2010/1/23 Tom Lane :
> I was just poking at the test case provided by Allen Johnson in bug
> #5294. The essence of the complaint is that the planner is choosing
> a sort-and-GroupAggregate plan when a HashAggregate-and-then-sort
> plan would be faster, because the aggregation steps are roughly the
I was just poking at the test case provided by Allen Johnson in bug
#5294. The essence of the complaint is that the planner is choosing
a sort-and-GroupAggregate plan when a HashAggregate-and-then-sort
plan would be faster, because the aggregation steps are roughly the
same speed either way while
