Add feature flags for metadata providers (#23)

Author: shefeek-jinnah

Cargo.toml

@@ -21,12 +21,12 @@ object_store = { version = "0.12.4", features = ["aws"] }
 url = "2.5"
 async-trait = "0.1"
 tokio = { version = "1", features = ["full"] }
-duckdb = { version = "1.4.1", features = ["bundled"] }
 futures = "0.3"
 tracing = "0.1"
 thiserror = "2.0"
 
-# Multi-database metadata providers (optional)
+# Metadata providers (optional)
+duckdb = { version = "1.4.1", features = ["bundled"], optional = true }
 sqlx = { version = "0.8", features = ["runtime-tokio"], optional = true }
 
 [dev-dependencies]
@@ -39,10 +39,13 @@ aws-credential-types = "1.2"
 sqllogictest = "0.23"
 
 [features]
+default = ["metadata-duckdb"]
+
 # Allow skipping tests that use docker (in CI, macos doesn't support docker).
 skip-tests-with-docker = []
 
 # Metadata provider backends
-metadata-postgres = ["sqlx", "sqlx/postgres", "sqlx/chrono"]
+metadata-duckdb = ["dep:duckdb"]
+metadata-postgres = ["dep:sqlx", "sqlx/postgres", "sqlx/chrono"]
 # Future: metadata-sqlite = ["sqlx", "sqlx/sqlite", "sqlx/chrono"]
 # Future: metadata-mysql = ["sqlx", "sqlx/mysql", "sqlx/chrono"]

examples/basic_query.rs

@@ -32,11 +32,11 @@
 
 use datafusion::execution::runtime_env::RuntimeEnv;
 use datafusion::prelude::*;
+#[cfg(feature = "metadata-duckdb")]
+use datafusion_ducklake::DuckdbMetadataProvider;
 #[cfg(feature = "metadata-postgres")]
 use datafusion_ducklake::PostgresMetadataProvider;
-use datafusion_ducklake::{
-    DuckLakeCatalog, DuckdbMetadataProvider, MetadataProvider, register_ducklake_functions,
-};
+use datafusion_ducklake::{DuckLakeCatalog, MetadataProvider, register_ducklake_functions};
 use object_store::ObjectStore;
 use object_store::aws::AmazonS3Builder;
 use std::env;
@@ -78,11 +78,20 @@ async fn main() -> Result<(), Box<dyn std::error::Error>> {
             run_query(provider, snapshot_id, sql).await?;
         }
     } else {
-        println!("Connecting to DuckDB catalog: {}", catalog_source);
-        let provider = Arc::new(DuckdbMetadataProvider::new(catalog_source)?);
-        let snapshot_id = provider.get_current_snapshot()?;
-        println!("Current snapshot ID: {}", snapshot_id);
-        run_query(provider, snapshot_id, sql).await?;
+        #[cfg(feature = "metadata-duckdb")]
+        {
+            println!("Connecting to DuckDB catalog: {}", catalog_source);
+            let provider = Arc::new(DuckdbMetadataProvider::new(catalog_source)?);
+            let snapshot_id = provider.get_current_snapshot()?;
+            println!("Current snapshot ID: {}", snapshot_id);
+            run_query(provider, snapshot_id, sql).await?;
+        }
+
+        #[cfg(not(feature = "metadata-duckdb"))]
+        {
+            eprintln!("Error: DuckDB catalog requires the 'metadata-duckdb' feature");
+            exit(1);
+        }
     }
 
     Ok(())

src/error.rs

@@ -14,6 +14,7 @@ pub enum DuckLakeError {
     Arrow(#[from] arrow::error::ArrowError),
 
     /// DuckDB error
+    #[cfg(feature = "metadata-duckdb")]
     #[error("DuckDB error: {0}")]
     DuckDb(#[from] duckdb::Error),
 

src/lib.rs

@@ -40,14 +40,15 @@ pub mod delete_filter;
 pub mod error;
 pub mod information_schema;
 pub mod metadata_provider;
-pub mod metadata_provider_duckdb;
 pub mod path_resolver;
 pub mod schema;
 pub mod table;
 pub mod table_functions;
 pub mod types;
 
-// Multi-database metadata providers (optional, feature-gated)
+// Metadata providers (feature-gated)
+#[cfg(feature = "metadata-duckdb")]
+pub mod metadata_provider_duckdb;
 #[cfg(feature = "metadata-postgres")]
 pub mod metadata_provider_postgres;
 
@@ -58,11 +59,12 @@ pub type Result<T> = std::result::Result<T, DuckLakeError>;
 pub use catalog::DuckLakeCatalog;
 pub use error::DuckLakeError;
 pub use metadata_provider::MetadataProvider;
-pub use metadata_provider_duckdb::DuckdbMetadataProvider;
 pub use schema::DuckLakeSchema;
 pub use table::DuckLakeTable;
 pub use table_functions::register_ducklake_functions;
 
-// Re-export multi-database metadata providers (feature-gated)
+// Re-export metadata providers (feature-gated)
+#[cfg(feature = "metadata-duckdb")]
+pub use metadata_provider_duckdb::DuckdbMetadataProvider;
 #[cfg(feature = "metadata-postgres")]
 pub use metadata_provider_postgres::PostgresMetadataProvider;

tests/common/mod.rs

@@ -1,3 +1,4 @@
+#![cfg(feature = "metadata-duckdb")]
 //! Common test utilities for integration tests
 //!
 //! This module provides helper functions to generate DuckLake catalogs

tests/concurrent_tests.rs

@@ -1,3 +1,4 @@
+#![cfg(feature = "metadata-duckdb")]
 //! Concurrent query safety tests
 //!
 //! This test suite verifies that DataFusion-DuckLake is thread-safe and can handle

tests/delete_filter_tests.rs

@@ -1,3 +1,4 @@
+#![cfg(feature = "metadata-duckdb")]
 //! Integration tests for delete file filtering
 //!
 //! These tests verify that the delete file implementation correctly filters out

tests/hybrid_asyncdb.rs

@@ -1,3 +1,4 @@
+#![cfg(feature = "metadata-duckdb")]
 //! Hybrid AsyncDB adapter for running DuckDB DuckLake tests
 //!
 //! This adapter uses a hybrid approach:

tests/information_schema_test.rs

@@ -1,3 +1,4 @@
+#![cfg(feature = "metadata-duckdb")]
 //! Integration tests for information_schema virtual tables and table functions
 
 use datafusion::prelude::*;

tests/object_store_integration_test.rs

@@ -1,3 +1,4 @@
+#![cfg(feature = "metadata-duckdb")]
 //! Integration test for object store support (S3/MinIO)
 //!
 //! This test verifies that DataFusion-DuckLake works correctly with object stores