Enum Outcome 

pub enum Outcome {
    Pass {
        leak_probability: f64,
        effect: EffectEstimate,
        samples_used: usize,
        quality: MeasurementQuality,
        diagnostics: Diagnostics,
        theta_user: f64,
        theta_eff: f64,
        theta_floor: f64,
    },
    Fail {
        leak_probability: f64,
        effect: EffectEstimate,
        exploitability: Exploitability,
        samples_used: usize,
        quality: MeasurementQuality,
        diagnostics: Diagnostics,
        theta_user: f64,
        theta_eff: f64,
        theta_floor: f64,
    },
    Inconclusive {
        reason: InconclusiveReason,
        leak_probability: f64,
        effect: EffectEstimate,
        samples_used: usize,
        quality: MeasurementQuality,
        diagnostics: Diagnostics,
        theta_user: f64,
        theta_eff: f64,
        theta_floor: f64,
    },
    Unmeasurable {
        operation_ns: f64,
        threshold_ns: f64,
        platform: String,
        recommendation: String,
    },
    Research(ResearchOutcome),
}

Top-level outcome of a timing test.

The adaptive Bayesian oracle returns one of four outcomes:

• Pass: No timing leak detected (leak_probability < pass_threshold)
• Fail: Timing leak confirmed (leak_probability > fail_threshold)
• Inconclusive: Cannot reach a definitive conclusion
• Unmeasurable: Operation too fast to measure on this platform

See spec Section 4.1 (Result Types).

Variants

Pass

No timing leak detected.

The posterior probability of a timing leak is below the pass threshold (default 0.05), meaning we’re confident there is no exploitable leak.

Fields

leak_probability: f64

Posterior probability of timing leak: P(effect > theta | data). Will be < pass_threshold (default 0.05).

effect: EffectEstimate

Effect size estimate (shift and tail components).

samples_used: usize

Number of samples used in the analysis.

quality: MeasurementQuality

Measurement quality assessment.

diagnostics: Diagnostics

Diagnostic information for debugging.

theta_user: f64

User’s requested threshold in nanoseconds.

theta_eff: f64

Effective threshold used for inference (may be elevated due to measurement floor).

theta_floor: f64

Measurement floor at final sample count.

Fail

Timing leak confirmed.

The posterior probability of a timing leak exceeds the fail threshold (default 0.95), meaning we’re confident there is an exploitable leak.

Fields

leak_probability: f64

Posterior probability of timing leak: P(effect > theta | data). Will be > fail_threshold (default 0.95).

effect: EffectEstimate

Effect size estimate (shift and tail components).

exploitability: Exploitability

Exploitability assessment based on effect magnitude.

samples_used: usize

Number of samples used in the analysis.

quality: MeasurementQuality

Measurement quality assessment.

diagnostics: Diagnostics

Diagnostic information for debugging.

theta_user: f64

User’s requested threshold in nanoseconds.

theta_eff: f64

Effective threshold used for inference (may be elevated due to measurement floor).

theta_floor: f64

Measurement floor at final sample count.

Inconclusive

Cannot reach a definitive conclusion.

The posterior probability is between pass_threshold and fail_threshold, or the analysis hit a limit (timeout, sample budget, noise).

Fields

reason: InconclusiveReason

Reason why the result is inconclusive.

leak_probability: f64

Current posterior probability of timing leak.

effect: EffectEstimate

Effect size estimate (may have wide credible intervals).

samples_used: usize

Number of samples used in the analysis.

quality: MeasurementQuality

Measurement quality assessment.

diagnostics: Diagnostics

Diagnostic information for debugging.

theta_user: f64

User’s requested threshold in nanoseconds.

theta_eff: f64

Effective threshold used for inference (may be elevated due to measurement floor).

theta_floor: f64

Measurement floor at final sample count.

Unmeasurable

Operation too fast to measure reliably on this platform.

The operation completes faster than the timer’s resolution allows for meaningful measurement, even with adaptive batching.

Fields

operation_ns: f64

Estimated operation duration in nanoseconds.

threshold_ns: f64

Minimum measurable duration on this platform.

platform: String

Platform description (e.g., “Apple Silicon (cntvct)”).

recommendation: String

Suggested actions to make the operation measurable.

Research(ResearchOutcome)

Research mode result.

Returned when using AttackerModel::Research. Unlike Pass/Fail/Inconclusive which make threshold-based decisions, research mode characterizes the timing behavior relative to the measurement floor using CI-based semantics.

See ResearchOutcome for details on the stopping conditions.

Implementations

impl Outcome

pub fn passed(&self) -> bool

Check if the test passed (no timing leak detected).

pub fn failed(&self) -> bool

Check if the test failed (timing leak detected).

pub fn is_conclusive(&self) -> bool

Check if the result is conclusive (either Pass or Fail).

pub fn is_measurable(&self) -> bool

Check if the operation was measurable.

pub fn leak_probability(&self) -> Option<f64>

Get the leak probability if available.

Returns None for Unmeasurable and Research (research mode uses CI, not probability).

pub fn effect(&self) -> Option<&EffectEstimate>

Get the effect estimate if available.

pub fn quality(&self) -> Option<MeasurementQuality>

Get the measurement quality if available.

pub fn diagnostics(&self) -> Option<&Diagnostics>

Get the diagnostics if available.

pub fn samples_used(&self) -> Option<usize>

Get the number of samples used if available.

pub fn is_reliable(&self) -> bool

Check if the measurement is reliable enough for assertions.

Returns true if:

• Test is conclusive (Pass or Fail), AND
• Quality is not TooNoisy, OR posterior is very conclusive (< 0.1 or > 0.9)

The key insight: a very conclusive posterior is trustworthy even with noisy measurements - the signal overcame the noise.

For Research mode, reliability is based on whether the CI is clearly above or below the measurement floor.

pub fn unwrap_pass( self, ) -> (f64, EffectEstimate, MeasurementQuality, Diagnostics)

Unwrap a Pass result, panicking otherwise.

pub fn unwrap_fail( self, ) -> (f64, EffectEstimate, Exploitability, MeasurementQuality, Diagnostics)

Unwrap a Fail result, panicking otherwise.

pub fn handle_unreliable( self, test_name: &str, policy: UnreliablePolicy, ) -> Option<Self>

Handle unreliable results according to policy.

Returns Some(self) if the result is reliable. For unreliable results:

• FailOpen: prints warning, returns None
• FailClosed: panics

Example

let outcome = oracle.test(...);
if let Some(result) = outcome.handle_unreliable("test_name", UnreliablePolicy::FailOpen) {
    assert!(result.passed());
}

Trait Implementations

impl Clone for Outcome

fn clone(&self) -> Outcome

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Outcome

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for Outcome

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Display for Outcome

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Serialize for Outcome

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.