[AdaWorldAPI]

D-CODEGRAPH-0: a resident Codegraph, Charon → ruff → lance-graph + crai-rs

Plan: .claude/plans/d-codegraph-0-resident-codegraph-v1.md. Entry: .claude/board/entries/2026-11-02-d-codegraph-0-gap-map.md. Representation: Kythe semantic facts + R2IL executable semantics + DT64/DR binding.

Gap map G1–G9, checked against main of lance-graph, OGAR and ruff and the last 50 PRs of lance-graph and OGAR:

• no Charon intake in ruff and no Kythe producer in Rust;
• no Kythe ref/call edges;
• no Rust-body → R2IL lowering;
• no resident or incremental store;
• closure capped at 64 positions per tile;
• the method body anchor is open;
• the ruff → lance-graph sink is unmerged.

sag-ra-probe (lab crate, excluded from the workspace):

• Build fix. It did not build from a fresh resolution: ra_ap_hir_def pins unicode-ident =1.0.24 (Unicode 17) while unicode-properties 0.1.5 is Unicode 18, and ra-ap-rustc_lexer refuses the mismatch. Fixed by pinning unicode-properties = "=0.1.4".
• GRAPH=1 mode (with NO_MIR, FILES): typed rust-analyzer rows (file, node, occ with use/pubuse roles, call, impl, import, macro, diag), as the independent check for the Charon intake.
• Oracle fixture fixtures/codegraph/{rev1,rev2}: aliases, re-exports, trait dispatch, cfg, a call hidden in a macro, unresolved names, a module cycle, a deleted symbol.
• Hand-read pins: rev1_xrefs.pin.tsv (19) and rev1_calls.pin.tsv (3).
• tools/kythe_calls_check.py: serves Charon-derived Kythe entries and checks the call pins.

Measured here:

• Kythe v0.0.76 /xrefs on the fixture: 22/22 symbols, 19/19 pins; a pin moved by one column fails.
• The D-SAG-6 sessions' runs, repeated: the identity chain and the sag_fixture Kythe check reproduce exactly. The mask-risc Charon join reproduces except one body (a derived PartialEq::eq now joins EXACT instead of DERIVED; same item).
• Charon vs rust-analyzer on the fixture:
  • Charon witnesses the call inside assert_eq!, which rust-analyzer's call hierarchy misses.
  • rust-analyzer records a reference inside #[cfg(any())] code that the build does not contain.
• Slice A, ruff_charon (ruff, see below):
  • calls byte-identical to llbc_calls.py on the fixture and on mask-risc (459 functions, 441 MB LLBC, 5.6 s release);
  • receipt conserved (16,149 = 16,149 + 0);
  • Kythe call pins 3/3; 0/3 without file nodes.

Also on this branch (earlier in the session, never in a PR)

• D-DQM M1 timing arms:
  • the DuckDB reference arm (lance-graph-quack/tests/duckdb/timing.py);
  • the Quack timed arm in the Rust differential;
  • a board entry recording the exit gate as open, with no performance claim.
• D-DQM M2 input: a plan note on Lance 13's row-mask read (with_row_addr_prefilter).
• Ledger revision (f) over board: catch-up for #1660–#1725 (PR_ARC, INTEGRATION_PLANS, LATEST_STATE) #1727–docs(D-NAM-PZL): palette-native LUT lowering and capstone alignment (S0) #1784 and ndarray fix(gguf): restore 32-byte scaling on data_offset (chatgpt-codex catch on PR #367) #368– SMB tenant schema reference #376.
• Duplicate vocabulary issue: a fresh ISS-RECIPE-LOCO-VOCABULARY-EXISTS-TWICE entry, replacing a stale Wave 0 for the Cypher→mask lowering: 37.3 % of the corpus lowers fully, 0 % fails to lower, and the binding it needs already ships unconsumed #1240 update.
• CI: rust-test.yml now runs lance-graph-dt64 and lance-graph-xlog, workspace members whose tests no workflow ran.

Not in this PR

• The ruff side, crates/ruff_charon: tests on the fixture's LLBC against llbc_calls.py's output, and four disable runs each red in exactly its own test. Its PR on AdaWorldAPI/ruff is pending: the designated branch predates ruff's upstream replay (fix: remove 359 binary files from git (49 MB → 1.2 MB tracked) #121), so it shares no history with ruff main.
• Slice B (resident rows and incremental updates), the body → R2IL lowering, and MCP.

Validation

• The Python tools' own tests: test_tools.py 15/15, test_kythe_entries.py 9/9.
• entries_index.py --check, append_only_gate.py origin/main, and supersession_index.py regenerated.
• sag-ra-probe builds with cargo build --release.

---

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: abc884ec-64b2-49b4-98ee-cf25b46615c6

• 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 2026-11-02T05:53:28.531904Z	c0f81a2	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.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c0f81a2a8c

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".

[chatgpt-codex-connector]

Require call-edge kinds when checking call pins

If the producer emits ordinary /kythe/edge/ref edges at the pinned spans instead of /kythe/edge/ref/call, this checker still reports every pin as passing. query_xrefs requests ALL_REFERENCES and reduces the returned anchors to spans without retaining their edge kinds, so membership in refs does not establish that a call edge exists. Returning ordinary reference anchors at the three committed pin spans reproduces three successful checks. Filter for call-edge kinds before matching pins and add a negative case that replaces call edges with ordinary references, so the validation actually detects loss of the call semantics.

Useful? React with 👍 / 👎.