Horton

Re: using index or check in ALTER TABLE SET NOT NULL

Tomas Vondra <tomas.vondra@2ndquadrant.com>

View thread View on postgr.es
From: Tomas Vondra <tomas.vondra@2ndquadrant.com>
To: David Rowley <david.rowley@2ndquadrant.com>
Cc: Sergei Kornilov <sk@zsrv.org>, Tom Lane <tgl@sss.pgh.pa.us>, Andres Freund <andres@anarazel.de>, Robert Haas <robertmhaas@gmail.com>, David Steele <david@pgmasters.net>, PostgreSQL mailing lists <pgsql-hackers@postgresql.org>, Alvaro Herrera <alvherre@alvh.no-ip.org>, Stephen Frost <sfrost@snowman.net>, Ildar Musin <i.musin@postgrespro.ru>
Date: 2019-08-02T19:54:38Z
Lists: pgsql-hackers 
On Wed, Jun 12, 2019 at 08:34:57AM +1200, David Rowley wrote:
>On Tue, 11 Jun 2019 at 03:35, Sergei Kornilov <sk@zsrv.org> wrote:
>> > Does anyone think we shouldn't change the INFO message in ATTACH
>> > PARTITION to a DEBUG1 in PG12?
>>
>> Seems no one wants to vote against this change.
>
>I'm concerned about two things:
>
>1. The patch reduces the test coverage of ATTACH PARTITION. We now
>have no way to ensure the constraint was used to validate the rows in
>the partition.
>2. We're inconsistent with what we do for SET NOT NULL and ATTACH
>PARTITION. We raise an INFO message when we use a constraint for
>ATTACH PARTITION and only a DEBUG1 for SET NOT NULL.
>
>Unfortunately, my two concerns conflict with each other.
>

We're getting close to beta3/rc1, and this thread was idle for ~1 month.

I think there's a consensus to change INFO to DEBUG1 in pg12, and then
maybe imlpement something like VERBOSE mode in the future. Objections?

As for the reduction of test coverage, can't we deduce whether a
constraint was used from data in pg_stats or something like that?

regards

-- 
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Commits

  1. Avoid using INFO elevel for what are fundamentally debug messages.

  2. Revert setting client_min_messages to 'debug1' in new tests.

  3. Allow ALTER TABLE .. SET NOT NULL to skip provably unnecessary scans.

  4. Improve predtest.c's internal docs, and enhance its functionality a bit.