luda posch <ludaludaluda(at)gmail(dot)com> writes:
> I am not an expert on the official sql spec, but I think it would be useful
> if distinct on() could be used within an aggregate when supplied with
> another column name. For example:
> select sum(distinct on(id) order_price) from order_table;
> select sum(distinct on(email_address) bounce_count) from email_bounces;
> select sum(distinct on(county) area) from state_county_area;
Use another level of sub-select, eg
select sum(order_price) from
(select distinct on(id) order_price from order_table ...) ss;
The usefulness of this doesn't really seem high enough to justify
inventing and maintaining a more compact way to handle it.
regards, tom lane