PLplot

On Thursday, March 04, 2004, at 09:53AM, Andrew Ross <andrewross@...> wrote:

>On Thu, Mar 04, 2004 at 08:24:30AM +0000, Andrew Ross wrote:
>
>OK. A quick google search shows this to be a known bug in the cmath
>header file on Mac OSX. I'll try and write a configure check to work
>around it.
>

Andrew, Koen:
I dug up this check I wrote back then:
| due to a broken <cmath>, the following testcase should detect it:
| AC_TRY_COMPILE([#include <cmath>], [isnan(1.0);], 
| [AC_MSG_RESULT("Pass")], [AC_MSG_RESULT("Fail")])
| Just to check, I tried
| AC_TRY_COMPILE([#include <math.h>], [isnan(1.0);], 
| [AC_MSG_RESULT("Pass")], [AC_MSG_RESULT("Fail")])
| which does pass as expected.

maybe it can be of some help.

/Per

* Per Persson <persquare@...> [2004-03-04 10:02]:

> Andrew, Koen:
> I dug up this check I wrote back then:
> | due to a broken <cmath>, the following testcase should detect it:
> | AC_TRY_COMPILE([#include <cmath>], [isnan(1.0);], 
> | [AC_MSG_RESULT("Pass")], [AC_MSG_RESULT("Fail")])
> | Just to check, I tried
> | AC_TRY_COMPILE([#include <math.h>], [isnan(1.0);], 
> | [AC_MSG_RESULT("Pass")], [AC_MSG_RESULT("Fail")])
> | which does pass as expected.
> 
> maybe it can be of some help.

Andrew: if you are going to write a configure test for that, remember that
AC_TRY_COMPILE is deprecated in Autoconf 2.59.  The following should work
(untested):

    AC_LANG_PUSH([C++])
    
    AC_MSG_CHECKING([whether $CXX's <cmath> defines isnan])
    AC_COMPILE_IFELSE([
#include <cmath>
main () {isnan(1.0);}
],
        [AC_MSG_RESULT(yes)],
        [AC_DEFINE(CMATH_ISNAN_BROKEN, [],
                   [Define if <cmath> included by CXX does not define isnan])
         AC_MSG_RESULT(no)])

    AC_LANG_POP([C++])

-- 
Rafael

On Thu, Mar 04, 2004 at 10:02:12AM +0100, Per Persson wrote:
>  
> Andrew, Koen:
> I dug up this check I wrote back then:
> | due to a broken <cmath>, the following testcase should detect it:
> | AC_TRY_COMPILE([#include <cmath>], [isnan(1.0);], 
> | [AC_MSG_RESULT("Pass")], [AC_MSG_RESULT("Fail")])
> | Just to check, I tried
> | AC_TRY_COMPILE([#include <math.h>], [isnan(1.0);], 
> | [AC_MSG_RESULT("Pass")], [AC_MSG_RESULT("Fail")])
> | which does pass as expected.
> 
> maybe it can be of some help.

Cheers

According to the documentation AC_TRY_COMPILE is now obsoleted, but I
have implemented something similar with AC_COMPILE_IFELSE.

This works still on my Linux system with gcc 3.3. Can people with affected 
systems please try this if possible?

Andrew