From:	Martin Stromberg <Martin DOT Stromberg AT lu DOT erisoft DOT se>
Message-Id:	<200003201649.RAA27339@lws256.lu.erisoft.se>
Subject:	Re: Unnormals???
To:	eliz AT is DOT elta DOT co DOT il (Eli Zaretskii)
Date:	Mon, 20 Mar 2000 17:49:49 +0100 (MET)
Cc:	djgpp-workers AT delorie DOT com (DJGPP-WORKERS)
In-Reply-To:	<Pine.SUN.3.91.1000320183221.26722P-100000@is> from "Eli Zaretskii" at Mar 20, 2000 06:34:32 PM
X-Mailer:	ELM [version 2.5 PL3]
MIME-Version:	1.0
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

> 
> 
> On Mon, 20 Mar 2000, Martin Stromberg wrote:
> 
> > So you, Eli, are saying that if we have a NaN we should print "nan"
> > even if the "+" flag is present?
> 
> Yes.  Is something wrong with that?

No, not if we decide that NaNs do not have a sign. Hence we'll never
print "-nan", only "nan" whatever the bitpattern is if it's a NaN.

> But I don't mind the current compromise, either.  Too bad it seems to be 
> against the standard.  But it seems that, amazingly enough, the standard 
> doesn't fit well to what Intel processors do, so perhaps we'd elect to 
> deviate from the standard on this one.

Perhaps the n-char-sequence shall be used to show the bitpattern of
the NaN? Then we would see the sign if we knew what bit it is...


Right,

							MartinS

- Raw text -