From: Tom Lane <tgl@sss.pgh.pa.us>
To: Michael Paquier <michael@paquier.xyz>
cc: Peter Eisentraut <peter.eisentraut@enterprisedb.com>,
        Tomas Vondra <tomas.vondra@enterprisedb.com>,
        PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Subject: Re: Shouldn't construct_array_builtin and deconstruct_array_builtin agree on types?
In-reply-to: <ZIeo2pVq0VLXQ6uI@paquier.xyz>
References: <bf166909-8a8d-2d5f-ed5e-4a4d5121bde7@enterprisedb.com> <3020223d-6cdc-8e15-56e6-bc8cfc6c2d81@enterprisedb.com> <ZIeo2pVq0VLXQ6uI@paquier.xyz>
Comments: In-reply-to Michael Paquier <michael@paquier.xyz>
	message dated "Tue, 13 Jun 2023 08:23:06 +0900"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1389828.1686616005.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Mon, 12 Jun 2023 20:26:45 -0400
Message-ID: <1389829.1686616005@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/1389829.1686616005%40sss.pgh.pa.us>
Precedence: bulk

Michael Paquier <michael@paquier.xyz> writes:
> On Mon, Jun 12, 2023 at 11:06:18PM +0200, Peter Eisentraut wrote:
>> They only support the types that they were actually being used with.  I=
f you
>> need another type, feel free to add it.

> FWIW, I agree with Tomas that this is an oversight that should be
> fixed in v16, saving from the need to have a copy of
> deconstruct_array_builtin() in extensions.

We don't want to bloat these functions indefinitely, so I understand
Peter's approach of only adding the cases actually being used.
At the same time, it's reasonable to take some thought for extensions
that might want slightly more functionality than the core code
happens to need at any given moment.

The idea of making both functions support the same set of types
does seem like a reasonable compromise.

			regards, tom lane