Horton

Re: psql - add SHOW_ALL_RESULTS option

Daniel Verite <daniel@manitou-mail.org>

View thread View on postgr.es
From: "Daniel Verite" <daniel@manitou-mail.org>
To: "Tom Lane" <tgl@sss.pgh.pa.us>
Cc: "Alvaro Herrera" <alvherre@2ndquadrant.com>,"Tomas Vondra" <tomas.vondra@2ndquadrant.com>,"Fabien COELHO" <coelho@cri.ensmp.fr>,"vignesh C" <vignesh21@gmail.com>,"Kyotaro Horiguchi" <horikyota.ntt@gmail.com>,peter.eisentraut@2ndquadrant.com,iwata.aya@jp.fujitsu.com,"PostgreSQL Developers" <pgsql-hackers@lists.postgresql.org>
Date: 2020-01-17T20:39:07Z
Lists: pgsql-hackers 
	Tom Lane wrote:

> I'm not really holding my breath for that to happen, considering
> it would involve fundamental breakage of the wire protocol.
> (For example, extended query protocol assumes that Describe
> Portal only needs to describe one result set.  There might be
> more issues, but that one's bad enough.)

I'm not sure that CALL can be used at all with the extended protocol
today  (just like multiple queries per statements, except that for these,
I'm sure).
My interpretation is that the extended protocol deliberately
lefts out the possibility of multiple result sets because it doesn't fit
with how it's designed and if you want to have this, you can just use
the old protocol's Query message. There is no need to break anything
or invent anything but on the contrary to use the older way.

Considering these 3 ways to use libpq to send queries:

1. using old protocol with PQexec: only one resultset.

2. using old protocol with PQsendQuery+looping on PQgetResult:
same as #1 except multiple result sets can be processed

3. using extended protocol: not for multiple result sets, not for copy,
possibly not for other things, but  can use bind parameters, binary format,
pipelining,...

The current patch is about using #2 instead of #1.
They have been patches about doing bits of #3 in some cases
(binary output, maybe parameters too?) and none got eventually in.
ISTM that the current situation is that psql is stuck at #1 since forever
so it's not fully using the capabilities of the protocol, both old and new.


Best regards,
-- 
Daniel Vérité
PostgreSQL-powered mailer: http://www.manitou-mail.org
Twitter: @DanielVerite

Commits

  1. psql: Show all query results by default

  2. psql: Refactor ProcessResult()

  3. psql: Refactor SendQuery()

  4. psql: Add test for psql behavior on server crash

  5. psql: Additional tests

  6. Improve some psql test code

  7. psql: Fix some scan-build warnings

  8. psql: Add test for handling of replication commands

  9. psql: More tests

  10. psql: Add various tests

  11. psql: Add test for query canceling

  12. Revert "psql: Show all query results by default"

  13. Extend a test case a little