From bug-octave-request at bevo dot che dot wisc dot edu Wed Jan 22 14:15:32 2003
Subject: Trivial 0ctave bug
From: "John W. Eaton" <jwe at bevo dot che dot wisc dot edu>
To: "Richard E. Borcherds" <reb at Math dot Berkeley dot EDU>
Cc: bug-octave at bevo dot che dot wisc dot edu
Date: Wed, 22 Jan 2003 14:15:30 -0600

On 18-Jan-2003, Richard E. Borcherds <reb at Math dot Berkeley dot EDU> wrote:

| Octave version 2.1.34 incorrectly gives __realmin__ the value
| 2^-1022 on machines that support IEEE 754 64 bit arithmetic.
| The correct value should be 2^-1074
| because IEEE 754 has "subnormal"numbers. The documentation
| for help __realmin__ needs fixing in the same way. (So does the numerical
| analysis textbook I am using...)

I'm not sure that this is a bug in Octave.  The variable realmin is
set in src/data.cc using the C macro DBL_MIN.  What do you suggest as
a portable fix?

BTW, Matlab also seems to behave this way, so any fix is likely to
break compatibility.

jwe



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------