Hello,
On Tue, 20 Sep 2022, Aldy Hernandez wrote:
> > FWIW, in IEEE, 'abs' (like 'copy, 'copysign' and 'negate') are not
> > arithmetic, they are quiet-computational. Hence they don't rise
> > anything, not even for sNaNs; they copy the input bits and appropriately
> > modify the bit pattern according to the specification (i.e. fiddle the
> > sign bit).
> >
> > That also means that a predicate like negative_p(x) that would be
> > implemented ala
> >
> > copysign(1.0, x) < 0.0
>
> I suppose this means -0.0 is not considered negative,
It would be considered negative if the predicate is implemented like
above:
copysign(1.0, -0.0) == -1.0
But really, that depends on what _our_ definition of negative_p is
supposed to be. I think the most reasonable definition is indeed similar
to above, which in turn is equivalent to simply looking at the sign bit
(which is what copysign() does), i.e. ...
> though it has
> the signbit set? FWIW, on real_value's real_isneg() returns true for
> -0.0 because it only looks at the sign.
... this seems the sensible thing. I just wanted to argue the case that
set_negative (or the like) which "sets" the sign bit does not make the
nan-ness go away. They are orthogonal.
> > deal with NaNs just fine and is required to correctly capture the sign of
> > 'x'. If frange::set_nonnegative is supposed to be used in such contexts
> > (and I think it's a good idea if that were the case), then set_nonnegative
> > does _not_ imply no-NaN.
> >
> > In particular I would assume that, given an VAYRING frange FR, that
> > FR.set_nonnegative() would result in an frange {[+0.0,+inf],+nan} .
>
> That was my understanding as well, and what my original patch did.
> But again, I'm just the messenger.
Ah, I obviously haven't followed the thread carefully then. If that's
what it was doing then IMO it was the right thing.
Ciao,
Michael.
