Re: BUG #4653: zero with negative sign returned on round(sin) function

"Marco Vieira" <maovieira@gmail.com> writes:
> If I query select round(sin(2.0*pi()*0.51)) I get "-0" as return but zero is
> unsigned.

In IEEE-standard float arithmetic, that isn't true --- zero and minus
zero are distinguishable values.  This is not a bug but just the way
your platform chooses to define the result of rounding a small
negative value.

(Or at least, if you are convinced it's a bug, take it up with the
glibc authors not us.)

            regards, tom lane