Обсуждение: Correlated subselect in regression test

The regression test file subselect.sql contains the following query under the 
heading "Correlated subselects":

SELECT '' AS five, f1 AS "Correlated Field" FROM SUBSELECT_TBL WHERE (f1, f2) IN (SELECT f2, CAST(f3 AS int4) FROM
SUBSELECT_TBL                   WHERE f3 IS NOT NULL);
 

It has been pointed out to me that there is no correlation in this query.  Is 
there some other special purpose to this query or is it simply categorized 
wrongly?

-- 
Peter Eisentraut
http://developer.postgresql.org/~petere/

Peter Eisentraut <peter_e@gmx.net> writes:
> The regression test file subselect.sql contains the following query under the 
> heading "Correlated subselects":

> SELECT '' AS five, f1 AS "Correlated Field"
>   FROM SUBSELECT_TBL
>   WHERE (f1, f2) IN (SELECT f2, CAST(f3 AS int4) FROM SUBSELECT_TBL
>                      WHERE f3 IS NOT NULL);

> It has been pointed out to me that there is no correlation in this query.  Is 
> there some other special purpose to this query or is it simply categorized 
> wrongly?

I think it's just mislabeled.  Looking at the CVS history, that
particular query hasn't been changed since Tom Lockhart first created
this test back in 6.3.  I don't see any reason to think it has a
special purpose in mind.
        regards, tom lane