Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT cygwin DOT com>
List-Archive: <http://sourceware.org/ml/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-help AT cygwin DOT com>, <http://sourceware.org/ml/#faqs>
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
Delivered-To: mailing list cygwin AT cygwin DOT com
From: "Dave Korn" <dave DOT korn AT artimi DOT com>
To: "'sorry for the on-topic post'" <cygwin-talk AT cygwin DOT com>
Cc: <cygwin AT cygwin DOT com>
Subject: RE: gcc-3.3.3 :fpclassify
Date: Thu, 30 Dec 2004 22:59:24 -0000
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
In-Reply-To: <1104440185.41d46b7970546@imap-serv.inrialpes.fr>
Message-ID: <NUTMEGQbKCO2ev2zosG00000b47@NUTMEG.CAM.ARTIMI.COM>
X-OriginalArrivalTime: 30 Dec 2004 22:59:24.0281 (UTC) FILETIME=[34AA0E90:01C4EEC3]

> -----Original Message-----
> From: cygwin-talk-owner On Behalf Of Jianguo.Zhang
> Sent: 30 December 2004 20:56

> The gcc/g++ version is 3.3.3
> 
> I am trying to compile the following program using g++ 3.3.3

>          if (FP_ZERO != std::fpclassify (a)) abort ();        
>                    
>                                                               
>    if (FP_NAN != std::fpclassify (b)) abort ();               
>                    
>                                                         
>   if (FP_NORMAL != std::fpclassify (c)) abort ();             

> When I use the following command to compile this file under cygwin:
> 
> $ g++ test.cpp
> 
> the following errors occurs:
> 
> test.cpp: In function `int main()':
> test.cpp:17: error: `fpclassify' undeclared in namespace `std'
> test.cpp:18: error: `fpclassify' undeclared in namespace `std'
> test.cpp:19: error: `fpclassify' undeclared in namespace `std'
> 
> there is no problem to compile the file under linux system 
> using g++ 3.3.1
> 
> Could anyone tell me why?


  It changed in 3.3.3; in earlier versions it was indeed in namespace std.

  It is a macro, defined in /usr/include/math.h, which is included by cmath, but
then (since 3.3.3) cmath has a section that undefines it again; this is
apparently for strict c++ standard conformance.

  In gnu libstdc++, the macro is placed into namespace __gnu_cxx before it is
undefined, used to make a template function that will accept any type, and then
imported from there to std:: later.  This is under the control of two #defines,
_GLIBCPP_USE_C99 and _GLIBCPP_USE_C99_FP_MACROS_DYNAMIC.  Unfortunately, these
can't just be set at the top of your code before the #includes, as then cstdlib
goes and breaks.  These defines live in
/usr/include/c++/3.3.3/i686-pc-cygwin/bits/c++config.h, and they are created at
the time the compiler itself is being configured and built.

  The cygwin compiler, it seems, was compiled without using the --enable-c99
option at configure time, whereas the linux version you've used was.

http://gcc.gnu.org/ml/libstdc++/2001-04/msg00345.html
and follow ups, and particularly the referenced thread
http://gcc.gnu.org/ml/libstdc++/2001-04/msg00120.html
make it clear what's going on.

  There are various answers and workarounds.  One would be to recompile gcc
(perhaps just libstdc++ would be enough) for cygwin with --enable-c99.  Another
would be to copy the definitions into your own code from math.h.

  Gerrit, does the output from "gcc -v" for 3.4.1 show "--enable-c99"?  Or has
someone fixed it in the headers somehow?

    cheers, 
      DaveK
-- 
Can't think of a witty .sigline today....


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/