Feedback-ID: ia2694551:Fastmail
Date: Tue, 6 Jun 2023 22:11:32 +0200
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Richard Guo <guofenglinux@gmail.com>,
	PostgreSQL-development <pgsql-hackers@postgresql.org>
Subject: Re: Assert failure of the cross-check for nullingrels
Message-ID: <20230606201132.i3xjwpsg36p6tii7@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <395264.1684698283@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/20230606201132.i3xjwpsg36p6tii7%40alvherre.pgsql>
Precedence: bulk

On 2023-May-21, Tom Lane wrote:

> Since we're hard up against the beta1 wrap deadline, I went ahead
> and pushed the v5 patch.  I doubt that it's perfect yet, but it's
> a small change and demonstrably fixes the cases we know about.
> 
> As I said in the commit message, the main knock I'd lay on v5
> is "why not use required_relids all the time?".

So, is this done?  I see that you made other commits fixing related code
several days after this email, but none seems to match the changes you
posted in this patch; and also it's not clear to me that there's any
test case where this patch is expected to change behavior.  (So there's
also a question of whether this is a bug fix or rather some icying on
cake.)

-- 
Álvaro Herrera               48°01'N 7°57'E  —  https://www.EnterpriseDB.com/