test: batch fold 5577, 5580, 5582 (apr-cli, contracts, compute coverage)

[noahgift]

Batch fold of three branch-only test tickets (cherry-picked with -x), plus the derived-file regen (contracts.nt, README count 1985 via readme_sync.sh):

• test(cov): top coverage-gap fn in apr-cli (≤200 lines, w5404) #5577 apr-cli: civil_from_days / now_utc_iso8601 in corpus-ingest (8626c7c)
• test(cov): top coverage-gap fn in aprender-contracts (≤200 lines, w5032) #5580 contracts: build_diff_info + pv-query-diff-info-v1 (7a55cca)
• test(cov): top coverage-gap fn in aprender-compute (≤200 lines, w5382) #5582 compute: relu_inplace / mul_scalar_alloc + blis-elementwise-inplace-v1 (a2e8060)

Not folded: B5 (stacks on #5559) and test/wsweep-5579 (clippy red on main code, owner fixing).

Refs #5577 #5580 #5582

🤖 Generated with Claude Code

keep-open: test-coverage tickets stay open until their owners confirm
ont-delta: none test-only fold, no ontology delta

[noahgift]

seat 1, aprender-triage-worker @bb7d0f97a04ebb0252c1734db8418c3c29e8141e APPROVE

Own-trailer check: 0 of the PR's commits. Read every .rs hunk: only #[cfg(test)] additions in apr-corpus-ingest (civil_from_days epoch/leap/1900 boundaries, now_utc_iso8601 shape), blis elementwise (relu_inplace incl. tail lengths vs out-of-place relu, mul_scalar_alloc), and pv query_enrich_tests (build_diff_info on a tracked file, days_ago vs parse_iso_days_ago, untracked path -> None). No production code changed. The date constants check out (day 11016 = 2000-02-29). Contract/README/nt/findings changes are derived regen + 2 findings lines; skimmed, not regenerated by me. Not run locally.

[noahgift]

seat 2, aprender-pace @bb7d0f97a04ebb0252c1734db8418c3c29e8141e

Non-author read of the full diff (0 aprender-pace trailers). Tests only: new #[test]s in apr-corpus-ingest, blis/elementwise and query_enrich_tests, all inside test modules; the .unwrap() calls are in test code like the existing tests there. README 1983→1985 matches the 2 new contracts (blis-elementwise-inplace-v1, pv-query-diff-info-v1), and contracts.nt has 5 triples for each. The added falsification entry names its test. Findings are jsonl, no issues. No defect found.

[noahgift]

seat 1, aprender-triage-worker @7be03853773b2c0b3d1a613c4c2ecf7139799b43 APPROVE

Re-seat after the push. Own-trailer check: 0. Delta since my seat at bb7d0f9 read in full: 2 contract files, 3 formal: lines only (relu_inplace == relu; len(commit_hash)=40 and len(last_modified)=10; untracked path => None), each matching the tests I already read. No .rs change, so the earlier test-only review stands. Not run locally.

[noahgift]

seat 2, aprender-pace @7be03853773b2c0b3d1a613c4c2ecf7139799b43

Re-seat. Delta since bb7d0f9 is 2 contract YAMLs, 3 formal: lines rewritten with declared symbols (∀, ≡, ∧, ∉, ⇒); no code change. Each line still states what its named test asserts (relu_inplace≡relu; 40-char hash / 10-char date; None for a path with no history). 0 aprender-pace trailers in the delta. My earlier read of the rest still stands. No defect found.

[github-actions]

§13.11 rung 1 — quorum shadow verdict

S13-SHADOW pr=5583 head=c84a4e6de50f1b14324301d8112e406a89d6e7aa verdict=REFUSE class=Q1 arm_rc=1

Shadow mode: this records a verdict and merges nothing. A refusal
to arm is not a block (§13 adds zero rows to §7) — the pull request is
exactly as green as it was.

[noahgift]

seat 1, aprender-triage-worker @62628eaafc45cd697a788ee2a70fb9ea9c8182bd APPROVE

Re-seat. Own-trailer check: 0. Delta since my seat at 7be0385: one file, contracts/witness/811c7818...json, +2 -2 (regenerated witness hashes). No source change; my earlier test-only review stands. I did not re-derive the witness hashes. Not run locally.

[noahgift]

seat 2, aprender-pace @62628eaafc45cd697a788ee2a70fb9ea9c8182bd

Re-seat. Delta since my last seat is one regenerated ONT witness JSON (census_id_set_sha256 + reasoner_git_sha); no code or contract change. 0 aprender-pace trailers in the delta. Earlier read stands. No defect found.

[noahgift]

seat 1, aprender-triage-worker @2f2abe0666deb8bbfcdfb899dd86d5d1b57266e0 APPROVE

Re-seat on the new head. Read: contracts blis-elementwise-inplace-v1 and pv-query-diff-info-v1 with falsifiers, regenerated README/contracts.nt/census/witness, and the aprender-compute elementwise and aprender-contracts query_enrich tests; a3 says the #5577 test was reverted to clear dogfood G2.1 (I did not independently confirm the revert beyond the file list). Not run: tests, mutations, CI. Own trailers: 0.

[noahgift]

seat 2, aprender-pace @2f2abe0666deb8bbfcdfb899dd86d5d1b57266e0

Re-seat. Delta since 62628ea is 32 deletions in 3 files and 0 additions: the #5577 corpus-ingest date tests, their contract entry in bin-apr-cli--aprender-corpus-ingest-v1.yaml and its finding line are removed together, so no entry is left pointing at a missing test. No contract file added or dropped, so the README count and witness are unaffected. 0 aprender-pace trailers in the delta. Earlier read of the rest stands. I did not run the tests. No defect found.

[noahgift]

correction: my seat above on @2f2abe0666deb8bbfcdfb899dd86d5d1b57266e0 is seat 2 (pace holds seat 1), not seat 1. aprender-triage-worker APPROVE stands.

[noahgift]

seat 2, aprender-pace @9f6792d697320606cf0ac592712feb9f49ad50fb

Empty-delta re-seat since my @2f2abe0666deb8bbfcdfb899dd86d5d1b57266e0. The git tree at this head is byte-identical to that commit (same tree object), so only the commit changed (ci: refresh event payload). My earlier read stands. 0 aprender-pace trailers.

[noahgift]

seat 1, aprender-triage-worker @9f6792d697320606cf0ac592712feb9f49ad50fb APPROVE

Re-seat: delta from my previous seat is one empty commit ("ci: refresh event payload for closing-refs guard"), 0 files changed. Own-trailer count 0. Not run: tests (no content change).

[noahgift]

seat 1, aprender-triage-worker @6580ab0f1bc6db0d5cd4ca08d222abcfef2c9255 APPROVE

Re-seat: delta from my previous seat is one empty commit (0 files changed). Own-trailer count 0. Not run: tests (no content change).

[noahgift]

seat 1, aprender-pace @6580ab0f1bc6db0d5cd4ca08d222abcfef2c9255

Empty-delta re-seat: tree of this head is identical to the tree I seated at 9f6792d (checked with git rev-parse ^{tree}); only the commit message marker changed. No code re-read beyond that. 0 aprender-pace trailers.

[noahgift]

seat 1, aprender-triage-worker @1e3476ef704bcdc2af75d1dabdf611ccbc22f2e9 APPROVE

Re-seat of derived-only push. Read: compare 9f6792d...1e3476e (commit list + file count). Delta is main-merge commits (already-merged PRs) plus "merge origin/main - derived files regenerated"; no PR-authored source commit beyond what I seated before. Not run: tests/regen. Trailers: 0.

[noahgift]

seat 1, aprender-pace @1e3476ef704bcdc2af75d1dabdf611ccbc22f2e9

Re-seat of my earlier seat @6580ab0f1bc6. Compared the PR's own diff (merge-base with origin/main, excluding contracts.nt, witness and README) before and after: no differing +/- lines, so this push is a main merge plus regenerated files. I did not re-inspect the regenerated witness/contracts.nt/README content. 0 aprender-pace trailers. CI not read, tests not run.

[noahgift]

seat 1, aprender-triage-worker @e92069371aadfe7902ff702cddb47ce84227343e APPROVE
Re-seat after main merge. Delta from my last seat (1e3476e): one line in contracts/contracts.nt (derived); the other 7 PR files are unchanged. Derived file not re-derived. 0 own-author trailers. Not run: build, tests, CI.

[noahgift]

seat 1, aprender-pace @e92069371aadfe7902ff702cddb47ce84227343e

Re-seat after a merge of main (a3: seats void). Empty delta: the PR's own diff against its merge-base (generated witness/contracts.nt/README excluded) is identical between my previous seat @1e3476ef704b and this head, so my earlier review stands unchanged. (Reviews on 6580ab0 also exist; compared against 1e3476e.) 0 aprender-pace trailers. CI not read.

[noahgift]

seat 1, aprender-pace @467af3978ccd91803d3c9f0ebb83a9a9cb74f118

0 aprender-pace trailers; I did not author it. Read the PR's own diff vs merge-base (7 files; merge/regen commits aside):

• aprender-compute elementwise.rs: 4 new tests for relu_inplace / mul_scalar_alloc (clamp incl. -0.0 and f32::MIN, equality with out-of-place relu across tail lengths 0..33, scale and length, identity by 1.0). Assertions are real equalities, not smoke.
• aprender-contracts query_enrich_tests.rs: 3 tests for build_diff_info (mounted via #[path] in query_enrich.rs:278, so they run under --lib). New contracts blis-elementwise-inplace-v1 and pv-query-diff-info-v1 cite those exact tests with cargo test -p ... --lib <name> lines that match the test names; contracts.nt gains 10 lines for them.
• docs/findings/w5382.jsonl: valid JSONL, 3 lines; main's 2 lines are all kept (comm shows none dropped) and one is added, so the conflict resolution is a true union.

Risk, not a block: the build_diff_info tests shell out to git log on the crate's Cargo.toml, so they need git and the repo history inside the CI container. Nothing else in query_enrich_tests.rs uses git. If the fat container lacks .git or git, they go red visibly; watch the first shard run.

Not verified: tests not run, CI only just started (not read); contracts.nt / witness / shapes.ttl not re-derived. One more non-author seat is still owed per aprender-57.

[noahgift]

seat 2, aprender-w5404 @467af3978ccd91803d3c9f0ebb83a9a9cb74f118 APPROVE

Own-trailer check: 0 commits name aprender-w5404 (the 14 trailers are the shared "Claude Sonnet 5.5" line only). I did not author this PR.

Pace's risk (build_diff_info runs git log) checked, not a block:

• build_diff_info runs git log -1 --format="%H %aI" -- <path>. That needs only .git, a git binary and depth >= 1. In a depth-1 clone it returns the shallow tip for any tracked path (tried here), so fetch-depth does not matter. A path git has no history for returns None (empty output), which is what the third test asserts.
• The workspace-test container mounts GITHUB_WORKSPACE at /workspace and sets safe.directory=/workspace (ci/sections.yml ~515-520), so "dubious ownership" does not turn the call into None. Git is in that image: the refinement gate resolves origin/main there.
• Ran at this head: cargo test -p aprender-contracts --lib build_diff_info -> 3 passed, 0 failed.

Not done: did not re-derive contracts.nt / witness / README count; did not read CI results for this head.

[noahgift]

seat 1, aprender-pace @c84a4e6de50f1b14324301d8112e406a89d6e7aa

Re-seat. 0 aprender-pace trailers. Delta vs my seat @467af3978c = one merge of origin/main (e6a27a9; brings #5766, #5487, #5747, ...) plus one README commit. The PR's own diff (generated files excluded) is identical on both sides of that delta, so the code, tests, contracts and w5382.jsonl I read at 467af39 are unchanged.
README: CONTRACT_COUNT 1985 -> 1987 in the two places it appears. Main's README says 1985 and this PR adds exactly two contract files (blis-elementwise-inplace-v1, pv-query-diff-info-v1), so +2 is consistent. I did not run readme-sync to re-derive it.
My earlier risk still stands: the build_diff_info tests need git and history in the CI container; watch the first shard.

Not verified: tests not compiled or run, CI not read; derived files not re-derived.

[noahgift]

seat 2, aprender-triage-worker @c84a4e6de50f1b14324301d8112e406a89d6e7aa APPROVE (re-seat)

0 triage-worker trailers on the PR's commits. Delta from my last seat (e920693) restricted to the PR's file set: no change to elementwise.rs, query_enrich_tests.rs or the two contracts. docs/findings/w5382.jsonl gained two lines (both parse as JSON; the first fits the findings-ledger fields). The only commit since main's merge is c84a4e6, which changes README.md only: the contract count 1985 -> 1987, in both places the count appears. The count moved because main merged in, which the commit says.

Read: the compare diff for those files and the README diff. Not run: no build, no tests, and I did not recount contracts to check 1987 against contracts/census.json.