From octave-maintainers-request at bevo dot che dot wisc dot edu  Tue Nov 24 17:31:20 1998
Subject: Unidentified subject!
From: "John W. Eaton" <jwe at bevo dot che dot wisc dot edu>
To: john <john at arrows dot demon dot co dot uk>
Cc: octave-maintainers at bevo dot che dot wisc dot edu
Date: Tue, 24 Nov 1998 17:29:57 -0600 (CST)

On 24-Nov-1998, john <john at arrows dot demon dot co dot uk> wrote:

| To: octave-maintainers at bevo dot che dot wisc dot edu
| Cc: john at arrows dot demon dot co dot uk
| Subject: airy(real) should return a real
| 
| Bug report for Octave 2.0.13.93 configured for i586-pc-linux-gnulibc1
| 
| Description:
| -----------
| 
| airy([0123], real )            should return an exact real
| airy([0123], positive-real,1)  should return an exact real
| 
| Repeat-By:
| ---------
| 
| octave:1> format long
| octave:2> airy(-5)
| ans = 3.50761009024113e-01 - 7.21326062346773e-16i
| octave:3> 
| 
| or
| 
| octave:5> gplot ([linspace(-10,1) ; airy(linspace(-10,1))]')
| Warning: empty x range [0:0], adjusting to [-1:1]
| Warning: empty y range [0:0], adjusting to [-1:1]
| 
| 
| Fix:
| ---
| 
| a cryptic line like
| 
|   if (zi==0.0 && ( !scaled  || zr > 0.0 ))   ai = 0.0
| 
| in subroutines airy and biry in lo-specfun.cc

I made this change.  If there are other cases where the Bessel
functions should produce real results, please let me know.  It's
unfortunate that the Fortran subroutines don't include the check
themselves.

Thanks,

jwe