Module report 

Validation report generation and display.

This module provides types for structured validation results and functions to generate human-readable output and JSON reports.

Report Structure

• ValidationReport - Top-level report containing all results
  • SuiteResult - Results for a test suite (e.g., “linear”, “nonlinear”)
    • TestResult - Results for a single test case
      • SignalResult - Results for each test signal
        • ComparisonMetrics - Detailed metric values

Example

use pedalkernel_validate::report::ValidationReport;
use std::collections::HashMap;

// After running tests, create a report
let report = ValidationReport::new(suite_results, 96000, 4);

// Print to terminal
report.print_summary();

// Print detailed metrics table
report.print_detailed();

// Save as JSON
report.save_json("report.json").unwrap();

JSON Format

The JSON output includes:

• Timestamp and git commit (if available)
• Sample rate and oversampling settings
• Per-suite and per-test results
• Summary statistics (pass/fail counts, pass rate)

Structs

ComparisonMetrics

Serializable version of ComparisonResult.

ProfileSummary

Summary of test results for one validation profile.

ReportSummary

Summary of all test results.

SignalResult

Result for a single signal within a test.

SuiteResult

Result for a single test suite.

TestResult

Result for a single test case.

ValidationReport

Full validation report.

Functions

format_boundary_disposition

Compact human-readable boundary-load DISPOSITION — FusedUpstream or Unloaded{reason} (namespaced reasons: env:*, gate:*, analysis:*).

format_boundary_model

Compact human-readable boundary-load MODEL summary — kind + element values + component ids, e.g. SeriesCapIntoLoad{10µF → 10kΩ; Cout,RL}.

render_ac_accuracy

Render the AC-accuracy table for a set of per-circuit results.

render_bias_accuracy

Render the bias-accuracy table for a set of per-circuit op-point results.

render_boundary_loads

Render the compact per-circuit boundary-load decision table: one row per analyzed stage output boundary (model + disposition), one explicit no-analysis (empty table) row per circuit whose table is EMPTY (that absence is itself a triage signal, distinct from Unloaded{reason}), and a trailing flag line per circuit whose output boundary is open (see unloaded_output_flag).

unloaded_output_flag

WSL/NaN triage flag: Some(flag) when a circuit’s OUTPUT boundary is left electrically open — either an analyzed boundary whose disposition is Unloaded{reason}, or an EMPTY table (no feedback group ever reached the general-MNA call site, e.g. the whole pedal compiled via blockwise/other paths — the boundary was never even analyzed). The two absence modes are surfaced DISTINCTLY because they need different fixes (widen the policy vs widen the analysis coverage).