Use this reference to select the C++ installation interface or look up a retained CEL signature. Operation links point to the CEL language book, which defines numeric behavior, error conditions, and examples.
The CMake package is cel-std; its exported target is cel::std.
The public header is <cel/std.hpp>.
Call cel::stdlib::install with a cel::op_lookup before constructing your parser.
See the tutorial for a compiled example.
The installer registers persistent ordinary and receiver-call signatures. The receiver supplies the first operand; no numeric conversions are introduced. Independent libraries can register disjoint overloads before or after installation. Install each exact standard-library signature only once per lookup.
Every f64 maps to C++ double. Scalar result types remain f64, including
rounding operations. sin_cos returns a structural tuple whose C++ adapter is
std::tuple<double, double>.
The following table is the entire port, not the book's larger Rust type set:
| CEL signature | Operation contract |
|---|---|
min(a: f64, b: f64) -> f64 |
min |
max(a: f64, b: f64) -> f64 |
max |
clamp(x: f64, lo: f64, hi: f64) -> f64 |
clamp |
abs(x: f64) -> f64 |
abs |
signum(x: f64) -> f64 |
signum |
sqrt(x: f64) -> f64 |
sqrt |
floor(x: f64) -> f64 |
floor |
ceil(x: f64) -> f64 |
ceil |
trunc(x: f64) -> f64 |
trunc |
fract(x: f64) -> f64 |
fract |
round_ties_even(x: f64) -> f64 |
round_ties_even |
round(x: f64) -> f64 |
round |
sin(x: f64) -> f64 |
sin |
cos(x: f64) -> f64 |
cos |
tan(x: f64) -> f64 |
tan |
asin(x: f64) -> f64 |
asin |
acos(x: f64) -> f64 |
acos |
atan(x: f64) -> f64 |
atan |
atan2(y: f64, x: f64) -> f64 |
atan2 |
sin_cos(x: f64) -> (f64, f64) |
sin_cos |
to_radians(x: f64) -> f64 |
to_radians |
to_degrees(x: f64) -> f64 |
to_degrees |
hypot(x: f64, y: f64) -> f64 |
hypot |
cbrt(x: f64) -> f64 |
cbrt |
powf(x: f64, exponent: f64) -> f64 |
powf |
exp(x: f64) -> f64 |
exp |
exp2(x: f64) -> f64 |
exp2 |
ln(x: f64) -> f64 |
ln |
log(x: f64, base: f64) -> f64 |
log |
log2(x: f64) -> f64 |
log2 |
log10(x: f64) -> f64 |
log10 |
recip(x: f64) -> f64 |
recip |
mul_add(x: f64, multiplier: f64, addend: f64) -> f64 |
mul_add |
rem_euclid(x: f64, divisor: f64) -> f64 |
rem_euclid |
copysign(magnitude: f64, sign: f64) -> f64 |
copysign |
No powi, integer, or f32 signature is installed. Those parts of the linked
book describe Rust capabilities outside this port.
Build for IEEE 754 binary64 without fast-math assumptions. The implementation uses C++ standard-library math and preserves the book's documented NaN, infinity, signed-zero, and rounding behavior. Transcendental results need not be bit-identical across implementations; compare them using a tolerance. NaN payloads and signs are unspecified.
round_ties_even is independent of the ambient rounding mode and does not
change that mode. The library does not promise to preserve floating-point
exception flags or errno.
During evaluation, invalid clamp bounds throw
std::invalid_argument("invalid clamp bounds"). Other numeric domain failures
produce the floating-point results documented in the book. Unsupported
signatures are parser compilation errors.