| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | "Voillequin, Jean-Marc" <Jean-Marc(dot)Voillequin(at)moodys(dot)com> |
| Cc: | Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org>, "pgsql-sql(at)lists(dot)postgresql(dot)org" <pgsql-sql(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: CTE materialized/not materialized |
| Date: | 2020-11-27 15:35:02 |
| Message-ID: | 704895.1606491302@sss.pgh.pa.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-sql |
"Voillequin, Jean-Marc" <Jean-Marc(dot)Voillequin(at)moodys(dot)com> writes:
> Let me rephrase.
> In the case the CTE can be inlined, why PG does not compute 2 plans (P1 with CTE inlined, P2 with CTE materialized) and compare them to choose the best?
It would be excessively expensive (even for just one CTE, never mind
several) and it wouldn't fit into the planner's structure very well.
As noted in the comment Alvaro quoted, the point where we make these
decisions is pretty far upstream of having any concrete cost estimates.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Voillequin, Jean-Marc | 2020-11-27 15:58:02 | RE: CTE materialized/not materialized |
| Previous Message | Voillequin, Jean-Marc | 2020-11-27 15:18:31 | RE: CTE materialized/not materialized |