neuwirth
/
epiphany
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
epiphany/crates/epiphany-bundle/src
History
Levi Neuwirth 0d8ec61a3c Pass 11 audit follow-up: honest LayoutObjectId status + doc/test gaps 
Independent audit of b2f2e20 / a7adbdc. The canonical, document-state byte
layouts were already correct and golden-locked; this closes the one real
spec-vs-code gap (in the non-canonical layout namespace) and the smaller
doc/test gaps around it.

LayoutObjectId (item 2.6): the spec normatively stated the id "is derived ...
with MUSCLOID" and the ratification log / layout DECISIONS header called it
"pinned" -- but no code uses MUSCLOID (stable_layout_id and
manifestation_layout_id are untagged; synthesized_layout_id borrows MUSCCONF),
and MUSCLOID cannot even be constructed: DomainTag::from_bytes accepts only
built-ins or MUSCS-prefixed tags, so wiring it requires changing the frozen
determinism crate (out of scope for a spec pass, and the spec itself says
MUSCLOID is not a canonical system tag). Kept MUSCLOID as the pinned Track-A
target and made every artifact honest that the v0 code is provisional:
  - spec: "is derived" -> "MUST be derived" (forward contract) + a note that
    the prototype mints provisional ids; changelog "pinned" -> "specified ...
    as the Track-A target".
  - records: ratification-log line 2.6 and layout DECISIONS header/body now
    state spec-pinned-but-code-provisional; stale "Pass 11 candidate 3"
    pointer now cites the ratified requirement.
  - provenance.rs comments name the MUSCLOID target and label the current
    derivation provisional.

Other fixes:
  - epiphany-core/DECISIONS.md: the Tuplet bullet still claimed degenerate
    ratios are caught by runtime invariant 16 "since a Tuplet is a plain
    struct" -- stale after the construction-time TupletRatio change. Rewritten.
  - codec.rs: added degenerate_tuplet_ratio_is_rejected_on_decode, guarding
    the TupletRatio::dec re-validation branch that no test exercised (the
    constructor was tested, the decode path was not).
  - spec: integrity-anomaly snippet kind.canonical_bytes() ->
    to_canonical_bytes() (the actual method).
  - ids.rs / reduce.rs: clarifying comments (ManifestId's intentional,
    golden-locked document_id/generation duplication; compute_promotions
    bucketing by voice == (staff_instance, original_voice) via Invariant 5).

Verification: cargo test --workspace (434 pass, +1), clippy --all-targets clean
(0 warnings), fmt clean; spec rebuilds (lualatex, 0 undefined refs, 254pp).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
 		2026-06-22 10:02:05 -04:00
..
block.rs A B C D F 2026-06-19 12:42:31 -04:00
bundle.rs Item 6 (part 1): Agent E honesty/correctness + Agent D extension preservation 2026-06-21 19:15:45 -04:00
chunk.rs Pass 11 follow-up: golden-lock the ratified discriminant tables; fix three spec-text inaccuracies 2026-06-22 09:22:55 -04:00
codec.rs A B C D F 2026-06-19 12:42:31 -04:00
crc.rs A B C D F 2026-06-19 12:42:31 -04:00
error.rs A B C D F 2026-06-19 12:42:31 -04:00
fuzz.rs A B C D F 2026-06-19 12:42:31 -04:00
header.rs A B C D F 2026-06-19 12:42:31 -04:00
ids.rs Pass 11 audit follow-up: honest LayoutObjectId status + doc/test gaps 2026-06-22 10:02:05 -04:00
lib.rs Item 6 (part 2): operation-block summaries (D-B, C/D integration) 2026-06-21 19:26:17 -04:00
manifest.rs Item 6 (part 2): operation-block summaries (D-B, C/D integration) 2026-06-21 19:26:17 -04:00
store.rs A B C D F 2026-06-19 12:42:31 -04:00
superblock.rs Pass 11 follow-up: golden-lock the ratified discriminant tables; fix three spec-text inaccuracies 2026-06-22 09:22:55 -04:00