Date: Thu, 16 Mar 2000 13:06:12 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
To: Martin Stromberg <Martin DOT Stromberg AT lu DOT erisoft DOT se>
cc: djgpp-workers AT delorie DOT com
Subject: Re: Unnormals???
In-Reply-To: <200003161038.LAA26422@lws256.lu.erisoft.se>
Message-ID: <Pine.SUN.3.91.1000316125955.3856B-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: dj-admin AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk


On Thu, 16 Mar 2000, Martin Stromberg wrote:

> If the FPU treats them as nans, aren't they nans? Why are you saying
> they aren't nans?

A NaN has a specific bit pattern, which these numbers lack.  So calling 
them NaNs would be misleading; in particular, library function `isnan' 
might as well return zero for these numbers (I didn't check, though).

> I suggest we print "nan(unnormal)" or "nan(unnormal0x<bit pattern>)"
> where <bit pattern> is the bits of the double float in hexadecimal.

I'm still not sure this is allowed, but if it is, I think that
"nan(0x<bit pattern>)" is enough.