Euler Taveira de Oliveira <euler@timbira.com> writes:
> Em 24-08-2011 11:27, Tom Lane escreveu:
>> Hmm. I agree we need to avoid executing 0/0 here, but should we force
>> the result to 0, or to NaN?
> If it returns NaN on other platforms, let's be consistent.
I kinda suspect that the NaN behavior was not designed but accidental.
What I'm wondering is whether it's really the "right", sensible,
behavior.
On reflection I suspect it isn't --- it'd bollix sum() or avg()
calculations over the function's results, for instance. But now
I'm not sure zero is the right thing to put in, either.
regards, tom lane