From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andrew Dunstan <andrew@dunslane.net>
cc: pgsql-committers@lists.postgresql.org
Subject: Re: pgsql: Declare assorted array functions using anycompatible not
 anyelem
In-reply-to: <fff35eb5-765c-3e60-3a2a-45ba5f88777d@dunslane.net>
References: <E1kaQ2c-0005lx-Eg@gemulon.postgresql.org>
 <fff35eb5-765c-3e60-3a2a-45ba5f88777d@dunslane.net>
Comments: In-reply-to Andrew Dunstan <andrew@dunslane.net>
	message dated "Mon, 09 Nov 2020 16:23:24 -0500"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1999747.1604957386.1@sss.pgh.pa.us>
Date: Mon, 09 Nov 2020 16:29:46 -0500
Message-ID: <1999748.1604957386@sss.pgh.pa.us>
Precedence: bulk

Andrew Dunstan <andrew@dunslane.net> writes:
> On 11/4/20 4:10 PM, Tom Lane wrote:
>> Declare assorted array functions using anycompatible not anyelement.

> This patch broke cross-version pg_upgrade testing.

Uh, yeah ... didn't you get my two prior messages about that?

> I have cured crake
> for the moment by having it execute this in the source database:

I think probably the right fix is just to change that test case to
use a different implementation function, per [1].  I'm holding off
pushing the fix till after this week's wraps, though.

> But I wonder if we really want to make it impossible to upgrade
> databases with aggregates that contain these functions?

If I thought that user-defined aggregates relying on array_cat were
really a thing (and not just a test case), I'd be more concerned about
this.  But it's hard to see why users shouldn't use array_agg() instead
of rolling their own.

			regards, tom lane

[1] https://www.postgresql.org/message-id/1915573.1604879242%40sss.pgh.pa.us