| From: | Josh Berkus <josh(at)agliodbs(dot)com> |
|---|---|
| To: | Betsy Barker <betsy(dot)barker(at)supportservicesinc(dot)com>, pgsql-novice(at)postgresql(dot)org |
| Subject: | Re: pgsql functions and transactions? |
| Date: | 2004-08-27 04:46:01 |
| Message-ID: | 200408262146.01453.josh@agliodbs.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-novice |
Betsy,
> I read the posts on pgsql functions and transactions from the Novice list,
> and I think they are saying "PGSQL functions do not support transactions".
> Is this correct? Or is the idea that functions are automatically in a
> transaction?
That's right. Soon (8.0 or 8.1) functions will support *sub-transactions*,
or savepoints, but that still won't do a checkpoint and synch, which is what
you need.
In my experience, there is a limit to the amount of processing you can
reasonably do in a single function in Postgres because of the need to stop
and synch (and possibly VACUUM). I often have "series" of functions (in
one case, about 18) which are executed in succession by a Perl script.
We've been discussing PROCEDURES on -hackers which are non-transactional (and
thus can contain several transactions). But nobody is coding this yet.
--
Josh Berkus
Aglio Database Solutions
San Francisco
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Noel Faux | 2004-08-27 08:31:15 | Re: Foreign keys |
| Previous Message | Tom Lane | 2004-08-27 03:09:12 | Re: pgsql functions and transactions? |