Re: MERGE/SPLIT PARTITIONS issues/questions

On Tue, Aug 18, 2026 at 6:48 AM jian he <jian(dot)universality(at)gmail(dot)com> wrote:
>
> On Mon, Aug 17, 2026 at 6:27 PM Alexander Korotkov <aekorotkov(at)gmail(dot)com> wrote:
> >
> > Any objections to pushing these 4 fixes?
>
> As mentioned in [1], RestrictSearchPath is called before
> MergePartitionsMoveRows and SplitPartitionMoveRows.
> This means that if an expression contains anything non-immutable, we can't
> evaluate it consistently for every row under a restricted search path,
> imagine function text_regclass.
> It would be better to add a comment directly above SplitPartitionMoveRows and
> MergePartitionsMoveRows to mention this situation.
> This will help future readers understand the implications.
>
> In MergePartitionsMoveRows, the `foreach(ltab, *wqueue)` can be removed,
> because ATExecMergePartitions->createPartitionTable doesn't call
> ATGetQueueEntry.
> Similarly, the foreach loop in deleteSplitPartitionContext can also be
> removed for the same reason,
> we can probably get rid of deleteSplitPartitionContext.
>
> [1]: https://www.postgresql.org/message-id/CACJufxHk0F%2B1UyvExHoMfBZrsUeGQiB8MBm1PC5Fd3MtAszLGw%40mail.gmail.com

Agree on your corrections expect for deleteSplitPartitionContext(): it
still have resources to free. The revised patchset is attached.

------
Regards,
Alexander Korotkov
Supabase