On Mon, 2007-02-26 at 22:56 +0000, Simon Riggs wrote:
> Proposal: Implement a new option for COMMIT, for enhancing performance,
> providing a MySQL-like trade-off between performance and robustness for
> *only* those that want it.
>
> COMMIT NOWAIT
>
> This form of COMMIT will *not* perform XLogFlush(), but will rely on a
> special background process to perform regular WAL fsyncs (see later).
>
I would think that the type of people who would want this feature would
want it to be a property of the connection, not per-statement. For one
thing, many applications use solitary SQL statements without surrounding
them in BEGIN/COMMIT.
Regards,
Jeff Davis