Usermode/libc - ftoa() fixes, current test set passes

diff --git a/Usermode/Libraries/Makefile.tpl b/Usermode/Libraries/Makefile.tpl
index 21f805c..f0084e2 100644 (file)
--- a/Usermode/Libraries/Makefile.tpl
+++ b/Usermode/Libraries/Makefile.tpl
@@ -78,7 +78,7 @@ $(OUTPUTDIR)Libs/%:
 
 obj-native/%.no: %.c
        @mkdir -p $(dir $@)
-       $(NCC) -c $< -o $@ -MD -MP -MF [email protected]
+       $(NCC) -c $< -o $@ -Wall -std=gnu99 -MD -MP -MF [email protected]
 
 TEST_%: obj-native/TEST_%.no obj-native/%.no
        $(NCC) -o $@ $^

diff --git a/Usermode/Libraries/libc.so_src/Makefile b/Usermode/Libraries/libc.so_src/Makefile
index 2e4e2b2..0fd1e5b 100644 (file)
--- a/Usermode/Libraries/libc.so_src/Makefile
+++ b/Usermode/Libraries/libc.so_src/Makefile
@@ -25,4 +25,4 @@ EXP_%.txt: TEST_%.native
        rm TEST_printf.native\r
 \r
 %.native: %.c\r
-       $(NCC) $< -o $@ -Wall -Werror\r
+       $(NCC) $< -o $@ -Wall -std=gnu99\r

diff --git a/Usermode/Libraries/libc.so_src/TEST_printf.c b/Usermode/Libraries/libc.so_src/TEST_printf.c
index ed78fd6..fe251d9 100644 (file)
--- a/Usermode/Libraries/libc.so_src/TEST_printf.c
+++ b/Usermode/Libraries/libc.so_src/TEST_printf.c
@@ -23,9 +23,10 @@ int main(int argc, char *argv[])
        TST("Integer", "%u", 1234);
        
        TST("Float", "%f", 3.1414926535);
+       TST("Float", "%f", 10.0);
        TST("Float", "%.10f", 3.1414926535);
-       TST("Float", "%e", 3.1414926535);
-       TST("Float", "%g", 3.1414926535);
+       TST("Float", "%e", 3.1415926535);
+       TST("Float", "%g", 3.1415926535);
        TST("Float", "%E", 1000000000.00);
        TST("Float", "%a", 16.0);
        TST("Float", "%a", 1024.0);

diff --git a/Usermode/Libraries/libc.so_src/printf.c b/Usermode/Libraries/libc.so_src/printf.c
index f4b13a4..bd67e00 100644 (file)
--- a/Usermode/Libraries/libc.so_src/printf.c
+++ b/Usermode/Libraries/libc.so_src/printf.c
@@ -240,6 +240,13 @@ EXPORT int _vcprintf_int(printf_putch_t putch_cb, void *putch_h, const char *for
                        pos += _printf_ftoa(putch_cb, putch_h, arg_f, 10, FPN_SCI,
                                precision, 0, bJustifyLeft, c == 'E');
                        break;
+               // Scientific Float
+               case 'g':
+               case 'G':
+                       arg_f = bLong ? va_arg(args, long double) : va_arg(args, double);
+                       pos += _printf_ftoa(putch_cb, putch_h, arg_f, 10, FPN_SHORTEST,
+                               precision, 0, bJustifyLeft, c == 'G');
+                       break;
                // Hexadecimal Scientific
                case 'a':
                case 'A':
@@ -632,12 +639,24 @@ size_t _printf_ftoa(printf_putch_t putch_cb, void *putch_h, long double num, siz
        {
                //TODO:
                //int   first_set_sig = BSL(significand);
-       //      TODO: if( num > pos(Base, 2+Precision+2+log_base(exponent) )
-               Notation = FPN_SCI;
+               // bSign+log10(num)+1+precision vs. bSign+1+1+precision+1+1+log10(exponent)
+                int    log10_num = exponent * 301 / 1000;      // log_10(2) = 0.30102999566...
+                int    log10_exp10 = 2;
+                int    sci_len = (signisneg || bForceSign) + 2 + (Precision-1) + 2 + log10_exp10;
+                int    std_whole_len = (log10_num > 0 ? log10_num : 1);
+                int    std_len = (signisneg || bForceSign) + std_whole_len + 1 + (Precision-std_whole_len);
+               if( sci_len > std_len ) {
+                       Precision -= std_whole_len;
+                       Notation = FPN_STD;
+               }
+               else {
+                       Precision -= 1;
+                       Notation = FPN_SCI;
+               }
        }
 
        double precision_max = 1;
-       while(Precision--)
+       for(int i = Precision; i--; )
                precision_max /= Base;
 
        // Determine scientific's exponent and starting denominator
@@ -669,7 +688,7 @@ size_t _printf_ftoa(printf_putch_t putch_cb, void *putch_h, long double num, siz
        }
        else
        {
-               while( den < num )
+               while( den <= num )
                        den *= Base;
                den /= Base;
        }
@@ -694,10 +713,10 @@ size_t _printf_ftoa(printf_putch_t putch_cb, void *putch_h, long double num, siz
        } while( den >= 1 );
 
        // Decimal point (if needed/forced)     
-       if( den >= precision_max || bForcePoint )
+       if( Precision > 0 || bForcePoint )
                _putch('.');
        // Decimal section
-       while( den >= precision_max )
+       for(int i = Precision; i--; )
        {
                num = _longdiv(num, den, &value);
                _putch(cDIGITS[value]);