Re: regress: query discrepancy between explain vs execution

On Mon, 19 Oct 2026, 02:46 David Rowley, <dgrowleyml(at)gmail(dot)com> wrote:
>
> On Sat, 3 Oct 2026 at 04:42, Matthias van de Meent <boekewurm(at)gmail(dot)com> wrote:
> > In the partition_aggregate tests, there's a case where the explained
> > query doesn't match the query that's executed just below. This has
> > existed since the file was first committed (e2f1eb0ee3), and doesn't
> > materially affect the plan, but if any issues were introduced it might
> > not show up in this part of the test, so we should probably fix it.
>
> I looked at this and noticed a couple more mistakes. Can you share why
> you picked the HAVING query over the version without?

Because that was more consistent with the rest of the queries in the
file. Every other explained-and-then-executed query on pagg_tab_ml
has a HAVING clause, except these.

> v2 attached with the other things I noticed. array_agg has supported
> partial aggregation since v16 (16fd03e95).

Thanks, LGTM.

Kind regards,

Matthias van de Meent
Databricks