On Sat, Oct 14, 2006 at 03:42:48PM -0400, Tom Lane wrote:
> David Fetter <david@fetter.org> writes:
> > On Sat, Oct 14, 2006 at 11:35:12AM -0700, Joshua D. Drake wrote:
> >> What is the use case for a READ ONLY transaction?
>
> > It would be handy for things like pgpool and Continuent, which could
> > reliably distinguish up front the difference between a transaction
> > that can write and one that can safely be sliced up and dispatched to
> > read-only databases.
>
> I don't think that works for PG's interpretation of READ ONLY, though.
> IIRC we let a "read only" transaction create and modify temp tables.
Am I missing something then?
test=> BEGIN READ ONLY;
BEGIN
test=> CREATE TEMPORARY TABLE foo (x integer);
ERROR: transaction is read-only
--
Michael Fuhr