| From: | Keyerror Smart <smartkeyerror(at)gmail(dot)com> |
|---|---|
| To: | Tender Wang <tndrwang(at)gmail(dot)com> |
| Cc: | PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>, Dean Rasheed <dean(dot)a(dot)rasheed(at)gmail(dot)com> |
| Subject: | Re: MERGE fails with RETURNING when an excluded inheritance parent has an extra-column child |
| Date: | 2026-11-09 13:18:58 |
| Message-ID: | CAD=-kXY2Bz0L1Xg09iRpBs8shq9AZNtQSbvwjnn2sTJM220eDw@mail.gmail.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Hi Tender Wang,
I can reproduce this on master, and the fix looks right to me.
With `missing_ok = true`, a child-only column just gets a 0 in the
map, and if an expression ever referenced it, `map_variable_attnos()`
would raise "unexpected varattno" instead of mis-mapping it.
ExecGetRootToChildMap() also passes missing_ok = true for
non-partition children, for the same reason:
/*
* When this child table is not a partition (!relispartition), it may
* have columns that are not present in the root table, which we ask
* to ignore by passing true for missing_ok.
*/
I also tried whole-row RETURNING (t, t.*, old/new), RLS as a
non-superuser, and a child with a dropped column. All work fine
with the patch.
Zhenglong Li,
Regards
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Robert Haas | 2026-11-09 13:28:58 | Re: Show 'Rows Removed by Filter' with two fractional digits in EXPLAIN |
| Previous Message | Bruce Momjian | 2026-11-09 13:08:19 | Re: First draft of PG 19 release notes |