Skip to content

chore(deps): bump Rust 1.98, crates, cookbook images, npm locks and CI actions - #1453

Merged
skakri merged 1 commit into
mainfrom
chore/deps-bump
Sep 23, 2026
Merged

skakri merged 1 commit into
mainfrom
chore/deps-bump

Conversation

@skakri

@skakri skakri commented Sep 23, 2026

Copy link
Copy Markdown
Member

Dependency audit and upgrade pass.

Security

  • cargo update clears RUSTSEC-2026-0306 (unsound faster-hex AVX2 path, via gix) by moving to faster-hex 0.10.1. The remaining cargo-audit output is unmaintained-crate warnings from transitive dependencies, which deny.toml already allows.

Rust

Crates

Crate Change Migration
fastembed 6.1 → 7.1 InitOptions → TextInitOptions
model2vec-rs 0.2.1 → 0.3.0 none
tree-sitter 0.26.11 → 0.27.0 see below
rmcp 3.3 → 3.4.1 ServerInfo/ClientInfo → ServerConfig/ClientConfig

tree-sitter 0.27: Node::kind() now returns a tree-borrowed &'tree str instead of &'static str. NodeSpans outlive their parse tree (they are kept in RefineMember), so walk_spanned interns kinds through a small leak-once set, which stays bounded because each grammar's node kinds are a fixed set. Also: the QueryMatch::captures() accessor, and redundant u32::try_from on child indices removed (indices are u32 end to end now).

Evidence that nothing changed:

  • Indexing the same corpus (rag-rat crates/, editors/, cookbook/ plus the time crate) with origin/main and with this branch gives byte-identical symbols, edges and chunks.
  • Embeddings from bge-small, MiniLM, jina-code and potion-retrieval-32M (including unicode, unknown-token and empty inputs) are bit-identical before and after, so no embedding model version bump is needed.

Cookbook

  • ollama 0.32.1 → 0.34.3: /v1/embeddings output for all-minilm is identical to 0.32.1 when run locally.
  • vLLM v0.25.1 → v0.30.0: --runner pooling, --max-model-len, --max-num-seqs, --host/--port and both served routes still exist in the v0.30.0 source. Not run on a GPU. The Modal vLLM weight cache is keyed by image, so the first box re-downloads once.
  • infinity is already at its latest release (0.0.77).
  • Refreshed the npm lockfile within existing ranges.

Editors and CI

  • VS Code extension lockfile refreshed within ranges. TypeScript stays on 5.x because typescript-eslint supports <6.1.
  • code-server 4.131 → 4.138.
  • tools/oracle-toolchain lock intentionally not refreshed: it would pull TypeScript 7 in under the pinned scip-typescript.
  • cargo-dist 0.32 → 0.33 (regenerated release.yml; only the installer URL changes).
  • release-plz action → v0.5.139 (SHA-pinned).
  • Current majors of checkout, upload-/download-artifact, setup-node, cache, github-script, codecov, dawidd6/action-download-artifact and webfactory/ssh-agent. Release notes checked against our usage: no require('@actions/github'), and codecov already uses files:.

Verification

  • cargo clippy --workspace --all-targets with --all-features and with --no-default-features, both -D warnings.
  • cargo nextest run --workspace --profile ci: 5846/5846 without default features, 5848/5848 with default features.
  • cargo +nightly fmt --all --check.
  • Cookbook: typecheck and tests. VS Code extension: types, lint, production build and tests.

…I actions

Rust
- MSRV 1.96 -> 1.98; bench container and docs track 1.98.0.
- Oracle rust-analyzer pin 1.96.0 -> 1.98.0: 1.98's rust-analyzer emits
  the rust-time SCIP without the next_solver panic from #553.

Crates
- cargo update, which clears RUSTSEC-2026-0306 (faster-hex 0.10.1).
- fastembed 7.1 (InitOptions -> TextInitOptions), model2vec-rs 0.3:
  embeddings from all four local models (bge-small, MiniLM, jina-code,
  potion-32M) are bit-identical to the previous versions, so no model
  version bump.
- tree-sitter 0.27: Node::kind() now borrows from the tree, so the clones
  normalizer interns kinds to keep NodeSpan independent of the tree;
  QueryMatch::captures() accessor; child indices are u32 end to end.
  Indexing the same corpus before and after yields byte-identical
  symbols, edges and chunks.
- rmcp 3.4 (ServerInfo/ClientInfo -> ServerConfig/ClientConfig).

Cookbook and editors
- ollama 0.32.1 -> 0.34.3 (embeddings identical to 0.32.1 locally),
  vLLM v0.25.1 -> v0.30.0 (every flag and route the recipe passes is
  present in v0.30.0); infinity is already at its latest release.
- Refreshed cookbook and VS Code extension npm locks; code-server 4.138.

CI
- cargo-dist 0.33 (regenerated release.yml), release-plz action v0.5.139,
  and current majors of checkout, upload/download-artifact, setup-node,
  cache, github-script, codecov, dawidd6 download-artifact and ssh-agent.
@skakri
skakri merged commit 61878b8 into main Sep 23, 2026
30 checks passed
@skakri
skakri deleted the chore/deps-bump branch September 23, 2026 20:11
@github-actions

Copy link
Copy Markdown
Contributor

SCIP oracle — resolution report

Heuristic→compiler edge resolution per corpus. Δ compares resolved-after to the main baseline (only when the corpus profile + tool version match).

corpus tool edges resolved (heuristic → compiler) precision recall monikers Δ vs main
c-libuv scip-clang 12894 37.9% → 48.6% 90.2% 55.3% 679 +0.0pp
cpp-yaml scip-clang 9423 43.2% → 72.9% 82.7% 41.0% 1056 +0.0pp
py-django scip-python 12937 28.8% → 73.5% 96.7% 84.1% 3701 +0.0pp
py-rich scip-python 8383 27.6% → 92.3% 95.0% 81.3% 1196 +0.0pp
rust-time rust-analyzer 11040 37.6% → 56.9% 82.2% 83.6% 574 — (baseline incomparable)
ts-rxjs scip-typescript 7875 39.6% → 94.9% 87.3% 77.4% 631 +0.0pp

resolved = Exact/Syntactic + compiler upgrades + resolved-external, over edge candidates with a callee range. precision/recall are the oracle eval metrics.

@github-actions

Copy link
Copy Markdown
Contributor

🐰 Bencher Report

Projectrag-rat
Branchchore/deps-bump
Testbedhetzner-bigmem

⚠️ WARNING: No Threshold found!

Without a Threshold, no Alerts will ever be generated.

Click here to create a new Threshold
For more information, see the Threshold documentation.
To only post results if a Threshold exists, set the --ci-only-thresholds flag.

Click to view all benchmark results
BenchmarkEstimated Cyclescycles x 1e6Instructionsinstructions x 1e6L1 Hitshits x 1e6LL Hitshits x 1e6RAM Hitshits x 1e3Total read+writereads/writes x 1e6
rag_pipeline::pipeline::index cargo_resolver:resolver_config()📈 view plot
⚠️ NO THRESHOLD
1,767.17 x 1e6📈 view plot
⚠️ NO THRESHOLD
1,100.88 x 1e6📈 view plot
⚠️ NO THRESHOLD
1,542.39 x 1e6📈 view plot
⚠️ NO THRESHOLD
36.12 x 1e6📈 view plot
⚠️ NO THRESHOLD
1,262.07 x 1e3📈 view plot
⚠️ NO THRESHOLD
1,579.77 x 1e6
rag_pipeline::pipeline::query_cold cargo_resolver:resolver_built_config()📈 view plot
⚠️ NO THRESHOLD
266.15 x 1e6📈 view plot
⚠️ NO THRESHOLD
176.16 x 1e6📈 view plot
⚠️ NO THRESHOLD
250.66 x 1e6📈 view plot
⚠️ NO THRESHOLD
2.88 x 1e6📈 view plot
⚠️ NO THRESHOLD
31.85 x 1e3📈 view plot
⚠️ NO THRESHOLD
253.57 x 1e6
rag_pipeline::pipeline::query_warm cargo_resolver:resolver_index()📈 view plot
⚠️ NO THRESHOLD
238.17 x 1e6📈 view plot
⚠️ NO THRESHOLD
157.22 x 1e6📈 view plot
⚠️ NO THRESHOLD
224.29 x 1e6📈 view plot
⚠️ NO THRESHOLD
2.55 x 1e6📈 view plot
⚠️ NO THRESHOLD
31.83 x 1e3📈 view plot
⚠️ NO THRESHOLD
226.88 x 1e6
🐰 View full continuous benchmarking report in Bencher

@github-actions

Copy link
Copy Markdown
Contributor

🐰 Bencher Report

Projectrag-rat
Branchchore/deps-bump
Testbedhetzner-bigmem
Click to view all benchmark results
BenchmarkLatencyBenchmark Result
seconds (s)
(Result Δ%)
Upper Boundary
seconds (s)
(Limit %)
index_time/full_rebuild_cargo📈 view plot
🚷 view threshold
6.41 s
(+4.26%)Baseline: 6.15 s
7.20 s
(88.97%)
🐰 View full continuous benchmarking report in Bencher

@codecov

codecov Bot commented Sep 23, 2026

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ All tests successful. No failed tests found.

📢 Thoughts on this report? Let us know!

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.

1 participant