Horton

Re: Eager aggregation, take 3

Tender Wang <tndrwang@gmail.com>

View thread View on postgr.es
From: Tender Wang <tndrwang@gmail.com>
To: Richard Guo <guofenglinux@gmail.com>
Cc: Radim Marek <radim@boringsql.com>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2026-06-01T07:57:39Z
Lists: pgsql-hackers 
Richard Guo <guofenglinux@gmail.com> 于2026年6月1日周一 15:19写道：
>
> Thanks for the report.  This is a bug.  We should never push a partial
> aggregation down to a relation on the inner (RHS) side of a semi/anti
> join.  A semi/anti join does not preserve its inner rows in the join
> output, so a partial aggregate computed on the inner side would not
> survive the join and could not be combined by the final aggregation.
>
> > I haven't thought about it too deeply yet. Maybe we can do something
> > in the make_grouped_join_rel().
> > ...
> > if (sjinfo->jointype == JOIN_ANTI || sjinfo->jointype == JOIN_SEMI)
> >     return;
> > ...
>
> That does fix the reported case, but I think it's too broad: it also
> disables pushing a partial aggregate to the outer side of a semi/anti
> join, which is valid.  And by the time we reach make_grouped_join_rel
> the grouped relation for the inner-side relation has already been
> built, so it would just go unused.

Yes, checking only the jointype and concluding that partial agg is not allowed
will cause us to miss some optimization opportunities for the outer
side of the semi or anti-join.


> So I'd rather fix it in eager_aggregation_possible_for_relation, right
> next to the existing outer-join check, by rejecting a relation that
> lies on the inner side of a semijoin/antijoin.  See attached.

The attached LGTM.



-- 
Thanks,
Tender Wang

Commits

  1. Fix eager aggregation for semi/antijoin inner rels

  2. Cover additional errors and corner conditions in repack.c

  3. Fix volatile function evaluation in eager aggregation

  4. Fix collation handling for grouping keys in eager aggregation

  5. Rename apply_at to apply_agg_at for clarity

  6. Fix comment in eager_aggregate.sql

  7. Remove unnecessary include of "utils/fmgroids.h"

  8. Implement Eager Aggregation

  9. Allow negative aggtransspace to indicate unbounded state size

  10. Add macros for looping through a List without a ListCell.

  11. Account for the effect of lossy pages when costing bitmap scans.

  12. Fix a thinko in join_is_legal: when we decide we can implement a semijoin