Re: Aggregate-function space leakage

Greg Stark <gsstark(at)mit(dot)edu> writes:
> Rereading your diagnosis of Merlin Moncure's original problem I'm a
> bit puzzled. Why do we have to rerun the final function when we rescan
> the hash table? Surely the logical thing to do is to store the final
> value in the hash table with some flag saying that value has been
> finalized rather than to reexecute the final function every time it's
> rescanned.

In the normal case where we're not going to do a rescan, this would very
likely make things slower because we'd have to make a never-used extra
copy of the function's output. It might be worth doing if we knew we
were likely to get rescanned; but I'm not eager to have two
significantly different operational modes for that. nodeAgg is
complicated enough already...

regards, tom lane