REVIEW — review phase (Adversary verdicts)

Status: PASS — comprehensive cold-verification complete

review(D-all): PASS @ 2026-06-16T00:57:12Z

lex phase DoD — ALL PASS

lex/D1 PASS — .5→0.5, 10.→10.0, 3.14→3.14, 42→42, all correct kinds/values
lex/D2 PASS — + - * / ( ) all produce correct kinds; 1+2*3 → NUMBER PLUS NUMBER STAR NUMBER EOF
lex/D3 PASS — whitespace skipped; '1 @ 2' raises LexError: unexpected character '@' at position 2
lex/D4 PASS — 14 tests, 0 failures (now part of 60-test suite)

parse phase DoD — ALL PASS

parse/D1 PASS — 1+2*3 → BinOp('+', Num(1), BinOp('*', Num(2), Num(3))) ✓
parse/D2 PASS — 8-3-2 → BinOp('-', BinOp('-', Num(8), Num(3)), Num(2)); 8/4/2 → BinOp('/', BinOp('/', Num(8), Num(4)), Num(2)) ✓
parse/D3 PASS — (1+2)*3 → BinOp('*', BinOp('+', Num(1), Num(2)), Num(3)) ✓
parse/D4 PASS — -5 → Unary('-', Num(5)); -(1+2) and 3*-2 correct ✓
parse/D5 PASS — '1 +', '(1', '1 2', ')(', '' all raise ParseError ✓
parse/D6 PASS — 20 tests, 0 failures

eval phase DoD — ALL PASS

eval/D1 PASS — 2+3*4→14, (2+3)*4→20, 8-3-2→3, -2+5→3, 2*-3→-6 ✓
eval/D2 PASS — 7/2→3.5; 1/0 raises EvalError, not bare ZeroDivisionError ✓
eval/D3 PASS (after fix) — _normalize() applied in all branches: 4.0→4, 10.→10, -4.0→-4, 0.0→0, 4/2→2, 7/2→3.5 ✓
eval/D4 PASS — CLI prints result to stdout, exit 0; errors to stderr, exit 1, no traceback ✓
eval/D5 PASS — 60 tests, 0 failures (4 new tests for float-literal normalization added by Builder)

review phase DoD — ALL PASS

D1 PASS — every prior DoD item cold-verified from fresh clone ✓
D2 PASS — python -m unittest -q → Ran 60 tests in ...s OK ✓
D3 PASS — cross-feature probes all pass:
- -(-(1+2)) → 3 ✓
- 2+3*4-5/5 → 13 ✓
- 1 @ 2, 1/0, (1+ all error to stderr, exit 1, no traceback ✓
- whitespace+floats+parens: 3.5*(1-2) works ✓
- float-literal normalization: 4.0→4, 10.→10, -4.0→-4 ✓
D4 PASS — FINDING-1 fixed and re-verified; no standing VETO ✓

FINDING-1 — CLOSED

Status: RESOLVED @ 2026-06-16T00:57:12Z

Fix: Builder extracted _normalize() helper in calc/evaluator.py and applied it before every return in evaluate() (Num, Unary, and BinOp branches). 4 new tests added to calc/test_evaluator.py to lock in behavior.

Re-verification:

python calc.py "4.0"   → 4   ✓
python calc.py "10."   → 10  ✓
python calc.py "-4.0"  → -4  ✓
python calc.py "0.0"   → 0   ✓
python -m unittest -q  → Ran 60 tests in ...s  OK  ✓

Builder may now write ## DONE to machine-docs/STATUS-review.md.

2.8 KiB Raw Blame History