On Thu, 2008-02-28 at 14:12 -0500, Robin Norwood wrote: > I assume we want all these operations to be atomic, and fail the > entire transaction if part of it doesn't work? Totally. Richard.