D-ACR-0: attention_mask audit — EXISTS-UNCALLED, and a rename register file wearing the name - #980
Conversation
…r file wearing the name
Report only, no code, per the deliverable. The falsifier asked for one of two
outcomes; the audit returns both, and the second is the load-bearing one.
EXISTS-UNCALLED: three hits outside the two files, all non-consumers -- two
`pub mod` lines, and mailbox_soa.rs:11, a doc comment stating the opposite
("wrap, NO AttentionMask/LRU"). MedCare-rs / OGAR / ndarray: 0 files each.
And it is a different mechanism. The shipped type is a complete implementation
of causaledge64-mailbox-rename-soa-v1.md §4 -- "the session-ephemeral rename
register file": wide identity (u32 OGIT domain / WitnessId / StyleId) into a
scarce narrow slot (5-bit G / 6-bit W / 8-bit style), LRU because slots are
scarce. That is compression. "Attention" there means which identities are
resident in the slot file, not where the eye looked. A residue carrier runs the
other direction, is keyed by the graph's own address rather than MailboxId, and
is discardable whole precisely because nothing is scarce.
Three properties settle it independently of provenance: keyed by MailboxId with
no NodeGuid/NiblePath/classid in the file; not a mask (Vec + linear scan per
operation, no bitset, no set algebra); records occupancy, never a trajectory
(last_touched_cycle is overwritten, so the previous look is gone).
This is the arc's fourth homonym collision after four "witness" surfaces, four
"nibble" encodings and three "hydration" meanings -- and the only one where the
shipped type is finished and correct for its own contract, which is what made
it read as available.
Piece E regrades from "shipped; unaudited for this use" to "shipped for a
DIFFERENT use; uncalled; not a basis for piece D". §1's six-of-nine count is
unchanged.
Measured in passing: plasticity_residual is declared, initialised to 0, and
never read or written non-zero (two grep hits total); BindReply carries three
fields to a NoOp handler; and the originating §4's singleton actor would
rebuild the singleton the V3 mailbox ruling removed, so "sprint-12+ work" is
superseded rather than pending.
Handed to D-ACR-1: WideFieldMask positions are u8 (universe capped at 256, loud
UniverseExceedsSocCap refusal above), while FieldMask is u64/MAX_FIELDS=64 and
silently drops >= 64. A row population is neither, so §6 Y2's parked basis
collision is a cardinality mismatch with a hard cap on one side, not a subtle
semantic worry.
Board hygiene in the same commit: EPIPHANIES prepend
(E-ATTENTION-MASK-IS-A-RENAME-REGISTER-FILE-NOT-A-RESIDUE-CARRIER-1),
STATUS_BOARD D-ACR-0 -> Shipped and D-ACR-1 -> Next with its basis constraint.
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Bugbot couldn't run - usage limit reachedBugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit. A user or team admin can review and increase usage limits in the Cursor dashboard. (requestId: serverGenReqId_682256e0-732e-4f7e-9bb5-2956e6c2189a) |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 46e5358ef9
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| ## 2. Callers — measured | ||
|
|
||
| ``` | ||
| grep -rn "AttentionMask|attention_mask" --include=*.rs . # workspace, minus target/ |
There was a problem hiding this comment.
The recorded command cannot reproduce this audit: ordinary grep treats | as a literal unless -E is supplied, so this exact invocation returns no matches rather than the three reported hits; it also does not exclude target/ despite the comment. Because the EXISTS-UNCALLED verdict is the deliverable's main falsifier, record the actual reproducible search, such as rg --glob '*.rs' --glob '!target/**' 'AttentionMask|attention_mask', before marking the audit shipped.
Useful? React with 👍 / 👎.
| The shipped type is not a half-built residue carrier. It is a **complete | ||
| implementation of a different design**, and that design is on the board: | ||
| `causaledge64-mailbox-rename-soa-v1.md` §4 — |
There was a problem hiding this comment.
When a later CE64 implementation session relies on this classification, it will incorrectly treat the rename mechanism as already implemented. AttentionMaskSoA::touch accepts both a MailboxId and an already-selected w_slot; it never allocates a slot, maps a WitnessId/StyleId/OGIT identity, or provides the plan's bind/lookup/resolve operations. The audit itself acknowledges the required slot tables and actor binding are absent in §5, so describe this as a partial LRU association scaffold rather than a complete implementation of the rename-register design.
Useful? React with 👍 / 👎.
| So the unfinished ractor binding is not merely unfinished — **finishing it as | ||
| specified would rebuild a singleton the substrate deliberately eliminated.** | ||
| Anyone picking `attention_mask_actor.rs` up should treat "sprint-12+ work" as | ||
| superseded, not queued. |
There was a problem hiding this comment.
The cited V3 ruling forbids a singleton CollapseGate/shared thoughtspace and enforces one writer per mailbox; it does not state that a session-scoped rename registry is forbidden. In fact, the active rename plan still explicitly specifies the singleton actor, and STATUS_BOARD.md still lists D-CE64-MB-5-impl as queued. Telling implementers to treat it as superseded therefore contradicts the repository's active plan and can cause required rename work to be skipped; this conclusion needs an explicit ruling that supersedes D-CE64-MB-5, or it should be removed.
Useful? React with 👍 / 👎.
2 participants
What this is
D-ACR-0 asked one question: is
attention_maskthe basis for the attention work, or does something else have to be built?The answer is neither of the two expected ones.
cognitive-shader-driver::attention_mask+attention_mask_actorare EXISTS-UNCALLED: they compile, they are tested, and nothing in the workspace calls them. More importantly, what they actually implement is a rename register file — the W-slot binding machinery fromcausaledge64-mailbox-rename-soa-v1.md§4 (touch/evict_lru/tick/is_activeoverMailboxId), which is LRU mailbox↔slot binding, not attention focus over rows.The name collides with the concept; the mechanism does not.
Why it matters
A session reaching for "the attention basis" would find these two files by name, read a plausible-looking
AttentionMaskSoA, and build focus semantics on top of an LRU eviction cache. That is the failure this audit exists to prevent. The report states the negative explicitly so the next reader does not have to re-derive it.The consequence for the sibling deliverable: D-ACR-1's basis decision could not be "extend
attention_mask". It became a reuse ofFacetCascadeinstead (shipped separately) — zero new bytes, prefix-containment composition, noFieldMask/WideFieldMaskcardinality inherited.What changed
.claude/ATTENTION_MASK_AUDIT_2026_08_21.md— the audit report: what each file implements, the call-site sweep that found zero callers, the provenance trace back to the rename-SoA plan, and the explicit non-claim that this is a basis for anything.No code changes. No types added. This is a measurement, and its whole value is that it is written down.
Verification
Call-site sweep run across the workspace; both modules have zero non-test callers. The
AttentionMaskBackendimpl forAttentionMaskSoA(added per a codex P2 review on #388) is the only production wiring, and it wires the actor to the SoA — not either to a consumer.🤖 Generated with Claude Code
https://claude.ai/code/session_01KCGhDYoQBXs3poaR7sFuqp
Generated by Claude Code