[jakeberv]

Finalize the CRAN submission comments and spelling dictionary, keep the RNG test within CRAN's two-core limit, and remove the Windows temporary-file NOTE.

The RNG test still compares two and three real Future chunks, scheduled on at most two workers, and asserts identical results and final caller RNG state. The Windows NOTE came from the separate test that deliberately kills a background worker: forced termination bypasses Rscript's normal temporary-file cleanup. That test now gives its workers a test-owned temporary directory with scoped cleanup. It retains real cancellation and all eight existing assertions.

No production fitting or RNG code changed. No tests, warnings, coverage requirements, or check stages were disabled. Temporary diagnostic instrumentation was removed.

Validation

• Spelling check including vignettes: zero findings after the dictionary update.
• RNG regression: original test fails on the three-worker warning; corrected test passes. A diagnostic mutation bypassing per-item seeding still makes it fail.
• Windows diagnostic: forced cancellation leaves Rscript… files; cooperative shutdown leaves none (log). After the fix, the full Windows check reports Status: OK, including the detritus check (log).
• Rebuilt source commit: 87b6d31444424a90f823aebcbcbd291bbf80b5d6, checked 2026-10-17 UTC.
• Full isolated R CMD check --as-cran --run-donttest, including PDF and HTML manuals: 0 errors, 0 warnings, 1 NOTE (unable to verify current time). 2,980 passes, zero test failures/warnings, 138 expected skips. The 101-page PDF passed qpdf --check.
• Archive SHA-256: 4a817a3ab4e3f6c98efc01342ebaf8fb0b47b61287a05294d943c596a78a00dc.
• Fresh isolated library; existing dependencies read-only; two advertised cores and one numerical thread per worker. Dependency DESCRIPTION hashes remained unchanged.
• git diff --check passed. The subsequent notes commit changes only files excluded from source packages.

Final head 310f14f8b99a8576ff2d079c251cf867355c1827: all CI checks passed. All five R CMD check jobs passed, including Windows with Status: OK; both parallel smoke jobs, pkgdown build, advisory audit, and Codecov patch check passed. Documentation deployment was correctly skipped for the PR. Whole-package coverage is exactly 100% with zero uncovered rows (strict gate passed); its test run recorded 4,958 assertions, zero warnings/failures, and 43 expected skips. Required reviewer approval remains outstanding.

Latest evidence and checklist: tools/releases/0.2.2.md. Local archive, logs and manual: output/release-0.2.2-pr252-check/run-yr9i7roc/. Windows diagnostic and targeted regression logs: output/windows-temp-investigation/. Earlier outputs are preserved; their source archives are superseded.

Issue #231 remains deferred. Local outputs and installations are preserved. No merge, CRAN submission, tag, or GitHub release has been performed.

[Copilot]

🟢 Approval recommended

The documentation-only update matches the verified evidence and leaves the checked package artifact unchanged.

No open issues 🎉

What changed in this PR

Finalizes the CRAN submission comments for bifrost 0.2.2 without changing package source.

Changes:

• Updates links to passing post-merge CI checks.
• Identifies the checked source archive and retains its checksum.
• Removes stale preparation instructions and troubleshooting details while preserving known limitations.

File	Description
cran-comments.md	Finalizes submission evidence and archive instructions.

_{Review effort: Balanced}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!