[AdaWorldAPI]

Summary

This PR changes board files only. It contains no code.

1. /coresearch extension 2: Semantic Grammar Recovery (entries/2026-11-02-coresearch-semantic-grammar-recovery.md)

The question, corrected by the operator: which semantic information do Kythe, Ruff/Charon, MLIR/DialEgg and XLOG/Soufflé already provide; which do we lose in the transcode; and how can it serve as typed grammar relations, ambiguity-resolution heuristics and controlled opaque reasoning boundaries in the existing DT64/F### model?

The earlier question, "one field reused by several facades", is closed: ClassView/DT64/F### owns it.

Process:

• A premise gate split the question. Each finding now gets a gap class plus a separate legitimacy verdict (admit / refuse / abstain). Destinations are existing owners; DT64/F### applies only to behaviour bindings.
• Six scouts and a 48-row crosswalk. Every row carries a knowledge-kind column. A Kythe edge is a code fact; an egglog equivalence is a justified transformation; flow runs fact → guard → transformation only.
• Three co-architects.
• Load-bearing citations re-read in the code.

Headlines:

• Every loss sits upstream of an existing owner. No row needs a new authority, engine, IR or parser. The only DT64/F### rows are the XLOG lens words and the missing per-key count word.
• Guards are scarce. Charon computes effect counts (asserts, writes through a dereferenced pointer) and drops them. FUNDECL and provenance lanes are written and never pushed. Residuals are written and read zero times, so the "dead function" diagnostic can report a reachable function as dead. Types are never harvested.
• Three wired hazards:
  • unnamed trait methods all collapse onto ?::?;
  • the C++ harvest emits an unresolved overload set as a bare name, the same as a resolved call;
  • an unstamped composed edge reads as a direct relation.
• Kythe export. It flattens "may dispatch" into ref/call. codegraph's handling of unresolved calls is richer than Kythe's schema.
• Reversed flow. A fact promoted straight to an equivalence occurs only in labs (edge ⇒ union in shared_relation; premise unions in egglog-r2il). A 2026-10-31 board sentence omits that precondition; a correction is proposed, not applied.
• The two grammars do not share a role vocabulary. Natural-language readings and code relations share only an abstention contract: certain ⊆ possible.

Also included:

• a 4-bucket recovery matrix: immediate / small polyfill connection / new guarded rules / not yet interpretable;
• N1–N8 graded against current code;
• pre-registered falsifiers, plus one discriminating probe (ruff_charon/tests/recover_vs_add.rs).

Operator decisions S1–S6 are listed. /5plus3 stays deferred and unsent.

2. #1930 post-merge record (PR_ARC_INVENTORY).

Test Plan

• Entries index regenerated.
• No new citation decay.
• Append-only gate OK (11 files checked).

Not to be merged automatically.

---

Generated by Claude Code

[coderabbitai]

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: 6f9b20ae-ba51-4962-8609-5135442f8e9a

• 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

• X
• Mastodon
• Reddit
• LinkedIn

[chatgpt-codex-connector]

Codex Review Summary

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

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-02T19:27:48.373852Z	619d48a	PR opened

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.