Dandandan commented on code in PR #20664:
URL: https://github.com/apache/datafusion/pull/20664#discussion_r2892038314
##########
datafusion/core/benches/sql_planner_extended.rs:
##########
@@ -212,21 +213,143 @@ fn build_test_data_frame(ctx: &SessionContext, rt:
&Runtime) -> DataFrame {
})
}
-fn criterion_benchmark(c: &mut Criterion) {
+/// Build a CASE-heavy dataframe over a non-inner join to stress
+/// planner-time filter pushdown and nullability/type inference.
+fn build_case_heavy_left_join_df(ctx: &SessionContext, rt: &Runtime) ->
DataFrame {
+ register_string_table(ctx, 100, 1000);
+ let query = build_case_heavy_left_join_query(30, 1);
+ rt.block_on(async { ctx.sql(&query).await.unwrap() })
+}
+
+fn build_case_heavy_left_join_query(predicate_count: usize, case_depth: usize)
-> String {
+ let mut query = String::from(
+ "SELECT l.c0, r.c0 AS rc0 FROM t l LEFT JOIN t r ON l.c0 = r.c0 WHERE
",
+ );
+
+ if predicate_count == 0 {
+ query.push_str("TRUE");
+ return query;
+ }
+
+ // Keep this deterministic so comparisons between profiles are stable.
+ for i in 0..predicate_count {
+ if i > 0 {
+ query.push_str(" AND ");
+ }
+
+ let mut expr = format!("length(l.c{})", i % 20);
Review Comment:
So the TPC-H/TPC-DS one is already a good one to optimize for.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
