From: Tom Lane <tgl@sss.pgh.pa.us>
To: natalya@timescale.com
cc: pgsql-bugs@lists.postgresql.org, Peter Geoghegan <pg@bowt.ie>
Subject: Re: BUG #19049: Assert failure when using skip arrays on an index key
 with DESC order
In-reply-to: <19049-b7df801e71de41b2@postgresql.org>
References: <19049-b7df801e71de41b2@postgresql.org>
Comments: In-reply-to PG Bug reporting form <noreply@postgresql.org>
	message dated "Thu, 11 Sep 2025 12:38:11 -0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1788649.1757605784.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Thu, 11 Sep 2025 11:49:45 -0400
Message-ID: <1788650.1757605785@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/1788650.1757605785%40sss.pgh.pa.us>
Precedence: bulk

PG Bug reporting form <noreply@postgresql.org> writes:
> -- This query crashes with an Assert
>  select distinct on(s0, s1) * from mv where t2 >=3D -1 and t1 < 10 order=
 by
> s0, s1, t1 desc, t2 desc;

> Assert is:
>                 if (ScanDirectionIsForward(dir) && array->low_compare)
> Assert(DatumGetBool(FunctionCall2Coll(&array->low_compare->sk_func,
> array->low_compare->sk_collation, tupdatum, array->low_compare->sk_argum=
ent)));

Indeed.  "git bisect" says this started at

b3f1a13f22f9e28842ee5fbd08b7ec805e27aaac is the first bad commit
commit b3f1a13f22f9e28842ee5fbd08b7ec805e27aaac
Author: Peter Geoghegan <pg@bowt.ie>
Date:   Fri Apr 4 14:14:08 2025 -0400

    Avoid extra index searches through preprocessing.

Peter, you want to take a look?

			regards, tom lane