0%

Section 05: Journey Corpus Expansion

Status: Not Started Goal: Add J21-J30+ journey programs targeting the leak-prone patterns that J1-J20 miss. Every pattern in tests/valgrind/fat_matrix/ that has no journey coverage gets a dedicated journey.

Success Criteria:

  • 10+ new journeys exist (J21-J30 minimum)
  • Break-in-loop with heap strings/lists covered
  • Nested closures capturing fat pointers covered
  • Partial iterator consumption (take/break) covered
  • ? operator with Option<str> / heap types covered
  • COW mutations with early exit covered
  • Deeply nested drops (3+ levels) covered
  • At least 1 canary journey with known leak (tests that the system detects it)
  • Satisfies mission criterion for corpus expansion

Context: The TPR review proved J1-J8 use scalar types only and J9-J20, while better, still miss critical patterns. The valgrind fat_matrix has 20 test files covering every fat pointer operation — but code journeys never exercise those patterns. The expanded corpus closes this gap.

Reference implementations:

  • tests/valgrind/fat_matrix/f19_break_continue.ori — break/continue with fat values in scope
  • tests/valgrind/fat_matrix/f04_closure_capture.ori — 11 closure capture patterns
  • tests/valgrind/fat_matrix/f15_question_mark.ori — ? with Option
  • tests/valgrind/cow/cow_leak_scenarios.ori — explicit leak scenarios

Depends on: Section 03 (orchestrator — needed to run the new journeys).

05.1 Gap Analysis

  • Map every tests/valgrind/fat_matrix/ pattern to existing J1-J20 coverage

  • Identify the uncovered patterns (from TPR findings):

    1. Break in for loop with str/[str]/Option<str> in scope
    2. Continue in for loop with fat values in scope
    3. Nested closures capturing heap types (2+ levels)
    4. map().take(N).collect() — partial iterator consumption
    5. for x in iter do if cond then break with fat iterator elements
    6. ? on Option<str> in scope with other heap values
    7. COW string mutation inside an if branch (uniqueness barrier)
    8. List push in loop with early break (COW + control flow)
    9. Struct field reassignment in loop (old value must be freed)
    10. Deeply nested struct drops (Container > Nested > [str])
    11. Panic path cleanup (values in scope when panic fires)

  • Assign each gap to a journey number and slug

  • Subsection close-out (05.1) — MANDATORY before starting 05.2:

    • All tasks above are [x] and gap map complete
    • Update this subsection’s status in section frontmatter to complete
    • Run /improve-tooling retrospectively on THIS subsection
    • Run /sync-claude on THIS subsection — check whether code changes invalidated any CLAUDE.md, .claude/rules/*.md, or canon.md claims. If no API/command/phase changes, document briefly. Fix any drift NOW.
    • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

05.2 Write J21-J25: Control Flow + Fat Values

File(s): plans/code-journeys/21-*.ori through 25-*.ori

Each journey follows the existing conventions: @main () -> int, standardized header comments, expected result comment, < 30 lines, focused on 1-2 interacting features.

  • J21: Break in for loop with heap string (21-break-fat-for.ori)

    • For loop over [str], break when condition met, verify remaining elements freed
    • Difficulty: complex, features: [loops, strings, break_continue, arc]
    • Inspired by: tests/valgrind/fat_matrix/f19_break_continue.ori

  • J22: Question mark with heap option (22-question-mark-heap.ori)

    • Function returns Option<str>, caller uses ?, verify cleanup on None path
    • Difficulty: complex, features: [option_type, strings, error_propagation, arc]
    • Inspired by: tests/valgrind/fat_matrix/f15_question_mark.ori

  • J23: Nested closure capture (23-nested-closure-capture.ori)

    • Outer closure captures str, inner closure captures outer closure, verify env cleanup
    • Difficulty: complex, features: [closures, strings, capture, higher_order, arc]
    • Inspired by: tests/valgrind/fat_matrix/f04_closure_capture.ori

  • J24: Struct field reassignment (24-field-reassign-loop.ori)

    • Loop reassigns a struct field holding str, verify old value freed each iteration
    • Difficulty: complex, features: [struct_construction, strings, loops, arc]
    • Inspired by: tests/valgrind/aims_drop_reuse.ori

  • J25: Deeply nested drop (25-deep-nested-drop.ori)

    • 3-level nesting: Outer { inner: Inner { items: [str] } } — verify cascading drops
    • Difficulty: complex, features: [nested_structs, lists, strings, arc]
    • Inspired by: tests/valgrind/recursion_stress.ori

  • Run all 5 new journeys through /code-journey to verify they produce valid results

  • Verify eval and AOT paths produce identical output for each

  • Subsection close-out (05.2) — MANDATORY before starting 05.3:

    • All tasks above are [x] and J21-J25 produce valid results
    • Update this subsection’s status in section frontmatter to complete
    • Run /improve-tooling retrospectively on THIS subsection
    • Run /sync-claude on THIS subsection — check whether code changes invalidated any CLAUDE.md, .claude/rules/*.md, or canon.md claims. If no API/command/phase changes, document briefly. Fix any drift NOW.
    • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

05.3 Write J26-J30: Closures, Iterators, COW

File(s): plans/code-journeys/26-*.ori through 30-*.ori

  • J26: Partial iterator consumption (26-partial-iter.ori)

    • iter.take(3).collect() on a 10-element list of strings — verify remaining 7 freed
    • Difficulty: complex, features: [iterators, strings, lists, arc]
    • Inspired by: tests/valgrind/iter_rc/

  • J27: COW mutation with early exit (27-cow-break.ori)

    • List push in loop with conditional break, verify COW buffer freed on exit
    • Difficulty: complex, features: [cow, lists, loops, break_continue, arc]
    • Inspired by: tests/valgrind/cow/cow_leak_scenarios.ori

  • J28: Continue in for-yield with fat values (28-continue-yield-fat.ori)

    • for x in items if cond yield transform(x) — verify skipped elements properly cleaned
    • Difficulty: complex, features: [loops, strings, for_yield, arc]

  • J29: Multiple shared references (29-multi-shared.ori)

    • Create 3 aliases to same list, mutate one (COW), verify others unchanged, all freed
    • Difficulty: complex, features: [cow, lists, strings, arc]
    • Inspired by: tests/valgrind/cow/cow_map_operations.ori

  • J30: Canary — known leak program (30-canary-leak.ori)

    • A program designed to trigger a specific known leak pattern
    • This journey’s EXPECTED verdict is “findings” not “clean” — it tests that the system detects issues
    • If the system reports “clean” on this journey, the detection is broken
    • Difficulty: complex, features: [arc, canary]

  • Run all 5 new journeys through /code-journey

  • Verify J30 canary produces findings (not clean)

  • Subsection close-out (05.3) — MANDATORY before marking section complete:

    • All tasks above are [x] and J26-J30 produce valid results
    • Update this subsection’s status in section frontmatter to complete
    • Run /improve-tooling retrospectively on THIS subsection
    • Run /sync-claude on THIS subsection — check whether code changes invalidated any CLAUDE.md, .claude/rules/*.md, or canon.md claims. If no API/command/phase changes, document briefly. Fix any drift NOW.
    • Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

05.R Third Party Review Findings

  • None.

05.N Completion Checklist

  • J21-J30 (10 journeys) exist and compile
  • Each journey exercises a previously uncovered leak-prone pattern
  • All journeys produce valid JSON results via /code-journey
  • J30 canary correctly detects the intended leak
  • timeout 150 ./test-all.sh green
  • /tpr-review — dual-source review of new journey programs
  • /impl-hygiene-review — verify naming conventions, no DRIFT from existing journey patterns
  • /improve-tooling section-close sweep