Re: Find non-unique "key" values via SQL

[Dan <dan27649@xxxxxxxxx>]

On 1/9/07, Chris Payne <CPayne@xxxxxxxxxxxxxxx> wrote:
> The reason for this is that the where conditions are applied before
> aggregation and the having clause is applied after aggregation, so if
> you wanted to use a sum() function and then drop all of the rows that
> did not have a certain sum, you would have to use a having clause. I
> assume that it has to come after the group by because it needs to count
> up the rows before it can check the having clause, and it needs the
> group by to know how to count up the rows.


That seems to make sense.  I was wondering how the SQL engine would
distinguish the pre- and post-aggregation when I had the count(*) test in
the WHERE clause.

Thanks,
Dan