SubhamSinghal commented on code in PR #24583:
URL: https://github.com/apache/datafusion/pull/24583#discussion_r3840907167
##########
benchmarks/sql_benchmarks/h2o/benchmarks/window/q14.benchmark:
##########
@@ -0,0 +1,26 @@
+subgroup window
+
+name Q14
+group h2o
+
+echo Loading ${H2O_BENCH_SIZE:-small} window ${H2O_FILE_TYPE:-csv} h2o data
+
+load
sql_benchmarks/h2o/init/load_window_${H2O_BENCH_SIZE:-small}_${H2O_FILE_TYPE:-csv}.sql
+
+assert I
+SELECT COUNT(*) > 0 FROM x
+----
+true
+
+run
+-- Window Top-N partition cardinality sweep (id3 % N gives N distinct
partitions).
+-- These exercise PartitionedTopKExec across cardinalities to validate it stays
Review Comment:
@kumarUjjawal is right, `enable_window_topn` defaults to `false` and the
rule hard-exits on it before anything else:
```rust
// physical-optimizer/src/window_topn.rs
if !config.optimizer.enable_window_topn {
return Ok(plan);
}
```
q13–q29 have been timing the SortExec + FilterExec baseline all along.
Just adding SET ... = true isn't right either: the comment's whole claim is
that these stay competitive with the baseline, so the sweep needs both arms,
not a flip to the other one.
--
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]
