From: Tom Lane <tgl@sss.pgh.pa.us>
To: Yura Sokolov <y.sokolov@postgrespro.ru>
cc: Alvaro Herrera <alvherre@alvh.no-ip.org>,
        pgsql-hackers@lists.postgresql.org
Subject: Re: Add PortalDrop in exec_execute_message
In-reply-to: <fb7b58d91dcc23b592ace457dc0f8c3a@postgrespro.ru>
References: <202106072107.d4i55hdscxqj@alvherre.pgsql>
 <fb7b58d91dcc23b592ace457dc0f8c3a@postgrespro.ru>
Comments: In-reply-to Yura Sokolov <y.sokolov@postgrespro.ru>
	message dated "Fri, 11 Jun 2021 08:21:01 +0300"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <1875432.1623418705.1@sss.pgh.pa.us>
Date: Fri, 11 Jun 2021 09:38:25 -0400
Message-ID: <1875433.1623418705@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/1875433.1623418705%40sss.pgh.pa.us>
Precedence: bulk

Yura Sokolov <y.sokolov@postgrespro.ru> writes:
> This makes me think, Close message were intended to be used
> but simply forgotten when libpq patch were made.
> Tom, could I be right?

You could argue all day about what the intentions were nearly twenty
years ago.  But the facts on the ground are that we don't issue Close
in those places, and changing it now would be a de facto protocol
change for applications.  So I'm a hard -1 on these proposals.

(Alvaro's proposed change isn't a protocol break, since pipeline
mode hasn't shipped yet.  It's trying to make some brand new code
act more like old code, which seems like a fine idea.)

I think that the actual problem here has been resolved in
commit bb4aed46a.  Perhaps we should reconsider my decision not to
back-patch that.  Unlike a protocol change, that one could possibly
be sane to back-patch.  I didn't think it was worth the trouble and
risk, but maybe there's a case for it.

			regards, tom lane