Re: BUG #3945: unexpected ON INSERT rule behaviour

The following bug has been logged online:

Bug reference: 3945
Logged by: Holger Klawitter
Email address: info(at)klawitter(dot)de
PostgreSQL version: 8.2.6
Operating system: Linux/i386
Description: unexpected ON INSERT rule behaviour
Details:

Well,
this is probably not really a bug, more a feature
deeply buried in the query-tree-concept worth placed as a pitfall warning in
the documentation :-)

The following code:
CREATE TABLE a (a int);
CREATE TABLE b (b int);

CREATE OR REPLACE RULE a_to_b
AS ON INSERT TO a
DO ALSO INSERT INTO b VALUES (NEW.a);

INSERT INTO a VALUES ( 1 );
INSERT INTO a VALUES ((SELECT max(a)+1 from a));

SELECT * from b;

Produces the following output:
b
---
1
3 <---- 2 expected here
(2 rows)

One would expect 1 and 2 to be stored in b.

The problem is that NEW.a is not the *value* being inserted. Instead, NEW
reproduces the unevaluated
selected statement.

Regards
Holger

On Feb 8, 2008 2:20 PM, Holger Klawitter <info(at)klawitter(dot)de> wrote:

>
> The following bug has been logged online:
>
> Bug reference: 3945
> Logged by: Holger Klawitter
> Email address: info(at)klawitter(dot)de
> PostgreSQL version: 8.2.6
> Operating system: Linux/i386
> Description: unexpected ON INSERT rule behaviour
> Details:
>
> Well,
> this is probably not really a bug, more a feature
> deeply buried in the query-tree-concept worth placed as a pitfall warning
> in
> the documentation :-)
>

You are correct, it's not a bug, but a feature request that won't be
entertained.

postgres=# explain INSERT INTO a VALUES ( 1 );
QUERY PLAN
------------------------------------------
Result (cost=0.00..0.01 rows=1 width=0)

Result (cost=0.00..0.01 rows=1 width=0)
(3 rows)

postgres=# explain INSERT INTO a VALUES ((SELECT max(a)+1 from a));
QUERY PLAN
-------------------------------------------------------------------
Result (cost=40.01..40.02 rows=1 width=0)
InitPlan
-> Aggregate (cost=40.00..40.01 rows=1 width=4)
-> Seq Scan on a (cost=0.00..34.00 rows=2400 width=4)

Result (cost=40.01..40.02 rows=1 width=0)
InitPlan
-> Aggregate (cost=40.00..40.01 rows=1 width=4)
-> Seq Scan on a (cost=0.00..34.00 rows=2400 width=4)
(9 rows)

The docs are pretty clear on this; Rules are applied at parse time (' The
rule system is located between the parser and the planner'), and hence can
only work with hard-coded values in the VALUES clause.

Personally, although much slower, I prefer using triggers if the logic
depends on NEW/OLD values. Rules are perfect fit only if
1) Your operations do not depend on NEW/OLD pseudo relations. For eg., plain
unconditional INSERT/UPDATE/DELETE on another relation.
2) You are *absolutely* sure that no app will use prepared statements, and
will always provide values in the statements, and not something like you did
above.

Best regards,
--
gurjeet[(dot)singh](at)EnterpriseDB(dot)com
singh(dot)gurjeet(at){ gmail | hotmail | indiatimes | yahoo }.com

EnterpriseDB http://www.enterprisedb.com

17° 29' 34.37"N, 78° 30' 59.76"E - Hyderabad
18° 32' 57.25"N, 73° 56' 25.42"E - Pune
37° 47' 19.72"N, 122° 24' 1.69" W - San Francisco *

http://gurjeet.frihost.net

Mail sent from my BlackLaptop device