Hmmm... Missed that!
I think you are looking for the feature that was introduced in
PostgreSQL 9.1 where you could have a non-group-by column in the select
list, but only if the group-by has a pkey to identify the actual row.
http://www.postgresql.org/docs/9.1/static/release-9-1.html
(Search for GROUP BY)
--
Robins Tharakan
On 11/08/2011 03:29 PM, Tarlika Elisabeth Schmitz wrote:
> Thank you for yuor suggestion, Robins. Unfortunately, it does not work;
> this returns:
> 1787 "Toomyvara" 0.5
> 1787 "Toomevara" 0.4
> 1188 "Toonybara" 0.4
> because while column "no" is identical, "name" isn't and you're
> grouping by both of them.
