TEST-NEEDS.md

Test & Benchmark Requirements

CRG Grade: C — ACHIEVED 2026-04-04

Current State

• Unit tests: NONE (no Haskell test files found, no test-suite in .cabal verified)
• Integration tests: 1 Zig integration test (ABI/FFI template)
• E2E tests: NONE
• Benchmarks: NONE (benchmark dir has only README placeholder)
• panic-attack scan: NEVER RUN

What's Missing

Point-to-Point (P2P)

• Data/A2ML.hs (main module) — no tests
• Data/A2ML/Types.hs — no tests
• Data/A2ML/Parser.hs — no tests (parser is most critical)
• Data/A2ML/Renderer.hs — no tests
• Idris2 ABI definitions — no verification tests
• Zig FFI — only template integration test
• tests/fuzz/ contains only placeholder.txt

End-to-End (E2E)

• Parse A2ML document and verify output
• Render A2ML and verify format
• Round-trip fidelity
• QuickCheck / Hedgehog property-based tests for parser correctness
• Error handling for malformed input

Aspect Tests

• Security (untrusted A2ML input)
• Performance (parsing large documents, lazy evaluation pitfalls)
• Concurrency (N/A for library)
• Error handling (partial parse, encoding issues)
• Accessibility (N/A)

Build & Execution

• cabal build — not verified
• cabal test — not verified (no test-suite section found)
• Self-diagnostic — none

Benchmarks Needed

• Parse throughput vs Rust and Elixir implementations
• Memory usage (Haskell lazy evaluation can hide space leaks)
• Criterion benchmarks for parser and renderer

Self-Tests

• panic-attack assail on own repo
• Built-in doctor/check command (if applicable)

Priority

• HIGH — 4 Haskell source modules with ZERO tests. For a Haskell library, the absence of property-based tests (QuickCheck/Hedgehog) is especially damning. The fuzz directory is empty. No benchmark code despite benchmark dir existing.