Skip to content

v0.5.0

Choose a tag to compare

View all tags
@acgetchell acgetchell released this 17 Oct 08:15
· 28 commits to main since this release
Immutable release. Only release title and notes can be modified.
v0.5.0
5d8f5b4
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

⚠️ Breaking Changes

  • changelog-unreleased now requires TAG DATE and leaves
    release metadata preparation separate. Replace the removed standalone
    archive-changelog, postprocess-changelog and tag-release entry points with
    research-repo-tools changelog commands. Tags must use v-prefixed SemVer;
    tagging requires today's UTC release date matching CITATION.cff.
  • contributor recipes use audit, shell-fix, toml-fix,
    and test-unit. Redundant aliases and help-workflows are removed; use
    bare just to discover the canonical commands.
  • la-stack now requires Rust 1.99.0 instead of Rust 1.98.1.

Merged Pull Requests

  • Publish reviewed GitHub releases to crates.io #282
  • Record algebraic floating-point rejection #281
  • Bump the github-actions group with 6 updates #280
  • Bump acgetchell/research-repo-tools/.github/workflows/dependabot-approve.yml #270
  • Bump the github-actions group across 1 directory with 5 updates #267

Added

  • Adopt shared CodeRabbit review commands c4fa23c

    • Add opt-in branch and uncommitted review recipes using the pinned
      research-repo-tools CLI and repository instructions.
    • Verify the default origin/main base against the live remote, preserve
      local-base overrides, and propagate review output and failures.
    • Document external CLI prerequisites, review scopes, and explicit agent
      invocation outside routine validation gates.
    • Advance the uv pin to 0.12.21 and refresh locked glam and python-dotenv
      dependencies.

  • Add stable unsigned vector angles 09020aa

    • Add allocation-free borrowed-slice and fixed-vector angle APIs with typed
      shape, zero-vector, and non-finite operand errors.
    • Preserve extreme scales and tiny separations with power-of-two scaling,
      compensated exterior products, and full-angle atan2 evaluation.
    • Add independent numerical regressions and properties, allocation checks,
      stable benchmark comparisons, and retained measurement provenance.
    • Document rounded semantics and the quadratic-time, constant-space tradeoff.
    • Permit Git mutations in agent guidance when explicitly approved by the user.
    • Validate with full just ci: 955 Rust tests, 502 Python tests, default/exact
      doctests, lint checks, benchmark compilation, and examples.

Changed

  • [breaking] Adopt pinned shared changelog commands ce72843

    • Delegate generation, normalization, minor archiving, release notes and
      local tag preparation to the published research-repo-tools 0.1.7 CLI.
    • Preserve retained release history and dependency-note links while moving
      completed series to docs/archives/changelog.
    • Keep the historical git-cliff policy as a documented consumer exception
      and validate published changelog links alongside the shared checks.

  • Stabilize disposable consumer fixtures 2372f81

    • Pin the synthetic Cargo identity independently of the checkout's release
      version while retaining the real consumer configuration and policy.
    • Disable inherited commit and tag signing only inside disposable repositories
      so fixture operations do not require user signing keys.

  • Adopt common template and draft reviews 1572957

    • Use the shared 0.1.7 changelog format with a bounded exception for historical deps-dev release-note bodies.
    • Preserve literal Rust code during generation and retain complete release history, dates and links.
    • Enable automatic CodeRabbit draft reviews and align zizmor-action with the existing scanner pin.

  • Complete shared maintenance adoption 8551e70

    • Route setup, managed execution, dependency updates, release metadata,
      and documentation and fixture checks through pinned research-repo-tools.
    • Declare managed Cargo and security tools centrally and reuse shared setup
      in hosted CI while preserving uncached release benchmark jobs.
    • Remove duplicate maintenance helpers and their dedicated tests, retaining
      scientific scripts, benchmark evidence, and consumer policy adapters.
    • Add shared OSV and Gitleaks recipes, upgrade PyJWT with crypto support,
      and narrowly allow a historical compatibility identifier false positive.
    • Document the current shared setup contract and deliberate update policies.

  • Adopt shared benchmark primitives 1c9c66d

    • Delegate Criterion parsing and comparisons, digest verification, archive
      extraction, and README section replacement to research-repo-tools.
    • Publish complete report, evidence, plot, and retained-summary output groups
      through shared transactions while preserving historical schemas and hashes.
    • Report invalid timings and publication failures with context, preserving
      existing artifacts when an operation fails.
    • Remove duplicate tooling and Semgrep rules, refresh subprocess policy,
      and update the Ruff and Semgrep pins.
    • Accept the benchmark-only paste maintenance advisory until January 1, 2027,
      while retaining checks for all other advisories.

  • Cover numerical edge cases and preserve benchmark evidence 1858b5e

    • Cover exact zero subtraction, tight square bounds, smallest interval
      determinants, and certified proof loss after underflow or range exhaustion.
    • Check independently derived scalar benchmark endpoint bits before timing.
    • Isolate each benchmark reproduction run's raw Criterion data and logs,
      resolve relative input paths, and document the export workflow.
    • Refresh development-tool pins and synchronize Cargo and Python locks.

  • Cover retained report promotion paths 73dd02f

    Verify that absent and empty retained-output mappings leave reports
    unchanged, and reject retained summaries that alias the current report
    with the complete path-collision diagnostic before mutation.

  • Expose borrowed vector angles through an extension trait 9b18f05

    • Replace the unreleased angle_between function with VectorAngle::angle on
      coordinate slices, retaining Vector::angle and root/prelude exports.
    • Require validated Vector inputs at the internal finite-angle boundary.
    • Expand independent rational-reference and typed error-path coverage.
    • Replace the angle benchmark study with fresh measurements and provenance
      for the extension API, and update API documentation and examples.
    • Refresh locked Rust/Python dependencies and the cargo-nextest tool pin.

  • Adopt shared performance and changelog workflows 279dbdf

    Pin published research-repo-tools 0.1.8 and replace the copied changelog template with the supported dependency-body policy. Delegate common-harness
    measurement, complete run retention, worktrees and transactional publication while retaining la-stack scientific selection and legacy evidence readers.

  • Replace Python adapters with shared tooling 6decbf8

    • Move benchmark commands, measurement inputs, provenance settings, and
      report paths into tooling configuration.
    • Use shared process, Criterion, and Just inspection APIs and remove
      duplicate local adapters and generic tests.
    • Preserve scientific report contracts, historical evidence replay, and
      configurable retained-history selection.
    • Refresh the uv pin and locked tooling dependencies with just update.

  • Compare archive symlink targets by filesystem identity 0d3532d

    • Preserve the observed link target across rejected report publication.
    • Cover absolute and relative symlinks without assuming Windows readlink
      returns the same path representation used when creating the link.

  • Standardize complete performance reports 1f33f14

    • Use shared rendering and configured publication for complete-run reports.
    • Preserve dimension plots, historical report bytes, and immutable evidence.
    • Align report guidance with retained provenance and make the Cargo shell
      fixture use LF on Windows.

  • Adopt shared performance tooling and Python 3.15 e64ff30

    • Replace six Python helpers and local packaging with research-repo-tools
      0.1.9 and declarative tooling configuration.
    • Retain only scientific policy and release adapters, with typed catalog
      parsing and source/harness publication checks.
    • Share benchmark discovery, baseline verification, reporting, plotting,
      and release publication while preserving historical artifacts.
    • Retire local console entry points and partial raw-Criterion plotting
      in favor of documented just recipes and complete-run evidence.
    • Target Python 3.15 while isolating Semgrep on locked Python 3.14 for
      local scans, fixtures, and SARIF.
    • Refresh development dependencies and document the migration.

Dependencies

Documentation

  • Expose all configured workflow badges 6933c15

    • Show CodeQL, zizmor, repository-rule analysis, coverage execution,
      benchmark, release-benchmark, and Dependabot approval status in the README.
    • Keep package metadata and workflow status in separate badge groups.

  • Align release roadmap and benchmark descriptions fb25684

    • Plan the stable-Rust v0.5.0 release sequence and conditional algebraic
      floating-point adoption, with compatible follow-up work in v0.5.1.
    • Defer expanded const generics until the required Generic Const Arguments
      capabilities are stable.
    • Clarify that research-repo-tools is a contributor development dependency.
    • Describe dense benchmark controls as row-permuted while preserving
      measured inputs and retained performance evidence.

  • Record algebraic floating-point rejection #281
    a275ddf

    • Retain strict arithmetic after the norm reduction prototype hid required overflow errors despite local timing gains.
    • Preserve the numerical counterexample, measurements, provenance, and reproducible experimental evidence.
    • Map floating-point, certified, interval, and exact arithmetic contracts to validation evidence and record closure of the remaining prototype work.
    • Update documentation navigation, release planning, and the reviewed README link inventory.

Fixed

  • Preserve consumer command and release contracts cb8c471

    • Forward preview options to the shared CLI and keep generated archives under
      the dedicated changelog formatter and final link check.
    • Accept valid included dependency groups while advancing only direct dev
      pins, leaving the shared tooling dependency under its own upgrade policy.
    • Keep declared release dates valid after a later merge and include canonical
      changelog archives in the release staging guidance.

  • Preserve zizmor failures and discard invalid SARIF 477acf1

    • Preserve the scanner exit status while checking its report before upload.
    • Remove empty or malformed SARIF output so failed scans cannot publish
      an unusable report; retain valid reports containing findings.

  • [breaking] Adopt shared approvals and generated command help
    fec147b

    • Replace owner-issued CodeRabbit requests with the pinned MCMC
      Dependabot approval workflow and document the repository settings.
    • Fail workflow security checks on findings before exporting SARIF.
    • Generate sorted Just help and keep usage, contributor guidance, and
      active documentation navigation with their respective owners.
    • Replace duplicate process and zizmor implementations with the
      published research-repo-tools APIs.

  • Honor inline Semgrep suppressions in uploaded SARIF c314554

    • Remove in-source-suppressed results before GitHub code scanning
      ingests the report so reviewed inline exceptions stay effective.
    • Preserve unsuppressed findings, their metadata, and the scanner's
      original exit status for the blocking security gate.

  • Preserve fixture line endings on Windows f748db0

    • Declare LF and CRLF benchmark snapshot inputs explicitly instead of
      inheriting the host's text-write behavior.
    • Check exact worktree bytes and staged contents against the selected
      line ending while preserving tracked-file and index guarantees.

  • Require a working Semgrep wheel on Windows 0e74cc3

    • Restore Semgrep 1.178.0, whose Windows wheel includes semgrep-core.exe.
    • Disable Semgrep source builds so missing platform wheels fail during
      installation instead of leaving an unusable scanner in the environment.

  • Restore Python checks with ty 0.0.85 0b97b4b

    Use Generator annotations for context managers and explicitly distinguish
    absent optional values in performance tooling. Preserve empty baseline
    names and retained-output mappings without suppressing diagnostics.

  • Preserve subnormal vector angles on Windows 7ee4bde

    • Use the direct exterior/dot ratio for sufficiently small positive angles,
      with a documented arctangent approximation bound, avoiding platform atan2
      underflow observed by the Windows CI regression tests.
    • Test the small-angle transition and normal/subnormal boundary across
      dimensions 2 through 6, 8, and 64 without relaxing existing tiny-angle tests.
    • Verify that exact-conversion error formatting propagates writer failures.
    • Retain the defensive non-finite-result guard without artificial coverage
      tests, exclusions, or coverage-threshold changes.
    • Refresh all 80 benchmark records and retain the original measurements by
      commit for a documented before/after comparison.
    • Validate with just check, just coverage-ci (876 tests), and isolated just ci
      (963 Rust tests, 502 Python tests, doctests, lints, benchmarks, and examples).

  • Preserve explicit same-version local comparisons 25cd362

    Keep the consumer-only local comparison exception while release promotion still requires distinct labels and sources remain independently identified. Cover
    explicit selection and current-phase gate failure before either timing phase.

  • Retain report test collection and release guidance deaecbc

    Restore the peer-context test prefix and describe complete-run README inputs, staged retention files, and the supported changelog dependency-body policy.

  • Harden shared performance publication and Windows fixtures 0ef7b78

    • Exclude unsupported norm cases from historical baseline inventories.
    • Preserve prior reports on archive conflicts and reject ambiguous identities.
    • Protect complete-run inputs during README publication and honor configured
      archive locations and equivalent input paths.
    • Keep immutable generated reports outside Markdown formatting and render
      curated provenance with valid list structure.
    • Compare command fixture directories by filesystem identity across Windows
      and POSIX shells, and align the managed rumdl pin to 0.2.79.

  • Preserve benchmark publication identity and diagnostics 86a2cf8

    • Keep README asset provenance tied to validated retained runs across
      scratch cleanup while preserving input freshness checks.
    • Reject non-canonical storage paths for release-report benchmarks.
    • Handle Git execution failures through the plotting CLI diagnostic.
    • Derive the minimum supported release diagnostic from the parsed policy.

Maintenance

  • [breaking] Require Rust 1.99.0 78c289a

    • Align the library, comparison workspace, contributor toolchain, and Clippy
      baseline with Rust 1.99.0 and LLVM 23.
    • Resolve new lint diagnostics without changing strict floating-point,
      exact-arithmetic, interval, or certified-error contracts.
    • Record paired compiler measurements, code-generation findings, and retained
      evidence as the strict-f64 baseline for future optimization work.
    • Update uv, Semgrep, dprint, and locked Rust/Python dependencies.

  • Publish reviewed GitHub releases to crates.io #282
    51868ea

ci: publish reviewed GitHub releases to crates.io

  • Gate root-crate publication on immutable release identity, exact-commit
    native checks, release notes, benchmark evidence, and deployment approval.
  • Use short-lived OIDC credentials, serialize uploads, and verify registry
    visibility without blindly retrying a potentially successful upload.
  • Leave benchmark-equipped release drafts for maintainer review and document
    the publishing setup, release sequence, and recovery procedure.
  • Keep publishing action updates under manual review so their pins stay
    aligned with the selected-actions policy.

Performance

  • Reduce interval and certified reduction overhead dd909a7

    • Reuse singleton interval arithmetic, select only needed sum endpoints,
      and size determinant workspaces to the compile-time dimension.
    • Use exact FMA residuals above the proved underflow threshold while
      preserving integer product comparison for smaller products.
    • Compact scalar certificates and reduce magnitude-accumulation overhead
      while preserving ordered estimates, finite endpoints, and signed zero.
    • Certify single normal unit products with zero error, including MAX.
    • Retain reproducible upstream and downstream performance evidence,
      adoption patches, rejected candidates, and measured tradeoffs.

  • Accelerate triangular rational solves 5921577

    • Recognize upper and lower triangular systems through row permutations
      and solve by exact substitution with one reduction per component.
    • Preserve the general Bareiss path, canonical rational inputs, and typed
      singularity diagnostics without changing the public API.
    • Add permanent structured benchmarks and retain Rust 1.99 measurements:
      11.15x faster prepared solves and 4.26x faster full adapters at D=8
      on the reproduced triangular family.
    • Refresh the staged cc, syn, uv, Ruff, and Ty dependency/tooling pins.
    • Validation: just ci passes, including 1,002 Rust tests, 371 Python
      tests, doctests, examples, and benchmark compilation.
Assets 4
Loading