Feedback-ID: ia2694551:Fastmail
Date: Wed, 15 Jun 2022 20:26:33 +0200
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Cc: daniele.varrazzo@gmail.com, pgsql-hackers@lists.postgresql.org
Subject: Re: Using PQexecQuery in pipeline mode produces unexpected Close
 messages
Message-ID: <20220615182633.ztcoloh7xbicggot@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="eq2sy33kqql2cp67"
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <20220610.152544.1597540405935332165.horikyota.ntt@gmail.com>
Archived-At: 
 <https://www.postgresql.org/message-id/20220615182633.ztcoloh7xbicggot%40alvherre.pgsql>
Precedence: bulk


--eq2sy33kqql2cp67
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit

On 2022-Jun-10, Kyotaro Horiguchi wrote:

> (Moved to -hackers)
> 
> At Wed, 8 Jun 2022 17:08:47 +0200, Alvaro Herrera <alvherre@alvh.no-ip.org> wrote in 
> > What that Close message is doing is closing the unnamed portal, which
> > is otherwise closed implicitly when the next one is opened.  That's how
> > single-query mode works: if you run a single portal, it'll be kept open.
> > 
> > I believe that the right fix is to not send that Close message in
> > PQsendQuery.
> 
> Agreed. At least Close message in that context is useless and
> PQsendQueryGuts doesn't send it. And removes the Close message surely
> fixes the issue.

So, git archaeology led me to this thread
https://postgr.es/m/202106072107.d4i55hdscxqj@alvherre.pgsql
which is why we added that message in the first place.

I was about to push the attached patch (a merged version of Kyotaro's
and mine), but now I'm wondering if that's the right approach.

Alternatives:

- Have the client not complain if it gets CloseComplete in idle state.
  (After all, it's a pretty useless message, since we already do nothing
  with it if we get it in BUSY state.)

- Have the server not send CloseComplete at all in the cases where
  the client is not expecting it.  Not sure how this would be
  implemented.

- Other ideas?

-- 
Álvaro Herrera               48°01'N 7°57'E  —  https://www.EnterpriseDB.com/
"That sort of implies that there are Emacs keystrokes which aren't obscure.
I've been using it daily for 2 years now and have yet to discover any key
sequence which makes any sense."                        (Paul Thomas)

--eq2sy33kqql2cp67
Content-Type: text/x-diff; charset=us-ascii
Content-Disposition: attachment;
	filename="v2-0001-PQsendQuery-Don-t-send-Close-in-pipeline-mode.patch"