diff --git a/.github/workflows/cargo-publish.yml b/.github/workflows/cargo-publish.yml
index 8172b1e584..0bcea7042d 100644
--- a/.github/workflows/cargo-publish.yml
+++ b/.github/workflows/cargo-publish.yml
@@ -8,7 +8,7 @@ on:
   workflow_dispatch:
     inputs:
       tag:
-        description: 'Tag to publish (e.g., v1.0.0)'
+        description: "Tag to publish (e.g., v1.0.0)"
         required: true
         type: string
 
@@ -24,7 +24,7 @@ jobs:
     env:
       # Need up-to-date compilers for kernels
       CC: clang-18
-      CXX: clang-18
+      CXX: clang++-18
     defaults:
       run:
         working-directory: .
@@ -53,5 +53,5 @@ jobs:
       - uses: albertlockett/publish-crates@v2.2
         with:
           registry-token: ${{ secrets.CARGO_REGISTRY_TOKEN }}
-          args: '--all-features'
+          args: "--all-features"
           path: .
diff --git a/.github/workflows/ci-benchmarks.yml b/.github/workflows/ci-benchmarks.yml
index 1b87ec69e0..bf6c4ee59f 100644
--- a/.github/workflows/ci-benchmarks.yml
+++ b/.github/workflows/ci-benchmarks.yml
@@ -13,7 +13,7 @@ jobs:
     env:
       # Need up-to-date compilers for kernels
       CC: clang-18
-      CXX: clang-18
+      CXX: clang++-18
     defaults:
       run:
         shell: bash
diff --git a/.github/workflows/python.yml b/.github/workflows/python.yml
index e716a0d6f6..f26a22b7de 100644
--- a/.github/workflows/python.yml
+++ b/.github/workflows/python.yml
@@ -39,7 +39,7 @@ jobs:
     env:
       # Need up-to-date compilers for kernels
       CC: clang-18
-      CXX: clang-18
+      CXX: clang++-18
     steps:
       - uses: actions/checkout@v4
         with:
@@ -67,8 +67,9 @@ jobs:
           sudo apt install -y protobuf-compiler libssl-dev
       - name: Lint Rust
         run: |
+          ALL_FEATURES=`cargo metadata --format-version=1 --no-deps | jq -r '.packages[] | .features | keys | .[]' | grep -v protoc | sort | uniq | paste -s -d "," -`
           cargo fmt --all -- --check
-          cargo clippy --locked --all-features --tests -- -D warnings
+          cargo clippy --locked --features ${ALL_FEATURES} --tests -- -D warnings
       - name: Build
         run: |
           python -m venv venv
diff --git a/.github/workflows/rust.yml b/.github/workflows/rust.yml
index 4a927d9a53..d9255d55ff 100644
--- a/.github/workflows/rust.yml
+++ b/.github/workflows/rust.yml
@@ -46,8 +46,9 @@ jobs:
           sudo apt install -y protobuf-compiler libssl-dev
       - name: Run clippy
         run: |
+          ALL_FEATURES=`cargo metadata --format-version=1 --no-deps | jq -r '.packages[] | .features | keys | .[]' | grep -v protoc | sort | uniq | paste -s -d "," -`
           cargo clippy --version
-          cargo clippy --locked --all-features --tests --benches -- -D warnings
+          cargo clippy --locked --features ${ALL_FEATURES} --tests --benches -- -D warnings
   linux-build:
     runs-on: "ubuntu-24.04"
     timeout-minutes: 45
@@ -59,7 +60,7 @@ jobs:
     env:
       # Need up-to-date compilers for kernels
       CC: clang
-      CXX: clang
+      CXX: clang++
     steps:
       - uses: actions/checkout@v4
       # pin the toolchain version to avoid surprises
@@ -81,13 +82,18 @@ jobs:
       - name: Run tests
         if: ${{ matrix.toolchain == 'stable' }}
         run: |
-          cargo llvm-cov  --locked --workspace --codecov --output-path coverage.codecov --all-features
+          ALL_FEATURES=`cargo metadata --format-version=1 --no-deps | jq -r '.packages[] | .features | keys | .[]' | grep -v protoc | sort | uniq | paste -s -d "," -`
+          cargo llvm-cov  --locked --workspace --codecov --output-path coverage.codecov --features ${ALL_FEATURES}
       - name: Build tests (nightly)
-        run: cargo test --locked --all-features --workspace --no-run
+        if: ${{ matrix.toolchain != 'stable' }}
+        run: |
+          ALL_FEATURES=`cargo metadata --format-version=1 --no-deps | jq -r '.packages[] | .features | keys | .[]' | grep -v protoc | sort | uniq | paste -s -d "," -`
+          cargo test --locked --features ${ALL_FEATURES} --workspace --no-run
       - name: Run tests (nightly)
         if: ${{ matrix.toolchain != 'stable' }}
         run: |
-          cargo test --all-features --workspace
+          ALL_FEATURES=`cargo metadata --format-version=1 --no-deps | jq -r '.packages[] | .features | keys | .[]' | grep -v protoc | sort | uniq | paste -s -d "," -`
+          cargo test --features ${ALL_FEATURES} --workspace
       - name: Upload coverage to Codecov
         if: ${{ matrix.toolchain == 'stable' }}
         uses: codecov/codecov-action@v4
@@ -113,20 +119,22 @@ jobs:
           sudo apt install -y protobuf-compiler libssl-dev pkg-config
       - name: Build tests
         run: |
-          cargo test --locked --all-features --no-run
+          ALL_FEATURES=`cargo metadata --format-version=1 --no-deps | jq -r '.packages[] | .features | keys | .[]' | grep -v protoc | sort | uniq | paste -s -d "," -`
+          cargo test --locked --features ${ALL_FEATURES} --no-run
       - name: Start DynamoDB local for tests
         run: |
           docker run -d -e AWS_ACCESS_KEY_ID=DUMMYKEY -e AWS_SECRET_ACCESS_KEY=DUMMYKEY -p 8000:8000 amazon/dynamodb-local
       - name: Run tests
         run: |
-          cargo test --locked --all-features
+          ALL_FEATURES=`cargo metadata --format-version=1 --no-deps | jq -r '.packages[] | .features | keys | .[]' | grep -v protoc | sort | uniq | paste -s -d "," -`
+          cargo test --locked --features ${ALL_FEATURES}
   build-no-lock:
     runs-on: ubuntu-24.04
     timeout-minutes: 30
     env:
       # Need up-to-date compilers for kernels
       CC: clang
-      CXX: clang
+      CXX: clang++
     steps:
       - uses: actions/checkout@v4
       # Remote cargo.lock to force a fresh build
@@ -139,7 +147,9 @@ jobs:
           sudo apt update
           sudo apt install -y protobuf-compiler libssl-dev
       - name: Build all
-        run: cargo build --benches --all-features --tests
+        run: |
+          ALL_FEATURES=`cargo metadata --format-version=1 --no-deps | jq -r '.packages[] | .features | keys | .[]' | grep -v protoc | sort | uniq | paste -s -d "," -`
+          cargo build --benches --features ${ALL_FEATURES} --tests
   mac-build:
     runs-on: "macos-14"
     timeout-minutes: 45
@@ -165,11 +175,14 @@ jobs:
         run: |
           rustup update ${{ matrix.toolchain }} && rustup default ${{ matrix.toolchain }}
       - name: Build tests
-        run: cargo test --locked --all-features --no-run
+        run: |
+          cargo test --locked --features fp16kernels,cli,tensorflow,dynamodb,substrait --no-run
       - name: Run tests
-        run: cargo test --all-features
+        run: |
+          cargo test --features fp16kernels,cli,tensorflow,dynamodb,substrait
       - name: Check benchmarks
-        run: cargo check --benches --all-features
+        run: |
+          cargo check --benches --features fp16kernels,cli,tensorflow,dynamodb,substrait
   windows-build:
     runs-on: windows-latest
     defaults:
@@ -203,7 +216,7 @@ jobs:
     env:
       # Need up-to-date compilers for kernels
       CC: clang
-      CXX: clang
+      CXX: clang++
     steps:
       - uses: actions/checkout@v4
         with:
@@ -218,4 +231,6 @@ jobs:
         with:
           toolchain: ${{ matrix.msrv }}
       - name: cargo +${{ matrix.msrv }} check
-        run: cargo check --workspace --tests --benches --all-features
+        run: |
+          ALL_FEATURES=`cargo metadata --format-version=1 --no-deps | jq -r '.packages[] | .features | keys | .[]' | grep -v protoc | sort | uniq | paste -s -d "," -`
+          cargo check --workspace --tests --benches --features ${ALL_FEATURES}
diff --git a/Cargo.lock b/Cargo.lock
index bde2af8f09..c428391884 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1313,6 +1313,15 @@ version = "0.7.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "afb84c814227b90d6895e01398aee0d8033c00e7466aca416fb6a8e0eb19d8a7"
 
+[[package]]
+name = "cmake"
+version = "0.1.52"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c682c223677e0e5b6b7f63a64b9351844c3f1b1678a68b7ee617e30fb082620e"
+dependencies = [
+ "cc",
+]
+
 [[package]]
 name = "colorchoice"
 version = "1.0.3"
@@ -3462,7 +3471,6 @@ dependencies = [
  "pretty_assertions",
  "prost 0.12.6",
  "prost 0.13.3",
- "prost-build 0.13.3",
  "prost-types 0.13.3",
  "rand",
  "random_word",
@@ -3617,6 +3625,7 @@ dependencies = [
  "prost 0.13.3",
  "prost-build 0.13.3",
  "prost-types 0.13.3",
+ "protobuf-src",
  "rand",
  "rand_xoshiro",
  "rstest",
@@ -3655,6 +3664,7 @@ dependencies = [
  "prost 0.13.3",
  "prost-build 0.13.3",
  "prost-types 0.13.3",
+ "protobuf-src",
  "rand",
  "snafu 0.7.5",
  "test-log",
@@ -3694,6 +3704,7 @@ dependencies = [
  "prost 0.13.3",
  "prost-build 0.13.3",
  "prost-types 0.13.3",
+ "protobuf-src",
  "rand",
  "roaring",
  "snafu 0.7.5",
@@ -3751,6 +3762,7 @@ dependencies = [
  "pprof",
  "prost 0.13.3",
  "prost-build 0.13.3",
+ "protobuf-src",
  "rand",
  "random_word",
  "rayon",
@@ -3800,7 +3812,6 @@ dependencies = [
  "pin-project",
  "pprof",
  "prost 0.13.3",
- "prost-build 0.13.3",
  "rand",
  "shellexpand",
  "snafu 0.7.5",
@@ -3894,6 +3905,7 @@ dependencies = [
  "prost 0.13.3",
  "prost-build 0.13.3",
  "prost-types 0.13.3",
+ "protobuf-src",
  "rand",
  "rangemap",
  "roaring",
@@ -5168,6 +5180,15 @@ dependencies = [
  "prost 0.13.3",
 ]
 
+[[package]]
+name = "protobuf-src"
+version = "2.1.0+27.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a7edafa3bcc668fa93efafcbdf58d7821bbda0f4b458ac7fae3d57ec0fec8167"
+dependencies = [
+ "cmake",
+]
+
 [[package]]
 name = "quanta"
 version = "0.12.3"
diff --git a/python/Cargo.lock b/python/Cargo.lock
index 3fe87b7edd..238c9f3e39 100644
--- a/python/Cargo.lock
+++ b/python/Cargo.lock
@@ -3063,7 +3063,6 @@ dependencies = [
  "pin-project",
  "prost 0.12.6",
  "prost 0.13.3",
- "prost-build 0.13.3",
  "prost-types 0.13.3",
  "rand",
  "roaring",
@@ -3330,7 +3329,6 @@ dependencies = [
  "path_abs",
  "pin-project",
  "prost 0.13.3",
- "prost-build 0.13.3",
  "rand",
  "shellexpand",
  "snafu 0.7.5",
diff --git a/python/python/lance/ray/sink.py b/python/python/lance/ray/sink.py
index 863a9fd014..20765f3e83 100644
--- a/python/python/lance/ray/sink.py
+++ b/python/python/lance/ray/sink.py
@@ -46,23 +46,8 @@ def _pd_to_arrow(
         tbl.schema = tbl.schema.remove_metadata()
         return tbl
     elif isinstance(df, pa.Table):
-        fields = df.schema.names
-        new_columns = []
-        new_fields = []
-        for field in fields:
-            col = df[field]
-            new_field = pa.field(field, col.type)
-            if (
-                pa.types.is_null(col.type)
-                and schema.field_by_name(field).type == pa.string()
-            ):
-                new_field = pa.field(field, pa.string())
-                col = pa.compute.if_else(pa.compute.is_null(col), NONE_ARROW_STR, col)
-            new_columns.append(col)
-            new_fields.append(new_field)
-        new_schema = pa.schema(fields=new_fields)
-        new_table = pa.Table.from_arrays(new_columns, schema=new_schema)
-        return new_table
+        if schema is not None:
+            return df.cast(schema)
     return df
 
 
@@ -439,6 +424,7 @@ def write_lance(
     output_uri: str,
     *,
     schema: Optional[pa.Schema] = None,
+    mode: Literal["create", "append", "overwrite"] = "create",
     transform: Optional[
         Callable[[pa.Table], Union[pa.Table, Generator[None, pa.Table, None]]]
     ] = None,
@@ -485,7 +471,9 @@ def write_lance(
         ),
         batch_size=max_rows_per_file,
     ).write_datasink(
-        LanceCommitter(output_uri, schema=schema, storage_options=storage_options)
+        LanceCommitter(
+            output_uri, schema=schema, mode=mode, storage_options=storage_options
+        )
     )
 
 
diff --git a/python/python/tests/test_ray.py b/python/python/tests/test_ray.py
index 54f1c42492..4c135c28be 100644
--- a/python/python/tests/test_ray.py
+++ b/python/python/tests/test_ray.py
@@ -138,3 +138,26 @@ def f(row):
     assert len(pylist) == 10
     for item in pylist:
         assert item is None
+
+
+@pytest.mark.filterwarnings("ignore::DeprecationWarning")
+def test_ray_write_lance_none_str_datasink(tmp_path: Path):
+    def f(row):
+        return {
+            "id": row["id"],
+            "str": None,
+        }
+
+    schema = pa.schema([pa.field("id", pa.int64()), pa.field("str", pa.string())])
+
+    sink = LanceDatasink(tmp_path, schema=schema)
+    (ray.data.range(10).map(f).write_datasink(sink))
+    ds = lance.dataset(tmp_path)
+    ds.count_rows() == 10
+    assert ds.schema == schema
+
+    tbl = ds.to_table()
+    pylist = tbl["str"].to_pylist()
+    assert len(pylist) == 10
+    for item in pylist:
+        assert item is None
diff --git a/rust/lance-encoding-datafusion/Cargo.toml b/rust/lance-encoding-datafusion/Cargo.toml
index 3cccdc9ec6..ffc608ed30 100644
--- a/rust/lance-encoding-datafusion/Cargo.toml
+++ b/rust/lance-encoding-datafusion/Cargo.toml
@@ -42,9 +42,17 @@ lance-datagen.workspace = true
 
 [build-dependencies]
 prost-build.workspace = true
+protobuf-src = { version = "2.1", optional = true }
 
 [target.'cfg(target_os = "linux")'.dev-dependencies]
 pprof = { workspace = true }
 
+[features]
+protoc = ["dep:protobuf-src"]
+
+[package.metadata.docs.rs]
+# docs.rs uses an older version of Ubuntu that does not have the necessary protoc version
+features = ["protoc"]
+
 [lints]
 workspace = true
diff --git a/rust/lance-encoding-datafusion/build.rs b/rust/lance-encoding-datafusion/build.rs
index 8d89a39ac3..9d0206e201 100644
--- a/rust/lance-encoding-datafusion/build.rs
+++ b/rust/lance-encoding-datafusion/build.rs
@@ -6,6 +6,10 @@ use std::io::Result;
 fn main() -> Result<()> {
     println!("cargo:rerun-if-changed=protos");
 
+    #[cfg(feature = "protoc")]
+    // Use vendored protobuf compiler if requested.
+    std::env::set_var("PROTOC", protobuf_src::protoc());
+
     let mut prost_build = prost_build::Config::new();
     prost_build.extern_path(".lance.encodings", "::lance_encoding::format::pb");
     prost_build.protoc_arg("--experimental_allow_proto3_optional");
diff --git a/rust/lance-encoding/Cargo.toml b/rust/lance-encoding/Cargo.toml
index e43a7c634e..27955b8340 100644
--- a/rust/lance-encoding/Cargo.toml
+++ b/rust/lance-encoding/Cargo.toml
@@ -56,10 +56,18 @@ rand_xoshiro = "0.6.0"
 
 [build-dependencies]
 prost-build.workspace = true
+protobuf-src = { version = "2.1", optional = true }
 
 [target.'cfg(target_os = "linux")'.dev-dependencies]
 pprof = { workspace = true }
 
+[features]
+protoc = ["dep:protobuf-src"]
+
+[package.metadata.docs.rs]
+# docs.rs uses an older version of Ubuntu that does not have the necessary protoc version
+features = ["protoc"]
+
 [[bench]]
 name = "decoder"
 harness = false
diff --git a/rust/lance-encoding/build.rs b/rust/lance-encoding/build.rs
index 1f030d6d7f..4c9929a978 100644
--- a/rust/lance-encoding/build.rs
+++ b/rust/lance-encoding/build.rs
@@ -6,6 +6,10 @@ use std::io::Result;
 fn main() -> Result<()> {
     println!("cargo:rerun-if-changed=protos");
 
+    #[cfg(feature = "protoc")]
+    // Use vendored protobuf compiler if requested.
+    std::env::set_var("PROTOC", protobuf_src::protoc());
+
     let mut prost_build = prost_build::Config::new();
     prost_build.protoc_arg("--experimental_allow_proto3_optional");
     prost_build.enable_type_names();
diff --git a/rust/lance-file/Cargo.toml b/rust/lance-file/Cargo.toml
index eabb950e08..17fd79801d 100644
--- a/rust/lance-file/Cargo.toml
+++ b/rust/lance-file/Cargo.toml
@@ -51,10 +51,18 @@ test-log.workspace = true
 
 [build-dependencies]
 prost-build.workspace = true
+protobuf-src = { version = "2.1", optional = true }
 
 [target.'cfg(target_os = "linux")'.dev-dependencies]
 pprof = { workspace = true }
 
+[features]
+protoc = ["dep:protobuf-src"]
+
+[package.metadata.docs.rs]
+# docs.rs uses an older version of Ubuntu that does not have the necessary protoc version
+features = ["protoc"]
+
 [[bench]]
 name = "reader"
 harness = false
diff --git a/rust/lance-file/build.rs b/rust/lance-file/build.rs
index dd004147ec..05b791fac3 100644
--- a/rust/lance-file/build.rs
+++ b/rust/lance-file/build.rs
@@ -6,6 +6,10 @@ use std::io::Result;
 fn main() -> Result<()> {
     println!("cargo:rerun-if-changed=protos");
 
+    #[cfg(feature = "protoc")]
+    // Use vendored protobuf compiler if requested.
+    std::env::set_var("PROTOC", protobuf_src::protoc());
+
     let mut prost_build = prost_build::Config::new();
     prost_build.protoc_arg("--experimental_allow_proto3_optional");
     prost_build.extern_path(".lance.encodings", "::lance_encoding::format::pb");
diff --git a/rust/lance-index/Cargo.toml b/rust/lance-index/Cargo.toml
index e6cf51d2d7..f28d900539 100644
--- a/rust/lance-index/Cargo.toml
+++ b/rust/lance-index/Cargo.toml
@@ -73,16 +73,22 @@ datafusion-sql.workspace = true
 random_word = { version = "0.4.3", features = ["en"] }
 
 [features]
+protoc = ["dep:protobuf-src"]
 tokenizer-lindera = ["lindera", "lindera-tantivy", "tokenizer-common"]
 tokenizer-jieba = ["jieba-rs", "tokenizer-common"]
 tokenizer-common = []
 
 [build-dependencies]
 prost-build.workspace = true
+protobuf-src = { version = "2.1", optional = true }
 
 [target.'cfg(target_os = "linux")'.dev-dependencies]
 pprof.workspace = true
 
+[package.metadata.docs.rs]
+# docs.rs uses an older version of Ubuntu that does not have the necessary protoc version
+features = ["protoc"]
+
 [[bench]]
 name = "find_partitions"
 harness = false
diff --git a/rust/lance-index/build.rs b/rust/lance-index/build.rs
index 8a31fbf600..402ef5012c 100644
--- a/rust/lance-index/build.rs
+++ b/rust/lance-index/build.rs
@@ -7,6 +7,10 @@ use std::io::Result;
 fn main() -> Result<()> {
     println!("cargo:rerun-if-changed=protos");
 
+    #[cfg(feature = "protoc")]
+    // Use vendored protobuf compiler if requested.
+    std::env::set_var("PROTOC", protobuf_src::protoc());
+
     let mut prost_build = prost_build::Config::new();
     prost_build.protoc_arg("--experimental_allow_proto3_optional");
     prost_build.compile_protos(&["./protos/index.proto"], &["./protos"])?;
diff --git a/rust/lance-io/Cargo.toml b/rust/lance-io/Cargo.toml
index c416f7556d..cd4c184eb7 100644
--- a/rust/lance-io/Cargo.toml
+++ b/rust/lance-io/Cargo.toml
@@ -53,9 +53,6 @@ tempfile.workspace = true
 test-log.workspace = true
 mockall.workspace = true
 
-[build-dependencies]
-prost-build.workspace = true
-
 [target.'cfg(target_os = "linux")'.dev-dependencies]
 pprof.workspace = true
 
diff --git a/rust/lance-io/src/lib.rs b/rust/lance-io/src/lib.rs
index 8e7a7694d8..b61c820f88 100644
--- a/rust/lance-io/src/lib.rs
+++ b/rust/lance-io/src/lib.rs
@@ -200,6 +200,16 @@ impl ReadBatchParams {
             )),
         }
     }
+
+    pub fn to_offsets_total(&self, total: u32) -> PrimitiveArray<UInt32Type> {
+        match self {
+            Self::Indices(indices) => indices.clone(),
+            Self::Range(r) => UInt32Array::from_iter_values(r.start as u32..r.end as u32),
+            Self::RangeFull => UInt32Array::from_iter_values(0_u32..total),
+            Self::RangeTo(r) => UInt32Array::from_iter_values(0..r.end as u32),
+            Self::RangeFrom(r) => UInt32Array::from_iter_values(r.start as u32..total),
+        }
+    }
 }
 
 #[cfg(test)]
diff --git a/rust/lance-table/Cargo.toml b/rust/lance-table/Cargo.toml
index f469676041..233073cb81 100644
--- a/rust/lance-table/Cargo.toml
+++ b/rust/lance-table/Cargo.toml
@@ -57,10 +57,16 @@ pprof = { workspace = true }
 
 [build-dependencies]
 prost-build.workspace = true
+protobuf-src = { version = "2.1", optional = true }
 
 [features]
 dynamodb = ["aws-sdk-dynamodb", "lazy_static"]
 dynamodb_tests = ["dynamodb"]
+protoc = ["dep:protobuf-src"]
+
+[package.metadata.docs.rs]
+# docs.rs uses an older version of Ubuntu that does not have the necessary protoc version
+features = ["protoc"]
 
 [[bench]]
 name = "row_id_index"
diff --git a/rust/lance-table/build.rs b/rust/lance-table/build.rs
index e0d0c15393..c4b2cc52dc 100644
--- a/rust/lance-table/build.rs
+++ b/rust/lance-table/build.rs
@@ -6,6 +6,10 @@ use std::io::Result;
 fn main() -> Result<()> {
     println!("cargo:rerun-if-changed=protos");
 
+    #[cfg(feature = "protoc")]
+    // Use vendored protobuf compiler if requested.
+    std::env::set_var("PROTOC", protobuf_src::protoc());
+
     let mut prost_build = prost_build::Config::new();
     prost_build.extern_path(".lance.file", "::lance_file::format::pb");
     prost_build.protoc_arg("--experimental_allow_proto3_optional");
diff --git a/rust/lance/Cargo.toml b/rust/lance/Cargo.toml
index d368654731..35eb75acbb 100644
--- a/rust/lance/Cargo.toml
+++ b/rust/lance/Cargo.toml
@@ -82,9 +82,6 @@ pprof.workspace = true
 # Need this so we can prevent dynamic linking in binaries (see cli feature)
 lzma-sys = { version = "0.1" }
 
-[build-dependencies]
-prost-build.workspace = true
-
 [dev-dependencies]
 lance-test-macros = { workspace = true }
 lance-datagen = { workspace = true }
@@ -111,6 +108,12 @@ tensorflow = ["tfrecord", "prost_old"]
 dynamodb = ["lance-table/dynamodb", "aws-sdk-dynamodb"]
 dynamodb_tests = ["dynamodb"]
 substrait = ["lance-datafusion/substrait"]
+protoc = [
+    "lance-encoding/protoc",
+    "lance-file/protoc",
+    "lance-index/protoc",
+    "lance-table/protoc",
+]
 
 [[bin]]
 name = "lq"
diff --git a/rust/lance/src/dataset/fragment.rs b/rust/lance/src/dataset/fragment.rs
index a3aa9af1c7..71f590498d 100644
--- a/rust/lance/src/dataset/fragment.rs
+++ b/rust/lance/src/dataset/fragment.rs
@@ -1981,12 +1981,7 @@ impl FragmentReader {
         let merged = if self.with_row_addr as usize + self.with_row_id as usize
             == self.output_schema.fields.len()
         {
-            let selected_rows = params
-                .slice(0, total_num_rows as usize)
-                .unwrap()
-                .to_offsets()
-                .unwrap()
-                .len();
+            let selected_rows = params.to_offsets_total(total_num_rows).len();
             let tasks = (0..selected_rows)
                 .step_by(batch_size as usize)
                 .map(move |offset| {
@@ -2389,6 +2384,71 @@ mod tests {
         }
     }
 
+    #[tokio::test]
+    async fn test_rowid_rowaddr_only() {
+        let test_dir = tempdir().unwrap();
+        let test_uri = test_dir.path().to_str().unwrap();
+        // Creates 400 rows in 10 fragments
+        let mut dataset = create_dataset(test_uri, LanceFileVersion::Legacy).await;
+        // Delete last 20 rows in first fragment
+        dataset.delete("i >= 20").await.unwrap();
+        // Last fragment has 20 rows but 40 addressable rows
+        let fragment = &dataset.get_fragments()[0];
+        assert_eq!(fragment.metadata.num_rows().unwrap(), 20);
+
+        // Test with take_range (all rows addressable)
+        for (with_row_id, with_row_address) in [(false, true), (true, false), (true, true)] {
+            let reader = fragment
+                .open(
+                    &fragment.schema().project::<&str>(&[]).unwrap(),
+                    FragReadConfig::default()
+                        .with_row_id(with_row_id)
+                        .with_row_address(with_row_address),
+                    None,
+                )
+                .await
+                .unwrap();
+            for valid_range in [0..40, 20..40] {
+                reader
+                    .take_range(valid_range, 100)
+                    .unwrap()
+                    .buffered(1)
+                    .try_collect::<Vec<_>>()
+                    .await
+                    .unwrap();
+            }
+            for invalid_range in [0..41, 41..42] {
+                assert!(reader.take_range(invalid_range, 100).is_err());
+            }
+        }
+
+        // Test with read_range (only non-deleted rows addressable)
+        for (with_row_id, with_row_address) in [(false, true), (true, false), (true, true)] {
+            let reader = fragment
+                .open(
+                    &fragment.schema().project::<&str>(&[]).unwrap(),
+                    FragReadConfig::default()
+                        .with_row_id(with_row_id)
+                        .with_row_address(with_row_address),
+                    None,
+                )
+                .await
+                .unwrap();
+            for valid_range in [0..20, 0..10, 10..20] {
+                reader
+                    .read_range(valid_range, 100)
+                    .unwrap()
+                    .buffered(1)
+                    .try_collect::<Vec<_>>()
+                    .await
+                    .unwrap();
+            }
+            for invalid_range in [0..21, 21..22] {
+                assert!(reader.read_range(invalid_range, 100).is_err());
+            }
+        }
+    }
+
     #[rstest]
     #[tokio::test]
     async fn test_fragment_take_range_deletions(
diff --git a/rust/lance/src/index/vector/utils.rs b/rust/lance/src/index/vector/utils.rs
index 0d8eaad5b4..34d01ec319 100644
--- a/rust/lance/src/index/vector/utils.rs
+++ b/rust/lance/src/index/vector/utils.rs
@@ -5,6 +5,7 @@ use std::sync::Arc;
 
 use arrow_array::{cast::AsArray, FixedSizeListArray};
 use lance_core::datatypes::Schema;
+use log::info;
 use snafu::{location, Location};
 use tokio::sync::Mutex;
 
@@ -86,11 +87,21 @@ pub async fn maybe_sample_training_data(
     let num_rows = dataset.count_rows(None).await?;
     let batch = if num_rows > sample_size_hint {
         let projection = dataset.schema().project(&[column])?;
-        dataset.sample(sample_size_hint, &projection).await?
+        let batch = dataset.sample(sample_size_hint, &projection).await?;
+        info!(
+            "Sample training data: retrieved {} rows by sampling",
+            batch.num_rows()
+        );
+        batch
     } else {
         let mut scanner = dataset.scan();
         scanner.project(&[column])?;
-        scanner.try_into_batch().await?
+        let batch = scanner.try_into_batch().await?;
+        info!(
+            "Sample training data: retrieved {} rows scanning full datasets",
+            batch.num_rows()
+        );
+        batch
     };
 
     let array = batch.column_by_name(column).ok_or(Error::Index {