> On Nov 1, 2026, at 01:17, Paul A Jungwirth <[email protected]>
> wrote:
>
> Hi Hackers,
>
> Here is a patch to fix some memory leaks in BEFORE/INSTEAD OF triggers.
>
> Andres reported the problem in [0] while reviewing FOR PORTION OF, but
> as he suspected, it is a problem on master too. It goes back to v14 at
> least. You can reach it if (1) the rel is not a regular table (e.g. a
> view) and (2) the trigger returns NEW without changing it (or NULL in
> ExecIRInsertTriggers).
>
> [0]
> https://www.postgresql.org/message-id/yfers7d2kizq6ockz4nu7e7avq3gj5anpd7ilxkazyvicv4qgy@3cesow2bbhnx
>
> Yours,
>
> --
> Paul ~{:-)
> [email protected]
> <v1-0001-Free-the-tuple-fetched-for-BEFORE-INSTEAD-OF-row-.patch>
V1 overall looks good to me. There seems a typo in the commit message:
```
heap slot) and (2) the trigger returned NEW unchanged (or NULL in
ExecIRInsertTriggers).
```
I think it should be “or NULL in ExecIRUpdateTriggers”.
Best regards,
--
Chao Li (Evan)
HighGo Software Co., Ltd.
https://www.highgo.com/
