pack about decimal_for_cpp HOT 2 CLOSED

Hi,

1. pack(123, 456), prec<2> is wrong. See documentation-in-comment for pack() function and "after" value description:

/// Combines two parts (before and after decimal point) into decimal value
/// \param[in] before value before decimal point
/// \param[in] after value after decimal point multiplied by 10^prec
/// \result Returns *this

In that case:

• calculated result of pack(123, 456) will be 123,56
• value 456 is not "after value" multiplied by 100

This function does not handle rounding - and it cannot do it.
It assumes you follow function requirements.

See example:

decimal<4> d;
/* 1 */ BOOST_CHECK(d.pack(1,1) == decimal<4>("1.0001")); // OK
/* 2 */ BOOST_CHECK(d.pack(1,/*000*/1) == decimal<4>("1.0001")); // OK
/* 3 */ BOOST_CHECK(d.pack(1,/*0000*/1) == decimal<4>("1.0001")); // OK, but if you meant "1.00001" or "1.01" - in that case it's an error

Also it's wrong to add any handling for "after" numbers > 10^prec, because it will work only for values with first digit > 0, for other numbers such a test will not work and error will go undetected.

I'll add clarification comment - "after" value cannot be higher or equal to 10^prec.

2. second case, pack(-12, 34), is not correct according to included unit tests, see:

  BOOST_CHECK(d.pack(-1,-5121) == decimal<4>(-1.5121));
  BOOST_CHECK(d.pack(-1,5121) != decimal<4>(-1.5121));
  BOOST_CHECK(d.pack(1,-5121) != decimal<4>(-1.5121));

You need to provide both numbers negative - I'll document this requirement in comment too.

from decimal_for_cpp.

For (1) see pack_rounded - where you specify input precision.
I managed also to correct "fromString" which from now performs rounding when precision of input value is higher then output precision.

from decimal_for_cpp.