Re: operation of 'round' function

[David Kastrup]

ronan.waide@euroconex.com (Ronan Waide) writes:

> On February 7, schwab@suse.de said:
> > 0.5 is exactly halfway between 0 and 1, so the additional rule is that in
> > case of a tie the even number is used.
> 
> Um.
> 
> That makes no sense whatsover.

It does not introduce a bias by rounding, which rounding up always
does.  It does round away from ambiguous values.  That way the
floating point equality
a+b-b = a+b-b+b-b
holds.  It wouldn't with rounding up.

It is the numerically preferred rounding method.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum
Email: David.Kastrup@t-online.de