# JOURNAL-eval — Adversary notes

## 2026-06-15T04:08Z — Initialized

Adversary starting for eval phase. Parse phase completed (all D1-D6 PASS).
Waiting for Builder to create STATUS-eval.md and claim gates.

Planned verification approach:
- D1: Run exact arithmetic expressions from plan, check output values
- D2: Verify true division (7/2=3.5), EvalError on divide-by-zero (not bare ZeroDivisionError)
- D3: Check output format — whole numbers no .0, fractions as float
- D4: CLI exit codes, stderr for errors, no tracebacks
- D5: Full test suite green, no regressions in lex+parse tests

---

## 2026-06-15T04:10Z — Builder implementation notes

Built evaluator, CLI, and tests from scratch in one pass.

### evaluator.py design
- EvalError wraps division-by-zero so the bare ZeroDivisionError never escapes the API.
- evaluate() is a simple recursive dispatch on Num/BinOp/Unary node types.
- True division via Python's `/` operator naturally produces float.

### calc.py CLI design
- _format() prints int for whole-valued floats (value.is_integer()), str(float) otherwise.
- All LexError/ParseError/EvalError caught → stderr message + exit(1), no traceback.

### Test run (commit 14db736)
```
$ python -m unittest -q
Ran 66 tests in 0.147s
OK
```

### CLI spot checks
```
$ python calc.py "2+3*4"    → 14
$ python calc.py "(2+3)*4"  → 20
$ python calc.py "7/2"      → 3.5
$ python calc.py "4/2"      → 2
$ python calc.py "1/0"      → stderr: error: division by zero, exit 1
$ python calc.py "1 +"      → stderr: error: unexpected token 'EOF' (None), exit 1
```