Possible bug concerning LASTOID in 7.2

From: Coax <coax(at)shell1(dot)cornernet(dot)com>
To: <pgsql-bugs(at)postgresql(dot)org>
Subject: Possible bug concerning LASTOID in 7.2
Date: 2002-02-22 22:47:11
Message-ID: Pine.LNX.4.30.0202221644510.16008-200000@shell1.cornernet.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

Hiya! I think i've found a bug in postgres 7.2.

I have tested this on 1 box, running redhat 7.2 w/ both pg7.1.2 and pg7.2.

I've attached a dumpfile that allows you to execute this query:

INSERT INTO forum (group_forum_id,posted_by,subject,body,date,is_followup_to,thread_id,most_recent_date) VALUES ('88760', '192470', 'asdfasdfsqdf', 'blah', '1014405669','0','625864','1014405669')

Here's the output of PG7.1.2:
test=# INSERT INTO forum (group_forum_id,posted_by,subject,body,date,is_followup_to,thread_id,most_recent_date) VALUES ('88760', '192470', 'asdfasdfsqdf', 'blah', '1014405669','0','625864','1014405669');
INSERT 80942390 1
test=# \echo :LASTOID
80942390

However, here's the output from PG7.2:

test=# INSERT INTO forum (group_forum_id,posted_by,subject,body,date,is_followup_to,thread_id,most_recent_date) VALUES ('88760', '192470', 'asdfasdfsqdf', 'blah', '1014405669','0','625864','1014405669')
test-# ;
UPDATE 0
test=# \echo :LASTOID
0

(Note: No OID, and actually returns the results from the UPDATE, and not
that of the INSERT!!!)

So, here's the problem:

My application looks for the OID of the record it just inserted, with PHP's
pg_getlastoid() function.

With postgres 7.1.2, it works great. However, with postgres 7.2, the function
returns 0 (because postgres returns 0!) - and screws up the application.

In the dumpfile, there are some rules generated:

CREATE RULE forum_insert_agg AS ON INSERT TO forum DO UPDATE forum_agg_msg_count SET count = (forum_agg_msg_count.count + 1) WHERE (forum_agg_msg_count.group_forum_id = new.group_forum_id);

CREATE RULE forum_delete_agg AS ON DELETE TO forum DO UPDATE forum_agg_msg_count SET count = (forum_agg_msg_count.count - 1) WHERE (forum_agg_msg_count.group_forum_id = old.group_forum_id);

These rules seem to trigger the behavior. it LOOKS like postgres is returning
the UPDATE procedure's results, rather than the INSERT results, which is what
i need it to return. (if the rules arent present, postgres returns the right
thing.)

Is this a bug?

(LASTOID is questionable here, because the rules in question really DID
generate an UPDATE sql, and run it... Therefore, the return, depending
on one's interpretation, could definitely be correct.)

Thanks for whatever help you might be able to provide.

FOOTNOTE: I'm not a dba. I'm just a sysadmin. So, if you need any further
information, please, do not hesitate to ask. (just dont know what to present
to start with.)

Thanks.

Chad Schwartz
CornerNet System Administration

Attachment Content-Type Size
testcase.dump text/plain 4.4 KB

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message pgsql-bugs 2002-02-22 23:07:56 Bug #601: libpq needs a lo_truncate() or INV_TRUNC
Previous Message Peter Eisentraut 2002-02-22 22:04:44 Re: CURRENT_TIMESTAMP not work correctly insinde a transaction.