X-Authentication-Warning: delorie.com: mailnull set sender to djgpp-workers-bounces using -f
From: Martin Str|mberg <ams AT ludd DOT luth DOT se>
Message-Id: <200201032140.WAA06212@father.ludd.luth.se>
Subject: Re: Function nan()
In-Reply-To: <1659-Thu03Jan2002212556+0200-eliz@is.elta.co.il> from Eli Zaretskii at "Jan 3, 2002 09:25:57 pm"
To: djgpp-workers AT delorie DOT com
Date: Thu, 3 Jan 2002 22:40:48 +0100 (MET)
Cc: Kbwms AT aol DOT com
X-Mailer: ELM [version 2.4ME+ PL54 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

According to Eli Zaretskii:
> K.B. Williams kindly agreed to work on adding to the DJGPP library
> long double math functions mandated by C9x.  One of the specifications
> of the standard says that the `nan' function accepts a text string
> which can modify the result to some extent.  But the standard is
> silent about the details, saying that the effect of that string is
> ``implementation-defined''.
> 
> I'm guessing that they allow to produce the different forms of NaN
> (QNaN etc.), but it would be nice to know what do other
> implementations do, and whether this is really the intent to begin
> with.
> 
> So could people please tell what they know about this, and perhaps
> look on other systems to see what they produce?

The standard says ``#include <math.h>
double nan(const char *tagp);
float nanf(const char *tagp);
long double nanl(const char *tagp);

Description

The call nan("n-char-sequence") is equivalent to
strtod("NAN(n-char-sequence)", (char**) NULL); the call nan("") is
equivalent to strtod("NAN()", (char**) NULL). If tagp does not point
to an n-char-sequence or an empty string, the call is equivalent to
strtod("NAN", (char**) NULL). Calls to nanf and nanl are equivalent to
the corresponding calls to strtof and strtold.

Returns

The nan functions returns a quiet NaN, if available, with content
indicated through tagp. If the implementation does not support quiet
NaNs, the functions return zero.''

Can tagp make a signaling NaN? If so it seems it must be forced into a
quiet one after conversion, I guess.


Another task needing to be done is make libm compile without warnings
while gcc.opt contains "-Wsign-compare" with a 3.x gcc. Please someone
who understands the code take a look.


Right,

						MartinS