
             output.plan
         };
 
-        let insert_exec = Arc::new(InsertExec::new_with_tracker(
-            ds_wrapper.clone(),
-            ds,
-            plan,
-            lance_params,
-            output.tracker.clone(),
-        ));
+        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 {
@@ -2655,7 +2624,6 @@ impl BaseTable for NativeTable {
         }
 
         let version = ds_wrapper.get().await?.manifest().version;
-        self.bump_freshness();
         Ok(AddResult { version })
     }
 
@@ -2706,9 +2674,7 @@ impl BaseTable for NativeTable {
 
     async fn update(&self, update: UpdateBuilder) -> Result<UpdateResult> {
         // Delegate to the submodule implementation
-        let result = update::execute_update(self, update).await?;
-        self.bump_freshness();
-        Ok(result)
+        update::execute_update(self, update).await
     }
 
     async fn create_plan(
@@ -2740,9 +2706,7 @@ impl BaseTable for NativeTable {
         params: MergeInsertBuilder,
         new_data: Box<dyn RecordBatchReader + Send>,
     ) -> Result<MergeResult> {
-        let result = merge::execute_merge_insert(self, params, new_data).await?;
-        self.bump_freshness();
-        Ok(result)
+        merge::execute_merge_insert(self, params, new_data).await
     }
 
     async fn set_unenforced_primary_key(&self, columns: &[&str]) -> Result<()> {
@@ -2763,9 +2727,7 @@ impl BaseTable for NativeTable {
 
     /// Delete rows from the table
     async fn delete(&self, predicate: Predicate<'_>) -> Result<DeleteResult> {
-        let result = delete::execute_delete(self, predicate).await?;
-        self.bump_freshness();
-        Ok(result)
+        delete::execute_delete(self, predicate).await
     }
 
     async fn tags(&self) -> Result<Box<dyn Tags + '_>> {
@@ -2784,30 +2746,22 @@ impl BaseTable for NativeTable {
         transforms: NewColumnTransform,
         read_columns: Option<Vec<String>>,
     ) -> Result<AddColumnsResult> {
-        let result = schema_evolution::execute_add_columns(self, transforms, read_columns).await?;
-        self.bump_freshness();
-        Ok(result)
+        schema_evolution::execute_add_columns(self, transforms, read_columns).await
     }
 
     async fn alter_columns(&self, alterations: &[ColumnAlteration]) -> Result<AlterColumnsResult> {
-        let result = schema_evolution::execute_alter_columns(self, alterations).await?;
-        self.bump_freshness();
-        Ok(result)
+        schema_evolution::execute_alter_columns(self, alterations).await
     }
 
     async fn update_field_metadata(
         &self,
         updates: &[FieldMetadataUpdate],
     ) -> Result<UpdateFieldMetadataResult> {
-        let result = schema_evolution::execute_update_field_metadata(self, updates).await?;
-        self.bump_freshness();
-        Ok(result)
+        schema_evolution::execute_update_field_metadata(self, updates).await
     }
 
     async fn drop_columns(&self, columns: &[&str]) -> Result<DropColumnsResult> {
-        let result = schema_evolution::execute_drop_columns(self, columns).await?;
-        self.bump_freshness();
-        Ok(result)
+        schema_evolution::execute_drop_columns(self, columns).await
     }
 
     async fn list_indices(&self) -> Result<Vec<IndexConfig>> {

rust/lancedb/src/table/add_data.rs

@@ -26,9 +26,6 @@ pub enum AddDataMode {
     #[default]
     Append,
     /// The existing table will be overwritten with the new data
-    ///
-    /// On overwrite, raw binary is not coerced into a blob struct. The input
-    /// must declare blob v2 for the column to stay a blob column.
     Overwrite,
 }
 

rust/lancedb/src/table/datafusion.rs

@@ -3,7 +3,6 @@
 
 //! This module contains adapters to allow LanceDB tables to be used as DataFusion table providers.
 
-mod blob_coerce;
 pub mod cast;
 pub mod insert;
 pub mod reject_nan;

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

@@ -1,495 +0,0 @@
-// SPDX-License-Identifier: Apache-2.0
-// SPDX-FileCopyrightText: Copyright The LanceDB Authors
-
-//! Coerces write-path input into blob v2 struct columns.
-//!
-//! [`super::cast::cast_to_table_schema`] calls [`coerce_blob_expr`].
-
-use std::sync::Arc;
-
-use arrow_schema::{DataType, Field, FieldRef};
-use datafusion::functions::core::{get_field, named_struct};
-use datafusion_common::ScalarValue;
-use datafusion_common::config::ConfigOptions;
-use datafusion_physical_expr::ScalarFunctionExpr;
-use datafusion_physical_expr::expressions::{CastExpr, Literal};
-use datafusion_physical_plan::PhysicalExpr;
-
-use crate::error::{Error, Result};
-
-/// Build a projection expression coercing `input_expr` into the blob struct
-/// declared by `table_field`, composing `named_struct` / `get_field` / `cast`.
-pub(super) fn coerce_blob_expr(
-    input_expr: Arc<dyn PhysicalExpr>,
-    input_field: &Field,
-    table_field: &FieldRef,
-    config: &Arc<ConfigOptions>,
-) -> Result<(Arc<dyn PhysicalExpr>, FieldRef)> {
-    let DataType::Struct(declared_fields) = table_field.data_type() else {
-        return Err(Error::InvalidInput {
-            message: format!(
-                "blob v2 column '{}' must be a struct, table declares {}",
-                table_field.name(),
-                table_field.data_type()
-            ),
-        });
-    };
-
-    let input_struct_children = match input_field.data_type() {
-        DataType::Binary | DataType::LargeBinary | DataType::BinaryView => None,
-        DataType::Struct(children) => {
-            if !children
-                .iter()
-                .any(|c| c.name() == "data" || c.name() == "uri")
-            {
-                return Err(Error::InvalidInput {
-                    message: format!(
-                        "blob struct input for column '{}' must contain a 'data' or 'uri' child",
-                        table_field.name()
-                    ),
-                });
-            }
-            Some(children)
-        }
-        other => {
-            return Err(Error::InvalidInput {
-                message: format!(
-                    "cannot coerce column '{}' with type {} into a blob v2 struct. \
-                     expected Binary, LargeBinary, BinaryView, or a Struct with a 'data' or 'uri' child",
-                    table_field.name(),
-                    other,
-                ),
-            });
-        }
-    };
-
-    let mut ns_args: Vec<Arc<dyn PhysicalExpr>> = Vec::with_capacity(declared_fields.len() * 2);
-    for declared in declared_fields.iter() {
-        ns_args.push(Arc::new(Literal::new(ScalarValue::from(
-            declared.name().as_str(),
-        ))));
-
-        let value: Arc<dyn PhysicalExpr> = match input_struct_children {
-            // Raw binary lands in `data` and everything else is a typed null.
-            None => {
-                if declared.name() == "data" {
-                    Arc::new(CastExpr::new(
-                        input_expr.clone(),
-                        declared.data_type().clone(),
-                        None,
-                    ))
-                } else {
-                    typed_null(declared.data_type())?
-                }
-            }
-            Some(children) => match children.iter().find(|c| c.name() == declared.name()) {
-                Some(child) => {
-                    let field_expr: Arc<dyn PhysicalExpr> = Arc::new(ScalarFunctionExpr::new(
-                        &format!("get_field({})", declared.name()),
-                        get_field(),
-                        vec![
-                            input_expr.clone(),
-                            Arc::new(Literal::new(ScalarValue::from(declared.name().as_str()))),
-                        ],
-                        Arc::new(child.as_ref().clone()),
-                        config.clone(),
-                    ));
-                    if child.data_type() == declared.data_type() {
-                        field_expr
-                    } else {
-                        Arc::new(CastExpr::new(
-                            field_expr,
-                            declared.data_type().clone(),
-                            None,
-                        ))
-                    }
-                }
-                None => typed_null(declared.data_type())?,
-            },
-        };
-        ns_args.push(value);
-    }
-
-    let expr: Arc<dyn PhysicalExpr> = Arc::new(ScalarFunctionExpr::new(
-        &format!("named_struct({})", table_field.name()),
-        named_struct(),
-        ns_args,
-        table_field.clone(),
-        config.clone(),
-    ));
-    Ok((expr, table_field.clone()))
-}
-
-fn typed_null(data_type: &DataType) -> Result<Arc<dyn PhysicalExpr>> {
-    let scalar = ScalarValue::try_from(data_type).map_err(|e| Error::InvalidInput {
-        message: format!("cannot build null literal for blob child type {data_type}: {e}"),
-    })?;
-    Ok(Arc::new(Literal::new(scalar)))
-}
-
-#[cfg(test)]
-mod tests {
-    use super::super::cast::cast_to_table_schema;
-    use super::*;
-    use crate::blob::blob;
-    use arrow_array::{
-        Array, ArrayRef, BinaryArray, BinaryViewArray, Int32Array, Int64Array, LargeBinaryArray,
-        RecordBatch, StringArray, StructArray, UInt8Array, UInt64Array,
-    };
-    use arrow_schema::Schema;
-    use datafusion::prelude::SessionContext;
-    use datafusion_catalog::MemTable;
-    use datafusion_physical_plan::ExecutionPlan;
-    use futures::TryStreamExt;
-    use lance_arrow::FieldExt;
-    use std::collections::HashMap;
-
-    fn wide_blob_field(name: &str) -> Field {
-        Field::new(
-            name,
-            DataType::Struct(
-                vec![
-                    Field::new("data", DataType::LargeBinary, true),
-                    Field::new("uri", DataType::Utf8, true),
-                    Field::new("position", DataType::UInt64, true),
-                    Field::new("size", DataType::UInt64, true),
-                ]
-                .into(),
-            ),
-            true,
-        )
-        .with_metadata(HashMap::from([(
-            "ARROW:extension:name".to_string(),
-            "lance.blob.v2".to_string(),
-        )]))
-    }
-
-    fn blob_table_schema() -> Schema {
-        Schema::new(vec![
-            Field::new("id", DataType::Int64, false),
-            blob("image", true),
-        ])
-    }
-
-    fn batch_with_image(image_field: Field, image: ArrayRef) -> RecordBatch {
-        let len = image.len();
-        RecordBatch::try_new(
-            Arc::new(Schema::new(vec![
-                Field::new("id", DataType::Int64, false),
-                image_field,
-            ])),
-            vec![Arc::new(Int64Array::from_iter_values(0..len as i64)), image],
-        )
-        .unwrap()
-    }
-
-    fn image_struct(batch: &RecordBatch) -> &StructArray {
-        batch
-            .column_by_name("image")
-            .unwrap()
-            .as_any()
-            .downcast_ref::<StructArray>()
-            .unwrap()
-    }
-
-    async fn plan_from_batch(batch: RecordBatch) -> Arc<dyn ExecutionPlan> {
-        let schema = batch.schema();
-        let table = MemTable::try_new(schema, vec![vec![batch]]).unwrap();
-        let ctx = SessionContext::new();
-        ctx.register_table("t", Arc::new(table)).unwrap();
-        let df = ctx.table("t").await.unwrap();
-        df.create_physical_plan().await.unwrap()
-    }
-
-    async fn coerce(batch: RecordBatch, table_schema: &Schema) -> RecordBatch {
-        let plan = plan_from_batch(batch).await;
-        let plan = cast_to_table_schema(plan, table_schema).unwrap();
-        let ctx = SessionContext::new();
-        let stream = plan.execute(0, ctx.task_ctx()).unwrap();
-        let batches: Vec<RecordBatch> = stream.try_collect().await.unwrap();
-        arrow_select::concat::concat_batches(&plan.schema(), &batches).unwrap()
-    }
-
-    async fn coerce_err(batch: RecordBatch, table_schema: &Schema) -> Error {
-        let plan = plan_from_batch(batch).await;
-        cast_to_table_schema(plan, table_schema).unwrap_err()
-    }
-
-    #[tokio::test]
-    async fn large_binary_coerces_to_declared_blob_struct() {
-        let batch = batch_with_image(
-            Field::new("image", DataType::LargeBinary, true),
-            Arc::new(LargeBinaryArray::from_iter_values([b"hello".as_slice()])),
-        );
-        let coerced = coerce(batch, &blob_table_schema()).await;
-        let image_field = coerced.schema().field_with_name("image").unwrap().clone();
-        assert!(image_field.is_blob_v2());
-        assert!(matches!(image_field.data_type(), DataType::Struct(_)));
-        let data = image_struct(&coerced).column_by_name("data").unwrap();
-        let data: &LargeBinaryArray = data.as_any().downcast_ref().unwrap();
-        assert_eq!(data.value(0), b"hello");
-    }
-
-    #[tokio::test]
-    async fn binary_coerces_to_declared_blob_struct() {
-        let batch = batch_with_image(
-            Field::new("image", DataType::Binary, true),
-            Arc::new(BinaryArray::from_iter_values([b"hi".as_slice()])),
-        );
-        let coerced = coerce(batch, &blob_table_schema()).await;
-        assert!(
-            coerced
-                .schema()
-                .field_with_name("image")
-                .unwrap()
-                .is_blob_v2()
-        );
-    }
-
-    #[tokio::test]
-    async fn binary_view_coerces_to_declared_blob_struct() {
-        let batch = batch_with_image(
-            Field::new("image", DataType::BinaryView, true),
-            Arc::new(BinaryViewArray::from_iter_values([b"view".as_slice()])),
-        );
-        let coerced = coerce(batch, &blob_table_schema()).await;
-        let data = image_struct(&coerced).column_by_name("data").unwrap();
-        let data: &LargeBinaryArray = data.as_any().downcast_ref().unwrap();
-        assert_eq!(data.value(0), b"view");
-    }
-
-    #[tokio::test]
-    async fn binary_nulls_stay_null_after_coercion() {
-        let batch = batch_with_image(
-            Field::new("image", DataType::Binary, true),
-            Arc::new(BinaryArray::from_iter(vec![
-                Some(b"present".as_slice()),
-                None,
-            ])),
-        );
-        let coerced = coerce(batch, &blob_table_schema()).await;
-        let image = image_struct(&coerced);
-        let data = image.column_by_name("data").unwrap();
-        assert!(!data.is_null(0));
-        assert!(data.is_null(1));
-    }
-
-    #[tokio::test]
-    async fn binary_coerces_into_four_child_blob_layout() {
-        let table_schema = Schema::new(vec![
-            Field::new("id", DataType::Int64, false),
-            wide_blob_field("image"),
-        ]);
-        let batch = batch_with_image(
-            Field::new("image", DataType::LargeBinary, true),
-            Arc::new(LargeBinaryArray::from_iter(vec![
-                Some(b"alpha".as_slice()),
-                None,
-            ])),
-        );
-        let coerced = coerce(batch, &table_schema).await;
-        let image = image_struct(&coerced);
-        assert_eq!(
-            image.num_columns(),
-            4,
-            "coerced struct keeps the declared layout"
-        );
-        assert!(image.column_by_name("position").unwrap().is_null(0));
-        assert!(image.column_by_name("size").unwrap().is_null(0));
-        assert!(!image.column_by_name("data").unwrap().is_null(0));
-        assert!(image.column_by_name("data").unwrap().is_null(1));
-    }
-
-    #[tokio::test]
-    async fn prebuilt_struct_gains_blob_field_metadata() {
-        let DataType::Struct(children) = blob("image", true).data_type().clone() else {
-            unreachable!("blob field is a struct")
-        };
-        let prebuilt = StructArray::new(
-            children,
-            vec![
-                Arc::new(LargeBinaryArray::from_iter_values([b"prebuilt".as_slice()])),
-                Arc::new(StringArray::from(vec![None::<&str>])),
-            ],
-            None,
-        );
-        let batch = batch_with_image(
-            Field::new("image", prebuilt.data_type().clone(), true),
-            Arc::new(prebuilt),
-        );
-        let coerced = coerce(batch, &blob_table_schema()).await;
-        assert!(
-            coerced
-                .schema()
-                .field_with_name("image")
-                .unwrap()
-                .is_blob_v2()
-        );
-    }
-
-    #[tokio::test]
-    async fn prebuilt_narrow_struct_widens_to_declared_layout() {
-        let DataType::Struct(narrow_children) = blob("image", true).data_type().clone() else {
-            unreachable!("blob field is a struct")
-        };
-        let prebuilt = StructArray::new(
-            narrow_children,
-            vec![
-                Arc::new(LargeBinaryArray::from_iter_values([b"prebuilt".as_slice()])),
-                Arc::new(StringArray::from(vec![None::<&str>])),
-            ],
-            None,
-        );
-        let table_schema = Schema::new(vec![
-            Field::new("id", DataType::Int64, false),
-            wide_blob_field("image"),
-        ]);
-        let batch = batch_with_image(
-            Field::new("image", prebuilt.data_type().clone(), true),
-            Arc::new(prebuilt),
-        );
-        let coerced = coerce(batch, &table_schema).await;
-        let image = image_struct(&coerced);
-        assert_eq!(image.num_columns(), 4);
-        assert!(image.column_by_name("position").unwrap().is_null(0));
-        assert!(image.column_by_name("size").unwrap().is_null(0));
-    }
-
-    #[tokio::test]
-    async fn external_reference_struct_preserves_uri_position_and_size() {
-        let prebuilt = StructArray::new(
-            vec![
-                Field::new("data", DataType::LargeBinary, true),
-                Field::new("uri", DataType::Utf8, true),
-                Field::new("position", DataType::UInt64, true),
-                Field::new("size", DataType::UInt64, true),
-            ]
-            .into(),
-            vec![
-                Arc::new(LargeBinaryArray::from(vec![None::<&[u8]>])) as ArrayRef,
-                Arc::new(StringArray::from(vec![Some("s3://bucket/blob.bin")])) as ArrayRef,
-                Arc::new(UInt64Array::from(vec![Some(7)])) as ArrayRef,
-                Arc::new(UInt64Array::from(vec![Some(6)])) as ArrayRef,
-            ],
-            None,
-        );
-        let table_schema = Schema::new(vec![
-            Field::new("id", DataType::Int64, false),
-            wide_blob_field("image"),
-        ]);
-        let batch = batch_with_image(
-            Field::new("image", prebuilt.data_type().clone(), true),
-            Arc::new(prebuilt),
-        );
-        let coerced = coerce(batch, &table_schema).await;
-        let image = image_struct(&coerced);
-
-        let uri: &StringArray = image
-            .column_by_name("uri")
-            .unwrap()
-            .as_any()
-            .downcast_ref()
-            .unwrap();
-        assert_eq!(uri.value(0), "s3://bucket/blob.bin");
-        let position: &UInt64Array = image
-            .column_by_name("position")
-            .unwrap()
-            .as_any()
-            .downcast_ref()
-            .unwrap();
-        assert_eq!(position.value(0), 7);
-        let size: &UInt64Array = image
-            .column_by_name("size")
-            .unwrap()
-            .as_any()
-            .downcast_ref()
-            .unwrap();
-        assert_eq!(size.value(0), 6);
-        assert!(image.column_by_name("data").unwrap().is_null(0));
-    }
-
-    #[tokio::test]
-    async fn descriptor_struct_without_value_child_is_rejected() {
-        let descriptor = StructArray::new(
-            vec![
-                Field::new("kind", DataType::UInt8, false),
-                Field::new("position", DataType::UInt64, false),
-                Field::new("size", DataType::UInt64, false),
-            ]
-            .into(),
-            vec![
-                Arc::new(UInt8Array::from(vec![0])),
-                Arc::new(UInt64Array::from(vec![0])),
-                Arc::new(UInt64Array::from(vec![0])),
-            ],
-            None,
-        );
-        let batch = batch_with_image(
-            Field::new("image", descriptor.data_type().clone(), true),
-            Arc::new(descriptor),
-        );
-        let err = coerce_err(batch, &blob_table_schema()).await;
-        assert!(err.to_string().contains("'data' or 'uri'"));
-        assert!(err.to_string().contains("image"));
-    }
-
-    #[tokio::test]
-    async fn unsupported_input_type_is_rejected_with_column_name() {
-        let batch = batch_with_image(
-            Field::new("image", DataType::Utf8, true),
-            Arc::new(StringArray::from(vec!["not bytes"])),
-        );
-        let err = coerce_err(batch, &blob_table_schema()).await;
-        assert!(matches!(err, Error::InvalidInput { .. }), "got {err:?}");
-        assert!(err.to_string().contains("image"));
-    }
-
-    #[tokio::test]
-    async fn blob_metadata_survives_cast_of_sibling_column() {
-        let batch = RecordBatch::try_new(
-            Arc::new(Schema::new(vec![
-                Field::new("id", DataType::Int32, false),
-                Field::new("image", DataType::LargeBinary, true),
-            ])),
-            vec![
-                Arc::new(Int32Array::from(vec![1])),
-                Arc::new(LargeBinaryArray::from_iter_values([b"x".as_slice()])),
-            ],
-        )
-        .unwrap();
-        let coerced = coerce(batch, &blob_table_schema()).await;
-
-        let image_field = coerced.schema().field_with_name("image").unwrap().clone();
-        assert!(
-            image_field.is_blob_v2(),
-            "expected blob marker on image field, got {:?}",
-            image_field.metadata()
-        );
-        assert_eq!(
-            coerced.schema().field_with_name("id").unwrap().data_type(),
-            &DataType::Int64
-        );
-    }
-
-    #[tokio::test]
-    async fn exact_blob_input_passes_through_unchanged() {
-        let DataType::Struct(children) = blob("image", true).data_type().clone() else {
-            unreachable!("blob field is a struct")
-        };
-        let image = StructArray::new(
-            children,
-            vec![
-                Arc::new(LargeBinaryArray::from_iter_values([b"exact".as_slice()])),
-                Arc::new(StringArray::from(vec![None::<&str>])),
-            ],
-            None,
-        );
-        let batch = batch_with_image(blob("image", true), Arc::new(image));
-        let table_schema = blob_table_schema();
-
-        let input = plan_from_batch(batch).await;
-        let input_ptr = Arc::as_ptr(&input);
-        let plan = cast_to_table_schema(input, &table_schema).unwrap();
-        assert_eq!(Arc::as_ptr(&plan), input_ptr, "no projection inserted");
-    }
-}

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

@@ -13,10 +13,8 @@ use datafusion_physical_expr::expressions::{CastExpr, Literal};
 use datafusion_physical_plan::expressions::Column;
 use datafusion_physical_plan::projection::ProjectionExec;
 use datafusion_physical_plan::{ExecutionPlan, PhysicalExpr};
-use lance_arrow::FieldExt;
 use lance_arrow::json::{is_arrow_json_field, is_json_field};
 
-use super::blob_coerce::coerce_blob_expr;
 use crate::{Error, Result};
 
 pub fn cast_to_table_schema(
@@ -79,17 +77,6 @@ fn build_field_exprs(
             continue;
         }
 
-        // Blob columns accept raw binary on write; exact matches pass through below.
-        if table_field.is_blob_v2() && input_field.as_ref() != table_field.as_ref() {
-            result.push(coerce_blob_expr(
-                input_expr,
-                input_field,
-                table_field,
-                &config,
-            )?);
-            continue;
-        }
-
         let expr = match (input_field.data_type(), table_field.data_type()) {
             // Both are structs: recurse into sub-fields to handle subschemas and casts.
             (DataType::Struct(in_children), DataType::Struct(tbl_children))

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

@@ -4,7 +4,6 @@
 //! DataFusion ExecutionPlan for inserting data into LanceDB tables.
 
 use std::any::Any;
-use std::sync::atomic::{AtomicU64, Ordering};
 use std::sync::{Arc, LazyLock, Mutex};
 
 use arrow_array::{RecordBatch, UInt64Array};
@@ -21,12 +20,11 @@ use datafusion_physical_plan::{
 use futures::TryStreamExt;
 use lance::Dataset;
 use lance::dataset::transaction::{Operation, Transaction};
-use lance::dataset::{CommitBuilder, InsertBuilder, WriteParams, WriteProgressFn};
+use lance::dataset::{CommitBuilder, InsertBuilder, WriteParams};
 use lance::io::exec::utils::InstrumentedRecordBatchStreamAdapter;
 use lance_table::format::Fragment;
 
 use crate::table::dataset::DatasetConsistencyWrapper;
-use crate::table::write_progress::WriteProgressTracker;
 
 pub(crate) static COUNT_SCHEMA: LazyLock<SchemaRef> = LazyLock::new(|| {
     Arc::new(ArrowSchema::new(vec![Field::new(
@@ -83,7 +81,6 @@ pub struct InsertExec {
     dataset: Arc<Dataset>,
     input: Arc<dyn ExecutionPlan>,
     write_params: WriteParams,
-    tracker: Option<Arc<WriteProgressTracker>>,
     properties: Arc<PlanProperties>,
     partial_transactions: Arc<Mutex<Vec<Transaction>>>,
     metrics: ExecutionPlanMetricsSet,
@@ -95,16 +92,6 @@ impl InsertExec {
         dataset: Arc<Dataset>,
         input: Arc<dyn ExecutionPlan>,
         write_params: WriteParams,
-    ) -> Self {
-        Self::new_with_tracker(ds_wrapper, dataset, input, write_params, None)
-    }
-
-    pub(crate) fn new_with_tracker(
-        ds_wrapper: DatasetConsistencyWrapper,
-        dataset: Arc<Dataset>,
-        input: Arc<dyn ExecutionPlan>,
-        write_params: WriteParams,
-        tracker: Option<Arc<WriteProgressTracker>>,
     ) -> Self {
         let schema = COUNT_SCHEMA.clone();
         let num_partitions = input.output_partitioning().partition_count();
@@ -120,7 +107,6 @@ impl InsertExec {
             dataset,
             input,
             write_params,
-            tracker,
             properties: Arc::new(properties),
             partial_transactions: Arc::new(Mutex::new(Vec::with_capacity(num_partitions))),
             metrics: ExecutionPlanMetricsSet::new(),
@@ -175,12 +161,11 @@ impl ExecutionPlan for InsertExec {
                 "InsertExec requires exactly one child".to_string(),
             ));
         }
-        Ok(Arc::new(Self::new_with_tracker(
+        Ok(Arc::new(Self::new(
             self.ds_wrapper.clone(),
             self.dataset.clone(),
             children[0].clone(),
             self.write_params.clone(),
-            self.tracker.clone(),
         )))
     }
 
@@ -191,11 +176,10 @@ impl ExecutionPlan for InsertExec {
     ) -> DataFusionResult<SendableRecordBatchStream> {
         let input_stream = self.input.execute(partition, context)?;
         let dataset = self.dataset.clone();
-        let mut write_params = self.write_params.clone();
+        let write_params = self.write_params.clone();
         let partial_transactions = self.partial_transactions.clone();
         let total_partitions = self.input.output_partitioning().partition_count();
         let ds_wrapper = self.ds_wrapper.clone();
-        let tracker = self.tracker.clone();
 
         let output_bytes = MetricBuilder::new(&self.metrics).output_bytes(partition);
         let input_schema = input_stream.schema();
@@ -211,20 +195,6 @@ impl ExecutionPlan for InsertExec {
             ));
 
         let stream = futures::stream::once(async move {
-            if let Some(tracker) = tracker
-                && write_params.write_progress.is_none()
-            {
-                let last_bytes = Arc::new(AtomicU64::new(0));
-                write_params.write_progress = Some(WriteProgressFn::new(move |stats| {
-                    let previous = last_bytes.swap(stats.bytes_written, Ordering::Relaxed);
-                    if stats.bytes_written > previous {
-                        let delta =
-                            usize::try_from(stats.bytes_written - previous).unwrap_or(usize::MAX);
-                        tracker.record_bytes(delta);
-                    }
-                }));
-            }
-
             let transaction = InsertBuilder::new(dataset.clone())
                 .with_params(&write_params)
                 .execute_uncommitted_stream(input_stream)

rust/lancedb/src/table/dataset.rs

@@ -518,10 +518,6 @@ mod tests {
 
         let wrapper = DatasetConsistencyWrapper::new_latest(ds, Some(Duration::from_millis(200)));
 
-        // Freeze `cached_at` on the mock clock so a slow external write below can't
-        // expire the TTL before the explicit advance_by() does (flake on loaded CI).
-        clock::pin();
-
         // Populate the cache
         let v1 = wrapper.get().await.unwrap().version().version;
         assert_eq!(v1, 1);

rust/lancedb/src/table/write_progress.rs

@@ -142,21 +142,11 @@ impl WriteProgressTracker {
         cb(&progress);
     }
 
-    /// Record wire bytes from the insert layer.
-    ///
-    /// These bytes may be IPC-encoded bytes for remote writes or bytes handed
-    /// to Lance's local writer. When wire bytes are recorded, they take
-    /// precedence over the in-memory Arrow bytes tracked by [`record_batch`].
+    /// 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);
-        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 progress = self.snapshot(guard.0, guard.1, false);
-        drop(guard);
-        cb(&progress);
     }
 
     /// Emit the final progress callback indicating the write is complete.
@@ -179,6 +169,8 @@ impl WriteProgressTracker {
         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(),
@@ -391,54 +383,6 @@ mod tests {
         }
     }
 
-    #[tokio::test]
-    async fn test_progress_uses_lance_write_bytes_for_local_tables() {
-        let dir = tempfile::tempdir().unwrap();
-        let db = connect(dir.path().to_str().unwrap())
-            .execute()
-            .await
-            .unwrap();
-
-        let batch = record_batch!(("id", Int32, [1, 2, 3])).unwrap();
-        let table = db
-            .create_table("local_write_bytes", batch)
-            .execute()
-            .await
-            .unwrap();
-
-        let new_data = record_batch!(("id", Int32, [4, 5, 6])).unwrap();
-        let in_memory_bytes = new_data.get_array_memory_size();
-        let final_bytes = Arc::new(AtomicUsize::new(0));
-        let seen_non_memory_bytes = Arc::new(std::sync::atomic::AtomicBool::new(false));
-        let final_bytes_cb = final_bytes.clone();
-        let seen_non_memory_bytes_cb = seen_non_memory_bytes.clone();
-
-        table
-            .add(new_data)
-            .write_parallelism(1)
-            .progress(move |p| {
-                if p.output_bytes() > 0 && p.output_bytes() != in_memory_bytes {
-                    seen_non_memory_bytes_cb.store(true, Ordering::SeqCst);
-                }
-                if p.done() {
-                    final_bytes_cb.store(p.output_bytes(), Ordering::SeqCst);
-                }
-            })
-            .execute()
-            .await
-            .unwrap();
-
-        assert!(
-            seen_non_memory_bytes.load(Ordering::SeqCst),
-            "progress should report Lance writer bytes, not only Arrow memory bytes"
-        );
-        assert_ne!(
-            final_bytes.load(Ordering::SeqCst),
-            in_memory_bytes,
-            "final progress bytes should come from Lance write stats"
-        );
-    }
-
     #[test]
     fn test_record_batch_recovers_from_poisoned_callback_lock() {
         use super::{ProgressCallback, WriteProgressTracker};

rust/lancedb/src/utils/background_cache.rs

@@ -329,15 +329,6 @@ pub mod clock {
         });
     }
 
-    /// Start mock time at the current instant if not already pinned.
-    pub fn pin() {
-        MOCK_NOW.with(|mock| {
-            if mock.get().is_none() {
-                mock.set(Some(Instant::now()));
-            }
-        });
-    }
-
     #[allow(dead_code)]
     pub fn clear_mock() {
         MOCK_NOW.with(|mock| mock.set(None));

rust/lancedb/tests/blob_integration.rs

@@ -1,380 +0,0 @@
-// SPDX-License-Identifier: Apache-2.0
-// SPDX-FileCopyrightText: Copyright The LanceDB Authors
-
-//! Integration tests for blob v2 columns.
-
-use std::sync::Arc;
-
-use arrow_array::{Array, BinaryArray, Int64Array, LargeBinaryArray, RecordBatch, StructArray};
-use arrow_schema::{DataType, Field, Schema};
-use futures::TryStreamExt;
-use lance_encoding::version::LanceFileVersion;
-use lancedb::{
-    Connection, Result, Table, blob::blob, connect,
-    database::listing::OPT_NEW_TABLE_ENABLE_STABLE_ROW_IDS, query::ExecutableQuery,
-};
-use tempfile::tempdir;
-
-fn blob_table_schema() -> Arc<Schema> {
-    Arc::new(Schema::new(vec![
-        Field::new("id", DataType::Int64, false),
-        blob("image", true),
-    ]))
-}
-
-fn binary_input_batch(ids: &[i64], payloads: &[Option<&[u8]>]) -> RecordBatch {
-    RecordBatch::try_new(
-        Arc::new(Schema::new(vec![
-            Field::new("id", DataType::Int64, false),
-            Field::new("image", DataType::LargeBinary, true),
-        ])),
-        vec![
-            Arc::new(Int64Array::from(ids.to_vec())),
-            Arc::new(LargeBinaryArray::from_iter(payloads.iter().copied())),
-        ],
-    )
-    .unwrap()
-}
-
-async fn create_inline_blob_table(
-    db: &Connection,
-    name: &str,
-    ids: &[i64],
-    payloads: &[Option<&[u8]>],
-) -> Result<Table> {
-    let table = db
-        .create_empty_table(name, blob_table_schema())
-        .execute()
-        .await?;
-    table
-        .add(binary_input_batch(ids, payloads))
-        .execute()
-        .await?;
-    Ok(table)
-}
-
-async fn storage_format_version(table: &Table) -> LanceFileVersion {
-    table
-        .as_native()
-        .unwrap()
-        .manifest()
-        .await
-        .unwrap()
-        .data_storage_format
-        .lance_file_version()
-        .unwrap()
-        .resolve()
-}
-
-async fn uses_stable_row_ids(table: &Table) -> bool {
-    table
-        .as_native()
-        .unwrap()
-        .manifest()
-        .await
-        .unwrap()
-        .uses_stable_row_ids()
-}
-
-async fn query_image_struct(table: &Table) -> StructArray {
-    let batches = table
-        .query()
-        .execute()
-        .await
-        .unwrap()
-        .try_collect::<Vec<_>>()
-        .await
-        .unwrap();
-    let batch = arrow_select::concat::concat_batches(&batches[0].schema(), &batches).unwrap();
-    batch
-        .column_by_name("image")
-        .expect("image column present")
-        .as_any()
-        .downcast_ref::<StructArray>()
-        .expect("blob column reads back as a descriptor struct")
-        .clone()
-}
-
-#[tokio::test]
-async fn declaring_blob_column_bumps_format_and_enables_stable_row_ids() -> Result<()> {
-    let tmp = tempdir().unwrap();
-    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
-    let table = db
-        .create_empty_table("t", blob_table_schema())
-        .execute()
-        .await?;
-
-    assert!(storage_format_version(&table).await >= LanceFileVersion::V2_2);
-    assert!(uses_stable_row_ids(&table).await);
-    Ok(())
-}
-
-#[tokio::test]
-async fn explicit_stable_row_id_setting_wins_over_blob_default() -> Result<()> {
-    let tmp = tempdir().unwrap();
-    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
-    let table = db
-        .create_empty_table("t", blob_table_schema())
-        .storage_option(OPT_NEW_TABLE_ENABLE_STABLE_ROW_IDS, "false")
-        .execute()
-        .await?;
-
-    assert!(
-        storage_format_version(&table).await >= LanceFileVersion::V2_2,
-        "format bump still applies; the schema cannot be written below 2.2"
-    );
-    assert!(!uses_stable_row_ids(&table).await);
-    Ok(())
-}
-
-#[tokio::test]
-async fn non_blob_table_keeps_default_format_and_row_id_setting() -> Result<()> {
-    let tmp = tempdir().unwrap();
-    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
-    let schema = Arc::new(Schema::new(vec![Field::new("id", DataType::Int64, false)]));
-    let table = db.create_empty_table("t", schema).execute().await?;
-
-    assert!(storage_format_version(&table).await < LanceFileVersion::V2_2);
-    assert!(!uses_stable_row_ids(&table).await);
-    Ok(())
-}
-
-#[tokio::test]
-async fn creating_with_blob_data_bumps_format() -> Result<()> {
-    let tmp = tempdir().unwrap();
-    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
-
-    // Batch already declares the blob field (pre-built struct).
-    let blob_field = blob("image", true);
-    let DataType::Struct(children) = blob_field.data_type().clone() else {
-        unreachable!("blob field is a struct")
-    };
-    let image = StructArray::new(
-        children,
-        vec![
-            Arc::new(LargeBinaryArray::from_iter_values([b"payload".as_slice()])),
-            Arc::new(arrow_array::StringArray::from(vec![None::<&str>])),
-        ],
-        None,
-    );
-    let batch = RecordBatch::try_new(
-        Arc::new(Schema::new(vec![
-            Field::new("id", DataType::Int64, false),
-            blob_field,
-        ])),
-        vec![Arc::new(Int64Array::from(vec![1])), Arc::new(image)],
-    )
-    .unwrap();
-    let table = db.create_table("t", batch).execute().await?;
-
-    assert!(storage_format_version(&table).await >= LanceFileVersion::V2_2);
-    assert!(uses_stable_row_ids(&table).await);
-    assert_eq!(table.count_rows(None).await?, 1);
-    Ok(())
-}
-
-#[tokio::test]
-async fn add_coerces_large_binary_into_blob_column() -> Result<()> {
-    let tmp = tempdir().unwrap();
-    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
-    let table =
-        create_inline_blob_table(&db, "t", &[1, 2], &[Some(b"cat".as_slice()), Some(b"dog")])
-            .await?;
-
-    assert_eq!(table.count_rows(None).await?, 2);
-    let image = query_image_struct(&table).await;
-    assert_eq!(image.len(), 2);
-    // Table schema still has the blob marker after append.
-    let schema = table.schema().await?;
-    let field = schema.field_with_name("image").unwrap();
-    assert_eq!(
-        field
-            .metadata()
-            .get("ARROW:extension:name")
-            .map(String::as_str),
-        Some("lance.blob.v2")
-    );
-    Ok(())
-}
-
-#[tokio::test]
-async fn add_coerces_binary_into_blob_column() -> Result<()> {
-    let tmp = tempdir().unwrap();
-    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
-    let table = db
-        .create_empty_table("t", blob_table_schema())
-        .execute()
-        .await?;
-
-    let batch = RecordBatch::try_new(
-        Arc::new(Schema::new(vec![
-            Field::new("id", DataType::Int64, false),
-            Field::new("image", DataType::Binary, true),
-        ])),
-        vec![
-            Arc::new(Int64Array::from(vec![1])),
-            Arc::new(BinaryArray::from_iter_values([b"small".as_slice()])),
-        ],
-    )
-    .unwrap();
-    table.add(batch).execute().await?;
-
-    assert_eq!(table.count_rows(None).await?, 1);
-    Ok(())
-}
-
-#[tokio::test]
-async fn add_accepts_null_blob_rows() -> Result<()> {
-    let tmp = tempdir().unwrap();
-    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
-    let table = create_inline_blob_table(
-        &db,
-        "t",
-        &[1, 2, 3],
-        &[Some(b"first".as_slice()), None, Some(b"third")],
-    )
-    .await?;
-
-    assert_eq!(table.count_rows(None).await?, 3);
-    let image = query_image_struct(&table).await;
-    assert_eq!(image.len(), 3);
-    Ok(())
-}
-
-#[tokio::test]
-async fn add_rejects_uncoercible_blob_input() -> Result<()> {
-    let tmp = tempdir().unwrap();
-    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
-    let table = db
-        .create_empty_table("t", blob_table_schema())
-        .execute()
-        .await?;
-
-    let batch = RecordBatch::try_new(
-        Arc::new(Schema::new(vec![
-            Field::new("id", DataType::Int64, false),
-            Field::new("image", DataType::Utf8, true),
-        ])),
-        vec![
-            Arc::new(Int64Array::from(vec![1])),
-            Arc::new(arrow_array::StringArray::from(vec!["not bytes"])),
-        ],
-    )
-    .unwrap();
-    let err = table.add(batch).execute().await.unwrap_err();
-    assert!(err.to_string().contains("image"), "got: {err}");
-    Ok(())
-}
-
-#[tokio::test]
-async fn connection_level_stable_row_id_setting_wins_over_blob_default() -> Result<()> {
-    let tmp = tempdir().unwrap();
-    let db = connect(tmp.path().to_str().unwrap())
-        .storage_option(OPT_NEW_TABLE_ENABLE_STABLE_ROW_IDS, "false")
-        .execute()
-        .await?;
-    let table = db
-        .create_empty_table("t", blob_table_schema())
-        .execute()
-        .await?;
-
-    assert!(storage_format_version(&table).await >= LanceFileVersion::V2_2);
-    assert!(!uses_stable_row_ids(&table).await);
-    Ok(())
-}
-
-#[tokio::test]
-async fn namespace_create_applies_blob_defaults() -> Result<()> {
-    let tmp = tempdir().unwrap();
-    let mut properties = std::collections::HashMap::new();
-    properties.insert("root".to_string(), tmp.path().to_str().unwrap().to_string());
-    let db = lancedb::connect_namespace("dir", properties)
-        .execute()
-        .await?;
-    let table = db
-        .create_empty_table("t", blob_table_schema())
-        .execute()
-        .await?;
-
-    assert!(storage_format_version(&table).await >= LanceFileVersion::V2_2);
-    assert!(uses_stable_row_ids(&table).await);
-    Ok(())
-}
-
-// Overwrite takes the input schema as-is (same as cast skip). Raw binary
-// overwrite drops the blob marker unless the input declares blob v2.
-#[tokio::test]
-async fn overwrite_replaces_blob_schema_with_input_schema() -> Result<()> {
-    use lancedb::table::AddDataMode;
-
-    let tmp = tempdir().unwrap();
-    let db = connect(tmp.path().to_str().unwrap()).execute().await?;
-    let table = create_inline_blob_table(&db, "t", &[1], &[Some(b"blob".as_slice())]).await?;
-
-    // Raw binary overwrite. Plain LargeBinary replaces the blob declaration.
-    let raw_schema = Arc::new(Schema::new(vec![
-        Field::new("id", DataType::Int64, false),
-        Field::new("image", DataType::LargeBinary, true),
-    ]));
-    let raw_batch = RecordBatch::try_new(
-        raw_schema.clone(),
-        vec![
-            Arc::new(Int64Array::from(vec![2])),
-            Arc::new(LargeBinaryArray::from_iter_values([b"plain".as_slice()])),
-        ],
-    )
-    .unwrap();
-    table
-        .add(raw_batch)
-        .mode(AddDataMode::Overwrite)
-        .execute()
-        .await?;
-    let schema = table.schema().await?;
-    assert_eq!(schema, raw_schema);
-    assert!(
-        !schema
-            .field_with_name("image")
-            .unwrap()
-            .metadata()
-            .contains_key("ARROW:extension:name"),
-        "raw binary overwrite leaves a plain binary column"
-    );
-
-    // Overwrite with a declared blob struct keeps the blob column.
-    let blob_field = blob("image", true);
-    let DataType::Struct(children) = blob_field.data_type().clone() else {
-        unreachable!("blob field is a struct")
-    };
-    let image = StructArray::new(
-        children,
-        vec![
-            Arc::new(LargeBinaryArray::from_iter_values([b"declared".as_slice()])),
-            Arc::new(arrow_array::StringArray::from(vec![None::<&str>])),
-        ],
-        None,
-    );
-    let declared_batch = RecordBatch::try_new(
-        Arc::new(Schema::new(vec![
-            Field::new("id", DataType::Int64, false),
-            blob_field,
-        ])),
-        vec![Arc::new(Int64Array::from(vec![3])), Arc::new(image)],
-    )
-    .unwrap();
-    table
-        .add(declared_batch)
-        .mode(AddDataMode::Overwrite)
-        .execute()
-        .await?;
-    let schema = table.schema().await?;
-    assert_eq!(
-        schema
-            .field_with_name("image")
-            .unwrap()
-            .metadata()
-            .get("ARROW:extension:name")
-            .map(String::as_str),
-        Some("lance.blob.v2")
-    );
-    Ok(())
-}