Re: distinct vs group by

Ron Mayer <rm_pg@cheapcomplexdevices.com> writes:
> Should the expressions
>   select distinct x from t
> and
>   select          x from t group by x
> have the same effect?

Offhand I think they are theoretically equivalent.

> It seems the optimizer sometimes chooses different plans for those
> expressions.   Could the select distinct have used the slightly
> faster hash aggregate?

This is partly historical (the DISTINCT code hasn't been rewritten in a
long time) and partly intentional --- you can choose one phrasing or the
other to control what plan you get.

            regards, tom lane