Remove Arrow engine

Signed-off-by: Adam Gutglick <adam@spiraldb.com>

Author: AdamGS

benchmarks/random-access-bench/src/main.rs

@@ -269,7 +269,7 @@ async fn benchmark_random_access(
     let timing = TimingMeasurement {
         name: measurement_name.to_string(),
         storage: storage.to_string(),
-        target: Target::new(format_to_engine(format), format),
+        target: Target::new(Engine::default(), format),
         runs,
     };
     Ok(RandomAccessRun {
@@ -326,17 +326,6 @@ fn push_v3_random_access_record(records: &mut Vec<v3::V3Record>, run: &RandomAcc
     records.push(v3::random_access_record(&run.timing, &dataset));
 }
 
-/// Map format to the appropriate engine for random access benchmarks.
-fn format_to_engine(format: Format) -> Engine {
-    match format {
-        Format::OnDiskVortex | Format::VortexCompact => Engine::Vortex,
-        Format::Parquet => Engine::Arrow,
-        #[cfg(feature = "lance")]
-        Format::Lance => Engine::Arrow, // Is this right here?
-        _ => Engine::default(),
-    }
-}
-
 /// Open a random accessor for any supported format.
 ///
 /// For Vortex and Parquet, the path comes from [`BenchDataset::path`].
@@ -528,7 +517,7 @@ mod tests {
         RandomAccessRun {
             timing: TimingMeasurement {
                 name: format!("random-access/{dataset}/parquet-tokio-local-disk"),
-                target: Target::new(Engine::Arrow, Format::Parquet),
+                target: Target::new(Engine::Vortex, Format::Parquet),
                 storage: STORAGE_NVME.to_string(),
                 runs: vec![Duration::from_nanos(10)],
             },

benchmarks/random-access-bench/src/render.rs

@@ -198,7 +198,7 @@ mod tests {
         RandomAccessRun {
             timing: TimingMeasurement {
                 name: format!("random-access/{dataset}/{}-tokio-local-disk", format.ext()),
-                target: Target::new(Engine::Arrow, format),
+                target: Target::new(Engine::Vortex, format),
                 storage: "nvme".to_string(),
                 runs: vec![Duration::from_micros(micros)],
             },
@@ -273,10 +273,6 @@ mod tests {
             rendered.contains("vortex-cached") && rendered.contains("vortex-reopen"),
             "expected ext-based column headers, got:\n{rendered}"
         );
-        assert!(
-            !rendered.contains("arrow"),
-            "expected no engine header row, got:\n{rendered}"
-        );
         assert!(
             rendered.contains("random-access/taxi")
                 && rendered.contains("random-access/taxi/uniform"),

vortex-bench/src/lib.rs

@@ -214,7 +214,6 @@ impl Format {
 pub enum Engine {
     #[default]
     Vortex,
-    Arrow,
     #[clap(name = "datafusion")]
     #[serde(rename = "datafusion")]
     DataFusion,
@@ -229,7 +228,6 @@ impl Display for Engine {
             Engine::DataFusion => write!(f, "datafusion"),
             Engine::DuckDB => write!(f, "duckdb"),
             Engine::Vortex => write!(f, "vortex"),
-            Engine::Arrow => write!(f, "arrow"),
         }
     }
 }

vortex-bench/src/measurements.rs

@@ -352,8 +352,8 @@ impl ToJson for CompressionTimingMeasurement {
     fn to_json(&self) -> serde_json::Value {
         let (name, engine) = match self.format {
             Format::OnDiskVortex => (self.name.to_string(), Engine::Vortex),
-            Format::Parquet => (format!("parquet_rs-zstd {}", self.name), Engine::Arrow),
-            Format::Lance => (format!("lance {}", self.name), Engine::Arrow),
+            Format::Parquet => (format!("parquet_rs-zstd {}", self.name), Engine::Vortex),
+            Format::Lance => (format!("lance {}", self.name), Engine::Vortex),
             _ => vortex_panic!(
                 "CompressionTimingMeasurement only supports vortex, lance, and parquet formats"
             ),
@@ -397,8 +397,8 @@ impl ToJson for CustomUnitMeasurement {
     fn to_json(&self) -> serde_json::Value {
         let engine = match self.format {
             Format::OnDiskVortex | Format::VortexCompact => Engine::Vortex,
-            Format::Parquet => Engine::Arrow,
-            Format::Lance => Engine::Arrow,
+            Format::Parquet => Engine::Vortex,
+            Format::Lance => Engine::Vortex,
             _ => Engine::Vortex, // Default to Vortex for other formats.
         };
 

vortex-bench/src/v3.rs

@@ -141,7 +141,7 @@ pub struct QueryMeasurementRecord {
     pub query_idx: u32,
     /// Storage backend the run targeted (`nvme` or `s3`).
     pub storage: String,
-    /// Query engine (`datafusion`, `duckdb`, `vortex`, `arrow`).
+    /// Query engine (`datafusion`, `duckdb`, `vortex`).
     pub engine: String,
     /// On-disk format (`parquet`, `vortex-file-compressed`, `lance`, ...).
     pub format: String,
@@ -519,7 +519,6 @@ fn duration_as_ns(d: std::time::Duration) -> u64 {
 fn engine_label(engine: Engine) -> &'static str {
     match engine {
         Engine::Vortex => "vortex",
-        Engine::Arrow => "arrow",
         Engine::DataFusion => "datafusion",
         Engine::DuckDB => "duckdb",
     }
@@ -663,7 +662,7 @@ mod tests {
     fn snapshot_random_access_time() -> anyhow::Result<()> {
         let timing = TimingMeasurement {
             name: "random-access/taxi/uniform/parquet-tokio-local-disk".to_string(),
-            target: Target::new(Engine::Arrow, Format::Parquet),
+            target: Target::new(Engine::Vortex, Format::Parquet),
             storage: "nvme".to_string(),
             runs: vec![
                 Duration::from_nanos(800_000),