RatioLab 0.1.15

RatioLab 0.1.15 fixes temporary profile isolation used by the built-in self-tests and installer preflight checks. Built from commit d2ce9e6ce576a05e28bcde41ee0047a12d428b61.

• Keep the native C environment stable after GTK/font initialization while isolating Python-side settings and export paths.
• Restore the original Python environment mapping and home-path policy when a self-test finishes or raises an error.
• Run the profile regression tests in all four native build jobs. Geometry and motion formulas are unchanged.

Native CI and Python distribution CI completed successfully for this exact commit. All four native and three Python logical jobs passed. Both new profile regression tests passed in each native job. Eight artifact ZIPs matched API SHA-256/size before bounded CRC/path validation; six binaries matched CI checksums. Linux tar/run/Debian manifest and embedded metadata, Windows PE/version and macOS disk-image structures were checked. See SHA256SUMS and build-evidence.json for actual package, GUI, MCP and native-update results.

Controlled local testing of the published 0.1.14 runtime reproduced an environment-access race. The earlier uninstrumented CI crashes have not been matched to that same native stack.

Post-release verification: A later tag-triggered Windows job passed installer, runtime and MCP checks but its update-helper readiness check exceeded the five-second wait. The readiness marker appeared after cancellation; the target update was never armed. The cause of the delay is not established. The original successful source CI remains the provenance of these unchanged downloads.

Hosted GUI rendering remains unverified on: Windows. Explicit display/GL absence is recorded separately from successful runtime/MCP and update checks.

Native update CI used the new helper to upgrade public 0.1.12 to 0.1.15; GUI restart was disabled. Those CI checks alone do not establish old-client helper or privileged Debian installation acceptance. Windows Setup is unsigned; macOS uses ad-hoc signing without Apple notarization.

Coordinator validation of final 0.1.15 source: 46 targeted self-test, bundle and release-verification tests passed; 23 release-note, updater and entrypoint tests passed; explicit baseline publication-version check passed. No complete repository-suite pass is claimed.

Additional isolated Linux checks upgraded actual public 0.1.13 and 0.1.14 installations through their immutable helpers to these exact installer bytes. Both waited for the old process, reopened at the same location, preserved three project/preferences sentinels and retained future update eligibility. The 0.1.14 receipt was consumed by the restarted GUI; no new receipt protocol is attributed to 0.1.13. The target passed real OpenGL 4.6 workspace checks for all 11 families and six feature areas. A separate coordinator installation passed three self-tests without display variables and was successfully uninstalled. These checks used temporary copies and profiles, leaving the daily installation untouched.

Downloads: Windows 11 x86-64 Setup, macOS 15+ Apple Silicon/Intel DMG, and x86-64 Linux .run, .tar.gz or .deb (glibc 2.39+). Python and GTK are included; a compatible desktop OpenGL driver is required. Changelog.

RatioLab 0.1.14

RatioLab 0.1.14 improves desktop responsiveness, updates and workspace feedback. Built from commit 12d79abb1efc4f342d21b895f54a630ba93e0eae. See the changelog.

Changes since 0.1.13

• Eligible native installations can download, verify, install and reopen an update from one action, in their existing installation location. Save, Discard and Cancel protect unsaved projects; the result is shown once after reopening. Installation type and manual fallbacks remain explicit. Release notes omit Unreleased development entries.
• Ordinary geometry and scene preparation run in a worker, keeping the interface responsive while calculations finish. Only current results are applied; outdated jobs cannot replace another document, model, parameter set or preview configuration. GTK and OpenGL updates remain on the main thread. This changes responsiveness, not geometry formulas.
• Export completion remains visible with the output name and destination, Open folder and Dismiss. Failures and cancellations are reported. Selected component exports retain their original source identities.
• RPM-only changes preserve component selection and isolated view while updating motion. Geometry edits and new documents still clear obsolete selections.
• Expand tool and Restore workspace make room for the current tool while retaining its state, scene, document and split positions.
• Component labels are easier to read. Original names, roles, source indices and BOM codes remain available in details; part names and exported identities are unchanged.

Downloads

• Windows 11 x86-64: RatioLab-0.1.14-windows-x86_64-setup.exe.
• macOS 15+: RatioLab-0.1.14-macos-arm64.dmg for Apple Silicon or RatioLab-0.1.14-macos-x86_64.dmg for Intel.
• Ubuntu 24.04+ / compatible x86-64 Linux: ratiolab_0.1.14_amd64.deb, RatioLab-0.1.14-linux-x86_64.run or RatioLab-0.1.14-linux-x86_64.tar.gz. Linux requires glibc 2.39+.

Python, GTK and required runtime libraries are included; graphics drivers come from the operating system. A compatible desktop OpenGL driver is required. Windows Setup is unsigned. macOS applications use ad-hoc signing without Apple notarization; systems may warn or block launch.

Verification and limits

Post-release update: A later Ubuntu 24.04 CI job rebuilt the same application source and encountered another SIGSEGV during the Linux installer's headless runtime preflight. The installer aborted before committing the installation. This was a new CI build; its failed bytes were not retained for comparison with the published installer. A separate controlled thread-scheduling experiment on Ubuntu 24.04 reproduced a SIGSEGV in the published Linux runtime: a native font worker read an environment array while the self-test changed it. This demonstrates a reachable failure mechanism, but the original CI crashes have no native stack confirming the same cause. A correction is being tested privately; it is not included in these immutable 0.1.14 assets.

Native CI and Python distribution CI completed successfully for the exact release commit. All four native platform/architecture jobs and all three Python jobs passed. The final native run retained the successful Windows and macOS results and reran only the failed Linux job on unchanged source.

All eight native artifact ZIPs matched their GitHub API SHA-256 digests and sizes before bounded, CRC-verified extraction with checked paths and member types. All six binaries matched the included CI checksums. Linux tar and Debian payloads matched the complete bundle manifest, exact source version and changelog; the .run contains the same tar payload. Windows installer version/PE structure and macOS UDIF block maps were checked. SHA256SUMS contains independently recalculated binary hashes. build-evidence.json records actual jobs, installed GUI/MCP results and native update checks without private runner paths or handoff directories.

Linux CI installed and uninstalled the .run, passed installed GUI/MCP checks and tested the extracted Debian runtime. No privileged Debian system installation was performed. Windows CI installed and uninstalled Setup and passed installed runtime/MCP checks. Both macOS architectures mounted the DMG, copied the application and passed runtime/MCP and strict deep code-signature checks.

Installed GUI checks on Linux, macOS Apple Silicon, macOS Intel passed all 11 model families, six feature groups, actual framebuffer geometry and workspace export. Windows GUI rendering remains unverified: the hosted runner could not create an OpenGL context. Installed MCP checks passed both supported protocols on all four platform/architecture jobs, with 30 requests per protocol and the existing Wolfrom geometry, export and signed/zero-speed checks.

All four native jobs also used the new release production helper to upgrade a public 0.1.12 installation into the actual 0.1.14 package at the same location. Baseline and target hashes matched the verified installers. User-data sentinels were preserved, the new installation remained eligible for its next update, and the terminal receipt was consumed exactly once. GUI restart was disabled in this harness; it does not establish the old public 0.1.13 client flow. No user daily installation was used as a QA target.

A separate disposable local Linux check used the public 0.1.13 helper to install the exact final 0.1.14 installer at the same location and reopen it once. Three project/preferences sentinels were preserved; future native updates remained eligible. The new installed application passed real GTK/OpenGL checks for all 11 model families, six feature groups and workspace export. The old 0.1.13 helper has no new receipt protocol, so this check does not prove the new once-only receipt behavior. It does not establish older-client upgrade/restart acceptance on Windows or macOS, and no daily installation was used.

The first Linux packaging attempt on this release source exited with SIGSEGV (-11) in a second frozen headless self-test, after its initial bundle GUI check had passed. It produced no explanatory stdout/stderr. The unchanged-source Linux rerun passed the complete build, package, installed GUI/MCP and native-update checks. The initial crash remains unexplained.

The exact final Linux installer also passed a separate disposable installation, version check and three actual headless self-tests with DISPLAY, WAYLAND_DISPLAY and XAUTHORITY removed, followed by successful uninstallation. These three local passes do not explain or establish the cause of the original CI SIGSEGV.

An intervening Linux retry was canceled during an apt mirror stall before the application build or self-test began; that infrastructure wait did not reproduce the crash.

The full local repository suite completed 914 tests in 1040.970 s, 910 passed and four GUI assertion failures (worm readiness timeout, two unsolicited release-link events before the primary update click, one theme scale assertion), no errors. With unchanged source and assertions, an isolated real-display rerun passed all 34 tests in 108.261 s including all four failed cases. The broad run is not claimed fully green and the intermittent cause is unproven. After scoped CI/updater fixes, overlapping targeted slices were run: 128 checks in 26.446 s, 59 checks in 6.985 s with no skips, and 60 checks in 7.188 s; all passed. These include actual PowerShell fixtures, foreign utility-module and tampered-package checks, and canonical Windows installation-directory identity checks. The broad 914-test suite was not rerun after these fixes.

The final Python wheel and source distribution passed artifact checks (104 wheel members and 429 source-distribution members), including verification of the current helper bytes. These checks do not establish performance or compatibility on every computer.

Older Windows clients may need to run Setup 0.1.14 manually once: fixes in the new helper do not retroactively repair an already installed older updater. Some older permissive Linux installations may similarly need the current installer once at the same location. Protected, shared and source installations retain manual fallback.

Gear geometry, measurements and motion remain nominal/theoretical. Existing calculation and model-specific limits continue to apply. These desktop UX changes do not add strength, manufacturing-fit, durability or conformity validation.

RatioLab 0.1.13

RatioLab 0.1.13 adds a nominal two-rim spur Wolfrom / 3K reducer to the desktop application and browser app. It is built from commit 35f0694669b4fd00a5e4b12e386a266072fb337a. See the changelog.

Changes since 0.1.12

• Create a Wolfrom reducer with the sun as input, a fixed first ring, a second ring as output, a free carrier and rigid compound planets. The default example has a signed reduction ratio of +600:1; the derived second module preserves the shared center distance.
• Preview both planet rims, their hubs, carrier plates, pins and shafts. The default assembly contains 18 nominal solids. Tooth-count, equal-spacing, singular-ratio and nominal-clearance checks reject incompatible configurations.
• Use the existing projects, signed/zero-speed motion, reports, component selection, exports and headless MCP workflows. The desktop and browser catalogues now contain 11 model families.
• Read illustrated parameter help and the Wolfrom technical notes for the equations, supported topology and calculation limits.

Downloads

• Windows 11 x86-64: RatioLab-0.1.13-windows-x86_64-setup.exe.
• macOS 15+: choose RatioLab-0.1.13-macos-arm64.dmg for Apple Silicon or RatioLab-0.1.13-macos-x86_64.dmg for Intel.
• Ubuntu 24.04+ / compatible x86-64 Linux: use ratiolab_0.1.13_amd64.deb, RatioLab-0.1.13-linux-x86_64.run or RatioLab-0.1.13-linux-x86_64.tar.gz. Linux requires glibc 2.39+.

Python, GTK and required runtime libraries are included; graphics drivers come from the operating system. A compatible desktop OpenGL driver is required. Windows Setup is unsigned. macOS applications use ad-hoc signing without Apple notarization; systems may warn or block launch.

Verification and limits

Native CI and Python distribution CI completed successfully for the exact release commit. All four native jobs and all three Python distribution jobs passed.

All eight native artifact ZIPs matched their GitHub SHA-256 digests and sizes before bounded, CRC-verified extraction. All six binaries matched their included CI checksums. The Linux tar and Debian payloads matched the complete bundle manifest (1,222 regular files and 25 symlinks), including the exact source version and changelog; the .run contains the same tar payload. Windows installer version/PE structure and macOS UDIF block maps were checked. SHA256SUMS contains independently recalculated binary hashes; build-evidence.json records the actual jobs, package checks and installed GUI/MCP results without private runner paths.

Linux CI installed and uninstalled the .run package, passed installed GUI and MCP checks, and passed a runtime self-test from an extracted Debian package. It did not perform a system Debian package installation. Windows CI installed and uninstalled Setup and passed installed runtime and MCP checks, but the hosted runner could not create an OpenGL context: Windows GUI rendering remains unverified by this CI. Both macOS architectures mounted the DMG, copied the application, and passed runtime, GUI, MCP and strict deep code-signature verification checks.

Linux using Mesa software rendering and both macOS architectures using Apple Software Renderer passed installed GUI checks for all 11 model families, six feature groups and workspace export. Installed MCP checks passed both supported protocols on all four platforms, with 30 requests per protocol. The Wolfrom checks verified the 18-component default, +600:1 ratio, fixed ring, derived second module, reverse/zero-speed operation with unchanged STL hashes, preservation after a singular edit, and project save/open. Seven real STL files per export (six selected solids and their assembly) were checked against manifest hashes.

The local repository suite completed 823 tests in 1162.018 seconds: 822 passed, one GUI assertion failed, and there were no errors or skips. The failed worm-diameter refresh assertion (20.0 versus 25) is also present in the independently recorded 0.1.12 full-suite log; the earlier suite was not rerun for this comparison. All 23 dedicated Wolfrom tests passed, including seven real GTK/OpenGL cases. The full local suite was not clean. All 26 website tests passed. The Python wheel and source distribution passed their artifact checks (99 wheel members and 414 source-distribution members); rebuilding a wheel from the source distribution reproduced all 99 wheel member bytes.

The installed GUI/MCP self-tests do not exercise an actual update from an older native installation. No native update installation was performed on the user computer. Windows/macOS updater installation acceptance remains unperformed. These checks do not establish performance or compatibility on every computer.

Gear geometry, measurements and motion remain nominal/theoretical. Wolfrom support currently covers the documented two-rim, zero-shift spur topology. General interference checks and measurements use sampled surfaces. Manufacturing fits, materials, strength, service life, loss models, physical backdrivability and standards conformity are outside these nominal checks. Overall efficiency is a user-supplied assumption. Read unresolved results and calculation-report limits before using exported models.

RatioLab 0.1.12

RatioLab 0.1.12 adds a complete desktop update flow: manual checks, a persistent automatic-check preference, cancellable downloads and release-checksum verification. It is built from commit 3277a09984ac95c273fca2d9da9a36d564ea117d. See the changelog.

Changes since 0.1.11

• Check for updates manually and see progress, up-to-date results or errors. Manual checks work with automatic checking disabled and can show skipped releases, while respecting update policy and GitHub rate limits.
• Download compatible packages inside the application, cancel downloads, and verify them against SHA256SUMS from the same release before preparing an installation handoff.
• Update now prepares Linux per-user bundle or Windows Setup updates and offers the normal Save / Discard / Cancel protection before closing. macOS opens the verified DMG; Debian packages use the system installer. Source, Python and unmarked portable installations retain manual updates.

Install 0.1.12 using the appropriate download below to get the new update flow.

Downloads

• Windows 11 x86-64: RatioLab-0.1.12-windows-x86_64-setup.exe.
• macOS 15+: choose RatioLab-0.1.12-macos-arm64.dmg for Apple Silicon or RatioLab-0.1.12-macos-x86_64.dmg for Intel.
• Ubuntu 24.04+ / compatible x86-64 Linux: use ratiolab_0.1.12_amd64.deb, RatioLab-0.1.12-linux-x86_64.run or RatioLab-0.1.12-linux-x86_64.tar.gz. Linux requires glibc 2.39+.

Python, GTK and required runtime libraries are included; graphics drivers come from the operating system. A compatible desktop OpenGL driver is required. Windows Setup is unsigned. macOS applications use ad-hoc signing without Apple notarization; systems may warn or block launch.

Verification and limits

Native CI and Python distribution CI completed successfully for the exact release commit. All eight native artifact ZIPs matched their GitHub SHA-256 digests and sizes before bounded, CRC-verified extraction. All six binaries matched their included CI checksums. SHA256SUMS contains independently recalculated binary hashes; build-evidence.json records the actual jobs, package checks and installed GUI/MCP results without private runner paths.

Linux CI installed and uninstalled the .run package, passed the installed GUI and MCP checks, and passed a runtime self-test from an extracted Debian package. It did not perform a system Debian package installation. Windows CI installed and uninstalled Setup and passed installed runtime and MCP checks, but the hosted runner could not create an OpenGL context: Windows GUI rendering remains unverified by this CI. Both macOS architectures mounted the DMG, copied the application, and passed runtime, GUI, MCP and strict deep code-signature verification checks.

Linux using Mesa software rendering and both macOS architectures using Apple Software Renderer passed installed GUI checks for all ten model families, six feature groups and workspace export. Installed MCP checks passed both supported protocols on all four platforms. These self-tests do not exercise an actual update from an older native installation, and do not establish performance or compatibility on every computer. Windows/macOS updater installation acceptance remains unperformed.

The local repository suite completed 797 tests with four GUI failures, no errors and no skips; all 92 updater tests passed. Diameter-refresh and theme failures were reproduced on unchanged 0.1.11. The backlash and divider classes passed unchanged on both baseline and candidate when rerun as complete classes. GUI timing/focus instability remains a recorded limitation; the full local suite was not clean.

Gear geometry, measurements and motion remain nominal/theoretical. General interference checks and measurements use sampled surfaces. Read unresolved results and calculation-report limits before using exported models; these checks do not certify manufacturing, strength, service life or standards conformity.

RatioLab 0.1.11

RatioLab 0.1.11 adds persistent Light and Dark desktop themes, a continuous Explode slider and corrected planetary internal-ring geometry. This downloadable release is built from commit c69a07d1af7c566f90859b5e2778ecf57cb090a5 and includes the source changes since the previous downloadable release, 0.1.9. See the changelog.

Changes since 0.1.9

• Choose Light or Dark in View; the selection persists and applies to dialogs and the 3D preview. Text-size changes retain the selected theme.
• Adjust the desktop Explode amount continuously from 0 to 1, including component picking at intermediate poses. Changing the amount pauses motion and clears measurement points. Starting motion or choosing Initial pose restores the assembled pose. Geometry exports and package previews retain nominal assembly coordinates.
• Derive planetary ring tooth height and root clearance from the actual mating planet profile. Reports expose the derived geometry and contact ratio. Regenerating older projects intentionally changes the inner ring contour; tooth counts, ratios, sun/planet profiles and outer assembly dimensions stay unchanged. Existing exported meshes and CAD bodies are not updated automatically.
• Check generated planetary transverse polygons over the complete neutral-backlash tooth period using bounded phase subdivision. Detected collisions are rejected, while zero/tiny-backlash and exhausted checks retain an explicit unresolved result.
• Correct differential component-list equations and speed/axis references in reports and complete exports for wrist and classic configurations.

The 0.1.10 source changes also introduced the separately hosted browser workspace, with a pinned runtime, WebGL preview, animation, project history/recovery, standalone downloads and analysis tools. Browser fixes include validated search proposals, legacy-project import, pending-speed saves, stale-operation rejection, Fit behavior and responsive help/menus. The installers below provide the native desktop application; the browser workspace is served separately.

Downloads

• Windows 11 x86-64: RatioLab-0.1.11-windows-x86_64-setup.exe.
• macOS 15+: choose RatioLab-0.1.11-macos-arm64.dmg for Apple Silicon or RatioLab-0.1.11-macos-x86_64.dmg for Intel.
• Ubuntu 24.04+ / compatible x86-64 Linux: use ratiolab_0.1.11_amd64.deb, RatioLab-0.1.11-linux-x86_64.run or the portable .tar.gz. Linux requires glibc 2.39+.

Python, GTK and required runtime libraries are included; graphics drivers come from the operating system. A compatible desktop OpenGL driver is required. Windows Setup is unsigned. macOS applications use ad-hoc signing without Apple notarization; systems may warn or block launch.

Verification and limits

Native CI and Python distribution CI completed successfully for the exact release commit. All eight native artifact ZIPs matched their GitHub SHA-256 digests before extraction. The six binaries matched their included checksum files; SHA256SUMS contains independently recalculated hashes. build-evidence.json records file hashes, jobs and the actual installed GUI/MCP results without private runner paths.

Linux using Mesa software rendering and both macOS architectures using Apple Software Renderer passed installed GUI checks for all ten families and workspace export. Windows passed installer, installed runtime and MCP checks, but its hosted runner could not create an OpenGL context: the Windows GUI remains unverified by this CI. Installed MCP checks passed on all four platforms. These checks do not establish performance or compatibility on every computer.

Gear geometry, measurements and motion remain nominal/theoretical. Planetary checks cover generated neutral-backlash transverse polygons and supported ideal axial sections. They do not certify unsampled analytic curves, exported axial interpolation, loaded contact, the entire backlash range, manufacturing tooling, strength, service life or standards conformity. General interference checks and measurements use sampled surfaces; read unresolved results and calculation-report limits before using exported models.

RatioLab 0.1.9

RatioLab 0.1.9 provides offline parameter guides, motion-speed updates without regenerating geometry, nominal involute gear measurements, working-backlash proposals and precise numeric editing. See the changelog for the source changes since 0.1.3.

This release includes the packaging repair that keeps warnings and errors separate from machine-readable native test reports. The tag identifies commit 2fde5c6bdbaffb549cf1486ce574f8550413679f; the earlier 0.1.9 source push did not provide a downloadable release. The browser application and other pending local development are not included.

Downloads

• Windows 11 x86-64: RatioLab-0.1.9-windows-x86_64-setup.exe.
• macOS 15+: choose the arm64.dmg for Apple Silicon or x86_64.dmg for Intel.
• Ubuntu 24.04+ / compatible x86-64 Linux: use ratiolab_0.1.9_amd64.deb, the .run installer or the portable .tar.gz. Linux requires glibc 2.39+.

Python, GTK and required runtime libraries are included; graphics drivers come from the operating system. Windows Setup is unsigned. macOS applications use ad-hoc signing without Apple notarization. A compatible OpenGL driver is required.

Verification

Native installer checks and Python distributions were built for the exact tag commit: native CI, Python CI. build-evidence.json records job outcomes, installed GUI/MCP reports and file hashes. SHA256SUMS contains independently recalculated hashes for the six downloadable binaries. Linux (Mesa software rendering) and both macOS architectures (Apple Software Renderer) passed installed GUI checks for all ten families and workspace export. Windows passed installer/runtime/MCP checks; its hosted runner could not create an OpenGL context, so the full Windows GUI remains unverified by this CI. The evidence records this limit explicitly. These checks do not establish performance or compatibility on every computer.

Gear measurements and motion are nominal/theoretical. Surface sampling and construction checks do not certify strength, service life or manufacturing readiness.

RatioLab 0.1.3

RatioLab 0.1.3, a stable numbered release of the open-source gear and reducer design application. Includes the features and distribution fixes recorded in versions 0.1.1–0.1.3.

Downloads

Python, GTK and the required application libraries are included.

System	Package
Windows 11, Intel/AMD 64-bit	RatioLab-0.1.3-windows-x86_64-setup.exe
macOS 15+, Apple Silicon	RatioLab-0.1.3-macos-arm64.dmg
macOS 15+, Intel	RatioLab-0.1.3-macos-x86_64.dmg
Ubuntu 24.04+ / compatible Linux x86-64, glibc 2.39+	ratiolab_0.1.3_amd64.deb, or RatioLab-0.1.3-linux-x86_64.run; portable .tar.gz also available

Open Windows Setup, drag the macOS app from the DMG to Applications, or install the Linux package. See the installation guide.

Changes since Preview 2

• Bevel differentials with wrist and classic configurations, straight or spiral teeth, interface/carrier variants, ratio presets and independent signed motion controls. External belts, pulleys and motors remain part of the user's CAD work.
• Matching worm-wheel and strain-wave tooth envelopes, revised bevel geometry, shared preview/export surfaces, and documented reference equations and numerical limits.
• Clearer finite combination-search controls, differential motion continuity, cycloidal eccentric visibility and more efficient exact sampled-surface distance searches.
• Consistent RatioLab icons and native window identity. Model notices follow actual generated metadata; meaningful geometry warnings remain visible.
• Automatic background release checks in installed builds. Update to [version] appears only for a newer eligible GitHub release; click it for release notes and the platform download. No update dialog opens automatically. Downloads open in the browser; installation remains a user action.
• Distribution checks now require every supported family and exact packaged resources/changelog. Stable version identities and download documentation are consistent across platforms.

Verification

All four native installer jobs and three Python distribution jobs passed for commit 1b0c85fec2b961cffa3a1d8520784fdf993acfe7.

Installed rendering, all ten family previews and workspace exports passed on Linux and both macOS architectures. The Windows hosted runner could not create an OpenGL context: its installer, installed runtime and offline MCP checks passed, but GUI rendering was not verified on that runner. Windows CI uses Windows Server 2025; this does not cover every Windows 11 system or graphics driver.

Both MCP protocol versions were verified on all four native targets. Hosted macOS checks use Cairo for GTK shell composition while still exercising the actual OpenGL preview. These checks do not cover every hardware/driver configuration or establish Fusion interoperability on untested platforms.

Native installer checks · Python distribution checks

SHA256SUMS covers all six installer/archive files. build-evidence.json binds their hashes and installed checks to the exact source commit and CI runs.

Limits and notices

Windows Setup is unsigned. macOS uses ad-hoc signing without Apple notarization; the operating system may warn about or block launch.

Geometry and motion checks do not certify elastic response, strength, service life or continuous clearance. Interference checks and measurements use sampled surfaces; strain-wave deformation is prescribed. Read the model limits and full changelog.

Licensed under GNU GPL version 3 only. Copyright © 2026 NoOneExceptEnderk. Third-party notices are included with the application.

RatioLab 0.1.0 Preview 2

Second public preview of RatioLab, an open-source parametric gear and reducer generator by NoOneExceptEnderk, licensed under GNU GPL version 3 only.

Fusion integration for Windows and Linux — 2026-10-29

Updated Windows Setup and Linux .run, .tar.gz and .deb downloads include CAD integrations → Install Fusion add-in / Connect / Create in Fusion.

• Fusion receives individual STEP BRep solids in a new component group, with source parameters, roles, visibility and BOM metadata. The bridge checks millimetre dimensions and volume and removes its new group on a failed import. Native Fusion feature timelines and joints are not reconstructed; RGB values are retained as metadata.
• Windows verified on a local PC: Fusion 2705.1.30, all nine default families and 74 components, failure cleanup, installation and the complete application-button workflow. The newly installed Windows bundle passed its GUI/export/MCP checks; 1083 manifest entries matched. Desktop OpenGL 3.3 / GLES 3.0 minimums are retained.
• Linux: the native RatioLab app can connect to Windows Fusion running through a community Wine/Proton port. Prefix/profile discovery and C-drive staging avoid requiring a Z: mapping. Set RATIOLAB_FUSION_WINEPREFIX (and RATIOLAB_FUSION_WINEUSER for ambiguous profiles) for custom installations. Linux installers and native runtime/rendering/export checks passed on Ubuntu CI. Wine path/protocol tests passed; a live Fusion-under-Wine import has not been verified.
• All seven GitHub Actions jobs passed for 2bf46071. See the Fusion setup guide and the attached checksum and build-evidence files.

Updated Windows/Linux runtime source: 2bf460713f667fd7c7738fe9262b573668f23e3b (source archive). macOS downloads keep their original Preview 2 binaries. The published tag and GitHub's automatic source archives still refer to the original preview revision.

Earlier Windows graphics hotfix — 2026-10-29

The Windows installer has been replaced with a build that fixes the "Application does not support OpenGL API" failure and black numeric text on the dark GTK workspace.

• GTK can negotiate desktop OpenGL 3.3+ or OpenGL ES 3.0+, with separate minimum-version requests. Shader dialect follows the realized context.
• Workspace controls inherit the intended foreground while retaining semantic and primary-button colors.
• Verified on a local Windows PC with NVIDIA GeForce GT 1030 / desktop OpenGL 4.6: 25 targeted tests passed; the installed application rendered all nine families, exported the workspace and passed both MCP protocol checks with external runtimes removed from PATH. All 1,079 installed manifest files matched the verified bundle.
• An additional 55 regression tests had 50 passes, one skip and four failures/errors reproduced unchanged on the original commit. GLES-only rendering is unavailable on this PC; its fallback and shader selection are covered by regression tests.

Windows runtime source: be73d21c437ba5c5ee562d8bce6cc610eab0d1cc (source archive). At the time of that hotfix, Linux and macOS downloads retained the original Preview 2 binaries. The published tag and its automatic source archives still refer to the original preview commit.

Repository cleanup

The public repository now separates application code in src/, automated checks in tests/, and user/developer documentation in docs/. The README starts with platform downloads and simple installation steps. Internal agent notes and working records remain outside the public source. Catalogs and branding have one canonical location. Runtime imports, project formats and calculation logic remain compatible.

Downloads

Python, GTK and the required application libraries are included. You do not need to install them separately.

System	Download
Windows 11 x86-64	RatioLab-0.1.0-windows-x86_64-setup.exe
macOS 15+, Apple Silicon	RatioLab-0.1.0-macos-arm64.dmg
macOS 15+, Intel	RatioLab-0.1.0-macos-x86_64.dmg
Linux x86-64, glibc 2.39+ (Ubuntu 24.04 baseline)	ratiolab_0.1.0_amd64.deb for Debian/Ubuntu, or RatioLab-0.1.0-linux-x86_64.run for a user installation; .tar.gz also available

Open Setup on Windows. On macOS, open the DMG and drag RatioLab to Applications. On Linux, open the .deb in the software manager or run sh RatioLab-0.1.0-linux-x86_64.run.

Included

• Nine gear/reducer families and an animated 3D workspace.
• Standalone STEP, STL and SVG/DXF, complete portable exports and BOMs.
• Projects with Undo/Redo, recent files, atomic saves and recovery.
• Generated tooth-root profiles and geometric worm/strain-wave study models.
• Discrete interference/distance checks and finite parameter searches.
• Component selection, isolation, sampled measurements and subset exports.
• An offline RatioLab-MCP companion (RatioLab-MCP.exe on Windows) with 13 agent tools, requiring no open GUI. MCP setup.

Repository/example filenames, code comments and exported report suffixes are in English. Source modules and tests are organized into dedicated folders; tests are excluded from installed applications. Earlier configuration formats remain readable; older preferences are preserved and migrate to ui-preferences.json when saved. The source includes a contributor guide and project license/attribution; the installers include those project notices and separate dependency notices.

Verification and preview limits

Original cross-platform build, before this Windows hotfix: 362 source tests passed, with no failures or skips. All four native installer jobs and all three Python distribution jobs passed. Installed GUI rendering, nine family previews, workspace exports and both supported MCP protocols were checked on Linux, Windows and both macOS architectures. The Windows hosted runner is Windows Server 2025; testing on individual Windows 11 systems is welcome. Hosted graphics checks do not cover every GPU/driver.

Windows Setup has no publisher signature. macOS uses ad hoc signing without Apple notarization, so the operating system may display a warning. The original Linux/macOS binaries need desktop OpenGL 3.3+. The updated Windows installer negotiates desktop OpenGL 3.3+ or OpenGL ES 3.0+, with ANGLE included as a GLES backend.

Interference checks use discrete samples; searches and measurements use the documented finite domains/sampled surfaces. Worm and strain-wave geometry are study models. This release does not certify strength, service life or manufacturing readiness.

SHA256SUMS covers all six installers. build-evidence.json records their exact hashes, native jobs and installed GUI/MCP results. See installation, verification evidence, license and contributing.

Native installer checks · Python distribution checks.

Original cross-platform runtime and release source commit: 870835e63fcf78e01b4c08acec0f271848cc7dca. Updated Windows/Linux runtime source commit: 2bf460713f667fd7c7738fe9262b573668f23e3b.

RatioLab 0.1.0 Preview 1

First public preview of RatioLab, an open-source parametric gear and reducer generator by NoOneExceptEnderk, licensed under GNU GPL version 3 only.

Downloads

Python, GTK and the required application libraries are included. You do not need to install them separately.

System	Download
Windows 11 x86-64	RatioLab-0.1.0-windows-x86_64-setup.exe
macOS 15+, Apple Silicon	RatioLab-0.1.0-macos-arm64.dmg
macOS 15+, Intel	RatioLab-0.1.0-macos-x86_64.dmg
Linux x86-64, glibc 2.39+ (Ubuntu 24.04 baseline)	ratiolab_0.1.0_amd64.deb for Debian/Ubuntu, or RatioLab-0.1.0-linux-x86_64.run for a user installation; .tar.gz also available

Open Setup on Windows. On macOS, open the DMG and drag RatioLab to Applications. On Linux, open the .deb in the software manager or run sh RatioLab-0.1.0-linux-x86_64.run.

Included

• Nine gear/reducer families and an animated 3D workspace.
• Standalone STEP, STL and SVG/DXF, complete portable exports and BOMs.
• Projects with Undo/Redo, recent files, atomic saves and recovery.
• Generated tooth-root profiles and geometric worm/strain-wave study models.
• Discrete interference/distance checks and finite parameter searches.
• Component selection, isolation, sampled measurements and subset exports.
• An offline RatioLab-MCP companion (RatioLab-MCP.exe on Windows) with 13 agent tools, requiring no open GUI. MCP setup.

Repository/example filenames, code comments and exported report suffixes are now in English. Earlier configuration formats remain readable; older preferences are preserved and migrate to ui-preferences.json when saved. The source includes a contributor guide and project license/attribution; the installers include those project notices and separate dependency notices.

Verification and preview limits

362 source tests passed, with no failures or skips. All four native installer jobs and all three Python distribution jobs passed. Installed GUI rendering, nine family previews, workspace exports and both supported MCP protocols were checked on Linux, Windows and both macOS architectures. The Windows hosted runner is Windows Server 2025; testing on individual Windows 11 systems is welcome. Hosted graphics checks do not cover every GPU/driver.

Windows Setup has no publisher signature. macOS uses ad hoc signing without Apple notarization, so the operating system may display a warning. Linux/macOS need desktop OpenGL 3.3+; Windows uses the bundled ANGLE runtime and OpenGL ES 3.0+ through Direct3D.

Interference checks use discrete samples; searches and measurements use the documented finite domains/sampled surfaces. Worm and strain-wave geometry are study models. This release does not certify strength, service life or manufacturing readiness.

SHA256SUMS covers all six installers. build-evidence.json records their exact hashes, native jobs and installed GUI/MCP results. See installation, verification, license and contributing.

Native installer checks · Python distribution checks.

Runtime source commit: b76ab88652cb5609e8e7a845bac4d45ec43e1320. Release source commit: a25e48af582bc60d067046160f0e49ad9014866b; later changes are documentation only.