From 344f120e268a608fdb16994ab90512ac21adac0f Mon Sep 17 00:00:00 2001 From: Andrew Lamb Date: Thu, 15 Jan 2026 13:31:50 -0500 Subject: [PATCH] Use correct setting for click bench queries in sql_planner benchmark --- datafusion/core/benches/sql_planner.rs | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/datafusion/core/benches/sql_planner.rs b/datafusion/core/benches/sql_planner.rs index 7cce7e0bd7db7..ea4e3154326f3 100644 --- a/datafusion/core/benches/sql_planner.rs +++ b/datafusion/core/benches/sql_planner.rs @@ -118,6 +118,11 @@ fn register_clickbench_hits_table(rt: &Runtime) -> SessionContext { let sql = format!("CREATE EXTERNAL TABLE hits STORED AS PARQUET LOCATION '{path}'"); + // ClickBench partitioned dataset was written by an ancient version of pyarrow that + // that wrote strings with the wrong logical type. To read it correctly, we must + // automatically convert binary to string. + rt.block_on(ctx.sql("SET datafusion.execution.parquet.binary_as_string = true;")) + .unwrap(); rt.block_on(ctx.sql(&sql)).unwrap(); let count =