incorrect constructing from fp

[mmkbbt]

incorrect constructing from fp

dec::decimal2(9999999999999998.0); // 9999999999999997.44
dec::decimal4(9999999999999998.0); // -92233720368547758.08

incorrect round dec::decimal2(3.1549999999999998); // must be 3.15, but 3.16

test condition and test condition is incorrect

[vpiotr]

Case 1: dec::decimal2(9999999999999998.0) is not representable in this data type. Maximum value is 2^63-1 divided by 10^precision, so something closer to 9,223372036854775808e16 in this case.

Case 2: See case 1.

Case 3: Using double constructor is not recommended and only useful for numbers with not too many significant digits.

In this case you are trying to use 17 digits with double which supports 15 to 17 significant decimal digits. In addition, construction from double requires additional digits in the middle of conversion (= precision). So in this case 17 + 2 = 19 digits are required from double. See sample code: https://ideone.com/GNTtuF Wiki describing double limits: https://en.wikipedia.org/wiki/Double-precision_floating-point_format

For such case a special constructor has been created which supports better so many digits (string constructor).

Code:

    dec::decimal<2> d1(3.1549999999999998);
    std::cout << "Value #1 is: " << d1 << std::endl;
    dec::decimal<2> d2("3.1549999999999998");
    std::cout << "Value #2 is: " << d2 << std::endl;

outputs:

    Value #1 is: 3.16
    Value #2 is: 3.15

[mmkbbt]

posible partial fix for line is

   void init(double value) {
        dec_storage_t t_v0 = static_cast<dec_storage_t>(value);
        m_value = RoundPolicy::round(
                static_cast<double>(DecimalFactor<Prec>::value) * (value - t_v0)) +
                DecimalFactor<Prec>::value * t_v0;
    }

[vpiotr]

Implemented fix as suggested, corrected unit tests, thanks!

[mmkbbt]

same bug in line. pls fix too

[vpiotr]

Assignment corrected too.