| From: | Michael Fuhr <mike(at)fuhr(dot)org> |
|---|---|
| To: | Ralf Wiebicke <ralf(dot)wiebicke(at)exedio(dot)com> |
| Cc: | pgsql-general(at)postgresql(dot)org |
| Subject: | Re: in failed sql transaction |
| Date: | 2006-09-24 23:40:09 |
| Message-ID: | 20060924234009.GA43233@winnie.fuhr.org |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general |
On Sun, Sep 24, 2006 at 12:03:59PM +0200, Ralf Wiebicke wrote:
> I just realized the following behaviour in postgresql: when I violate any
> constraint (unique constraint in my case) then the transaction is not usable
> anymore. Any other sql command returns a "in failed sql transaction" error.
Transactions are all-or-nothing: all statements must succeed or the
transaction fails (but see below regarding savepoints).
> All other databases I used up to now just ignore the statement violating the
> constraint, but leave the transaction intact.
Which databases behave that way? Does COMMIT succeed even if some
statements failed?
> Is this intended behaviour or rather a bug? Or is there any way to "switch on"
> the behaviour I'd like to see?
This is intended behavior. You can use savepoints to roll back
part of a transaction so the transaction can continue after an
error.
--
Michael Fuhr
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Markus Grabner | 2006-09-25 00:56:47 | query rewrite rules for updateable views? |
| Previous Message | Bob Pawley | 2006-09-24 23:31:20 | Re: serial column |