| From: | Sam Mason <sam(at)samason(dot)me(dot)uk> | 
|---|---|
| To: | pgsql-general(at)postgresql(dot)org | 
| Subject: | Re: Embarassing GROUP question | 
| Date: | 2009-10-03 17:31:26 | 
| Message-ID: | 20091003173126.GG5407@samason.me.uk | 
| Views: | Whole Thread | Raw Message | Download mbox | Resend email | 
| Thread: | |
| Lists: | pgsql-general | 
On Sat, Oct 03, 2009 at 01:05:49PM -0400, Tom Lane wrote:
> What you might be after is something like Postgres' DISTINCT ON
> feature
Yup, looks that way to me as well.
> I have never really played around with this aspect of MySQL ...
Me neither.
> but looking at this example, and presuming that you find that
> it actually does something useful, I wonder whether they interpret
> the combination of GROUP BY and ambiguous-per-spec ORDER BY
> in some fashion similar to DISTINCT ON.
Yup, does look that way doesn't it.  It's still a weird pair of
semantics to conflate.
Hum, if they were assuming that you'd always have to implement GROUP BY
by doing a sort step first then I can see why they'd end up with this.
But if you want to do *anything* else (i.e. hash aggregate in PG) then
you want to keep the semantics of GROUP BY and ORDER BY separate as the
spec and indeed PG does.
-- 
  Sam  http://samason.me.uk/
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Tom Lane | 2009-10-03 20:14:21 | Re: Procedure for feature requests? | 
| Previous Message | Sam Mason | 2009-10-03 17:24:42 | Re: Procedure for feature requests? |