| From: | shihao zhong <zhong950419(at)gmail(dot)com> |
|---|---|
| To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
| Cc: | link_xliv(at)hotmail(dot)com, pgsql-bugs(at)lists(dot)postgresql(dot)org |
| Subject: | Re: BUG #19408: Bad plan for UNION ALL subquery with outer WHERE, ORDER BY, LIMIT, and separate indexes |
| Date: | 2026-11-02 05:19:14 |
| Message-ID: | CAGRkXqRjd-t0HqsaJJdEcnwUDrrQNixJGyMn8mA-NsXvad3Gvw@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-bugs |
> Another reason why this is only WIP quality is that it causes a bunch
> of plan changes in the regression tests. I haven't looked to see if
> they're all reasonable.
I went through them. Two defeat the test they are in. Most of the rest
turn Sort over Append into MergeAppend over per-child Sorts.
The report only goes wrong with a LIMIT, so the attached keeps just
the fractional part of your patch. One existing plan still changes.
I feel it is master only. Would you rather take this narrow version,
or the full one?
Thanks,
Shihao
| Attachment | Content-Type | Size |
|---|---|---|
| nocfbot-v1-0001-Consider-sorting-the-cheapest-child-path-in-generate.patch | application/octet-stream | 7.6 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Richard Guo | 2026-11-02 05:51:25 | Re: Wrong results: hashed SubPlan referenced twice after OR-qual extraction reuses a stale hash table (13 to 19beta4) |
| Previous Message | Kirill Reshke | 2026-11-02 02:40:37 | Re: BUG #19817: CREATE UNIQUE INDEX on a partitioned table can silently attach an invalid, not-ready child index, le |