On Mon, Aug 20, 2012 at 09:33:45PM -0700, Jeff Davis wrote:
> On Mon, 2012-08-20 at 19:17 -0400, David Johnston wrote:
> > Ideally the decision of whether to do so could be a client
> > decision. Not storing intra-transaction changes is easier than
> > storing all changes. At worse you could stage up all changed then
> > simply fail to store all intermediate results within a given
> > relation. It that case you gain nothing in execution performance
> > but safe both storage and interpretative resources. So the
> > question becomes is it worth doing without the ability to store
> > intermediate results? If you were to ponder both which setup
> > would the default be? If the default is the harder one (all
> > statements) to implement then to avoid upgrade issues the syntax
> > should specify that it is logging transactions only.
>
> I think the biggest question here is what guarantees can be offered?
> What if the transaction aborts after having written some data, does
> the audit log still get updated?
There are definitely use cases for this, but until we have autonomous
transactions, a totally separate project, I don't think we should
attempt them in the first version.
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