Alexander MedvednikovandAlbris W
8a3f84415a
Preserve decimal coefficient and scale internally, trim normal string output, and provide explicit fixed-place formatting. Keep default division at sixteen fractional places regardless of stored operand scale. Supersedes #31215 with the requested Go-style semantics. Validated with literal ./v self, decimal tests under normal GC and GC none, formatting, Markdown and the executable README example. Independent agent review approved the exact source and validation evidence. Co-authored-by: Albris W <admin@albrisllc.com>
Description
math provides commonly used mathematical functions for
trigonometry, logarithms, etc.
factorial(n) returns n! for nonnegative integers and uses gamma(n + 1) for other
values. When the result overflows f64, it returns positive infinity, including for
n >= 171 and positive infinity. For nonnegative inputs, use is_inf(result, 1)
to detect this overflow; factorial(170) remains finite. The gamma extension also
returns positive infinity at the pole n = -1, so infinity alone does not indicate
overflow for arbitrary inputs.
For floating point values, math.abs maps both signs of zero to positive zero.
math.expm1 preserves its input's zero sign, so math.expm1(-0.0) returns negative zero.
math.min and math.max return NaN when either floating point argument is NaN.
For equal non-NaN values, including opposite signs of zero, both return their second argument.
modf preserves the input sign on both its integer and fractional results, including zero.
log_gamma returns the logarithm of the absolute value of Gamma; either positive or negative
infinity as input returns positive infinity.
log2(x) returns an exact integer exponent for every representable positive power of two,
including subnormal inputs.
log1p(x) returns positive infinity for positive infinity, negative infinity for -1,
and NaN for values below -1.
logf(x) provides the natural logarithm for f32 inputs, including in pure V builds.
Its special cases are the same as for log.
sqrt(x) rounds its result to the nearest f64, including in pure V builds and for
subnormal inputs. It preserves signed zero and returns NaN for negative inputs.
The trigonometric functions sin, cos, sincos, tan and cot reduce large finite angles
with multiword arithmetic before evaluation, including inputs near a multiple of pi/2 and
values up to math.max_f64. Infinite angles produce NaN.
lgamma aliases log_gamma. remainder(x, y) uses the IEEE nearest-integer,
ties-to-even remainder and preserves the dividend's sign on zero. On the C backend,
fma(x, y, z) computes one rounded multiply-add, including intermediate product overflow.
The C backend uses its math library for both operations. Other backends use portable
remainder and a two-round multiply-add fallback. floor_mod(x, y) instead follows the
divisor's sign, including zero. to_degrees and to_radians convert angle units.
is_power_of_two and next_power_of_two operate on u64; the latter returns zero when the
next power cannot fit. classify distinguishes NaN, infinite, normal, subnormal and zero.
wrapping(value) and saturating(value) create integer arithmetic wrappers for
i8, i16, i32, i64, u8, u16, u32, u64, and int.
Their +, -, and * operators wrap modulo the integer width or clamp to its bounds,
respectively. neg() applies the corresponding negation policy.
For signed minimum divided by -1, wrapping returns the minimum and saturation returns
the maximum. Both % operators return zero for that pair without overflowing the backend's
division operation. Zero divisors follow ordinary integer division behavior.
gcd(a, b) and lcm(a, b) accept matching integer types from i8 through i64,
u8 through u64, and int. They compute with unsigned magnitudes, including signed minimum
values, and clamp an unrepresentable result to the type maximum.
gcd_many(values) computes the common divisor before applying that clamp; its empty result is zero.
lcm_many(values) returns one for an empty input and zero when any element is zero.