The value of this variable controls the production of results from floating-point computations, in combination with the types of the arguments supplied to the relevant procedure. If it is false then only decimals (single precision floats) are produced. If it is the word "ddecimal" then ddecimal results are produced by arithmetic operators only if at least one of the arguments is ddecimal. If the value is anything else (e.g. true), then a ddecimal result can be forced if at least one of the arguments is integral or rational, even if the others are all decimals.
In NO case is there an increase in precision of floating point computations if all arguments are single-float decimal to start with. The default value of popdprecision is false.
Examples:
Case 1: Only single precision results
false -> popdprecision;
dataword(sqrt(3.5d0)) =>
** decimal
dataword(sqrt(3.5s0)) =>
** decimal
dataword(3.5d0 + 3.5s0) =>
** decimal
dataword(sqrt(2)) =>
** decimal
Case 2: Double precision whenever a ddecimal is involved initially
"ddecimal"-> popdprecision;
dataword(sqrt(3.5d0)) =>
** ddecimal
dataword(sqrt(3.5s0)) =>
** decimal
dataword(3.5d0 + 3.5s0) =>
** ddecimal
dataword(sqrt(2)) =>
** decimal
Case 3: Double precision produced if either ddecimal or non-decimal
number is involved initially.
true -> popdprecision
dataword(sqrt(3.5d0)) =>
** ddecimal
dataword(sqrt(3.5s0)) =>
** decimal
dataword(3.5d0 + 3.5s0) =>
** ddecimal
dataword(sqrt(2)) =>
** ddecimal
This means that if popdprecision is non false, then the evaluation of
arithmetical expressions in which intermediate floating point numbers
are produced will sometimes create temporary ddecimal numbers that are
then discarded. As these are compound items (explained above) this can
cause garbage collections, leading to reduced speed, the price of
greater accuracy.
This is normally true. Making it false prevents Pop-11's normal behaviour in which a ratio result is always reduced to its lowest common terms (and therefore to an integral result if the denominator becomes 1).