Message-Id: <3.0.6.32.19990228075720.008a8100@pop.globalserve.net>
X-Sender: derbyshire AT pop DOT globalserve DOT net
X-Mailer: QUALCOMM Windows Eudora Light Version 3.0.6 (32)
Date: Sun, 28 Feb 1999 07:57:20 -0500
To: egcs AT egcs DOT cygnus DOT com, djgpp AT delorie DOT com
From: Paul Derbyshire <pderbysh AT usa DOT net>
Subject: Re: Bug in libm or libstdc++.
In-Reply-To: <Pine.SUN.3.91.990228142826.5950V-100000@is>
References: <3 DOT 0 DOT 6 DOT 32 DOT 19990226012010 DOT 008e7bd0 AT pop DOT globalserve DOT net>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Reply-To: djgpp AT delorie DOT com

At 02:28 PM 2/28/99 +0200, you wrote:
>
>On Fri, 26 Feb 1999, Paul Derbyshire wrote:
>
>> You're wrong, I'm right. Long double versions of atan and friends ARE
>> demanded by the standard.

>Which standard is that?

The ISO C++ standard. The C++ compiling environment is what was under
discussion.

>libm.a which comes with DJGPP v2.02 is not a C++
>library, it's a C library, and AFAIK it conforms to current C
>standards.

Then it must be libstdc++ that is supposed to overload those functions for
floats and long doubles. As the subject says, the bug was in one of those
libraries, and your clarification has narrowed it down to libstdc++.


-- 
   .*.  "Clouds are not spheres, mountains are not cones, coastlines are not
-()  <  circles, and bark is not smooth, nor does lightning travel in a
   `*'  straight line."    -------------------------------------------------
        -- B. Mandelbrot  |http://surf.to/pgd.net
_____________________ ____|________     Paul Derbyshire     pderbysh AT usa DOT net
Programmer & Humanist|ICQ: 10423848|