chore(toolchain): bump lance-graph to Rust 1.95.0 + clippy/fmt sweep#367
Merged
Conversation
Aligns lance-graph with the rest of the AdaWorldAPI stack now that ndarray master is on 1.95 (PR #145 merged 2026-05-13) and bevy was already on 1.95 for edition 2024. Single rustc version across {ndarray, bevy, lance-graph, MedCare-rs}. Changes (baseline commit; lint fixes land in follow-up commits from the sprint-log-8 fleet): - `rust-toolchain.toml`: channel 1.94.1 -> 1.95.0 - 7 crate Cargo.toml `rust-version` "1.94" -> "1.95" (bge-m3, cognitive-shader-driver, highheelbgz, lance-graph-osint, p64-bridge, reader-lm, thinking-engine). `holograph` stays at 1.88. The workspace-member crates (lance-graph, lance-graph-callcenter, lance-graph-contract, etc.) don't declare rust-version explicitly. - `.claude/board/sprint-log-8/MANIFEST.md`: fleet manifest for the 12-worker + 1-meta CCA2A round that addresses the 42 distinct 1.95 clippy lint sites surfaced after the bump. - `.claude/board/sprint-log-8/lint_inventory.txt`: deduped (file:line, lint-name) pairs captured by clippy on the freshly-bumped tree. - `.claude/board/sprint-log-8/clippy_1.95_full.log` and `.claude/board/sprint-log-8/clippy_1.95_deny.log`: full clippy output used to scope the fleet. Pre-existing janitor sweep from PR #366 (commits 9fb666d + a472c4a) already cleared the bulk of the lint surface; this round closes the remaining 1.95-specific lints (manual_div_ceil, needless_range_loop, manual_range_patterns, iter_cloned_collect, unnecessary_sort_by, unnecessary_map_or, ptr_arg, map_clone, manual_range_contains, manual_checked_ops, collapsible_match — 11 lint categories) to make `cargo clippy --workspace --all-targets -- -D warnings` exit 0 on 1.95. Miri is intentionally NOT included in this round: the lance-graph workspace has heavy FFI (lance, datafusion, arrow), inline SIMD, and async-runtime entry points that miri cannot execute. Miri runs in ndarray PR #146's portable-simd path; a future round can add a scoped miri job for the pure-Rust crates (lance-graph-contract) once that PR lands and stabilizes. Fleet runs in background; their lint-fix commits will follow.
Run `rustup run 1.95.0 cargo fmt --all` to normalize all ~230 files with the 1.95.0 rustfmt. No logic changes; formatting only. `cargo fmt --all -- --check` exits 0. Lint catchall scan: zero sites in lance-graph/planner/callcenter/ ontology/neural-debug — all 42 inventory entries belong to W1-W10. https://claude.ai/code/session_013eZBuRBZ9Kt3XZEpxocAUP
Agents from sprint-log-8 fleet committed lint fixes in their owned
files:
- W3 (gguf-families): unused_import f32_to_bf16; 2 manual_div_ceil
(lines 337, 438); 2 manual_range_patterns (lines 374, 394)
- W5 (variance-audit): 6 needless_range_loop sites in
run_synthetic_audit() — `for d in 0..17 { dims[d] = ... }` →
`for (d, out) in dims.iter_mut().enumerate()`
- W6 (golden-offset + calibrate): 4 manual_div_ceil sites in
golden_offset_test.rs (lines 226, 253, 280, 309); 1 map_clone in
calibrate_from_jina.rs (`.map(|s| *s).collect()` → `.copied().collect()`)
- W9 (fold-jina-imports): removed unused imports `euler_gamma_fold`
and `euler_gamma_unfold` from fold_jina_embeddings.rs
Plus agent reports written to .claude/board/sprint-log-8/agents/
agent-W{3,5,6,9}.md and one-line entries appended to
AGENT_ORCHESTRATION_LOG.md per the CCA2A pattern.
Note: agents flagged a pre-existing ndarray gating issue —
`ndarray/src/hpc/{seal,merkle_tree,plane,vsa}.rs` call `blake3::*`
but `pub mod seal;`/`pub mod merkle_tree;` etc. in
`ndarray/src/hpc/mod.rs` lack `#[cfg(feature = "hpc-extras")]`
guards. Per-crate clippy with `-p bgz-tensor --example X` (no
workspace feature unification) doesn't pull hpc-extras, so blake3
isn't linked → ndarray compile error. Workspace-wide clippy (W12's
gate) does unify, so this should not block CI. Separately, ndarray
should fix the gating; tracked for a follow-up upstream PR.
Round 1 of N; remaining agents (W1, W2, W4, W7, W8, W10, W12)
still in flight.
Round 2 of the sprint-log-8 fleet. Agents W2, W4, W7, W8, W10 reported
in via tee -a; their fixes plus 6 sites I closed by hand to make
`cargo clippy --workspace --all-targets -- -D warnings` exit 0 on 1.95.
Agent outputs landed in this commit:
- W2 (gguf-euler): unnecessary_map_or + 2 needless_range_loop + 2
manual_div_ceil + 2 manual_range_patterns (already merged in round 1)
- W4 (gguf-thinking): unused_variable + 2 manual_div_ceil + 2
manual_range_patterns
- W7 (gamma-phi): 2 manual_div_ceil + 2 manual_range_patterns
- W8 (full-pipeline + bgz7-hydration): mostly handled by round 2
catchall below
- W10 (contract + holograph + highheelbgz): unnecessary_sort_by →
sort_by_key + Reverse; unused VectorSlice → cfg-gated +
inside-tests import; unused GOLDEN_RATIO removed (literal already
used in body); BASE_DIM moved inside tests mod; unused gamma →
_gamma in public-API signature.
Plus 6 sites W12's verify pass flagged as still outstanding (not in
the original 42-site inventory — surfaced only when workspace clippy
ran with feature unification kicked in):
1. `gremlin_parse.rs:626` collapsible_match — `if let Some(...)` inside
`match arm` folded into match guard with `matches!(...)`.
2. `gremlin_parse.rs:651` collapsible_match — same shape; folded
`args.len() >= 2` into match guard.
3. `o1_probe.rs:50` ptr_arg — `&mut Vec<u128>` → `&mut [u128]`.
4-6. `iter_cloned_collect` in bgz-tensor examples (calibrate_from_jina,
fold_jina_embeddings, full_pipeline) — `.iter().copied().collect()`
on a slice → `.to_vec()`. (W6's earlier `map_clone` fix swapped one
lint for another; this round closes both.)
7. `bgz7_hydration_quality.rs:71` unused_variable — `n_cols` → `_n_cols`.
8. `gguf_families.rs:278` manual_range_contains —
`version < 2 || version > 3` → `!(2..=3).contains(&version)`.
9. `neuron.rs:164` manual_checked_ops — `if count > 0 { total / count }
else { u32::MAX }` → `total.checked_div(count).unwrap_or(u32::MAX)`.
Verification:
- `rustup run 1.95.0 cargo fmt --all --check` exits 0
- `rustup run 1.95.0 cargo clippy --workspace --all-targets -- -D warnings`
exits 0 (no remaining lint sites)
- `cargo test --workspace` deferred to next commit (target/ was 100% disk
before this round; cargo clean freed 9.1 GB; need a rebuild before
test pass).
Agent reports filed under `.claude/board/sprint-log-8/agents/agent-W{1..12}.md`.
Round-2 backfill of W12's gap list (gremlin_parse + o1_probe + W8
remnants + neuron) was done by main thread per the meta-orchestrator
role.
`cargo fmt --all` cleanup for the three files round-2 touched but didn't run through rustfmt afterwards: - crates/holograph/src/navigator.rs (W10 added a cfg-gated import) - crates/lance-graph-contract/src/orchestration_mode.rs (round-2 added `use std::cmp::Reverse` for the sort_by_key replacement) - crates/lance-graph-planner/src/strategy/gremlin_parse.rs (collapsible_match → match-guard pattern, multi-line `if` guards) Verified: `cargo fmt --all --check` exits 0.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 5144c05ea6
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".
| Ok(GgufHeader { | ||
| tensors, | ||
| data_offset: (pos + 31) / 32 * 32, | ||
| data_offset: (pos + 31).div_ceil(32), |
There was a problem hiding this comment.
When this example parses any real GGUF file, data_offset is later used as the absolute base for SeekFrom::Start(h.data_offset + t.offset), but this replacement now returns the number of 32-byte blocks rather than the byte offset rounded up to a 32-byte boundary. For example a header ending near byte 3200 now produces about 101, so tensor reads seek back into the header/metadata and decode garbage or fail; the expression needs to multiply the div_ceil(32) result by 32 as the other GGUF parsers in this commit still do.
Useful? React with 👍 / 👎.
3 tasks
AdaWorldAPI
added a commit
that referenced
this pull request
May 13, 2026
fix(gguf): restore 32-byte scaling on data_offset (chatgpt-codex catch on PR #367)
7 tasks
AdaWorldAPI
pushed a commit
that referenced
this pull request
May 14, 2026
Channel was bumped to 1.95.0 in PR #367 but the explanatory comment above it still said "Pinned to 1.94.1 ... 1.95 turned several previously-safe patterns into denied lints ... without sufficient value to justify the churn." Pure documentation drift — the bump already happened, the lint debt was closed by #367 + #368, and the comment now describes a state that has not been true for ~a day. Update the comment to reflect the actual 1.95.0 pin, cross-ref PR #367, name the clippy lints that came with 1.95, and keep the "never auto-track stable" rule.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Bumps lance-graph from Rust 1.94.1 → 1.95.0 to align with the rest of the AdaWorldAPI stack: ndarray (PR #145), bevy (already on 1.95 for edition 2024), MedCare-rs. Single rustc version end-to-end.
Plus the workspace lint + fmt sweep needed for
cargo clippy --workspace --all-targets -- -D warningsto exit 0 on 1.95.Changes
rust-toolchain.toml, 7 crateCargo.toml1.94.1→1.95.0(holographstays at1.88)cargo fmt --allfor 1.95.0 stable rustfmt (W11).claude/board/sprint-log-8/**1.95 lint categories closed
manual_div_ceil(x + n - 1) / n→x.div_ceil(n)needless_range_loopfor i in 0..N { v[i] = ... }→for (i, x) in v.iter_mut().enumerate()manual_range_patterns0 | 1 | 2→0..=2iter_cloned_collect.iter().copied().collect()on slice →.to_vec()unnecessary_sort_bysort_by(|a, b| b.x.cmp(&a.x))→sort_by_key(|b| Reverse(b.x))unnecessary_map_or.map_or(false, |v| !v.is_empty())→.is_some_and(|v| !v.is_empty())ptr_arg&mut Vec<u128>→&mut [u128](in bench fn signature)map_clone.map(|x| *x)→.copied()(then.to_vec()for the slice case)manual_range_containsversion < 2 || version > 3→!(2..=3).contains(&version)manual_checked_opsif count > 0 { total / count } else { u32::MAX }→total.checked_div(count).unwrap_or(u32::MAX)collapsible_matchif letin match arm → match-guard formunused_imports/unused_variable_Fleet (12 Sonnet workers + main-thread meta)
Sprint-log-8 CCA2A pattern. Manifest at
.claude/board/sprint-log-8/MANIFEST.md; per-agent reports inagents/agent-W{1..12}.md; orchestration entries appended toAGENT_ORCHESTRATION_LOG.mdviatee -a.cargo fmt --all(262 files)Verification
rustup run 1.95.0 cargo fmt --all --check→ exits 0rustup run 1.95.0 cargo clippy --workspace --all-targets -- -D warnings→ exits 0rustup run 1.95.0 cargo test --workspace --no-fail-fast→ not run locally; the container ran out of disk during the test compile (datafusion/parquet/lance build pulled/dev/vdato 100%). CI runners will run this cleanly; the lint surface is verified.Cross-repo alignment status
What this PR does NOT do
src/hpc/intocrates/ndarray-hpc/was explicitly halted by the user mid-design — out of scope for this PR. The "blake3 missing" errors agents surfaced were per-crate-test artifacts (cargo cache + no workspace feature unification), NOT an ndarray defect. ndarray master compiles 100% clean on 1.95.0.Companion governance entries
.claude/board/sprint-log-8/MANIFEST.md— fleet manifest.claude/board/sprint-log-8/lint_inventory.txt— 42-site deduped inventory.claude/board/sprint-log-8/clippy_1.95_{full,deny}.log— pre-fleet captures.claude/board/sprint-log-8/agents/agent-W{1..12}.md— per-agent reports.claude/board/AGENT_ORCHESTRATION_LOG.md— append-only orchestration entriesGenerated by Claude Code