MIME-Version: 1.0
References: <CAHM0NXjz0kDwtzoe-fnHAqPB1qA8_VJN0XAmCgUZ+iPnvP5LbA@mail.gmail.com>
 <CAKFQuwY0vhNG8T+pC3BQJurFi3NN_L1KbEPGagRN3V5nKZcpDQ@mail.gmail.com> <931747.1721687375@sss.pgh.pa.us>
In-Reply-To: <931747.1721687375@sss.pgh.pa.us>
From: David Christensen <david@pgguru.net>
Date: Tue, 23 Jul 2024 08:37:02 -0500
Message-ID: <CAHM0NXhwmrAeUmg5syC=GjFkCDCvg0gaYBy_JbaEwgBinTmDtg@mail.gmail.com>
Subject: Re: [PATCH] GROUP BY ALL
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: "David G. Johnston" <david.g.johnston@gmail.com>, 
	pgsql-hackers <pgsql-hackers@postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: <https://www.postgresql.org/message-id/CAHM0NXhwmrAeUmg5syC%3DGjFkCDCvg0gaYBy_JbaEwgBinTmDtg%40mail.gmail.com>
Precedence: bulk

On Mon, Jul 22, 2024 at 5:29=E2=80=AFPM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> "David G. Johnston" <david.g.johnston@gmail.com> writes:
> > On Mon, Jul 22, 2024 at 1:55=E2=80=AFPM David Christensen <david@pgguru=
.net> wrote:
> >> I see that there'd been some chatter but not a lot of discussion about
> >> a GROUP BY ALL feature/functionality.  There certainly is utility in
> >> such a construct IMHO.
>
> > I strongly dislike adding this feature.  I'd only consider supporting i=
t if
> > it was part of the SQL standard.
>
> Yeah ... my recollection is that we already rejected this idea.
> If you want to re-litigate that, "throwing this out there" is
> not a sufficient argument.

Heh, fair enough.  I actually wrote the patch after encountering the
syntax in DuckDB and it seemed easy enough to add to Postgres while
providing some utility, then ended up seeing a thread about it later.
I did not get the sense that this had been globally rejected; though
there were definitely voices against it, it seemed to trail off rather
than coming to a conclusion.

> (Personally, I'd wonder exactly what ALL is quantified over: the
> whole output of the FROM clause, or only columns mentioned in the
> SELECT tlist, or what? And why that choice rather than another?)

My intention here was to basically be a shorthand for "group by
specified non-aggregate fields in the select list".  Perhaps I'm not
being creative enough, but what is the interpretation/use case for
anything else? :-)

While there are other ways to accomplish these things, making an easy
way to GROUP BY with aggregate queries would be useful in the field,
particularly when doing iterative discovery work would save a lot of
time with a situation that is both detectable and hits users with
errors all the time.

I'm not married to the exact syntax of this feature; anything else
short and consistent could work if `ALL` is considered to potentially
gain a different interpretation in the future.

David