Kevin Grittner wrote:
> contexts. I don't think the updates to the system tables have the
> same magnitude of performance hit as creating these tables, especially
> if write barriers are on.
>
Wouldn't it be cleaner just to defer creation of real files to support the
structures associated with a temp table until it i snecessary to spill the
data from the backend's RAM? This data doesn't need to be in
shared memory and the tables and data aren't visible to any other
session, so can't they run out of RAM most of the time (or all the
time if the data in them is short lived)?