Make ballista support an optional feature to datafusion-cli

.github/workflows/rust.yml

@@ -70,12 +70,20 @@ jobs:
         env:
           CARGO_HOME: "/github/home/.cargo"
           CARGO_TARGET_DIR: "/github/home/target/release"
-      - name: Check DataFusion Build without default features
+      - name: Check DataFusion builds without default features
         run: |
           cargo check --no-default-features -p datafusion
         env:
           CARGO_HOME: "/github/home/.cargo"
           CARGO_TARGET_DIR: "/github/home/target"
+      - name: Check DataFusion CLI builds with ballista
+        run: |
+          cargo check -p datafusion-cli
+          cargo check --no-default-features -p datafusion-cli
+          cargo check --features=ballista -p datafusion-cli
+        env:
+          CARGO_HOME: "/github/home/.cargo"
+          CARGO_TARGET_DIR: "/github/home/target"
 
   # test the crate
   linux-test:

datafusion-cli/Cargo.toml

@@ -33,5 +33,5 @@ rustyline = "9.0"
 tokio = { version = "1.0", features = ["macros", "rt", "rt-multi-thread", "sync", "parking_lot"] }
 datafusion = { path = "../datafusion", version = "7.0.0" }
 arrow = { version = "9.0.0" }
-ballista = { path = "../ballista/rust/client", version = "0.6.0" }
+ballista = { path = "../ballista/rust/client", version = "0.6.0", optional=true }
 env_logger = "0.9"

datafusion-cli/src/context.rs

@@ -17,8 +17,6 @@
 
 //! Context (remote or local)
 
-use ballista::context::BallistaContext;
-use ballista::prelude::BallistaConfig;
 use datafusion::dataframe::DataFrame;
 use datafusion::error::{DataFusionError, Result};
 use datafusion::execution::context::{ExecutionConfig, ExecutionContext};
@@ -28,18 +26,14 @@ use std::sync::Arc;
 pub enum Context {
     /// In-process execution with DataFusion
     Local(ExecutionContext),
-    /// Distributed execution with Ballista
+    /// Distributed execution with Ballista (if available)
     Remote(BallistaContext),
 }
 
 impl Context {
     /// create a new remote context with given host and port
     pub fn new_remote(host: &str, port: u16) -> Result<Context> {
-        let config: BallistaConfig = BallistaConfig::new()
-            .map_err(|e| DataFusionError::Execution(format!("{:?}", e)))?;
-        Ok(Context::Remote(BallistaContext::remote(
-            host, port, &config,
-        )))
+        Ok(Context::Remote(BallistaContext::try_new(host, port)?))
     }
 
     /// create a local context using the given config
@@ -55,3 +49,36 @@ impl Context {
         }
     }
 }
+
+// implement wrappers around the BallistaContext to support running without ballista
+
+#[cfg(feature = "ballista")]
+pub struct BallistaContext(ballista::context::BallistaContext);
+#[cfg(feature = "ballista")]
+impl BallistaContext {
+    pub fn try_new(host: &str, port: u16) -> Result<Self> {
+        use ballista::context::BallistaContext;
+        use ballista::prelude::BallistaConfig;
+        let config: BallistaConfig = BallistaConfig::new()
+            .map_err(|e| DataFusionError::Execution(format!("{:?}", e)))?;
+        Ok(Self(BallistaContext::remote(host, port, &config)))
+    }
+    pub async fn sql(&mut self, sql: &str) -> Result<Arc<dyn DataFrame>> {
+        self.0.sql(sql).await
+    }
+}
+
+#[cfg(not(feature = "ballista"))]
+pub struct BallistaContext();
+#[cfg(not(feature = "ballista"))]
+impl BallistaContext {
+    pub fn try_new(_host: &str, _port: u16) -> Result<Self> {
+        Err(DataFusionError::NotImplemented(
+            "Remote execution not supported. Compile with feature 'ballista' to enable"
+                .to_string(),
+        ))
+    }
+    pub async fn sql(&mut self, _sql: &str) -> Result<Arc<dyn DataFrame>> {
+        unreachable!()
+    }
+}