Re: math patches for OpenVMS.

[Bruno Haible]

John E. Malmberg wrote:
> This is a patch to work around that OpenVMS defines NAN in <fp.h> 
> instead of <math.h>

Looks halfway reasonable, but here are a few hints to provide better patches:

> address@hidden
> +The macros @code{NAN}, @code{HUGE_VALL} and @code{INFINITY} are defined
> +in @code{<fp.h>} instead of @code{<math.h>} on some platforms: OpenVMS

When writing the doc about a problem, you don't need to go into the details
about what the workaround is that gnulib implements. Just mention the problem
itself.

> +/* On OpenVMS, NAN, INFINITY, and HUGEVAL macros are fp.h */

Please write complete sentences. Terminate sentences with a period.

> +#if defined __VMS && ! defined NAN
> +# include <fp.h>
> +#endif

If <math.h> gets included twice, there is no need to include <fp.h> twice.
Include it once should be sufficient. So, put these 3 lines after the
double-inclusion guard. Like it's done e.g. in gnulib's <wchar.h> substitute.

> * lib/math.in.h : OpenVMS has NAN in <fp.h> instead of <math.h>.
> * m4/getdtablesize.m4 : OpenVMS allows users to create file descriptors ...

Statements about facts belong in comments, not in the ChangeLog entry.
In the ChangeLog entry, you should answer the question: "What is the difference
between the new code and the previous code?"


Here's what I'm committing in your name:

2017-07-15  John E. Malmberg  <address@hidden>
            Bruno Haible  <address@hidden>

        math: Add support for OpenVMS.
        * lib/math.in.h [__VMS]: Include <fp.h>.
        * doc/posix-headers/math.texi: Mention OpenVMS issues.

diff --git a/doc/posix-headers/math.texi b/doc/posix-headers/math.texi
index e640663..0cf56a0 100644
--- a/doc/posix-headers/math.texi
+++ b/doc/posix-headers/math.texi
@@ -35,13 +35,18 @@ glibc/HPPA, glibc/SPARC, AIX 5.1, IRIX 6.5, Solaris 9, MSVC 
9.
 The macros @code{FP_ILOGB0} and @code{FP_ILOGBNAN} are not defined on some
 platforms:
 NetBSD 5.1, AIX 5.1, IRIX 6.5, Solaris 9, MSVC 9.
+
address@hidden
+The macros @code{NAN}, @code{HUGE_VALL}, and @code{INFINITY} are not
+defined on some platforms:
+OpenVMS.
 @end itemize
 
 Portability problems not fixed by Gnulib:
 @itemize
 @item
 @code{NAN} is not a compile time constant with some compilers:
-OSF/1 with Compaq (ex-DEC) C 6.4.
+OSF/1 with Compaq (ex-DEC) C 6.4, OpenVMS.
 @item
 The macro or variable @code{math_errhandling} is not defined on some platforms:
 glibc 2.11, OpenBSD 4.9, NetBSD 5.1, UP-UX 11, IRIX 6.5, OSF/1 5.1, Cygwin 
1.7.9, mingw, MSVC 9.
diff --git a/lib/math.in.h b/lib/math.in.h
index c76b997..9e59278 100644
--- a/lib/math.in.h
+++ b/lib/math.in.h
@@ -28,6 +28,11 @@
 #ifndef address@hidden@_MATH_H
 #define address@hidden@_MATH_H
 
+/* On OpenVMS, NAN, INFINITY, and HUGEVAL macros are defined in <fp.h>.  */
+#if defined __VMS && ! defined NAN
+# include <fp.h>
+#endif
+
 #ifndef _GL_INLINE_HEADER_BEGIN
  #error "Please include config.h first."
 #endif