Struct Timing 

pub struct Timing { /* private fields */ }

Accumulates wait time separately from total elapsed time, so a caller can report CPU as elapsed - wait.

“CPU” here means “not waiting”: neither on a host call nor for a task to finish. Each task thread keeps its own timing, so with ADR 0008’s thread per task the separation is a concurrency measurement — one task’s CPU work and another’s wait are recorded against different contexts and can overlap in wall-clock time, which is what ADR 0001 asks a trace to make attributable. It is also why the two no longer sum to the run’s elapsed time: several tasks can be waiting at once.

Implementations

impl Timing

pub fn start() -> Self

Starts timing now.

pub fn add_wait(&mut self, wait: Duration)

Records wait as time spent waiting on a host call.

pub fn wait(&self) -> Duration

Total wait time recorded so far.

pub fn elapsed(&self) -> Duration

Total time elapsed since Timing::start.

pub fn cpu(&self) -> Duration

elapsed() - wait(): time spent doing anything other than waiting on a host call.