From octave-maintainers-request at bevo dot che dot wisc dot edu  Thu Feb  1 22:00:32 2001
Subject: Re: ISO C++ and Octave
From: "John W. Eaton" <jwe at bevo dot che dot wisc dot edu>
To: Mumit Khan <khan at NanoTech dot Wisc dot EDU>
Cc: octave-maintainers at bevo dot che dot wisc dot edu, khan@xraylith.wisc.edu
Date: Thu, 1 Feb 2001 22:00:26 -0600

On  1-Feb-2001, Mumit Khan <khan at NanoTech dot Wisc dot EDU> wrote:

| On Thu, 1 Feb 2001, John W. Eaton wrote:
| 
| > Nope.  That looks pretty bad.  Sort of like
| > 
| >   octave:6> sin (pi)
| >   ans =  1.2246e-16
| > 
| > (would it really hurt to make this a special case?).
| 
| I must confess that I don't have a opinion either way that I
| can back up. On one hand, we certainly can't expect exact
| precision from finite precision numerics, but on the other
| hand, users of software such as Octave may have come to expect
| certain "nice-ness". Don't know ...

I don't plan to add special cases like this to Octave.  I was just
wondering out loud why if we define M_PI, the libraries don't check
for a few special cases.  I don't see how it would hurt anything, but
maybe someone knows better.  Perhaps because it would be tougher to
guarantee that

  sin (N*pi) == 0 for all N = 0, 1, 2, ...

jwe