| From: | Michael Stone <mstone+postgres(at)mathom(dot)us> |
|---|---|
| To: | pgsql-performance(at)postgresql(dot)org |
| Subject: | Re: insert without oids |
| Date: | 2006-01-13 22:48:01 |
| Message-ID: | 20060113224800.GJ1408@mathom.us |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-performance |
On Fri, Jan 13, 2006 at 04:29:15PM -0500, Neil Conway wrote:
>There's really no additional operations required:
>INSERT INTO t2 VALUES (currval('t1_id_seq'), ...);
>You need a separate SELECT if you want to use the generated sequence
>value outside the database,
That would, of course, be the goal. IOW, if you have a table which has
data which is unique only for the serial column, the old syntax provided
a way to refer to the newly inserted row uniquely without any additional
operations.
>although the INSERT ... RETURNING extension will avoid that
That sounds promising. I'll have to put the TODO list on my todo list.
:)
Mike Stone
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Jim C. Nasby | 2006-01-13 23:34:54 | Re: Slow query with joins |
| Previous Message | Neil Conway | 2006-01-13 21:29:15 | Re: insert without oids |