Function ballistics_calculate_zero_angle 

#[unsafe(no_mangle)]
pub unsafe extern "C" fn ballistics_calculate_zero_angle(
    inputs: *const FFIBallisticInputs,
    wind: *const FFIWindConditions,
    atmosphere: *const FFIAtmosphericConditions,
    zero_distance: c_double,
) -> c_double

Calculate the zero angle for a target distance through the C ABI.

Solves to the classic sight-line zero contract (ZeroTargetFrame::SightLine, MBA-1412): the target height is sight geometry and the rifle zeroes LEVEL. inputs.shooting_angle applies to the subsequent shot, not to the zero.

How this relates to solve-json’s zero_distance_m

Both surfaces aim the search at the same place. This export passes the caller’s sight_height as the target height; solve-json, handed a zero_distance_m with no target_height_m, defaults to the line of sight, muzzle_height_m + sight_height_m (MBA-1537 — before that it defaulted to a flat 0 and the two disagreed by a full sight height at the zero). For a level rifle at the same station conditions the two now return the same angle; tests/zero_sight_line_default.rs pins that.

Two deliberate differences remain, neither of which is a defect — see the module documentation for the reasoning and for the numbers:

• Atmosphere. This export goes through crate::calculate_zero_angle_with_conditions, which applies the legacy default-sentinel heuristic to atmosphere. solve-json goes through crate::cli_api::TrajectorySolver::new_with_resolved_station_atmosphere and trusts a supplied station temperature/pressure as given. Feeding both surfaces 15 °C / 1013.25 hPa at 2000 m is about a 0.29 MOA disagreement on a 300 m zero.
• Target frame. solve-json’s documented contract (docs/SOLVE_JSON_V1.md) is WorldVertical — an absolute height above the ground datum, with inclined zeroing projected into the world frame — not this export’s SightLine, which zeroes level and treats inputs.shooting_angle as a shot-time condition. The frames coincide for a level rifle and diverge once shooting_angle is nonzero.

A caller that needs solve-json’s semantics over a C ABI should use the JSON bridge (crate::bridge, src/bridge/ffi.rs) rather than this export.

Safety

inputs may be null, in which case this function returns NaN. When non-null, it must point to a valid, properly aligned FFIBallisticInputs that remains readable and is not mutated for the duration of this call. wind and atmosphere may also be null; each non-null pointer has the same requirements for its corresponding type.