Fixes #153: Now applying roll-over for values rounded up using Banker…

…'s rounding in `get_components()`

* Also added a comment-instead-of-a-testcase for a command in which that roll-over is meaningful

src/printf/printf.c

@@ -609,9 +609,11 @@ static const floating_point_t powers_of_10[PRINTF_MAX_PRECOMPUTED_POWER_OF_10 +
 #define PRINTF_MAX_SUPPORTED_PRECISION (NUM_DECIMAL_DIGITS_IN_INT64_T - 1)
 
 
-// Break up a floating-point number - which is known to be a finite non-negative number -
-// into its base-10 parts: integral - before the decimal point, and fractional - after it.
-// Taken the precision into account, but does not change it even internally.
+// Break up a non-negative, finite, floating-point number into two integral
+// parts of its decimal representation: The number up to the decimal point,
+// and the number appearing after that point - whose number of digits
+// corresponds to the precision value. Example: The components of 12.621
+// are 12 and 621 for precision 3, or 12 and 62 for precision 2.
 static struct floating_point_components get_components(floating_point_t number, printf_size_t precision)
 {
   struct floating_point_components number_;
@@ -624,18 +626,14 @@ static struct floating_point_components get_components(floating_point_t number,
   floating_point_t remainder = scaled_remainder - (floating_point_t) number_.fractional;
   const floating_point_t one_half = (floating_point_t)  0.5;
 
-  if (remainder > one_half) {
+  if ((remainder > one_half) ||
+      // Banker's rounding, i.e. round half to even: 1.5 -> 2, but 2.5 -> 2
+      ((remainder == one_half) && (number_.fractional & 1U))) {
     ++number_.fractional;
-    // handle rollover, e.g. case 0.99 with precision 1 is 1.0
-    if ((floating_point_t) number_.fractional >= powers_of_10[precision]) {
-      number_.fractional = 0;
-      ++number_.integral;
-    }
   }
-  else if ((remainder == one_half) && (number_.fractional & 1U)) {
-    // Banker's rounding, i.e. round half to even:
-    // 1.5 -> 2, but 2.5 -> 2
-    ++number_.fractional;
+  if ((floating_point_t) number_.fractional >= powers_of_10[precision]) {
+    number_.fractional = 0;
+    ++number_.integral;
   }
 
   if (precision == 0U) {

test/test_suite_main_testcases.hpp

@@ -1111,8 +1111,17 @@ PRINTF_TEST_CASE(misc)
   PRINTING_CHECK("0.000000e+00",            ==, sprintf_, buffer, "%e", 0.0);
   PRINTING_CHECK("-0.000000e+00",           ==, sprintf_, buffer, "%e", -0.0);
 #endif
+  // What should we get with  sprintf_(buffer, "%.2f", 0.995) ?
+  // that number cannot be represented as a double, and the closest double
+  // is actually 0.9949999999999999955591... and so on. That should give
+  // us 0.99 as the output.
+  // still, we can't complain about that value being confused with real 0.995, and
+  // so we can't complain if we get the banker's rounding for 0.995 to 1.00 as the
+  // output
+  //PRINTING_CHECK("1.00",                    ==, sprintf_, buffer, "%.2f", 0.995);
 }
 
+
 PRINTF_TEST_CASE(extremal_signed_integer_values)
 {
   char buffer[base_buffer_size];