>>
>>Of course... that formula you gave has a risk of numeric overflow...
>
>The formula I gave was
>

>N = -N
>

>
>I find one exception where the negative value results in the same value.
>
>I'm not convinced overflow comes in
>
>(1) the value is -2^31 and (2) you force the value to be contained in a 4 byte signed integer
>

>n = 0x80000000
>?'n = ', n	&& 2147483648
>?'-n = ', -n	&& -2147483648
>n = bitor(n, 0) && force a 4 byte signed integer
>?'bitor = ', n	&& -2147483648
>x = -n
>?'-n = ', x	&& -2147483648
>?'abs', abs(n)	&& -2.1474E+9
>
>? ' x == -x ', x == -x  && TRUE
>
>

>
>Update - and yes - I know why

Sorry. Wasn't paying attention to who I was responding to. I should've made it clear that I was meaning the formula as given by the original poster:

N = N - 2*N

was the one with the problem.

Think I'll need more caffeine (opens fresh can of coffee -- takes spoon and begins munching away -- too impatient to wait for brewing batch -- besides, why dilute it with water?)