[PATCH] Fix printf %e .9999999999999999

[Jakub Jelinek <jakub at redhat dot com>]

Hi!

We print ATM 0.999999999 with "%e" as "1.000000e-00", while 1.0 as
"1.000000e+00".  While the standard doesn't talk about the sign of the
exponent when it is zero, as IEEE754 numbers don't have positive vs.
negative zero exponents, just exponent 0, I'd say we should always
print e+00 instead of e-00. 

2007-04-30  Jakub Jelinek  <jakub@redhat.com>

	* stdio-common/printf_fp.c (___printf_fp): Don't print negative sign
	for exponent 0.
	* stdio-common/tfformat.c (sprint_doubles): Add a new test.

--- libc/stdio-common/printf_fp.c.jj	2007-04-23 10:54:03.000000000 +0200
+++ libc/stdio-common/printf_fp.c	2007-04-30 22:06:04.000000000 +0200
@@ -793,7 +793,7 @@ ___printf_fp (FILE *fp,
   else
     {
       /* This is a special case.  We don't need a factor because the
-	 numbers are in the range of 0.0 <= fp < 8.0.  We simply
+	 numbers are in the range of 1.0 <= |fp| < 8.0.  We simply
 	 shift it to the right place and divide it by 1.0 to get the
 	 leading digit.	 (Of course this division is not really made.)	*/
       assert (0 <= exponent && exponent < 3 &&
@@ -1013,6 +1013,12 @@ ___printf_fp (FILE *fp,
 		  {
 		    *wstartp = '1';
 		    exponent += expsign == 0 ? 1 : -1;
+
+		    /* The above exponent adjustment could lead to 1.0e-00,
+		       e.g. for 0.999999999.  Make sure exponent 0 always
+		       uses + sign.  */
+		    if (exponent == 0)
+		      expsign = 0;
 		  }
 		else if (intdig_no == dig_max)
 		  {
--- libc/stdio-common/tfformat.c.jj	2007-04-23 10:54:04.000000000 +0200
+++ libc/stdio-common/tfformat.c	2007-04-30 22:21:42.000000000 +0200
@@ -4021,6 +4021,8 @@ sprint_double_type sprint_doubles[] =
   {__LINE__, 0.000098,			"0.0001", "%#.0g"},
   {__LINE__, 0.0000996,			"0.00010", "%#.2g"},
   {__LINE__, 9.999999999999999e-05,	"0.0001", "%g"},
+  {__LINE__, 1.0,			"1.000000e+00", "%e"},
+  {__LINE__, .9999999999999999,		"1.000000e+00", "%e"},
 
   {0 }
 

	Jakub