toolchain: Rust 1.99, lance 13, lancedb 0.40 (+ clippy 1.99 / EdgeFacet fixes)

[AdaWorldAPI]

Moves q2 to Rust 1.99, Lance 13 and lancedb 0.40, matching lance-graph main, and fixes everything that bump broke.

Commits

• 5495210f: version pins.
  • rust-toolchain.toml and the Dockerfile move from 1.98.1 to 1.99.0.
  • In cockpit-server, lance goes from =12.0.0 to =13.0.0 and lancedb from =0.39.0 to =0.40.0; lancedb 0.40 still needs remote.
  • The stub crate q2-ndarray is bumped too. arrow 58 and datafusion 54 are unchanged.
• f5075c6c: Cargo.lock. Re-resolved: every lance-* crate moves from 12.0.0 to 13.0.0, and Lance 13's new transitive dependencies are added (foyer-*, datasketches, lru, mixtrics).
• efa42863: the workspace now passes cargo clippy -- -D warnings on 1.99.
  • double_must_use is allowed in [workspace.lints.clippy], because #[async_trait] adds #[must_use] to methods that already return a must-use future.
  • quarto-publish gets the same allow at crate level, since it doesn't inherit [workspace.lints].
  • In pampa's readers/qmd.rs, the stream is now reborrowed (&mut *output_stream) instead of taking &mut of a &mut, so the binding no longer needs mut.
  • In osint-bake and cockpit-server/osint_gotham, lance-graph main removed EdgeBlock's in_family/out_family fields. EdgeBlock is now EdgeFacet, an opaque [u8; 16] (lance-graph ISS-EDGE-BLOCK-WAS-A-SECOND-TYPE-FOR-THE-SAME-FACET).
  • Both crates already treated the block as one flat array of 16 mixin bytes: in_family[0..12] followed by out_family[0..4]. They now write as_bytes_mut()[k] and read as_bytes().iter(). The stored bytes are identical.
• 1e422e0a: quarto-publish enables url's serde feature. It serializes url::Url, but compiled only inside the full workspace, where another crate turns url/serde on. cargo test -p quarto-publish failed with five E0277 errors. This is also missing on quarto-dev/q2 upstream.

Verified locally

Run on 1.99.0, with debug info off and incremental compilation off, against lance-graph main 7db81de.

cargo xtask verify --skip-hub-build results:

• Steps 1–3 pass: custom lints plus clippy (-D warnings), the cargo fmt check, and the workspace build with warnings denied.
• Step 4 (tree-sitter grammars) could not run because the tree-sitter CLI isn't installed in this environment. This PR doesn't touch the grammars.

The full cargo nextest run --workspace could not run either: linking every test binary needs more disk than this container has. I ran the changed crates instead:

crate	result
cockpit-server (lance 13)	205 passed, 3 skipped
quarto-publish	101 passed
pampa + osint-bake	4046 passed, 4 failed

The 4 failures are pandoc comparison tests (test_html_writer, test_json_writer, unit_test_corpus_matches_pandoc_{markdown,commonmark}). They panic with "Pandoc version is not suitable for testing"; this container has pandoc 3.1.3.

Upstream note

Compared with quarto-dev/q2: the fork diverged at df029875b (2026-06-29). Since then upstream has 1931 commits and the fork has 431. A trial git merge-tree of upstream into this branch conflicts in 4 files: test-suite.yml, .gitignore, Cargo.lock (regenerable), and pampa/src/readers/qmd.rs. Backups of this branch are at backup/ccr-edc85bd9-89tiwe-{1,2}.

🤖 Generated with Claude Code

https://claude.ai/code/session_017YPZhKpSxN7eSxk7VH2vzB

---

Generated by Claude Code

[coderabbitai]

Note

Currently processing new changes in this PR. This may take a few minutes, please wait...

⚙️ Run configuration

• Configuration used: Organization UI
• Review profile: CHILL
• Plan: Essentials
• Run ID: 5cf32cd0-2f5b-472d-bed9-adc8e945e675

---

📥 Commits

Reviewing files that changed from the base of the PR and between 138ef5a and 1e422e0.

---

⛔ Files ignored due to path filters (1)

• Cargo.lock is excluded by !**/*.lock

---

📒 Files selected for processing (10)

• Cargo.toml
• Dockerfile
• crates/cockpit-server/Cargo.toml
• crates/cockpit-server/src/osint_gotham.rs
• crates/osint-bake/src/lib.rs
• crates/pampa/src/readers/qmd.rs
• crates/quarto-publish/Cargo.toml
• crates/quarto-publish/src/lib.rs
• crates/stubs/q2-ndarray/Cargo.toml
• rust-toolchain.toml

---

 ___________________
< Look ma, no bugs! >
 -------------------
  \
   \   \
        \ /\
        ( )
      .( o ).

✨ Finishing Touches

📝 Generate docstrings

• Commit to this branch
• Create a new PR

---

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Warning

Billing warning: we have not been able to collect payment for this subscription for more than 72 hours. Please update the payment method or pay any pending invoices in Billing to avoid service interruption.

---

_{Comment @coderabbitai help to get the list of available commands.}

[AdaWorldAPI]

Run test suite (ubuntu-latest) fails at the Rustfmt (check) step, but not on q2 code. Every diff it reports is in the lance-graph sibling checkout, in 13 files under lance-graph/crates/lance-graph-lane-fold and lance-graph/crates/lance-graph-arm-discovery. Those files aren't rustfmt-clean on lance-graph main (they came in with #1381). cargo fmt --all reaches them, as the comment above the step in test-suite.yml explains.

I opened a fix in AdaWorldAPI/lance-graph#1451. It only reformats those 13 files, with no code changes. With that branch checked out as the sibling, cargo fmt --all -- --check in q2 exits 0 locally. I can't port it into this PR, because the sibling is checked out from lance-graph main. Once #1451 merges, re-running this job should clear the gate.

---

Generated by Claude Code

[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 2026-10-10T14:13:07.572527Z	1e422e0	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.