pub trait ClockRead {
// Required methods
fn wall_ns(&self) -> Result<i128, ClockError>;
fn mono_s(&self) -> Result<f64, ClockError>;
// Provided method
fn wall_parts(&self) -> Result<(i64, u32), ClockError> { ... }
}Expand description
Read the platform clocks.
Required Methods§
Sourcefn wall_ns(&self) -> Result<i128, ClockError>
fn wall_ns(&self) -> Result<i128, ClockError>
Wall-clock nanoseconds since the Unix epoch.
Sourcefn mono_s(&self) -> Result<f64, ClockError>
fn mono_s(&self) -> Result<f64, ClockError>
Monotonic seconds from an arbitrary origin, unaffected by steps.
Provided Methods§
Sourcefn wall_parts(&self) -> Result<(i64, u32), ClockError>
fn wall_parts(&self) -> Result<(i64, u32), ClockError>
Wall clock as (seconds, nanoseconds) since the Unix epoch.
This is the shape an NTP timestamp actually wants, and the shape the
OS already has: every platform’s clock call returns a seconds/nanos
pair. Going through wall_ns multiplies those together into an i128
and the caller then divides them straight back apart — and a 128-bit
division is a software routine, not an instruction. On the server’s
per-reply path callgrind attributed 5% of everything the server did to
__divti3, for a round trip that produces the numbers it started with.
The default implementation keeps that round trip, so implementations that only have nanoseconds still work; the platform clocks override it.
Dyn Compatibility§
This trait is dyn compatible.
In older versions of Rust, dyn compatibility was called "object safety".