Alvaro Herrera <alvherre@2ndquadrant.com> writes:
> I'm in favor of hash_mem_multiplier. I think a >1 default is more
> sensible than =1 in the long run, but if strategic vote is what we're
> doing, then I support the =1 option.
FWIW, I also think that we'll eventually end up with >1 default.
But the evidence to support that isn't really there yet, so
I'm good with 1.0 default to start with.
regards, tom lane