neural/trident/audit/current-workspace-continuation.md

Current workspace continuation after FINAL5

2026-09-12. Live-source test receipts following the owner-only deep-stack move, alternate XField-width repair and formal helper support. This is not a frozen candidate, an installed-platform smoke, or a replacement for the separate FINAL5 198-proof gate. No ignored test was enabled. Production files were not edited during this validation; the formal owner corrected one stale test.

Trident

Command:

CARGO_BUILD_JOBS=2 RAYON_NUM_THREADS=4 cargo test --manifest-path trident/Cargo.toml --workspace --release --features neural --locked -- --test-threads=1

First receipt /tmp/current-trident-workspace-continuation.log: exit 101, 723 passed, one failed in the library harness. The existing verify::solve::tests::test_inlined_function_verification expected Unknown for a now-supported helper asserting x + 0 == x; the solver correctly returned Safe. The formal owner updated that expectation and added a false helper x + 1 == x called with zero, requiring StaticViolation. The owner first validated the focused regression separately.

Complete repeated command /tmp/current-trident-workspace-continuation-final.log: exit 0; 894 passed, 0 failed, 0 ignored across 19 test/doc result groups. No Rust compilation warning appeared. Three intentionally rendered language warning messages occur in the diagnostic unit tests (unused x, unused y, redundant range check); zero compiler warnings does not mean an empty log.

Trisha

Command:

CARGO_BUILD_JOBS=2 RAYON_NUM_THREADS=4 cargo test --manifest-path trisha/Cargo.toml --workspace --all-features --release --locked -- --test-threads=1

/tmp/current-trisha-workspace-continuation.log: all 57 completed ordinary runtime/unit/integration result groups passed, 450 passed, 0 failed, 5 ignored. The overall command then exited 101 during Honeycrisp doctest compilation, before a doctest result: E0463 could not find crate trident, with secondary Prover/Verifier name-resolution errors.

The exact --extern artifact named by that failing invocation, target/release/deps/libtrident-97717fb0d6e8c982.rlib, was absent when inspected. Another agent had compiled a different Trisha integration/feature selection in the same target directory during the broad run. Artifact turnover is therefore a plausible cause, not a demonstrated compiler/source defect or a proven race reproduction. No cleanup was performed by this reviewer; root confirmed no cleanup on its side. No dependency, implementation or doctest was disabled.

An isolated Honeycrisp --all-features --release --locked --doc retry rebuilt its dependency selection and passed (zero doctests); metadata diagnostics were retained in /tmp/current-honeycrisp-doctest-diagnosis.log. The exact workspace all-features documentation gate is repeated after other Trisha builds stop; its final receipt is recorded below. The earlier 450 ordinary test passes are preserved rather than overwritten by a zero-test doctest receipt.

The five explicitly ignored gates are the genuine custom-lock transaction prepare/mock gateway test, its transaction/binding mutation counterpart, the typed-entry proof, imported-generic proof, and full recursive outer proof. They remain separate gated proofs with their own required fixtures/resources.

No Rust compilation warnings appeared in the ordinary suite. Four source warnings were intentionally printed by existing fixture compilation: vm.core.convert twice, vm.crypto.hash once, os.neptune.xfield once. These are distinguished from Rust warnings and are not silently omitted.

Final documentation retry:

CARGO_BUILD_JOBS=2 RAYON_NUM_THREADS=4 cargo test --manifest-path trisha/Cargo.toml --workspace --all-features --release --locked --doc -- --test-threads=1

/tmp/current-trisha-workspace-doc-retry.log: exit 0, all four documentation harnesses (trisha_honeycrisp, trisha_neptune, trisha_rs, trisha_wgpu) compile successfully, each containing zero doctest examples. No warning was emitted. This closes the earlier failed documentation compilation separately from the preserved 450 ordinary test passes; it does not invent four test cases.

Literal per-harness counts, exit codes and SHA256 log identities are in current-workspace-continuation.json. Tools were host Homebrew Rust/Cargo/rustdoc 1.95.0 on macOS ARM64. Frozen candidate native MSRV/platform/proof receipts remain separate evidence. No source defect beyond the reviewed stale formal test was repaired in this run.

Final all-target/all-feature checks

Both commands completed with exit 0 and no Rust warnings:

CARGO_BUILD_JOBS=2 cargo check --manifest-path trident/Cargo.toml --workspace --all-targets --all-features --release --locked
CARGO_BUILD_JOBS=2 cargo check --manifest-path trisha/Cargo.toml --workspace --all-targets --all-features --release --locked

Logs /tmp/current-trident-all-target-check.log (57.64s Cargo duration) and /tmp/current-trisha-all-target-check.log (58.66s) are hash-bound in the JSON receipt. Trisha check began only after the tagged genuine proof owner confirmed its command/monitor completed and released the shared target directory. No production code changes were required by these final checks.

External dependency deltas were additionally reviewed in external-dependency-continuation.md. Their separate cryptographic/source-freeze requirements are not replaced by a successful Cargo check.

Homonyms

warriors/joy/audit/current-workspace-continuation
Current workspace continuation — 2026-09-12 This receipt concerns the live Joy workspace after the bounded artifact readers and scalar-helper formal analysis changes. Frozen FINAL5 sources and binaries are unchanged. It is separate from the root's frozen 198-proof baseline run. Commands and…

Graph