miroir/notes/bf-1ccqa.md
jedarden c85da7de24 docs(bf-1ccqa): post Mode A partitioning coverage close-out comments
Step-4 close-out of the Mode A partitioning coverage investigation
(umbrella bf-1zn8a, grandparent bf-30m2j). Aggregates the now-closed
sibling findings (bf-qvrmh test evidence, bf-59oct coverage map,
bf-4nybh gap verdicts) into one authoritative comment posted on both
umbrellas (bf-30m2j:[19], bf-1zn8a:[20]).

The close-out corrects bf-30m2j's stale premise that the disabled
acceptance_4 test is the sole coverage, and steers later children at the
genuine narrow gaps — (a) survivor-stability assertion, (b) scale-up
case, (c) golden-vector test, plus the trivial stale-comment fix —
instead of a redundant acceptance_4 rewrite.

Co-Authored-By: Claude <noreply@anthropic.com>
2026-07-09 09:51:47 -04:00

2.7 KiB

bf-1ccqa — Post coverage map + gap analysis as comments (close-out)

Step 4 / close-out of the Mode A partitioning coverage investigation. Parent umbrella: bf-1zn8a. Grandparent umbrella: bf-30m2j.

What this bead did

Aggregated the now-closed sibling findings into ONE authoritative close-out comment and posted it on both umbrellas so the record is durable and later children attack the real (narrow) gaps instead of rewriting the redundant disabled acceptance_4 test.

  • Posted comment [19] on bf-30m2j (grandparent) — corrects its stale premise that acceptance_4 is the only/sole coverage.
  • Posted comment [20] on bf-1zn8a (umbrella) — durable close-out record.

(These are distinct from the parent's earlier preliminary comments [17]/[18], "investigative step 1", which predated the siblings closing. The close-out comments cite the closed siblings bf-qvrmh / bf-59oct / bf-4nybh and their notes files as the source of truth.)

Sources aggregated

  • bf-qvrmh — cargo test run evidence (notes/bf-qvrmh.md)
  • bf-59oct — per-criterion coverage map (notes/bf-59oct.md)
  • bf-4nybh — gap confirm/refute verdicts (notes/bf-4nybh.md)

Line numbers re-verified against the current working tree before posting

  • mode_a_coordinator.rs: ModeACoordinator::new :66, rendezvous_score :112 (private), separator/seed :113/115, owns_shard :151, set_peer_set_for_test :306, test_owns_exactly_one_peer_per_item :422. Module is #[cfg(feature = "peer-discovery")] at lib.rs:42; default = [] at Cargo.toml:66 — so these tests do NOT run in plain cargo test, only under --features peer-discovery / make test (--all-features). Criterion 4 is PARTIALLY met.
  • anti_entropy.rs: three-pods test :1541 (owner_count==1 :1623, total_owned==64 :1632), pod-reassignment :1665 (pod3_owned_initial :1705, re-check :1724, re-homed-to-one-survivor :1734, all-64-owned :1752), partitioning-wiring :1761, with_mode_a :199 (call sites :1581/1584/1587/1790).

Genuine gaps (the narrow holes for later children — NOT a test rewrite)

  • (a) No minimal-reshuffling / survivor-stability assertion in test_mode_a_pod_reassignment (anti_entropy.rs:1665). CONFIRMED.
  • (b) No 3->4 scale-UP case; only 3->2 scale-down tested. CONFIRMED.
  • (c) No pinned golden-vector test locking the top1_by_score(hash(s||pid)) contract (rendezvous_score is private). CONFIRMED, lower priority.
  • (5) Stale comment at p13_8_anti_entropy.rs:240-244 still references removed with_mode_a_scaling — trivial orthogonal edit.

Directive to children: do NOT rewrite acceptance_4 (redundant). Extend pod-reassignment with survivor stability, add a scale-up test, add a golden vector, and fix the stale comment.