Date: Wed, 20 Jan 1999 09:02:31 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
To: Kbwms AT aol DOT com
cc: moshier AT mediaone DOT net, djgpp-workers AT delorie DOT com
Subject: Re: Bug when printing long doubles
In-Reply-To: <58ca22ce.36a4c038@aol.com>
Message-ID: <Pine.SUN.3.91.990120090005.2569A-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp-workers AT delorie DOT com


On Tue, 19 Jan 1999 Kbwms AT aol DOT com wrote:

> I see nothing wrong with printing an item of poorly formed bits as NaN.
> In the final analysis, that's what it is, isn't it?

No, they aren't, at least not as far as the x87 processor is concerned.  
NaNs have sevral well-defined bit patterns, which this example doesn't 
fit.  When the x87 FPU meets a number such as the one in Robert's 
example, the bits in the x87 status and tag words are NOT set as they are 
when it sees a NaN.