Skip to content

D-BAG-FOLD v3: Arm L against production witness contracts (lab) - #1938

Merged
AdaWorldAPI merged 2 commits into
mainfrom
ccr-85acaff0-oszq40
Nov 3, 2026
Merged

AdaWorldAPI merged 2 commits into
mainfrom
ccr-85acaff0-oszq40

Conversation

@AdaWorldAPI

@AdaWorldAPI AdaWorldAPI commented Nov 3, 2026 •

Copy link
Copy Markdown
Owner

This PR is the bounded validation of Arm L that followed #1937. It merged with v3 only (commits 7c4053c61 and ea5313389). The v4 commit was pushed after the merge and is in #1939.

v3 replaces the probe-local mirrors with production witness contracts:

  • SealedFacetLane / OrderedLaneWitness (seal, validate, seek_from, bound);
  • FacetCascade keys;
  • lance_graph_dt64::dr::FrameStore.

Everything lives in the isolated crate .claude/probes/2026-11-03-d-bag-fold/probe-v3/, which has its own [workspace].

  • No production code changes. MaskOp::Gather runs only inside the lab crate.
  • Arm L is not promoted, EXO-140 is unchanged, and no DT64 behaviour is added.

Results (26/26 checks)

  • Oracle parity on every root of two fixtures (340 and 4,680 nodes), and on the timed root at every benchmark size.
  • Production refuses:
    • a stale lane version;
    • another lane's witness;
    • a foreign frame;
    • a stale frame, when the presence writer advances it.
  • No production witness exists for the following. Each is closed only by a lab guard, and each guard's contract-only disable run gives a silent wrong answer:
    • presence-plane binding;
    • key uniqueness;
    • pointer bounds;
    • pointer provenance.
  • A lens mismatch cannot be constructed, because SemanticLens has only one variant.
  • Execution: zero allocations, in microseconds. It is 25–250× slower than a gather-free baseline, because each Gather is gated by T_in rather than by the candidate range.
  • Preparation: tracks the production seal's cost.

The decision matrix is in REPORT-V3.md. Board entry: D-BAG-FOLD-V3.

🤖 Generated with Claude Code

https://claude.ai/code/session_01XAqKMNZqKXSaeb9g2ypywF

New isolated probe crate .claude/probes/2026-11-03-d-bag-fold/probe-v3/ (own
[workspace]). Arm L now uses the production SealedFacetLane /
OrderedLaneWitness (seal, validate, seek_from, bound), FacetCascade keys and
lance_graph_dt64::dr::FrameStore instead of probe-local mirrors. The v1/v2
probes are unchanged.

26/26 checks: oracle parity on every root of two fixtures; negative controls
for permuted planes, stale presence, duplicate keys, pointer bounds, wrong
ClassId / foreign witness / foreign frame and value-derived pointers; each lab
guard has a contract-only disable that gives a silent wrong answer.

Production refuses stale versions, foreign witnesses and foreign or stale
frames. Missing: presence-plane binding, key uniqueness, pointer bounds and
pointer provenance; lens mismatch is not constructible. Execution: 0 allocs,
microseconds; preparation tracks the production seal's cost, which crosses
the no-ms gate at 69,904 nodes (printed as FAILED-GATE). Decision matrix in
REPORT-V3.md. Arm L is not promoted; EXO-140 unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XAqKMNZqKXSaeb9g2ypywF
@coderabbitai

coderabbitai Bot commented Nov 3, 2026 •

Copy link
Copy Markdown

Important

  • Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

Run configuration
  • Source mode: Auto
  • Repository: Unavailable
  • Branch: Unavailable
  • File: No configuration file from this repository applied
  • Commit: Unavailable
  • Layers: Organization UI
  • Review profile: CHILL
  • Plan: Advanced
  • Run ID: 9af4f7d2-a96c-4f96-8d5d-bca07abbdd89
  • Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

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.

Share

CodeRabbit Verify

  • Run CodeRabbit Verify to test the changes in this pull request.

entries_index.py --check failed with DUPLICATE ID: D-BAG-FOLD, because the
v3 entry's heading resolved to the same ID as the earlier D-BAG-FOLD entry.
The heading now names D-BAG-FOLD-V3 and the index row is regenerated;
--check passes locally.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XAqKMNZqKXSaeb9g2ypywF
@AdaWorldAPI
AdaWorldAPI marked this pull request as ready for review November 3, 2026 15:19
@AdaWorldAPI
AdaWorldAPI merged commit a942b3e into main Nov 3, 2026
4 checks passed
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Nov 3, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-11-03T15:20:21.391465Z ea53133 Draft marked ready
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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@AdaWorldAPI AdaWorldAPI changed the title D-BAG-FOLD v3: Arm L against production witness contracts (lab) D-BAG-FOLD v3 + v4: Arm L against production witnesses, attested lens, gated execution (lab) Nov 3, 2026
@AdaWorldAPI AdaWorldAPI changed the title D-BAG-FOLD v3 + v4: Arm L against production witnesses, attested lens, gated execution (lab) D-BAG-FOLD v3: Arm L against production witness contracts (lab) Nov 3, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants