On Fri, Sep 03, 2010 at 01:28:15PM -0400, Alvaro Herrera wrote:
> Excerpts from David Fetter's message of vie sep 03 10:21:00 -0400 2010:
>
> > > How about using a SAVEPOINT before the cast? Wouldn't fail your
> > > transaction..
> >
> > For unattended operation, there are some issues:
> >
> > * Generating appropriate SAVEPOINT names
> > * Keeping track of same
> > * Detecting errors
> > * Issuing ROLLBACKs to the aforementioned SAVEPOINT
> >
> > None of this works super well for a bulk load.
>
> Why do you need any of this? The logic is pretty simple:
>
> SAVEPOINT my_savepoint;
> select ...::regclass;
>
> -- if it works
> RELEASE my_savepoint;
>
> -- if it fails
> ROLLBACK TO my_savepoint;
> RELEASE my_savepoint;
>
> Granted, it's not super-performant ...
We have no mechanism to do this, or any other check, during a bulk
load. It'd be a great feature to have :)
Cheers,
David.
--
David Fetter <david@fetter.org> http://fetter.org/
Phone: +1 415 235 3778 AIM: dfetter666 Yahoo!: dfetter
Skype: davidfetter XMPP: david.fetter@gmail.com
iCal: webcal://www.tripit.com/feed/ical/people/david74/tripit.ics
Remember to vote!
Consider donating to Postgres: http://www.postgresql.org/about/donate