Sequences in transaction context

From: "Erik Pearson" <erik(at)cariboulake(dot)com>
To: pgsql-general(at)postgresql(dot)org
Subject: Sequences in transaction context
Date: 2001-06-06 02:22:43
Message-ID: 9fk43l$2s8m$1@news.tht.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-general

I searched through mailing list archives but was unable to find full
coverage of this question -- my apologies if this is a reposted question.

As in the FAQ, I am trying to retrieve the value of a sequence value from a
newly inserted row. So, first I call something like:

insert into foobar (foo, bar)
values (nextval('foobar_foo_seq'), 'whatever');

Then, I want to retrieve the value that generated from the sequence and
inserted into the table, so I use a call to currval:

insert into foobar_rel_table(foo_fk, baz)
values (currval('foobar_foo_seq', 'something else');

This is (one of the methods that is) prescribed in the FAQ. However, I'm
concerned that another transaction attempting to insert into the same table
might make a call to nextval('foobar_foo_seq') between the two operations
above. This would mean that my second statement would use the wrong value
from the sequence.

I've tested this scenario with different transaction isolation levels, and
it appears that any state changes to sequences become immediately visible to
other transactions (obviously, a read-commited type strategy wouldn't work,
however, serializing access to sequences, or explicit locking would solve
this problem).

Has anyone else come across this problem, and is there a workaround? If
not, are there any alternate suggestions for generating a PK on insert and
immediately retrieving it that is free from concurrency issues?

Any help would be appreciated. Thanks! -- Erik

--
Erik Pearson
erik(at)cariboulake(dot)com
http://www.cariboulake.com

Responses

Browse pgsql-general by date

  From Date Subject
Next Message Doug McNaught 2001-06-06 02:24:14 Re: ORDER BY Problem
Previous Message Eric G. Miller 2001-06-06 01:23:54 Re: Vacuum problems ... HELP !!!