Section 01: Runtime COW Protocol Centralization

Status: Not Started Goal: Reduce 17+ inline copies of the COW uniqueness formula, 3 copies of propagate_elem_header, 3 copies of write_collection_struct, and 4 copies of the iterator consumer loop harness into single canonical functions. Zero behavioral change.

Context: The COW uniqueness check !is_slice_cap(cap) && (cow_mode == 1 || (cow_mode != 2 && ori_rc_is_unique(data))) is inlined at 17 call sites across 7 files. The cow_mode semantics (0=dynamic, 1=static unique, 2=static shared) have no canonical home. propagate_elem_header exists as two identical functions with different names (propagate_elem_header in cow.rs, propagate_header in cow_sort/mod.rs) plus a third inline copy in cow_structural.rs. write_list_output, write_map_struct, write_set_struct are identical 3-line functions.

01.1 Centralize COW Uniqueness Check

File(s): compiler/ori_rt/src/cow_helpers.rs (new), all COW files

Create a single canonical function:

/// Determine whether a COW mutation can proceed in-place.
///
/// cow_mode semantics:
/// - 0: dynamic — check runtime uniqueness via ori_rc_is_unique
/// - 1: static unique — always mutate in-place (compiler proved uniqueness)
/// - 2: static shared — always copy (compiler proved sharing)
///
/// Slices are never mutated in-place (they are views into another allocation).
#[inline(always)]
pub(crate) fn cow_can_mutate_in_place(data: *const u8, cap: i64, cow_mode: i32) -> bool {
    !is_slice_cap(cap) && (cow_mode == 1 || (cow_mode != 2 && ori_rc_is_unique(data)))
}

01.2 Unify propagate_elem_header / propagate_header

File(s): compiler/ori_rt/src/cow_helpers.rs, compiler/ori_rt/src/list/cow.rs, compiler/ori_rt/src/list/cow_sort/mod.rs, compiler/ori_rt/src/list/cow_structural.rs

01.3 Unify write_collection_struct

File(s): compiler/ori_rt/src/cow_helpers.rs, compiler/ori_rt/src/list/mod.rs, compiler/ori_rt/src/map/mod.rs, compiler/ori_rt/src/set/mod.rs

write_list_output, write_map_struct, write_set_struct are identical:

pub(crate) unsafe fn write_collection_struct(out: *mut u8, len: i64, cap: i64, data: *mut u8) {
    out.cast::<i64>().write(len);
    out.cast::<i64>().add(1).write(cap);
    out.add(16).cast::<*mut u8>().write(data);
}

01.4 Extract Iterator Consumer Loop Harness

File(s): compiler/ori_rt/src/iterator/consumers.rs

ori_iter_any, ori_iter_all, ori_iter_find, ori_iter_for_each share identical loop harness: null check, cast to IterState, loop calling state.next(), test element, cleanup via Box::from_raw.

Extract consume_iter() higher-order function:

unsafe fn consume_iter<F, R>(
    iter: *mut u8,
    elem_size: i64,
    init: R,
    f: F,
) -> R
where
    F: FnMut(R, &[u8]) -> ControlFlow<R, R>,

Rewrite ori_iter_any, ori_iter_all, ori_iter_find, ori_iter_for_each using consume_iter
Extract collect_to_reverse_vec() for shared collection phase in ori_iter_rfold and ori_iter_rfind
Verify: timeout 150 cargo test -p ori_rt passes

Cleanup (fix while touching these files)

[WASTE] compiler/ori_rt/src/iterator/consumers.rs:232 — Remove decorative unicode dash banner // ── Backward consumers ...──────────, replace with plain // Backward consumers (require double-ended iterators)
[WASTE] compiler/ori_rt/src/list/cow.rs — write_list_output is still defined here even after write_collection_struct extraction; ensure it’s fully removed, not left as dead code
/tpr-review passed — independent review found no critical or major issues (or all findings triaged)
/impl-hygiene-review passed — hygiene review clean. MUST run AFTER /tpr-review is clean.
Subsection close-out (01.4) — MANDATORY before starting the next subsection. Run /improve-tooling retrospectively on THIS subsection’s debugging journey (per .claude/skills/improve-tooling/SKILL.md “Per-Subsection Workflow”): which diagnostics/ scripts you ran, where you added dbg!/tracing calls, where output was hard to interpret, where test failures gave unhelpful messages, where you ran the same command sequence repeatedly. Forward-look: what tool/log/diagnostic would shorten the next regression in this code path by 10 minutes? Implement improvements NOW (zero deferral) and commit each via SEPARATE /commit-push using a valid conventional-commit type (build(diagnostics): ... — surfaced by section-01.4 retrospective — build/test/chore/ci/docs are valid; tools(...) is rejected by the lefthook commit-msg hook). Mandatory even when nothing felt painful. If genuinely no gaps, document briefly: “Retrospective 01.4: no tooling gaps”. Update this subsection’s status in section frontmatter to complete.
/sync-claude section-close doc sync — verify Claude artifacts across all section commits. Map changed crates to rules files, check CLAUDE.md, canon.md. Fix drift NOW.
Repo hygiene check — run diagnostics/repo-hygiene.sh --check and clean any detected temp files.

01.R Third Party Review Findings

None.

01.T Test Strategy

This section is pure structural refactoring with zero behavioral change. The test strategy focuses on:

Existing test suite as regression gate: ./test-all.sh must pass identically before and after each sub-section.
Unit tests for new canonical functions: Each extracted function gets direct unit tests verifying it matches the old inline behavior.
Structural invariant tests: Grep-based tests that verify no inline copies remain.

Add unit tests for cow_can_mutate_in_place() in compiler/ori_rt/src/cow_helpers/tests.rs:
- cow_mode=1 (static unique) returns true regardless of RC
- cow_mode=2 (static shared) returns false regardless of RC
- cow_mode=0 (dynamic) returns true when ori_rc_is_unique returns true
- cow_mode=0 (dynamic) returns false when refcount > 1
- Slice cap always returns false regardless of cow_mode
Add unit tests for write_collection_struct(): verify struct layout matches the (len, cap, data) triple at correct offsets (0, 8, 16)
Add unit tests for consume_iter(): verify ControlFlow::Break stops iteration, ControlFlow::Continue processes all elements
Verify timeout 150 cargo test -p ori_rt passes after each sub-section (01.1, 01.2, 01.3, 01.4)
Verify timeout 150 ./test-all.sh passes after all sub-sections complete
Verify ORI_CHECK_LEAKS=1 reports zero leaks on COW-heavy test programs (e.g., tests/spec/collections/cow/)