Declared Numerics

Declared Numerics

The :gbbopen-tools module contains a set of declared-numeric and pseudo probability operators that provide convenient shorthands for declaring fixnum, short-float, single-float, double-float, long-float, and pseudo-probability numeric operations. If the feature :full-safety is present at compile time, these operators do not make their fixnum, single-float, double-float, and long-float declarations.

The names of the declared-numeric and pseudo-probability operators include these “type-indicator” characters:

Thus, the declared-numeric + operators are:

Most Common Lisp implementations map double-float numbers to the 64-bit IEEE 754 double format and single-float numbers to the 32-bit IEEE 754 single format. The four types of floating-point declared numeric operators (short-float, single-float, double-float, and long-float) are always defined, even if the Common Lisp implementation provides fewer distinct internal float representations. The features :has-short-float, :has-single-float, :has-double-float, and :has-long-float are added when the implementation provides that distinct internal float representation. (The feature :has-single-float is always defined.)

The following two features related to fixnum capabilities are defined when appropriate: :fixnum-size-below-29 and :fixnum-size-supports-unsigned-byte-32.

Some examples of declared-numeric operators include:

  (& x) ==> (the fixnum x)
  (+& x y) ==> (the fixnum (+ (the fixnum x) (the fixnum y))
  (>$ a b) ==> (> (the single-float a) (the single-float b))
  (minusp$$ value) ==> (minusp (the double-float value))
  (truncate& x y)) ==> (the (values fixnum fixnum) (truncate (& x) (& y)))

• Fixnum operators
• Short-float operators
• Single-float operators
• Double-float operators
• Long-float operators

Declared Numerics