Re: regress test db

On Tue, 13 Nov 2001 21:27:12 -0500 (EST)
Vince Vielhaber <vev(at)michvhf(dot)com> wrote:

> AFAIK the regress test database is working again. If anyone's having
> a problem with it let me know.

Hi, Vince


Not a problem, but...

I pondered whether the several examples, which is used nested
EXCEPT/INTERSECT/UNION among three SELECTs at the "Operator precedence
and (((((extra))))) parentheses" in the ../regress/sql/union.sql,
have the case like a query returning an incorrect result in v7.1.*
(e.g. See below) or not. Finally, the reconfirmation may lead me to
conclude that the examples can't reflect the behavior of the fixed
nested EXCEPT/INTERSECT(Thanks Tom). I wish there was something new
to be able to demonstrate the fixed.

SELECT q1 FROM int8_tbl
INTERSECT ALL
SELECT q2 FROM int8_tbl
INTERSECT
SELECT q2 FROM int8_tbl;

q1
-----
123 <--NG
(1 row)

SELECT q1 FROM int8_tbl
INTERSECT
SELECT q2 FROM int8_tbl
INTERSECT
SELECT q2 FROM int8_tbl;

q1
-----
123 <--NG
(1 row)

SELECT q1 FROM int8_tbl
INTERSECT ALL
SELECT q2 FROM int8_tbl
EXCEPT
SELECT q2 FROM int8_tbl WHERE q2 < 1000;

q1
----
(0 rows) <--NG

SELECT q1 FROM int8_tbl
INTERSECT
SELECT q2 FROM int8_tbl
EXCEPT ALL
SELECT q2 FROM int8_tbl WHERE q2 < 1000;

q1
----
(0 rows) <--NG

TIA,
Masaru Sugawara