Re: Document aggregate functions better w.r.t. ORDER BY

On Thu, Oct 26, 2023 at 03:09:26PM -0700, David G. Johnston wrote:
> On Thu, Oct 26, 2023 at 2:56 PM Bruce Momjian <bruce(at)momjian(dot)us> wrote:
>
> On Wed, Oct 25, 2023 at 10:34:10PM -0700, David G. Johnston wrote:
> > I would reword the existing note to be something like:
> >
> > The SQL Standard defines specific aggregates and their properties,
> including
> > which of DISTINCT and/or ORDER BY is allowed.  Due to the extensible
> nature of
> > PostgreSQL it accepts either or both clauses for any aggregate.
>
> Uh, is this something in my patch or somewhere else?  I don't think
> PostgreSQL extensible is an example of syntax flexibility.
>
>
> https://www.postgresql.org/docs/current/sql-expressions.html#SYNTAX-AGGREGATES
>
> Note
> The ability to specify both DISTINCT and ORDER BY in an aggregate function is a
> PostgreSQL extension.
>
> I am pointing out that the first sentence of the existing note above seems to
> be factually incorrect.  I tried to make it correct - while explaining why we
> differ.  Though in truth I'd probably rather just remove the note.

Agreed, removed, patch attached. This is just too complex to specify.

> > We get enough complaints regarding "apparent ordering" that I would like
> to
> > add:
> >
> > As a reminder, while some DISTINCT processing algorithms produce sorted
> output
> > as a side-effect, only by specifying ORDER BY is the output order
> guaranteed.
>
> Well, we need to create a new email thread for this and look at all the
> areas is applies to since this is a much larger issue.
>
> I was hoping to sneak this one in regardless of the bigger picture issues,
> since this specific combination is guaranteed to output ordered presently.

No sneaking. ;-) It would be bad to document this unevenly because it
sets expectations in other parts of the system if we don't mention it.

--
Bruce Momjian <bruce(at)momjian(dot)us> https://momjian.us
EDB https://enterprisedb.com

Only you can decide what is important to you.