Crate perfgate_types

Expand description

Shared types for perfgate.

Design goal: versioned, explicit, boring. These structs are used for receipts, PR comments, and (eventually) long-term baselines.

Part of the perfgate workspace.

§Examples

Round-trip a ToolInfo through JSON:

use perfgate_types::ToolInfo;

let tool = ToolInfo { name: "perfgate".into(), version: "1.0.0".into() };
let json = serde_json::to_string(&tool).unwrap();
let back: ToolInfo = serde_json::from_str(&json).unwrap();
assert_eq!(tool, back);

§Feature Flags

arbitrary: Enables Arbitrary derive for structure-aware fuzzing with cargo-fuzz.

Structs§

BaselineServerConfig: Configuration for the baseline server connection.
BenchConfigFile
BenchMeta
Budget
BudgetOverride
Capability: A capability with its status and optional reason.
CompareReceipt: A versioned receipt comparing baseline vs current (perfgate.compare.v1).
CompareRef
ConfigFile: Top-level configuration file (perfgate.toml).
DefaultsConfig
Delta
F64Summary
FindingData: Data associated with a metric finding.
HostInfo
HostMismatchInfo: Details about a detected host mismatch between baseline and current runs.
NoiseDiagnostics: Diagnostics about noise in paired benchmark measurements.
PairedBenchMeta
PairedDiffSummary
PairedRunReceipt
PairedSample
PairedSampleHalf
PairedStats
PerfgateReport: A performance report wrapping compare results in a cockpit-compatible envelope.
ReportFinding: A single finding from the performance check.
ReportSummary: Summary counts and key metrics for the report.
RunMeta
RunReceipt: A versioned receipt from a single benchmark run (perfgate.run.v1).
Sample
ScalingConfig: Configuration for computational complexity validation.
SensorArtifact: An artifact produced by the sensor.
SensorCapabilities: Capabilities available to the sensor.
SensorFinding: A finding from the sensor.
SensorReport: The sensor.report.v1 envelope for cockpit integration.
SensorRunMeta: Run metadata for the sensor report.
SensorVerdict: Verdict for the sensor report.
SensorVerdictCounts: Verdict counts for the sensor report.
Significance
SignificancePolicy: Policy for statistical significance testing.
Stats: Aggregated statistics for a benchmark run.
ToolInfo
U64Summary
Verdict: Overall verdict for a comparison, with pass/warn/fail counts.
VerdictCounts

Enums§

BenchNameValidationError
CapabilityStatus: Capability status for “No Green By Omission” principle.
ConfigValidationError
Direction
HostMismatchPolicy: Policy for handling host mismatches when comparing receipts from different machines.
Metric
MetricStatistic
MetricStatus
NoiseLevel: Noise level classification for paired benchmark results.
NoisePolicy
PerfgateError
SensorSeverity: Severity level for sensor findings (cockpit vocabulary).
SensorVerdictStatus: Verdict status for the sensor report.
Severity: Severity level for a finding.
SignificanceTest
VerdictStatus

Constants§

Functions§

validate_bench_name

Crate perfgate_types

Crate perfgate_types Copy item path

§Examples

§Feature Flags

Structs§

Enums§

Constants§

Functions§

Crate perfgate_types