develooper Front page | perl.perl5.porters | Postings from August 2001

Re: inf handling in Perl incomplete

Thread Previous | Thread Next
From:
Kurt D. Starsinic
Date:
August 8, 2001 15:19
Subject:
Re: inf handling in Perl incomplete
Message ID:
20010808151839.D20088@wolfetech.com
On Wed, Aug 08, 2001 at 11:26:42PM +0200, Tels wrote:
> On 08-Aug-01 Kurt D. Starsinic tried to scribble about:
> > On Wed, Aug 08, 2001 at 09:11:56AM +0100, Piers Cawley wrote:
> >> Jarkko Hietaniemi <jhi@iki.fi> writes:
> >> > While that someone is at it, "NaN" also needs fixing.  In there one
> >> > needs to move more carefully, since there usually are two kinds:
> >> > "signaling" and "quiet" NaNs.  The first kind gives SIGFPEs, the
> >> > second doesn't.
> >> > 
> >> >    $x = NaN;
> >> >    print "goofy\n" if 0/0 == NaN;  # should always print

    My reading of IEEE is that 0/0 is NaN, and semantically, NaN != NaN, so
the above should never print.

> >> >    print "spooky\n" if NaN == NaN; # should never print
> >>     print "ooky\n" if NaN == 1;     # should never print. But does.
> > 
> >     I don't believe that the Perl documentation claims that the bareword
> > `NaN' becomes an IEEE NaN in numeric context.  It just suggests that
> > an IEEE NaN is `NaN' in string context.  Do you believe otherwise,
> > or are you proposing a change?
> 
> Incidentily, up until today (well, 2 hours ago), I believed and would have
> swonr that 'NaN' becomes an IEEE NaN. Oups. Can (read: is it possible) to
> change that?

    Anything's possible.  I'm not in favor of it, because it will surprise
too many numerically-casual people who expect any string that doesn't look
like an everyday number to be numerically zero.

> Problem arise also with use Math::BigInt :constant; It won't touch NaN, nor
> inf because these are strings (barewords), and are thus not integer
> constants. But Math::BigInt->new('inf') works fine, so :constant could
> (should) cover them. Or so I think.

    Correct on both counts, but I wouldn't characterize the first as a
problem.  The overloading mode of Math::BigInt is designed to be an easy
upgrade to an already-working script.  If you have a script that has bareword
Nan and inf in it, it's already broken.

    I don't want to see bareword NaN and Inf upgrade automagically.  I think
it's more cruft, and there isn't just one NaN anyway.  I wouldn't mind a
pragma that made, e.g., 0/0 or sqrt(-1) return the appropriate NaN instead
of throwing an exception, though.

    - Kurt


Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About