Sather Home Page

Section 8.12.1.12:
$IEEE_FLOAT

min_exp : INT ; -- This feature is the minimum negative integer x such that b^(x-1) is -- in the range of normalized floating point numbers. min_exp10 : INT ; -- This feature is the minimum negative integer x such that 10^x is in -- the range of normalized floating point numbers. max_exp : INT ; -- This feature is the maximum permissible binary exponent value for -- numbers of this class max_exp10 : INT ; -- This feature is the maximum integer value x s.t 10^x is in the range -- of values in this class. epsilon : NTP ; -- This feature is the minimum positive value x s.t 1.0 + x /= x digits : INT ; -- This feature is the number of decimal digits of precision for values -- in this class. mantissa_bits : INT ; -- This feature is the number of bits in the significand, including the -- implied bit referrred to in the specification. Max_Precision : CARD ; -- This feature is the number of decimal digits in a number above which -- rounding problems occur in arithmetic operations. Num_Bits : CARD ; -- This feature is the total number of binary digits in the representation of values of this class. is_nan : BOOL ; -- This returns true if and only if self is not the representation of -- a number in the real number value domain, otherwise false. quiet_NaN( sig : INT ) : NTP ; -- This routine returns the representation which is interpreted by -- the IEEE arithmetic model as being a quiet (ie non-interrupting) NaN. -- The argument is not used in this implementation. Built-in to this -- implementation. signalling_NaN( sig : INT ) : NTP ; -- This routine returns the representation which is interpreted by -- the IEEE arithmetic model as a signalling (ie interrupt generating) NaN. -- The argument is unused in this implementation. infinity : NTP ; -- This routine returns the representation of infinity -- which is -- implementation-dependent. min_normal : NTP ; -- This routine returns the smallest normalized positive number -- in this class. max_normal : NTP ; -- This routine returns the largest normalized positive number -- in this class. min_subnormal : NTP ; -- This routine returns the smallest un-normalized positive number -- in this class. max_subnormal : NTP ; -- This routine returns the largest un-normalized positive number -- in this class. is_normal : BOOL ; -- This predicate returns true if and only if self is a normalised -- number. Built-in to this implementation. is_subnormal : BOOL ; -- This predicate returns true if and only if self is an un-normalised -- number. signbit_set : BOOL ; -- This predicate returns true if and only if the sign bit of self is -- set. The number, however, can be one of the special values in addition -- to being a negative number if this is set. is_finite : BOOL ; -- This predicate returns true if and only if self is zero, subnormal -- or normal. is_inf : BOOL ; -- This predicate returns true if and only if self is infinite. unbiassed_exponent : INT ; -- This routine returns the unbiassed exponent of self. -- -- NOTE The IEEE 754 standard specifies that exponent storage shall be -- unsigned - being given an offset or bias in order to achieve this. copysign( other : NTP ) : NTP ; -- This routine returns self with the sign bit set to be the same as -- the sign bit of other. This routine is provided to overcome some of -- the limitations of certain IEEE 754 implementations which have two values -- of zero - in order to make comparison for zero possible. nextup : NTP ; -- This routine returns the value in the IEEE model number domain which -- is the successor in that domain of the value of self. nextdown : NTP ; -- This routine returns the value in the IEEE model number domain which -- is the predecessor in that domain of the value of self. -- get_representation( -- out sign : BOOL, -- out exp : INT, -- out mantissa : INT -- ) ; -- This routine splits up the floating point coding into its sign, -- exponent and mantissa components which are returned in the arguments. -- BECAUSE OF DIFFERING MANTISSA SIZES THIS CANNOT BE IN THE ABSTRACTION!! remainder( other : NTP ) : NTP ; -- This feature returns a remainder of self with respect to other; that -- is, the result is one of the numbers that differs from self by an integral -- multiple of other -- '(self - result)/other' is an integral value.
Language Index Library Index Numeric Index
Comments or enquiries should be made to Keith Hopper.
Page last modified: Wednesday, 21 February 2001.
Produced with Amaya