agent-orchestrator-benchmark/calculators/builder-adversary-min/run-05/machine-docs/REVIEW-eval.md

REVIEW-eval

Phase: eval Adversary cold-verified at commit 3ff8ae9.

Gates

Gate	Status
D1 — arithmetic	PASS @2026-06-15T
D2 — division / EvalError	PASS @2026-06-15T
D3 — result type formatting	PASS @2026-06-15T
D4 — CLI	PASS @2026-06-15T
D5 — tests green + end-to-end	PASS @2026-06-15T

---

review(D1): PASS @2026-06-15

Cold run of all D1 expressions:

Expression	Expected	Got	Type
2+3*4	14	14	int ✓
(2+3)*4	20	20	int ✓
8-3-2	3	3	int ✓
-2+5	3	3	int ✓
2*-3	-6	-6	int ✓

Additional edge cases probed: --5→5, -(3+2)→-5, 0*100→0. All correct.

review(D2): PASS @2026-06-15

• 7/2 → 3.5 (true division, not integer) ✓
• 1/0 raises EvalError("division by zero"), not bare ZeroDivisionError ✓
• 0/5 → 0 ✓
• 8/4/2 → 1 (left-associative) ✓

review(D3): PASS @2026-06-15

• 4/2 → 2 type=int ✓ (whole result coerced to int)
• 7/2 → 3.5 type=float ✓ (fractional stays float)
• 2+3*4 → 14 type=int ✓ (integer arithmetic stays int)
• CLI output: python calc.py "4/2" prints 2 (no trailing .0) ✓
• CLI output: python calc.py "7/2" prints 3.5 ✓

review(D4): PASS @2026-06-15

• python calc.py "2+3*4" → stdout 14, exit 0 ✓
• python calc.py "(2+3)*4" → stdout 20, exit 0 ✓
• python calc.py "1/0" → error: division by zero to stderr, exit 1 ✓ (no traceback)
• python calc.py "1 +" → error: unexpected token 'EOF' (None) to stderr, exit 1 ✓ (no traceback)

review(D5): PASS @2026-06-15

python -m unittest -q
Ran 52 tests in 0.001s
OK

52 tests, 0 failures, 0 errors. Covers lex + parse (prior phases) + evaluator (D1–D3) + CLI. No regressions.