Christian Breimann <chr@math.uni-muenster.de> writes:
> select distinct S.customer_name
> from depositor as S
> where not exists(
> (select branch_name
> from branch
> where branch_city='Brooklyn')
> except
> (select R.branch_name
> from depositor as T, account as R
> where T.account_number = R.account_number
> and S.customer_name = T.customer_name)
> );
Ah. In current sources I get correct (I think) results:
customer_name
----------------------
Johnson
(1 row)
customer_name
----------------------
Hayes
Jones
Lindsay
Smith
Turner
(5 rows)
Your query is running into the same bug someone else discovered a few
days ago:
2001-05-08 15:47 tgl
* src/backend/executor/: nodeAppend.c, nodeSubqueryscan.c
(REL7_1_STABLE), nodeAppend.c, nodeSubqueryscan.c: Append and
SubqueryScan nodes were not passing changed-parameter signals down
to their children, leading to misbehavior if they had any children
that paid attention to chgParam (most plan node types don't).
Append's bug has been there a long time, but nobody had noticed
because it used to be difficult to create a query where an Append
would be used below the top level of a plan; so there were never
any parameters getting passed down. SubqueryScan is new in 7.1 ...
and I'd modeled its behavior on Append :-(
The fix will be in 7.1.2 (due out real soon now), or you can pull
current REL7_1_STABLE sources from our CVS server if you are in a hurry.
Thanks for the report!
regards, tom lane
