Re: New feature-testing macros in <sys/cdefs.h>

["Kamil Rytarowski" <n54%gmx.com@localhost>]

Me wrote:
> Hello,
> 
> Clang ships with fine-grained useful macros for checking
> whether a feature is supported. Instead of redefining them
>  in the other places, I propose to put them in a central
> place in <sys/cdefs.h>.
> 
> GCC 5.0 ships with __has_include, for now it's skipped
> as GCC 5.0 isn't released.
> 
> Regards,
> 
> diff --git a/sys/cdefs.h b/sys/cdefs.h
> index 02ea4b3..f850655 100644
> --- a/sys/cdefs.h
> +++ b/sys/cdefs.h
> @@ -55,6 +55,41 @@
>  #define	__GNUC_PREREQ__(x, y)	0
>  #endif
>  
> +/*
> + * Macro to test if we're using a compiler with specific:
> + * - features,
> + * - extensions,
> + * - attributes,
> + * - builtins.
> + *
> + * Usage:
> + *
> + *	#if __has_builtin(__builtin_popcount)
> + *		return __builtin_popcount(x);
> + *	#else
> + *		int count = 0;
> + *		for (; x != 0; x &= x - 1)
> + *			count++;
> + *		return count;
> + *	#endif
> + */
> +
> +#ifndef __has_feature
> +#define __has_feature(x) 0
> +#endif
> +
> +#ifndef __has_extension
> +#define __has_extension(x) 0
> +#endif
> +
> +#ifndef __has_attribute
> +#define __has_attribute(x) 0
> +#endif
> +
> +#ifndef __has_builtin
> +#define __has_builtin(x) 0
> +#endif
> +
>  #include <machine/cdefs.h>
>  #ifdef __ELF__
>  #include <sys/cdefs_elf.h>
> 

Hello,

Any comments? Someone for or against of use of these macros?

I'm about to propose a library for overflows, which reuses these symbols,
instead of depending on the hard-coded Clang checks.

Every compiler is free to support this or that builtin,
without playing a role of GCC/Clang.

Actually __has_include() is fine, as it's already in Clang.

Regards,