On K, 2005-06-01 at 11:31 -0400, Tom Lane wrote:
> Hannu Krosing <hannu@skype.net> writes:
> > I think this should be a decision done when creating a table, just like
> > TEMP tables. So you always know if a certain table is or is not
> > safe/replicated/recoverable.
> > This has also the advantage of requiring no changes to actual COPY and
> > INSERT commands.
>
> That doesn't seem right to me; the scenario I envision is that you are
> willing to do the initial data loading over again (since you presumably
> still have the source data available). But once you've got it loaded
> you want full protection.
What I mean, was that as it can't be safely replicated using log-
shipping, It should be visible as such.
> Perhaps it could work to use an ALTER TABLE command to flip the state.
No. It would be the same as flipping a TEMP table to an ordinary table,
which we don't support, and IMHO for a good reason
> But I'm not really seeing the point compared to treating it as a COPY
> option.
The point is having a separate (sub)type of storage - non-WAL/non-
replicated table and its indexes.
> I do not believe that anyone needs this to work on individual
> INSERT commands --- if you are after max speed, why aren't you using
> COPY? And treating it as an ALTER property opens the possibility of
> forgetting to ALTER the table back to normal behavior, which would be
> a foot-gun of large caliber indeed :-(
That's what I'm trying to avoid - If it is obvious, that the whole table
is quasi-stable (in PITR/log-shipping sense) it is more clearly a user
choice what kinds of data can be stored there. Same as TEMP tables
again.
--
Hannu Krosing <hannu@tm.ee>
