convert TPC-H to use Utf8View

Author: a10y

Cargo.lock

@@ -35,15 +35,15 @@ rust-version = "1.76"
 ahash = "0.8.11"
 allocator-api2 = "0.2.16"
 arrayref = "0.3.7"
-arrow = { version = "52.0.0", features = ["pyarrow"] }
-arrow-array = "52.0.0"
-arrow-buffer = "52.0.0"
-arrow-cast = "52.0.0"
-arrow-csv = "52.0.0"
-arrow-data = "52.0.0"
-arrow-ipc = "52.0.0"
-arrow-schema = "52.0.0"
-arrow-select = "52.0.0"
+arrow = { version = "52.1.0", features = ["pyarrow"] }
+arrow-array = "52.1.0"
+arrow-buffer = "52.1.0"
+arrow-cast = "52.1.0"
+arrow-csv = "52.1.0"
+arrow-data = "52.1.0"
+arrow-ipc = "52.1.0"
+arrow-schema = "52.1.0"
+arrow-select = "52.1.0"
 async-trait = "0.1"
 bindgen = "0.69.4"
 bytes = "1.6.0"
@@ -53,12 +53,14 @@ criterion = { version = "0.5.1", features = ["html_reports"] }
 croaring = "2.0.0"
 csv = "1.3.0"
 object_store = "0.10.1"
-datafusion = "40.0.0"
-datafusion-common = "40.0.0"
-datafusion-execution = "40.0.0"
-datafusion-expr = "40.0.0"
-datafusion-physical-expr = "40.0.0"
-datafusion-physical-plan = "40.0.0"
+
+datafusion = { path = "/Volumes/Code/datafusion/datafusion/core" }
+datafusion-common = { path = "/Volumes/Code/datafusion/datafusion/common" }
+datafusion-execution = { path = "/Volumes/Code/datafusion/datafusion/execution" }
+datafusion-expr = { path = "/Volumes/Code/datafusion/datafusion/expr" }
+datafusion-physical-expr = { path = "/Volumes/Code/datafusion/datafusion/physical-expr" }
+datafusion-physical-plan = { path = "/Volumes/Code/datafusion/datafusion/physical-plan" }
+
 divan = "0.1.14"
 duckdb = "1.0.0"
 enum-iterator = "2.0.0"
@@ -115,3 +117,18 @@ warnings = "deny"
 [workspace.lints.clippy]
 all = { level = "deny", priority = -1 }
 or_fun_call = "deny"
+
+[patch.crates-io]
+arrow = { path = "/Volumes/Code/arrow-rs/arrow" }
+arrow-array = { path = "/Volumes/Code/arrow-rs/arrow-array" }
+arrow-buffer = { path = "/Volumes/Code/arrow-rs/arrow-buffer" }
+arrow-cast = { path = "/Volumes/Code/arrow-rs/arrow-cast" }
+arrow-csv = { path = "/Volumes/Code/arrow-rs/arrow-csv" }
+arrow-data = { path = "/Volumes/Code/arrow-rs/arrow-data" }
+arrow-ipc = { path = "/Volumes/Code/arrow-rs/arrow-ipc" }
+arrow-schema = { path = "/Volumes/Code/arrow-rs/arrow-schema" }
+arrow-select = { path = "/Volumes/Code/arrow-rs/arrow-select" }
+arrow-string = { path = "/Volumes/Code/arrow-rs/arrow-string" }
+arrow-ord = { path = "/Volumes/Code/arrow-rs/arrow-ord" }
+arrow-flight = { path = "/Volumes/Code/arrow-rs/arrow-flight" }
+parquet = { path = "/Volumes/Code/arrow-rs/parquet" }

Cargo.toml

@@ -19,13 +19,13 @@ async fn main() {
     // The formats to run against (vs the baseline)
     let formats = [
         Format::Arrow,
-        Format::Parquet,
-        Format::Vortex {
-            disable_pushdown: false,
-        },
-        Format::Vortex {
-            disable_pushdown: true,
-        },
+        // Format::Parquet,
+        // Format::Vortex {
+        //     disable_pushdown: false,
+        // },
+        // Format::Vortex {
+        //     disable_pushdown: true,
+        // },
     ];
 
     // Load datasets
@@ -53,8 +53,8 @@ async fn main() {
     // Send back a channel with the results of Row.
     let (rows_tx, rows_rx) = sync::mpsc::channel();
     for i in 1..=22 {
-        if i == 15 {
-            continue;
+        if vec![2, 7, 8, 9, 10, 15, 18, 21].contains(&i) {
+            continue
         }
         let _ctxs = ctxs.clone();
         let _tx = rows_tx.clone();
@@ -123,6 +123,7 @@ async fn main() {
                     .push(Cell::new(&format!("{} us", measure.as_micros())).style_spec(style_spec));
             }
 
+            println!("QUERY {i} complete");
             _tx.send((i, Row::new(cells))).unwrap();
         });
     }