fix(ci): fix ci by install protobuf-compiler

.github/workflows/ci.yml

@@ -38,6 +38,8 @@ jobs:
         with:
           components: clippy
       - uses: Swatinem/rust-cache@v2
+      - name: Install dependencies
+        run: sudo apt-get update && sudo apt-get install -y protobuf-compiler
       - name: Clippy
         run: cargo clippy --all-targets -- -D warnings
 
@@ -52,6 +54,8 @@ jobs:
       - uses: actions/checkout@v4
       - uses: actions-rust-lang/setup-rust-toolchain@v1
       - uses: Swatinem/rust-cache@v2
+      - name: Install dependencies
+        run: sudo apt-get update && sudo apt-get install -y protobuf-compiler
       - name: Run tests
         run: cargo test
       - name: Run C/C++ compilation tests
@@ -65,6 +69,8 @@ jobs:
       - uses: actions/checkout@v4
       - uses: actions-rust-lang/setup-rust-toolchain@v1
       - uses: Swatinem/rust-cache@v2
+      - name: Install dependencies
+        run: brew install protobuf
       - name: Run tests
         run: cargo test
       - name: Run C/C++ compilation tests
@@ -78,6 +84,8 @@ jobs:
       - uses: actions/checkout@v4
       - uses: actions-rust-lang/setup-rust-toolchain@v1
       - uses: Swatinem/rust-cache@v2
+      - name: Install dependencies
+        run: sudo apt-get update && sudo apt-get install -y protobuf-compiler
       - name: Check documentation
         run: RUSTDOCFLAGS="-D warnings" cargo doc --no-deps
 
@@ -91,6 +99,8 @@ jobs:
         with:
           toolchain: "1.91.0"
       - uses: Swatinem/rust-cache@v2
+      - name: Install dependencies
+        run: sudo apt-get update && sudo apt-get install -y protobuf-compiler
       - name: Check MSRV
         run: cargo check --all-targets
 

AGENTS.md

@@ -0,0 +1,29 @@
+# AGENTS.md
+
+## Structure
+Rust FFI source: `src/`
+C header (stable ABI): `include/lance.h`
+C++ RAII wrappers (header-only): `include/lance.hpp`
+Tests (Rust): `tests/c_api_test.rs`
+Tests (C/C++): `tests/cpp/`
+Historical test data: `test_data/`
+
+## Commands
+check: `cargo check --all-targets`
+format: `cargo fmt`
+lint: `cargo clippy --all-targets -- -D warnings`
+test: `cargo test`
+test C/C++ compilation: `cargo test --test compile_and_run_test -- --ignored`
+
+## Key Patterns
+- Opaque handles with `lance_*_open`/`lance_*_close` lifecycle.
+- Thread-local error handling via `ffi_try!` macro.
+- Arrow C Data Interface for zero-copy data exchange.
+- `panic = "abort"` in release to prevent unwinding across FFI.
+
+## Adding New APIs
+1. Add `extern "C"` function in `src/`.
+2. Add declaration to `include/lance.h`.
+3. Add C++ wrapper to `include/lance.hpp`.
+4. Add test in `tests/c_api_test.rs`.
+

CLAUDE.md

@@ -0,0 +1 @@
+AGENTS.md

src/async_dispatcher.rs

@@ -8,7 +8,7 @@
 //! sequentially, avoiding reentrancy and Tokio thread blocking.
 
 use std::ffi::c_void;
-use std::sync::{mpsc, LazyLock};
+use std::sync::{LazyLock, mpsc};
 
 /// C callback function pointer type for async operations.
 /// - `ctx`: opaque pointer passed back to the caller

src/dataset.rs

@@ -9,11 +9,11 @@ use std::sync::Arc;
 use arrow::ffi::FFI_ArrowSchema;
 use arrow::ffi_stream::FFI_ArrowArrayStream;
 use arrow_schema::Schema as ArrowSchema;
-use lance::dataset::builder::DatasetBuilder;
 use lance::Dataset;
+use lance::dataset::builder::DatasetBuilder;
 use lance_core::Result;
 
-use crate::error::{ffi_try, set_last_error, LanceErrorCode};
+use crate::error::{LanceErrorCode, ffi_try, set_last_error};
 use crate::helpers;
 use crate::runtime::block_on;
 

src/error.rs

@@ -7,7 +7,7 @@
 //! the caller retrieves the error code and message from thread-local storage.
 
 use std::cell::RefCell;
-use std::ffi::{c_char, CString};
+use std::ffi::{CString, c_char};
 use std::ptr;
 
 /// Error codes returned by `lance_last_error_code()`.

src/helpers.rs

@@ -4,7 +4,7 @@
 //! C string parsing utilities.
 
 use std::collections::HashMap;
-use std::ffi::{c_char, CStr};
+use std::ffi::{CStr, c_char};
 
 use lance_core::{Error, Result};
 use snafu::location;

src/lib.rs

@@ -27,6 +27,6 @@ mod scanner;
 pub use batch::*;
 pub use dataset::*;
 pub use error::{
-    lance_free_string, lance_last_error_code, lance_last_error_message, LanceErrorCode,
+    LanceErrorCode, lance_free_string, lance_last_error_code, lance_last_error_message,
 };
 pub use scanner::*;

src/scanner.rs

@@ -12,18 +12,18 @@ use std::task::{Context, Poll, RawWaker, RawWakerVTable, Waker};
 use arrow::ffi_stream::FFI_ArrowArrayStream;
 use arrow_schema::SchemaRef;
 use futures::{Stream, StreamExt};
-use lance::dataset::scanner::DatasetRecordBatchStream;
 use lance::Dataset;
+use lance::dataset::scanner::DatasetRecordBatchStream;
 use lance_core::Result;
 use lance_io::ffi::to_ffi_arrow_array_stream;
 use lance_io::stream::RecordBatchStream;
 
 use crate::async_dispatcher::{self, LanceCallback};
 use crate::batch::LanceBatch;
 use crate::dataset::LanceDataset;
-use crate::error::{clear_last_error, ffi_try, set_lance_error, set_last_error, LanceErrorCode};
+use crate::error::{LanceErrorCode, clear_last_error, ffi_try, set_lance_error, set_last_error};
 use crate::helpers;
-use crate::runtime::{block_on, RT};
+use crate::runtime::{RT, block_on};
 
 /// Opaque scanner handle. Stores configuration until stream materialization.
 pub struct LanceScanner {
@@ -294,12 +294,12 @@ pub unsafe extern "C" fn lance_scanner_next(
     let s = unsafe { &mut *scanner };
 
     // Lazily materialize the stream on first call.
-    if s.stream.is_none() {
-        if let Err(err) = s.materialize_stream() {
-            set_lance_error(&err);
-            unsafe { *out = ptr::null_mut() };
-            return -1;
-        }
+    if s.stream.is_none()
+        && let Err(err) = s.materialize_stream()
+    {
+        set_lance_error(&err);
+        unsafe { *out = ptr::null_mut() };
+        return -1;
     }
 
     let stream = s.stream.as_mut().unwrap();
@@ -431,12 +431,12 @@ pub unsafe extern "C" fn lance_scanner_poll_next(
     let s = unsafe { &mut *scanner };
 
     // Lazily materialize the stream.
-    if s.stream.is_none() {
-        if let Err(err) = s.materialize_stream() {
-            set_lance_error(&err);
-            unsafe { *out = ptr::null_mut() };
-            return LancePollStatus::Error;
-        }
+    if s.stream.is_none()
+        && let Err(err) = s.materialize_stream()
+    {
+        set_lance_error(&err);
+        unsafe { *out = ptr::null_mut() };
+        return LancePollStatus::Error;
     }
 
     let stream = s.stream.as_mut().unwrap();

tests/c_api_test.rs

@@ -10,8 +10,8 @@ use std::ffi::CString;
 use std::ptr;
 use std::sync::Arc;
 
-use arrow::ffi::from_ffi;
 use arrow::ffi::FFI_ArrowSchema;
+use arrow::ffi::from_ffi;
 use arrow::ffi_stream::ArrowArrayStreamReader;
 use arrow::ffi_stream::FFI_ArrowArrayStream;
 use arrow::record_batch::RecordBatchReader;