zhuqi-lucas commented on PR #22518:
URL: https://github.com/apache/datafusion/pull/22518#issuecomment-5047341591
Force-pushed a full rebase onto current main (was 483 commits behind after
the #22710 aggregation-stream migration).
The three commits map onto the new structure:
1. **`e4d1cba17` config knobs** — `skip_partial_aggregation_use_cost_model`
+ `skip_partial_aggregation_ab_sampling_rows` + configs.md regen.
2. **`c01757dfe` SkipAggregationProbe A/B state machine** — extends the
probe (now in the standalone `skip_partial.rs` since #22729) with `Partial →
AbSampling → Locked { should_skip }`. Four lazily-registered diagnostic gauges.
Callers in `grouped_hash_stream.rs` + `hash_stream.rs` updated.
3. **`f1182e841` state-machine wiring** — new `ExecutionState::AbSampling` +
transitions.
**Status: draft-ish.** 123/123 existing `aggregates::*` tests pass locally
on this branch. Still TODO before this is ready for another round of review:
- [ ] Migrate the seven A/B-specific unit tests from the original commit
series onto the new `skip_partial.rs` module.
- [ ] SLT snapshots for `push_down_filter_regression.slt` /
`information_schema.slt` regen for the new gauges + config rows.
- [ ] Rerun ClickBench on current main to confirm the Q18 1.12× / Q19 1.43×
wins from the original benchmark still hold.
cc @2010YOUY01 @alamb — reviving this after the row_hash.rs →
grouped_hash_stream.rs migration you flagged has settled.
--
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]
