templated operators do not work about decimal_for_cpp HOT 5 CLOSED

decimal_diff.txt
This fixes the issue, I think (credits go to my collegue Clemens Misch)

from decimal_for_cpp.

Hi, thanks for this fix.

This is good direction but there are some issues to be considered:

• there should be compilation error in this case: p4 += p8
• this compilation error must be meaningful
• m_value is preferred over getUnbiased, the latter is an API method for persistence

from decimal_for_cpp.

Hi,
thanks for the reply. If I understand you correct, p4 += p8 should give a compilation error because you possibly lose precision? This might depend on the use case, for my task the (pre-existing) code that rounds the result is fine. Ideally one could configure the behaviour with a template parameter.
We had to use getUnbiased() because else the compiler would complain about m_value being private.
We forgot assignment operators, I have it fixed here, it's done in a similar way.

from decimal_for_cpp.

Hi,

There are the following common rules for this lib:

• on value copy execution without data loss is preferred (see constructor with string argument)
• when p4 is mixed with another p4, rounding occurs automatically
• when p4 is mixed with p8, it is possible only if there is no rounding, otherwise you should explicit specify (using decimal_cast) when and how the rounding should be executed.
• so if you mix p8 and p4 and there is no requirement for rounding, statement can be compiled correctly (only if rounding detection can be performed at compile time)
• and if you mix p8 and p4 without decimal_cast and rounding can happen, compilation error should occur

This is kind of "strong typing" which is helpful for decimal calculations in financial code.

from decimal_for_cpp.

Updated code accordingly. Please see unit test decimalTypeLevel and readme for more information.

from decimal_for_cpp.