Fix for bug #11708 (real function returns wrongly rounded decimal)

mysql-test/r/type_newdecimal.result:
  test result fixed
mysql-test/t/type_newdecimal.test:
  test case added
sql/item_func.cc:
  Item_real_func::val_decimal implemented
sql/item_func.h:
  Item_real_func::val_decimal declared

mysql-test/r/type_newdecimal.result

@@ -937,3 +937,6 @@ drop table t1;
 select cast('1.00000001335143196001808973960578441619873046875E-10' as decimal(30,15));
 cast('1.00000001335143196001808973960578441619873046875E-10' as decimal(30,15))
 0.000000000100000
+select ln(14000) c1, convert(ln(14000),decimal(2,3)) c2, cast(ln(14000) as decimal(2,3)) c3;
+c1	c2	c3
+9.5468126085974	9.547	9.547

mysql-test/t/type_newdecimal.test

@@ -978,3 +978,8 @@ drop table t1;
 # Bug #10891 (converting to decimal crashes server)
 #
 select cast('1.00000001335143196001808973960578441619873046875E-10' as decimal(30,15));
+
+#
+# Bug #11708 (conversion to decimal fails in decimal part)
+#
+select ln(14000) c1, convert(ln(14000),decimal(2,3)) c2, cast(ln(14000) as decimal(2,3)) c3;

sql/item_func.cc

@@ -566,6 +566,17 @@ String *Item_real_func::val_str(String *str)
 }
 
 
+my_decimal *Item_real_func::val_decimal(my_decimal *decimal_value)
+{
+  DBUG_ASSERT(fixed);
+  double nr= val_real();
+  if (null_value)
+    return 0; /* purecov: inspected */
+  double2my_decimal(E_DEC_FATAL_ERROR, nr, decimal_value);
+  return decimal_value;
+}
+
+
 void Item_func::fix_num_length_and_dec()
 {
   decimals= 0;

sql/item_func.h

@@ -187,6 +187,7 @@ class Item_real_func :public Item_func
   Item_real_func(Item *a,Item *b) :Item_func(a,b) {}
   Item_real_func(List<Item> &list) :Item_func(list) {}
   String *val_str(String*str);
+  my_decimal *val_decimal(my_decimal *decimal_value);
   longlong val_int()
     { DBUG_ASSERT(fixed == 1); return (longlong) val_real(); }
   enum Item_result result_type () const { return REAL_RESULT; }