float/src/VAX/Real.i3


Copyright (C) 1994, Digital Equipment Corp.

INTERFACE Real;
Properties of REAL (for the VAX).

This package defines some basic properties of the built-in float type REAL, for the VAX architecture. If you want to instantiate a generic type with real numbers, you should import the RealType interface instead.

Index: REAL; floating-point; generics

TYPE T = REAL;

CONST
  Base: INTEGER = 2;
  (* The radix of the floating-point representation for T *)

  Precision: INTEGER = 24;
  (* The number of digits of precision in the given Base for T. *)

  MaxFinite: T = 1.70141173e+38;
  (* The maximum finite value in T.  For non-IEEE implementations,
     this is the same as LAST(T). *)

  MinPos: T = 2.93873588e-39;
  (* The minimum positive value in T. *)

  MinPosNormal: T = 2.93873588e-39;
  (* The minimum positive "normal" value in T; differs from MinPos
     only for implementations with denormalized numbers. *)

CONST
  MaxExpDigits = 2;
  MaxSignifDigits = 9;
MaxExpDigits is the smallest integer with the property that every finite number of type T can be written in base-10 scientific notation using an exponent with at most MaxExpDigits. MaxSignifDigits is the smallest integer with the property that floating-decimal numbers with MaxSignifDigits are more closely spaced, all along the number line, than are numbers of type T. Typically,

      MaxExpDigits    = ceiling(log_10(log_10(MaxFinite)))
      MaxSignifDigits = ceiling(log_10(Base^Precision)) + 1.
END Real.