fix c99-float-1 on solaris

[Richard Henderson]

The system float.h doesn't provide FLT_EVAL_METHOD or DECIMAL_DIG.
This as of Solaris 2.8.  I don't have access to the 2.9 betas to
know if its been fixed there.


r~


        * config.gcc (sparc*-solaris): Use float_format=sparc.

Index: config.gcc
===================================================================
RCS file: /cvs/gcc/gcc/gcc/config.gcc,v
retrieving revision 1.167.2.4
diff -c -p -d -u -r1.167.2.4 config.gcc
--- config.gcc	2002/03/22 00:22:02	1.167.2.4
+++ config.gcc	2002/03/27 02:03:46
@@ -3080,7 +3080,7 @@ sparcv9-*-solaris2* | sparc64-*-solaris2
 		tmake_file="$tmake_file t-slibgcc-sld"
 	fi
 	extra_parts="crt1.o crti.o crtn.o gcrt1.o crtbegin.o crtend.o"
-	float_format=none
+	float_format=sparc
 	if test x${enable_threads} = x ; then
 	    enable_threads=$have_pthread_h
 	    if test x${enable_threads} = x ; then
@@ -3110,7 +3110,7 @@ sparc-hal-solaris2*)
                 float_format=i128
         ;;
         *)
-                float_format=none
+                float_format=sparc
                 ;;
         esac
 	thread_file='solaris'
@@ -3147,7 +3147,7 @@ sparc-*-solaris2*)
 		float_format=i128
 		;;
 	*)
-		float_format=none
+		float_format=sparc
 		;;
 	esac
 	if test x${enable_threads} = x; then