[PATCH] make printf output for Infinity and NaN match C99 standard

[Jeff Johnston]

Patch checked in.

-- Jeff J.

Bob Wilson wrote:
> As you can probably tell, I'm working on contributing a bunch of changes 
> from Tensilica's copy of newlib.  Here's another one, done by David 
> Weatherford. I've also got a bunch of documentation fixes that I'll try 
> to sort through next week.
> 
> Anyway, this patch makes the output of Infinity and NaN floating-point 
> values match the C99 specification: "inf" and "nan" for e/f/g formats 
> and uppercase versions of those for E/F/G formats.  The 'F' format 
> wasn't even recognized, so the patch fixes that, too.
> 
> 2005-10-06  David Weatherford  <weath@tensilica.com>
> 
>     * libc/stdio/vfprintf.c (_VFPRINTF_R): Recognize 'F' format.
>     Print "inf" and "nan" in lowercase for e/f/g formats and in uppercase
>     for E/F/G formats.
> 
> 
> ------------------------------------------------------------------------
> 
> Index: libc/stdio/vfprintf.c
> ===================================================================
> RCS file: /cvs/src/src/newlib/libc/stdio/vfprintf.c,v
> retrieving revision 1.40
> diff -u -r1.40 vfprintf.c
> --- libc/stdio/vfprintf.c	16 Jun 2005 19:14:01 -0000	1.40
> +++ libc/stdio/vfprintf.c	6 Oct 2005 21:23:21 -0000
> @@ -800,6 +800,7 @@
>  		case 'e':
>  		case 'E':
>  		case 'f':
> +		case 'F':
>  		case 'g':
>  		case 'G':
>  			if (prec == -1) {
> @@ -819,12 +820,18 @@
>  			if (isinf (_fpvalue)) {
>  				if (_fpvalue < 0)
>  					sign = '-';
> -				cp = "Inf";
> +				if (ch == 'E' || ch == 'F' || ch == 'G')
> +					cp = "INF";
> +				else
> +					cp = "inf";
>  				size = 3;
>  				break;
>  			}
>  			if (isnan (_fpvalue)) {
> -				cp = "NaN";
> +				if (ch == 'E' || ch == 'F' || ch == 'G')
> +					cp = "NAN";
> +				else
> +					cp = "nan";
>  				size = 3;
>  				break;
>  			}