SYMPTOM:
The Big Decimal datatype in Strategy handles numbers up to a precision of 38 digits. The Big Decimal datatype allows for the proper display and passing of high-precision numbers.
In Strategy 9.x, when the Analytical Engine performs calculations on Big Decimal metrics, the precision of the metric is lost. For example, when a subtotal of a Big Decimal metric is displayed, the subtotal has a precision of 15 digits, as illustrated below:
CAUSE:
The Big Decimal datatype is implemented internally as a string datatype. With high-precision numbers converted to a string datatype, the Analytical Engine is able to pass around high-precision numbers without losing precision.
However, mathematical calculations cannot be performed on string datatypes. Therefore, if the Analytical Engine is forced to perform a calculation on a Big Decimal datatype, the numbers must be internally converted to a double precision floating-point number before the calculation is performed.
Start with Strategy 10 Hotfix 1, more functions to use an implementation of double-double arithmetic are supported. Because the double-double calculations performed have a maximum precision of 31 digits, only 31 digits of precision can be maintained.
WORKAROUND:
To maintain consistency in the report display, set the metric formatting of Big Decimal metrics such that the metric values are displayed with a precision of 15 digits or less.
