On Wed, Jul 12, 2017 at 10:42 AM, Heikki Linnakangas <hlinnaka(at)iki(dot)fi> wrote:
>> FWIW, memtupcount is allowed to go negative.
>
>
> Huh, say what?
Well, there are defenses against that case, both in assertions and in
code like tuplesort_heap_delete_top(). In that sense, it's allowed to
go negative.
--
Peter Geoghegan