1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
/// RoundingMode determines how a Float value is rounded to the
/// desired precision
/// Accuracy describes the rounding error produced by the most recent
/// operation that generated a Float value, relative to the exact value.
/// Internal representation: The mantissa bits x.mant of a nonzero finite
/// Float x are stored in a nat slice long enough to hold up to x.prec bits;
/// the slice may (but doesn't have to) be shorter if the mantissa contains
/// trailing 0 bits. x.mant is normalized if the msb of x.mant == 1 (i.e.,
/// the msb is shifted all the way "to the left"). Thus, if the mantissa has
/// trailing 0 bits or x.prec is not a multiple of the Word size _W,
/// x.mant[0] has trailing zero bits. The msb of the mantissa corresponds
/// to the value 0.5; the exponent x.exp shifts the binary point as needed.
///
/// A zero or non-finite Float x ignores x.mant and x.exp.
///
/// x form neg mant exp
/// ----------------------------------------------------------
/// ±0 zero sign - -
/// 0 < |x| < +Inf finite sign mantissa exponent
/// ±Inf inf sign - -
/// A form value describes the internal representation.
/// The form value order is relevant - do not change!
pub