Re: SQL SAVEPOINT command ON ROLLBACK option
- From: Jonathan Bailey <jonathan.bailey@xxxxxxxxx>
- Date: Fri, 03 Aug 2007 04:24:15 -0700
On Aug 2, 4:43 pm, "walker.l2" <walker...@xxxxxxxxxxxxxx> wrote:
Hmm, maybe.
I'm worried that neither option seems to fully release SQL resources.
If my program hits an SQL error, I can rollback to the savepoint okay,
but will just doing that release all the SQL resources my program
might have open at the point of the initial error? (Or are these SQL
resources released automatically by the OS when my program ends?)
I suppose I could look into BEGIN ATOMIC ... END instead, but at first
glance SAVEPOINT seemed the simpler solution.
Surely to fully release resources you have to end the work unit with a
final commit, or rollback with no savepoint specified?
It does seem odd that you must specify 'on rollback retain cursors'
when adding a savepoint unless there is a plan to allow specification
without this sometime in the future. I'm on V5.3 .
Jonathan
.
- Follow-Ups:
- Re: SQL SAVEPOINT command ON ROLLBACK option
- From: walker.l2
- Re: SQL SAVEPOINT command ON ROLLBACK option
- References:
- Re: SQL SAVEPOINT command ON ROLLBACK option
- From: Kent Milligan
- Re: SQL SAVEPOINT command ON ROLLBACK option
- From: walker.l2
- Re: SQL SAVEPOINT command ON ROLLBACK option
- Prev by Date: Re: difference between 10 and 15 cpi
- Next by Date: Re: command syntax
- Previous by thread: Re: SQL SAVEPOINT command ON ROLLBACK option
- Next by thread: Re: SQL SAVEPOINT command ON ROLLBACK option
- Index(es):
Relevant Pages
|