| From: | William Bernbaum <wbernbaum(at)dwdev(dot)com> |
|---|---|
| To: | "pgsql-hackers(at)lists(dot)postgresql(dot)org" <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Cc: | "guofenglinux(at)gmail(dot)com" <guofenglinux(at)gmail(dot)com> |
| Subject: | RE: Plan an inner join as a semijoin under eager deduplication |
| Date: | 2026-08-31 01:05:50 |
| Message-ID: | PH0PR18MB44434B2841710BA0CA162BF4A6A92@PH0PR18MB4443.namprd18.prod.outlook.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hi hackers,
Here is v2:
- 0008 no longer sets aside and restores cached selectivity. It now marks
the pass on PlannerInfo and bypasses the cache while the mark is set.
This covers the per-subclause selectivities an OR clause caches, which
v1 of this series missed.
- EagerAggMode replaces the booleans and empty list checks previously
threaded through the series.
- Deduplication paths are created with AGGSPLIT_SIMPLE, so they are
costed as plain groupings and print without a Partial label. The function
that relabeled them is gone.
- Because the mode is now explicit, 0015 also reaches a plain
deduplication. Two plans lose a repeated grouping.
- 0012 is new. It splits the construction of GroupingExprInfos out of
create_grouping_expr_infos(). This is the only change in patch count.
Thanks,
Will
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Xuneng Zhou | 2026-08-31 01:47:20 | Re: timeout value overflow in wait for lsn |
| Previous Message | Sami Imseih | 2026-08-31 00:21:07 | Re: Add starelid, attnum to pg_stats and leverage this in pg_dump |