MIME-Version: 1.0
Date: Wed, 28 Aug 2024 16:58:56 +1000
From: Justin Clift <justin@postgresql.org>
To: David Rowley <dgrowleyml@gmail.com>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, nikhil raj <nikhilraj474@gmail.com>,
 "pgsql-generallists.postgresql.org" <pgsql-general@lists.postgresql.org>,
 NIKITA PATEL <patelnikita1411@gmail.com>, Patel Khushbu
 <patelkhushbu2067@gmail.com>
Subject: Re: Significant Execution Time Difference Between PG13.14 and PG16.4
 for Query on information_schema Tables.
In-Reply-To: <CAApHDvr6igx9SPQs3n_K5WwvTdYwhZ1LzP6qOfiGFNbUHXOCyQ@mail.gmail.com>
References: <CALNXOfqDr5jVRZHThyXyNftDuezjxOJB1mqw=4x1T8YBibi6uw@mail.gmail.com>
 <CAG1ps1w61XD1LxUf3EBCZ=VR3S_orRj_dk6cxmO5KusBVyrgMg@mail.gmail.com>
 <CAAvG1pChk3W9MWkK-_JPyUdThSoJhv4MueH5i8cZS2T_HUP5dA@mail.gmail.com>
 <CAG1ps1xvnTZceKK24OUfMKLPvDP2vjT-d+F2AOCWbw_v3KeEgg@mail.gmail.com>
 <2962669.1724722813@sss.pgh.pa.us>
 <CAApHDvpRxyw1G2HVkA=PM5Uqh59moOFqPXyMTsEDUadKo2fQTA@mail.gmail.com>
 <fa70f9fbd7a0bbfb78316dc8ac2ffddb@postgresql.org>
 <CAApHDvr6igx9SPQs3n_K5WwvTdYwhZ1LzP6qOfiGFNbUHXOCyQ@mail.gmail.com>
User-Agent: Roundcube Webmail/1.4.15
Message-ID: <7b97171e276c41ec6ae4eeb8c2e8b47d@postgresql.org>
Content-Type: text/plain; charset=US-ASCII;
 format=flowed
Content-Transfer-Encoding: 7bit
Archived-At: <https://www.postgresql.org/message-id/7b97171e276c41ec6ae4eeb8c2e8b47d%40postgresql.org>
Precedence: bulk

On 2024-08-27 20:14, David Rowley wrote:
> On Tue, 27 Aug 2024 at 18:00, Justin Clift <justin@postgresql.org> 
> wrote:
>> As a general thought, seeing that this might be an actual problem
>> should some kind of automated testing be added that checks for
>> performance regressions like this?
> 
> We normally try to catch these sorts of things with regression tests.
> Of course, that requires having a test that would catch a particular
> problem, which we don't seem to have for this particular case.  A
> performance test would also require testing a particular scenario, so
> I don't see why that's better.  A regression test is better suited as
> there's no middle ground between pass and fail.

Yeah, that's the kind of thing I was thinking.

Any idea who normally does those, and if it would be reasonable to add
test(s) for the internal information tables?

Regards and best wishes,

Justin Clift