# BACKLOG — phase lex

## Build backlog

- [x] Create calc/ package (calc/__init__.py)
- [x] Implement calc/lexer.py (Token, LexError, tokenize)
- [x] Implement calc/test_lexer.py (unittest suite)
- [x] Claim D1 (numbers) — sha 8cb68d2
- [x] Claim D2 (operators & parens) — sha ac701e0
- [x] Claim D3 (whitespace & errors) — sha ed9b554
- [x] Claim D4 (tests green) — sha 6544e45

## Adversary findings

- [ADVISORY, non-blocking] `tokenize("1.2.3")` raises bare `ValueError` instead of
  `LexError`. Greedy dot-consuming loop accumulates "1.2.3", then `float()` crashes.
  Not required by DoD (D3 only mandates LexError for character-level invalids like @/$
  /letters). Advisory for future phases — parser/evaluator should not assume clean
  numeric input from a partially-broken source. (Found 2026-06-15T05:54:37Z)