Function sin 

pub fn sin(x: f64) -> f64

sin x for |x| ≤ 0.25 by its Taylor polynomial through x⁹.

The pole is evaluated only while |θ| ≤ THETA_LIMIT (≈ 0.2094); on |x| ≤ 0.25 the truncation error is below 0.25¹¹ / 11! < 6.1e-15, and the Horner evaluation adds a few units in the last place, so the polynomial is within 1e-12 of f64::sin there, as checked by the polynomial test. It uses only multiplication and addition, so every platform rounds it identically.