Re: Extended statistics improvement: multi-column MCV missing values

From: Enrique Sánchez <enriqueesanchz(at)gmail(dot)com>
To: Zsolt Parragi <zsolt(dot)parragi(at)percona(dot)com>
Cc: pgsql-hackers(at)lists(dot)postgresql(dot)org
Subject: Re: Extended statistics improvement: multi-column MCV missing values
Date: 2026-06-04 07:00:23
Message-ID: CAOCkzAntraDudo=zghSecb4d7f4fnveUiXDHUYVWAAhrhE2QSg@mail.gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi, thank you for reviewing it.

Tue, 02 Jun 2026 at 0:18, Zsolt Parragi (<zsolt(dot)parragi(at)percona(dot)com>) wrote:

> > The strongest
> > case is a top-level AND condition with equality clauses for all columns
>
> I see one issue related to that with the current patch, having the
> same amount of conditions as the table doesn't mean that we have
> conditions for all columns.

I've fixed this in the v2 patch attached above (also attached here for
completeness). It focuses on the basic full-dimensional top-level AND
equality miss
Now we construct a Bitmapset with the columns the clause covers and compare
it to `stat->keys` that contains the columns covered by the MCV statistic.

Best regards,
Enrique.

Attachment Content-Type Size
v2-0001-Cap-selectivity-when-values-are-not-in-multi-colu.patch text/x-patch 10.4 KB

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Jingtang Zhang 2026-06-04 07:19:24 Re: Optimize CPU usage of dropping buffers during recovery
Previous Message Zhijie Hou (Fujitsu) 2026-06-04 06:58:31 RE: Parallel Apply