Rally Coach 0.1.0-dev.1967+g16da3d7c114b — alpha

Rally Coach 0.1.0-dev.1967+g16da3d7c114b — alpha

Build commit: 16da3d7c114b326d5096c5eb931803c99fc7a0db

Publication commit: 16da3d7c114b326d5096c5eb931803c99fc7a0db

Added

• Show a heading-based vehicle-width band and cursor car rectangle from installed body dimensions on telemetry maps and local recording plots, with remembered layer controls and explicit unknown position offset and missing geometry reasons.
• Local Rallys pages search and sort RaceStat rally name claims and show driven attempts in heuristic same-car passes with at most three days between runs. Repeated stages carry attempt numbers; source names are decoded for display. No official overall result or championship is inferred.
• Rally history offers pass sorting, car and repeated-pass filters and bounded continuation through every name. Passes show distinct-stage catalog kilometres and missing lengths, fastest finishes with at least one same-stage/car peer (ties share the mark), and links to available telemetry recordings.
• Stages includes installed and catalog routes without driving history, with a typed not-driven state, no invented times and working search, sorting and detail links. Local map descriptors and installed personal pacenotes provide declarative search filters using native stage identity resolution.
• Add an English/German static website with a six-step authored screenshot tour, platform download links, unsigned-alpha guidance and local-processing privacy information. The website does not change desktop behavior.
• Runs have a reloadable page linked from the latest recorded run and stage/car history, driver attempt labels, finish state, a single penalty summary, and recorded setup, conditions and splits. Debrief uses the shared exact recording-match verdict rather than an unconditional contradictory disclaimer; no finish tolerance was added.
• Show immediate accessible library loading with consistent page gutters, plain bilingual index phases and measured recording counts through one nonblocking web/native progress contract; retain the saved index and show a typed failure message when refresh fails.
• Telemetry now includes a grip scatter for the visible distance window and a measured gearing sheet for one attempt. Rust clips source acceleration spikes and calculates native powered RPM per km/h, sample counts and speed at observed maximum RPM. Typed absences and German/English notes retain unverified acceleration scale and recorded gear encoding; matching setup gearbox sections link to an attempt sheet.
• Keep bounded local diagnostics with release build identity and prepare a redacted feedback file from Help; open a prefilled GitHub issue only on request, with manual attachment and no app upload.
• Discover protected stage imagery from installed native stage files, send bounded read-only stage proofs without login, explain missing or unmatched versions, and retain cached maps offline while preserving Martin compatibility.
• Private release builds can provide a default imagery endpoint at compile time; saved settings override it, builds without it keep imagery unconfigured, and release manifests record only injection status and SHA256.
• After explicit distance-assumption consent, debrief offers up to three consistency review spots with measured spread, attempt counts and thin support for two measurements. Drivers can choose a spot, keep the existing topics or finish; unavailable evidence and the consistency-only scope remain explicit.
• Five reviewed built-in numeric mappings resolve only matching symbolic pacenote sections in the selected installation, with disclosed provenance. Conflicting installed sections remain explicit. Read-only note diagnostics linked from Settings and recce list installation and built-in resolutions and remaining undefined IDs with counts.
• Maps can display personal pacenotes from the selected RBR installation along a cached host route. Local Rust interpolates unchanged native distances with explicit unverified correspondence, preserves route/roadbook/notation hashes and opaque variant identity, and withholds unsupported positions with typed reasons. Drivers can choose an installed roadbook and reading system using the existing installed reader; a latest-modified file is only a current candidate, never a historical default. Pacenotes remain separate from host-derived corners and crests. No metre scale, fitted offset, authenticated direction, driving benefit or recce/coaching behavior change is claimed.
• Add a persisted bilingual Debug log switch, on by default for now and effective immediately, with private command/route/index/settings/panic diagnostics and map view/tile detail only at debug level; keep normal faults available with debug off and retain read-only settings changes in memory.
• Read static car footprints from the selected RSF installation with bounded persistent caching, source provenance and typed unknown geometry or pose correspondence.
• Run pages link directly to preselected telemetry with large stacked charts, bounded distance zoom and pan, shared chart/map cursor and a small route/driven map. Rust owns channel selection, typed missingness, native held gear, source units, acceleration signs and fixed percentile spike clipping; German and English explain unverified scales.
• Telemetry can overlay two recordings on one Rust-selected common distance grid, including partial attempts. A recording picker starts with the existing proposed reference; two labelled colours share chart zoom, pan and cursor. Typed coverage and stage refusals prevent extrapolation or incompatible overlays.
• The bundled German and English manual now has chapters, searchable topics and back/related-topic navigation. Contextual question marks show a short hover or keyboard-focus hint and open the matching topic in one movable, resizable in-page window. Escape closes it and restores focus; optional browser storage remembers its bounded position and size.
• Keep runs-library filters, sorting and search in the URL with Back/Forward and reload restoration. Copy the current link from the header without comparison consent; invalid filter syntax uses defaults with a localized status. Recording links can use opaque, root-independent recording references resolved locally with typed absence reasons. Moving a recording within telemetry changes its reference. Desktop links can be copied; link import is not supported.
• Unsigned desktop candidate build tooling for macOS DMG, Linux AppImage/deb and Windows NSIS, with source/tool manifests, checksums, license inventories and supplemental asset checks. Windows configuration embeds the WebView2 offline installer; real offline Windows acceptance remains required.
• Telemetry without imagery or route now draws a Rust-projected local recording line with distance marks and a shared chart cursor. Raw XY coordinates retain declared units or explicit unknown units, without claiming geographic placement; missing or invalid positions remain typed absences, alongside the stage-specific missing imagery reason.
• Generate the b4r changelog from validated per-bead fragments and check freshness plus explicit-base Git change coverage before landing.
• Read worked CLI examples generated from executed Rust Given/When/Then scenarios, with a live freshness check and a combined pre-landing documentation and changelog gate.
• Follow a real-recording CLI walkthrough with a reusable private request builder, exact input identities and observed analysis, saved-result and history outputs, including justified abstention.
• Browse from generated documentation back to BDD catalog entries, exact executable scenario functions, generators, Rust entry points and change fragments, with checked source targets and regenerated line anchors.
• Define BDD-first development and independent native Vega review with a target-owned project rubric, different-model reviewers, explicit evidence requirements and no routine review bypass.
• Add a private-input corpus audit and safe observations for ten distinct original recordings, with a fixed prospective development/held-out split, source-declared units and explicit capture/synchrony limitations. No product behavior or verified coaching claim changes.
• Inspect explicitly selected offline stage and car catalog claims in the headless workflow, preserving ambiguity and source identity across reopen without authenticating the recorded stage or car.
• Add a reusable offline reader for explicitly selected stage and car JSON catalog snapshots, retaining raw fields and source hashes, rejecting malformed or oversized inputs, and exposing duplicate, conflicting, missing and stale ID selections without authenticating a capture. Workflow integration is separate.
• Add the reusable b4r-analysis library for explicit inclusive native-window signed speed summaries and source raw brake falls/throttle transitions, preserving unit and provenance evidence, gaps, deterministic extrema/change endpoints and censored boundaries without inferring duration, physical pedal calibration or technique quality. Workflow and CLI integration are separate.
• Expose bounded optional recorded-signal question plans through the offline workflow and CLI, retaining typed source-derived reports, explicit unavailable outcomes and native-window review exercises across reopen. Public BDD covers the three source-signal questions and workbook-derived window-minimum baseline; numerical observations do not select a driving correction or establish benefit. Summaries distinguish raw baselines from relative extrema and withhold physical interpretations of negative recorded speed.
• Add the reusable relative_position question (Q-RAW-BASELINE): subtract the earliest observed-finite minimum of an explicit inclusive native window from source-scaled unverified driveline values. Compact reports retain the original baseline, relative extrema, raw adjacent changes, unknown units, partial coverage and censored boundar...

Rally Coach 0.1.0-dev.1961+g13073136f765 — alpha

Rally Coach 0.1.0-dev.1961+g13073136f765 — alpha

Build commit: 13073136f765651a690cf161fb914f3e24a8553e

Publication commit: 13073136f765651a690cf161fb914f3e24a8553e

Added

• Show a heading-based vehicle-width band and cursor car rectangle from installed body dimensions on telemetry maps and local recording plots, with remembered layer controls and explicit unknown position offset and missing geometry reasons.
• Local Rallys pages search and sort RaceStat rally name claims and show driven attempts in heuristic same-car passes with at most three days between runs. Repeated stages carry attempt numbers; source names are decoded for display. No official overall result or championship is inferred.
• Rally history offers pass sorting, car and repeated-pass filters and bounded continuation through every name. Passes show distinct-stage catalog kilometres and missing lengths, fastest finishes with at least one same-stage/car peer (ties share the mark), and links to available telemetry recordings.
• Stages includes installed and catalog routes without driving history, with a typed not-driven state, no invented times and working search, sorting and detail links. Local map descriptors and installed personal pacenotes provide declarative search filters using native stage identity resolution.
• Add an English/German static website with a six-step authored screenshot tour, platform download links, unsigned-alpha guidance and local-processing privacy information. The website does not change desktop behavior.
• Runs have a reloadable page linked from the latest recorded run and stage/car history, driver attempt labels, finish state, a single penalty summary, and recorded setup, conditions and splits. Debrief uses the shared exact recording-match verdict rather than an unconditional contradictory disclaimer; no finish tolerance was added.
• Show immediate accessible library loading with consistent page gutters, plain bilingual index phases and measured recording counts through one nonblocking web/native progress contract; retain the saved index and show a typed failure message when refresh fails.
• Telemetry now includes a grip scatter for the visible distance window and a measured gearing sheet for one attempt. Rust clips source acceleration spikes and calculates native powered RPM per km/h, sample counts and speed at observed maximum RPM. Typed absences and German/English notes retain unverified acceleration scale and recorded gear encoding; matching setup gearbox sections link to an attempt sheet.
• Keep bounded local diagnostics with release build identity and prepare a redacted feedback file from Help; open a prefilled GitHub issue only on request, with manual attachment and no app upload.
• Discover protected stage imagery from installed native stage files, send bounded read-only stage proofs without login, explain missing or unmatched versions, and retain cached maps offline while preserving Martin compatibility.
• Private release builds can provide a default imagery endpoint at compile time; saved settings override it, builds without it keep imagery unconfigured, and release manifests record only injection status and SHA256.
• After explicit distance-assumption consent, debrief offers up to three consistency review spots with measured spread, attempt counts and thin support for two measurements. Drivers can choose a spot, keep the existing topics or finish; unavailable evidence and the consistency-only scope remain explicit.
• Five reviewed built-in numeric mappings resolve only matching symbolic pacenote sections in the selected installation, with disclosed provenance. Conflicting installed sections remain explicit. Read-only note diagnostics linked from Settings and recce list installation and built-in resolutions and remaining undefined IDs with counts.
• Maps can display personal pacenotes from the selected RBR installation along a cached host route. Local Rust interpolates unchanged native distances with explicit unverified correspondence, preserves route/roadbook/notation hashes and opaque variant identity, and withholds unsupported positions with typed reasons. Drivers can choose an installed roadbook and reading system using the existing installed reader; a latest-modified file is only a current candidate, never a historical default. Pacenotes remain separate from host-derived corners and crests. No metre scale, fitted offset, authenticated direction, driving benefit or recce/coaching behavior change is claimed.
• Add a persisted bilingual Debug log switch, on by default for now and effective immediately, with private command/route/index/settings/panic diagnostics and map view/tile detail only at debug level; keep normal faults available with debug off and retain read-only settings changes in memory.
• Read static car footprints from the selected RSF installation with bounded persistent caching, source provenance and typed unknown geometry or pose correspondence.
• Run pages link directly to preselected telemetry with large stacked charts, bounded distance zoom and pan, shared chart/map cursor and a small route/driven map. Rust owns channel selection, typed missingness, native held gear, source units, acceleration signs and fixed percentile spike clipping; German and English explain unverified scales.
• Telemetry can overlay two recordings on one Rust-selected common distance grid, including partial attempts. A recording picker starts with the existing proposed reference; two labelled colours share chart zoom, pan and cursor. Typed coverage and stage refusals prevent extrapolation or incompatible overlays.
• The bundled German and English manual now has chapters, searchable topics and back/related-topic navigation. Contextual question marks show a short hover or keyboard-focus hint and open the matching topic in one movable, resizable in-page window. Escape closes it and restores focus; optional browser storage remembers its bounded position and size.
• Keep runs-library filters, sorting and search in the URL with Back/Forward and reload restoration. Copy the current link from the header without comparison consent; invalid filter syntax uses defaults with a localized status. Recording links can use opaque, root-independent recording references resolved locally with typed absence reasons. Moving a recording within telemetry changes its reference. Desktop links can be copied; link import is not supported.
• Unsigned desktop candidate build tooling for macOS DMG, Linux AppImage/deb and Windows NSIS, with source/tool manifests, checksums, license inventories and supplemental asset checks. Windows configuration embeds the WebView2 offline installer; real offline Windows acceptance remains required.
• Telemetry without imagery or route now draws a Rust-projected local recording line with distance marks and a shared chart cursor. Raw XY coordinates retain declared units or explicit unknown units, without claiming geographic placement; missing or invalid positions remain typed absences, alongside the stage-specific missing imagery reason.
• Generate the b4r changelog from validated per-bead fragments and check freshness plus explicit-base Git change coverage before landing.
• Read worked CLI examples generated from executed Rust Given/When/Then scenarios, with a live freshness check and a combined pre-landing documentation and changelog gate.
• Follow a real-recording CLI walkthrough with a reusable private request builder, exact input identities and observed analysis, saved-result and history outputs, including justified abstention.
• Browse from generated documentation back to BDD catalog entries, exact executable scenario functions, generators, Rust entry points and change fragments, with checked source targets and regenerated line anchors.
• Define BDD-first development and independent native Vega review with a target-owned project rubric, different-model reviewers, explicit evidence requirements and no routine review bypass.
• Add a private-input corpus audit and safe observations for ten distinct original recordings, with a fixed prospective development/held-out split, source-declared units and explicit capture/synchrony limitations. No product behavior or verified coaching claim changes.
• Inspect explicitly selected offline stage and car catalog claims in the headless workflow, preserving ambiguity and source identity across reopen without authenticating the recorded stage or car.
• Add a reusable offline reader for explicitly selected stage and car JSON catalog snapshots, retaining raw fields and source hashes, rejecting malformed or oversized inputs, and exposing duplicate, conflicting, missing and stale ID selections without authenticating a capture. Workflow integration is separate.
• Add the reusable b4r-analysis library for explicit inclusive native-window signed speed summaries and source raw brake falls/throttle transitions, preserving unit and provenance evidence, gaps, deterministic extrema/change endpoints and censored boundaries without inferring duration, physical pedal calibration or technique quality. Workflow and CLI integration are separate.
• Expose bounded optional recorded-signal question plans through the offline workflow and CLI, retaining typed source-derived reports, explicit unavailable outcomes and native-window review exercises across reopen. Public BDD covers the three source-signal questions and workbook-derived window-minimum baseline; numerical observations do not select a driving correction or establish benefit. Summaries distinguish raw baselines from relative extrema and withhold physical interpretations of negative recorded speed.
• Add the reusable relative_position question (Q-RAW-BASELINE): subtract the earliest observed-finite minimum of an explicit inclusive native window from source-scaled unverified driveline values. Compact reports retain the original baseline, relative extrema, raw adjacent changes, unknown units, partial coverage and censored boundar...

Rally Coach 0.1.0-dev.1954+g5de1482107da — alpha

Rally Coach 0.1.0-dev.1954+g5de1482107da — alpha

Build commit: 5de1482107da55f8e2f525cd910b9bdc725f1bea

Publication commit: 5de1482107da55f8e2f525cd910b9bdc725f1bea

Added

• Show a heading-based vehicle-width band and cursor car rectangle from installed body dimensions on telemetry maps and local recording plots, with remembered layer controls and explicit unknown position offset and missing geometry reasons.
• Local Rallys pages search and sort RaceStat rally name claims and show driven attempts in heuristic same-car passes with at most three days between runs. Repeated stages carry attempt numbers; source names are decoded for display. No official overall result or championship is inferred.
• Rally history offers pass sorting, car and repeated-pass filters and bounded continuation through every name. Passes show distinct-stage catalog kilometres and missing lengths, fastest finishes with at least one same-stage/car peer (ties share the mark), and links to available telemetry recordings.
• Stages includes installed and catalog routes without driving history, with a typed not-driven state, no invented times and working search, sorting and detail links. Local map descriptors and installed personal pacenotes provide declarative search filters using native stage identity resolution.
• Add an English/German static website with a six-step authored screenshot tour, platform download links, unsigned-alpha guidance and local-processing privacy information. The website does not change desktop behavior.
• Runs have a reloadable page linked from the latest recorded run and stage/car history, driver attempt labels, finish state, a single penalty summary, and recorded setup, conditions and splits. Debrief uses the shared exact recording-match verdict rather than an unconditional contradictory disclaimer; no finish tolerance was added.
• Show immediate accessible library loading with consistent page gutters, plain bilingual index phases and measured recording counts through one nonblocking web/native progress contract; retain the saved index and show a typed failure message when refresh fails.
• Telemetry now includes a grip scatter for the visible distance window and a measured gearing sheet for one attempt. Rust clips source acceleration spikes and calculates native powered RPM per km/h, sample counts and speed at observed maximum RPM. Typed absences and German/English notes retain unverified acceleration scale and recorded gear encoding; matching setup gearbox sections link to an attempt sheet.
• Keep bounded local diagnostics with release build identity and prepare a redacted feedback file from Help; open a prefilled GitHub issue only on request, with manual attachment and no app upload.
• Discover protected stage imagery from installed native stage files, send bounded read-only stage proofs without login, explain missing or unmatched versions, and retain cached maps offline while preserving Martin compatibility.
• Private release builds can provide a default imagery endpoint at compile time; saved settings override it, builds without it keep imagery unconfigured, and release manifests record only injection status and SHA256.
• After explicit distance-assumption consent, debrief offers up to three consistency review spots with measured spread, attempt counts and thin support for two measurements. Drivers can choose a spot, keep the existing topics or finish; unavailable evidence and the consistency-only scope remain explicit.
• Five reviewed built-in numeric mappings resolve only matching symbolic pacenote sections in the selected installation, with disclosed provenance. Conflicting installed sections remain explicit. Read-only note diagnostics linked from Settings and recce list installation and built-in resolutions and remaining undefined IDs with counts.
• Maps can display personal pacenotes from the selected RBR installation along a cached host route. Local Rust interpolates unchanged native distances with explicit unverified correspondence, preserves route/roadbook/notation hashes and opaque variant identity, and withholds unsupported positions with typed reasons. Drivers can choose an installed roadbook and reading system using the existing installed reader; a latest-modified file is only a current candidate, never a historical default. Pacenotes remain separate from host-derived corners and crests. No metre scale, fitted offset, authenticated direction, driving benefit or recce/coaching behavior change is claimed.
• Add a persisted bilingual Debug log switch, on by default for now and effective immediately, with private command/route/index/settings/panic diagnostics and map view/tile detail only at debug level; keep normal faults available with debug off and retain read-only settings changes in memory.
• Read static car footprints from the selected RSF installation with bounded persistent caching, source provenance and typed unknown geometry or pose correspondence.
• Run pages link directly to preselected telemetry with large stacked charts, bounded distance zoom and pan, shared chart/map cursor and a small route/driven map. Rust owns channel selection, typed missingness, native held gear, source units, acceleration signs and fixed percentile spike clipping; German and English explain unverified scales.
• Telemetry can overlay two recordings on one Rust-selected common distance grid, including partial attempts. A recording picker starts with the existing proposed reference; two labelled colours share chart zoom, pan and cursor. Typed coverage and stage refusals prevent extrapolation or incompatible overlays.
• The bundled German and English manual now has chapters, searchable topics and back/related-topic navigation. Contextual question marks show a short hover or keyboard-focus hint and open the matching topic in one movable, resizable in-page window. Escape closes it and restores focus; optional browser storage remembers its bounded position and size.
• Keep runs-library filters, sorting and search in the URL with Back/Forward and reload restoration. Copy the current link from the header without comparison consent; invalid filter syntax uses defaults with a localized status. Recording links can use opaque, root-independent recording references resolved locally with typed absence reasons. Moving a recording within telemetry changes its reference. Desktop links can be copied; link import is not supported.
• Unsigned desktop candidate build tooling for macOS DMG, Linux AppImage/deb and Windows NSIS, with source/tool manifests, checksums, license inventories and supplemental asset checks. Windows configuration embeds the WebView2 offline installer; real offline Windows acceptance remains required.
• Telemetry without imagery or route now draws a Rust-projected local recording line with distance marks and a shared chart cursor. Raw XY coordinates retain declared units or explicit unknown units, without claiming geographic placement; missing or invalid positions remain typed absences, alongside the stage-specific missing imagery reason.
• Generate the b4r changelog from validated per-bead fragments and check freshness plus explicit-base Git change coverage before landing.
• Read worked CLI examples generated from executed Rust Given/When/Then scenarios, with a live freshness check and a combined pre-landing documentation and changelog gate.
• Follow a real-recording CLI walkthrough with a reusable private request builder, exact input identities and observed analysis, saved-result and history outputs, including justified abstention.
• Browse from generated documentation back to BDD catalog entries, exact executable scenario functions, generators, Rust entry points and change fragments, with checked source targets and regenerated line anchors.
• Define BDD-first development and independent native Vega review with a target-owned project rubric, different-model reviewers, explicit evidence requirements and no routine review bypass.
• Add a private-input corpus audit and safe observations for ten distinct original recordings, with a fixed prospective development/held-out split, source-declared units and explicit capture/synchrony limitations. No product behavior or verified coaching claim changes.
• Inspect explicitly selected offline stage and car catalog claims in the headless workflow, preserving ambiguity and source identity across reopen without authenticating the recorded stage or car.
• Add a reusable offline reader for explicitly selected stage and car JSON catalog snapshots, retaining raw fields and source hashes, rejecting malformed or oversized inputs, and exposing duplicate, conflicting, missing and stale ID selections without authenticating a capture. Workflow integration is separate.
• Add the reusable b4r-analysis library for explicit inclusive native-window signed speed summaries and source raw brake falls/throttle transitions, preserving unit and provenance evidence, gaps, deterministic extrema/change endpoints and censored boundaries without inferring duration, physical pedal calibration or technique quality. Workflow and CLI integration are separate.
• Expose bounded optional recorded-signal question plans through the offline workflow and CLI, retaining typed source-derived reports, explicit unavailable outcomes and native-window review exercises across reopen. Public BDD covers the three source-signal questions and workbook-derived window-minimum baseline; numerical observations do not select a driving correction or establish benefit. Summaries distinguish raw baselines from relative extrema and withhold physical interpretations of negative recorded speed.
• Add the reusable relative_position question (Q-RAW-BASELINE): subtract the earliest observed-finite minimum of an explicit inclusive native window from source-scaled unverified driveline values. Compact reports retain the original baseline, relative extrema, raw adjacent changes, unknown units, partial coverage and censored boundar...

Rally Coach 0.1.0-dev.1937+g6532d9686944 — alpha

Rally Coach 0.1.0-dev.1937+g6532d9686944 — alpha

Build commit: 6532d9686944ff8525b950195c81a040ad4d744a

Publication commit: 6532d9686944ff8525b950195c81a040ad4d744a

Added

• Show a heading-based vehicle-width band and cursor car rectangle from installed body dimensions on telemetry maps and local recording plots, with remembered layer controls and explicit unknown position offset and missing geometry reasons.
• Local Rallys pages search and sort RaceStat rally name claims and show driven attempts in heuristic same-car passes with at most three days between runs. Repeated stages carry attempt numbers; source names are decoded for display. No official overall result or championship is inferred.
• Rally history offers pass sorting, car and repeated-pass filters and bounded continuation through every name. Passes show distinct-stage catalog kilometres and missing lengths, fastest finishes with at least one same-stage/car peer (ties share the mark), and links to available telemetry recordings.
• Stages includes installed and catalog routes without driving history, with a typed not-driven state, no invented times and working search, sorting and detail links. Local map descriptors and installed personal pacenotes provide declarative search filters using native stage identity resolution.
• Add an English/German static website with a six-step authored screenshot tour, platform download links, unsigned-alpha guidance and local-processing privacy information. The website does not change desktop behavior.
• Runs have a reloadable page linked from the latest recorded run and stage/car history, driver attempt labels, finish state, a single penalty summary, and recorded setup, conditions and splits. Debrief uses the shared exact recording-match verdict rather than an unconditional contradictory disclaimer; no finish tolerance was added.
• Show immediate accessible library loading with consistent page gutters, plain bilingual index phases and measured recording counts through one nonblocking web/native progress contract; retain the saved index and show a typed failure message when refresh fails.
• Telemetry now includes a grip scatter for the visible distance window and a measured gearing sheet for one attempt. Rust clips source acceleration spikes and calculates native powered RPM per km/h, sample counts and speed at observed maximum RPM. Typed absences and German/English notes retain unverified acceleration scale and recorded gear encoding; matching setup gearbox sections link to an attempt sheet.
• Keep bounded local diagnostics with release build identity and prepare a redacted feedback file from Help; open a prefilled GitHub issue only on request, with manual attachment and no app upload.
• Discover protected stage imagery from installed native stage files, send bounded read-only stage proofs without login, explain missing or unmatched versions, and retain cached maps offline while preserving Martin compatibility.
• Private release builds can provide a default imagery endpoint at compile time; saved settings override it, builds without it keep imagery unconfigured, and release manifests record only injection status and SHA256.
• After explicit distance-assumption consent, debrief offers up to three consistency review spots with measured spread, attempt counts and thin support for two measurements. Drivers can choose a spot, keep the existing topics or finish; unavailable evidence and the consistency-only scope remain explicit.
• Five reviewed built-in numeric mappings resolve only matching symbolic pacenote sections in the selected installation, with disclosed provenance. Conflicting installed sections remain explicit. Read-only note diagnostics linked from Settings and recce list installation and built-in resolutions and remaining undefined IDs with counts.
• Maps can display personal pacenotes from the selected RBR installation along a cached host route. Local Rust interpolates unchanged native distances with explicit unverified correspondence, preserves route/roadbook/notation hashes and opaque variant identity, and withholds unsupported positions with typed reasons. Drivers can choose an installed roadbook and reading system using the existing installed reader; a latest-modified file is only a current candidate, never a historical default. Pacenotes remain separate from host-derived corners and crests. No metre scale, fitted offset, authenticated direction, driving benefit or recce/coaching behavior change is claimed.
• Add a persisted bilingual Debug log switch, on by default for now and effective immediately, with private command/route/index/settings/panic diagnostics and map view/tile detail only at debug level; keep normal faults available with debug off and retain read-only settings changes in memory.
• Read static car footprints from the selected RSF installation with bounded persistent caching, source provenance and typed unknown geometry or pose correspondence.
• Run pages link directly to preselected telemetry with large stacked charts, bounded distance zoom and pan, shared chart/map cursor and a small route/driven map. Rust owns channel selection, typed missingness, native held gear, source units, acceleration signs and fixed percentile spike clipping; German and English explain unverified scales.
• Telemetry can overlay two recordings on one Rust-selected common distance grid, including partial attempts. A recording picker starts with the existing proposed reference; two labelled colours share chart zoom, pan and cursor. Typed coverage and stage refusals prevent extrapolation or incompatible overlays.
• The bundled German and English manual now has chapters, searchable topics and back/related-topic navigation. Contextual question marks show a short hover or keyboard-focus hint and open the matching topic in one movable, resizable in-page window. Escape closes it and restores focus; optional browser storage remembers its bounded position and size.
• Keep runs-library filters, sorting and search in the URL with Back/Forward and reload restoration. Copy the current link from the header without comparison consent; invalid filter syntax uses defaults with a localized status. Recording links can use opaque, root-independent recording references resolved locally with typed absence reasons. Moving a recording within telemetry changes its reference. Desktop links can be copied; link import is not supported.
• Unsigned desktop candidate build tooling for macOS DMG, Linux AppImage/deb and Windows NSIS, with source/tool manifests, checksums, license inventories and supplemental asset checks. Windows configuration embeds the WebView2 offline installer; real offline Windows acceptance remains required.
• Telemetry without imagery or route now draws a Rust-projected local recording line with distance marks and a shared chart cursor. Raw XY coordinates retain declared units or explicit unknown units, without claiming geographic placement; missing or invalid positions remain typed absences, alongside the stage-specific missing imagery reason.
• Generate the b4r changelog from validated per-bead fragments and check freshness plus explicit-base Git change coverage before landing.
• Read worked CLI examples generated from executed Rust Given/When/Then scenarios, with a live freshness check and a combined pre-landing documentation and changelog gate.
• Follow a real-recording CLI walkthrough with a reusable private request builder, exact input identities and observed analysis, saved-result and history outputs, including justified abstention.
• Browse from generated documentation back to BDD catalog entries, exact executable scenario functions, generators, Rust entry points and change fragments, with checked source targets and regenerated line anchors.
• Define BDD-first development and independent native Vega review with a target-owned project rubric, different-model reviewers, explicit evidence requirements and no routine review bypass.
• Add a private-input corpus audit and safe observations for ten distinct original recordings, with a fixed prospective development/held-out split, source-declared units and explicit capture/synchrony limitations. No product behavior or verified coaching claim changes.
• Inspect explicitly selected offline stage and car catalog claims in the headless workflow, preserving ambiguity and source identity across reopen without authenticating the recorded stage or car.
• Add a reusable offline reader for explicitly selected stage and car JSON catalog snapshots, retaining raw fields and source hashes, rejecting malformed or oversized inputs, and exposing duplicate, conflicting, missing and stale ID selections without authenticating a capture. Workflow integration is separate.
• Add the reusable b4r-analysis library for explicit inclusive native-window signed speed summaries and source raw brake falls/throttle transitions, preserving unit and provenance evidence, gaps, deterministic extrema/change endpoints and censored boundaries without inferring duration, physical pedal calibration or technique quality. Workflow and CLI integration are separate.
• Expose bounded optional recorded-signal question plans through the offline workflow and CLI, retaining typed source-derived reports, explicit unavailable outcomes and native-window review exercises across reopen. Public BDD covers the three source-signal questions and workbook-derived window-minimum baseline; numerical observations do not select a driving correction or establish benefit. Summaries distinguish raw baselines from relative extrema and withhold physical interpretations of negative recorded speed.
• Add the reusable relative_position question (Q-RAW-BASELINE): subtract the earliest observed-finite minimum of an explicit inclusive native window from source-scaled unverified driveline values. Compact reports retain the original baseline, relative extrema, raw adjacent changes, unknown units, partial coverage and censored boundar...

Rally Coach 0.1.0-dev.1831+g54153bee6109 — alpha

Rally Coach 0.1.0-dev.1831+g54153bee6109 — alpha

Build commit: 54153bee61099d83852d0a3828cdfb1964cebb6c

Publication commit: 54153bee61099d83852d0a3828cdfb1964cebb6c

Added

• Show a heading-based vehicle-width band and cursor car rectangle from installed body dimensions on telemetry maps and local recording plots, with remembered layer controls and explicit unknown position offset and missing geometry reasons.
• Local Rallys pages search and sort RaceStat rally name claims and show driven attempts in heuristic same-car passes with at most three days between runs. Repeated stages carry attempt numbers; source names are decoded for display. No official overall result or championship is inferred.
• Rally history offers pass sorting, car and repeated-pass filters and bounded continuation through every name. Passes show distinct-stage catalog kilometres and missing lengths, fastest finishes with at least one same-stage/car peer (ties share the mark), and links to available telemetry recordings.
• Stages includes installed and catalog routes without driving history, with a typed not-driven state, no invented times and working search, sorting and detail links. Local map descriptors and installed personal pacenotes provide declarative search filters using native stage identity resolution.
• Add an English/German static website with a six-step authored screenshot tour, platform download links, unsigned-alpha guidance and local-processing privacy information. The website does not change desktop behavior.
• Runs have a reloadable page linked from the latest recorded run and stage/car history, driver attempt labels, finish state, a single penalty summary, and recorded setup, conditions and splits. Debrief uses the shared exact recording-match verdict rather than an unconditional contradictory disclaimer; no finish tolerance was added.
• Show immediate accessible library loading with consistent page gutters, plain bilingual index phases and measured recording counts through one nonblocking web/native progress contract; retain the saved index and show a typed failure message when refresh fails.
• Telemetry now includes a grip scatter for the visible distance window and a measured gearing sheet for one attempt. Rust clips source acceleration spikes and calculates native powered RPM per km/h, sample counts and speed at observed maximum RPM. Typed absences and German/English notes retain unverified acceleration scale and recorded gear encoding; matching setup gearbox sections link to an attempt sheet.
• Keep bounded local diagnostics with release build identity and prepare a redacted feedback file from Help; open a prefilled GitHub issue only on request, with manual attachment and no app upload.
• Discover protected stage imagery from installed native stage files, send bounded read-only stage proofs without login, explain missing or unmatched versions, and retain cached maps offline while preserving Martin compatibility.
• Private release builds can provide a default imagery endpoint at compile time; saved settings override it, builds without it keep imagery unconfigured, and release manifests record only injection status and SHA256.
• After explicit distance-assumption consent, debrief offers up to three consistency review spots with measured spread, attempt counts and thin support for two measurements. Drivers can choose a spot, keep the existing topics or finish; unavailable evidence and the consistency-only scope remain explicit.
• Five reviewed built-in numeric mappings resolve only matching symbolic pacenote sections in the selected installation, with disclosed provenance. Conflicting installed sections remain explicit. Read-only note diagnostics linked from Settings and recce list installation and built-in resolutions and remaining undefined IDs with counts.
• Maps can display personal pacenotes from the selected RBR installation along a cached host route. Local Rust interpolates unchanged native distances with explicit unverified correspondence, preserves route/roadbook/notation hashes and opaque variant identity, and withholds unsupported positions with typed reasons. Drivers can choose an installed roadbook and reading system using the existing installed reader; a latest-modified file is only a current candidate, never a historical default. Pacenotes remain separate from host-derived corners and crests. No metre scale, fitted offset, authenticated direction, driving benefit or recce/coaching behavior change is claimed.
• Add a persisted bilingual Debug log switch, on by default for now and effective immediately, with private command/route/index/settings/panic diagnostics and map view/tile detail only at debug level; keep normal faults available with debug off and retain read-only settings changes in memory.
• Read static car footprints from the selected RSF installation with bounded persistent caching, source provenance and typed unknown geometry or pose correspondence.
• Run pages link directly to preselected telemetry with large stacked charts, bounded distance zoom and pan, shared chart/map cursor and a small route/driven map. Rust owns channel selection, typed missingness, native held gear, source units, acceleration signs and fixed percentile spike clipping; German and English explain unverified scales.
• Telemetry can overlay two recordings on one Rust-selected common distance grid, including partial attempts. A recording picker starts with the existing proposed reference; two labelled colours share chart zoom, pan and cursor. Typed coverage and stage refusals prevent extrapolation or incompatible overlays.
• The bundled German and English manual now has chapters, searchable topics and back/related-topic navigation. Contextual question marks show a short hover or keyboard-focus hint and open the matching topic in one movable, resizable in-page window. Escape closes it and restores focus; optional browser storage remembers its bounded position and size.
• Keep runs-library filters, sorting and search in the URL with Back/Forward and reload restoration. Copy the current link from the header without comparison consent; invalid filter syntax uses defaults with a localized status. Recording links can use opaque, root-independent recording references resolved locally with typed absence reasons. Moving a recording within telemetry changes its reference. Desktop links can be copied; link import is not supported.
• Unsigned desktop candidate build tooling for macOS DMG, Linux AppImage/deb and Windows NSIS, with source/tool manifests, checksums, license inventories and supplemental asset checks. Windows configuration embeds the WebView2 offline installer; real offline Windows acceptance remains required.
• Telemetry without imagery or route now draws a Rust-projected local recording line with distance marks and a shared chart cursor. Raw XY coordinates retain declared units or explicit unknown units, without claiming geographic placement; missing or invalid positions remain typed absences, alongside the stage-specific missing imagery reason.
• Generate the b4r changelog from validated per-bead fragments and check freshness plus explicit-base Git change coverage before landing.
• Read worked CLI examples generated from executed Rust Given/When/Then scenarios, with a live freshness check and a combined pre-landing documentation and changelog gate.
• Follow a real-recording CLI walkthrough with a reusable private request builder, exact input identities and observed analysis, saved-result and history outputs, including justified abstention.
• Browse from generated documentation back to BDD catalog entries, exact executable scenario functions, generators, Rust entry points and change fragments, with checked source targets and regenerated line anchors.
• Define BDD-first development and independent native Vega review with a target-owned project rubric, different-model reviewers, explicit evidence requirements and no routine review bypass.
• Add a private-input corpus audit and safe observations for ten distinct original recordings, with a fixed prospective development/held-out split, source-declared units and explicit capture/synchrony limitations. No product behavior or verified coaching claim changes.
• Inspect explicitly selected offline stage and car catalog claims in the headless workflow, preserving ambiguity and source identity across reopen without authenticating the recorded stage or car.
• Add a reusable offline reader for explicitly selected stage and car JSON catalog snapshots, retaining raw fields and source hashes, rejecting malformed or oversized inputs, and exposing duplicate, conflicting, missing and stale ID selections without authenticating a capture. Workflow integration is separate.
• Add the reusable b4r-analysis library for explicit inclusive native-window signed speed summaries and source raw brake falls/throttle transitions, preserving unit and provenance evidence, gaps, deterministic extrema/change endpoints and censored boundaries without inferring duration, physical pedal calibration or technique quality. Workflow and CLI integration are separate.
• Expose bounded optional recorded-signal question plans through the offline workflow and CLI, retaining typed source-derived reports, explicit unavailable outcomes and native-window review exercises across reopen. Public BDD covers the three source-signal questions and workbook-derived window-minimum baseline; numerical observations do not select a driving correction or establish benefit. Summaries distinguish raw baselines from relative extrema and withhold physical interpretations of negative recorded speed.
• Add the reusable relative_position question (Q-RAW-BASELINE): subtract the earliest observed-finite minimum of an explicit inclusive native window from source-scaled unverified driveline values. Compact reports retain the original baseline, relative extrema, raw adjacent changes, unknown units, partial coverage and censored boundar...

Rally Coach 0.1.0-dev.1827+g5c3cfcf026bc — alpha

Rally Coach 0.1.0-dev.1827+g5c3cfcf026bc — alpha

Build commit: 5c3cfcf026bca8eb2199e7a2ebabfcd82c672d4c

Publication commit: 5c3cfcf026bca8eb2199e7a2ebabfcd82c672d4c

Added

• Show a heading-based vehicle-width band and cursor car rectangle from installed body dimensions on telemetry maps and local recording plots, with remembered layer controls and explicit unknown position offset and missing geometry reasons.
• Local Rallys pages search and sort RaceStat rally name claims and show driven attempts in heuristic same-car passes with at most three days between runs. Repeated stages carry attempt numbers; source names are decoded for display. No official overall result or championship is inferred.
• Rally history offers pass sorting, car and repeated-pass filters and bounded continuation through every name. Passes show distinct-stage catalog kilometres and missing lengths, fastest finishes with at least one same-stage/car peer (ties share the mark), and links to available telemetry recordings.
• Stages includes installed and catalog routes without driving history, with a typed not-driven state, no invented times and working search, sorting and detail links. Local map descriptors and installed personal pacenotes provide declarative search filters using native stage identity resolution.
• Add an English/German static website with a six-step authored screenshot tour, platform download links, unsigned-alpha guidance and local-processing privacy information. The website does not change desktop behavior.
• Runs have a reloadable page linked from the latest recorded run and stage/car history, driver attempt labels, finish state, a single penalty summary, and recorded setup, conditions and splits. Debrief uses the shared exact recording-match verdict rather than an unconditional contradictory disclaimer; no finish tolerance was added.
• Show immediate accessible library loading with consistent page gutters, plain bilingual index phases and measured recording counts through one nonblocking web/native progress contract; retain the saved index and show a typed failure message when refresh fails.
• Telemetry now includes a grip scatter for the visible distance window and a measured gearing sheet for one attempt. Rust clips source acceleration spikes and calculates native powered RPM per km/h, sample counts and speed at observed maximum RPM. Typed absences and German/English notes retain unverified acceleration scale and recorded gear encoding; matching setup gearbox sections link to an attempt sheet.
• Keep bounded local diagnostics with release build identity and prepare a redacted feedback file from Help; open a prefilled GitHub issue only on request, with manual attachment and no app upload.
• Discover protected stage imagery from installed native stage files, send bounded read-only stage proofs without login, explain missing or unmatched versions, and retain cached maps offline while preserving Martin compatibility.
• Private release builds can provide a default imagery endpoint at compile time; saved settings override it, builds without it keep imagery unconfigured, and release manifests record only injection status and SHA256.
• After explicit distance-assumption consent, debrief offers up to three consistency review spots with measured spread, attempt counts and thin support for two measurements. Drivers can choose a spot, keep the existing topics or finish; unavailable evidence and the consistency-only scope remain explicit.
• Five reviewed built-in numeric mappings resolve only matching symbolic pacenote sections in the selected installation, with disclosed provenance. Conflicting installed sections remain explicit. Read-only note diagnostics linked from Settings and recce list installation and built-in resolutions and remaining undefined IDs with counts.
• Maps can display personal pacenotes from the selected RBR installation along a cached host route. Local Rust interpolates unchanged native distances with explicit unverified correspondence, preserves route/roadbook/notation hashes and opaque variant identity, and withholds unsupported positions with typed reasons. Drivers can choose an installed roadbook and reading system using the existing installed reader; a latest-modified file is only a current candidate, never a historical default. Pacenotes remain separate from host-derived corners and crests. No metre scale, fitted offset, authenticated direction, driving benefit or recce/coaching behavior change is claimed.
• Add a persisted bilingual Debug log switch, on by default for now and effective immediately, with private command/route/index/settings/panic diagnostics and map view/tile detail only at debug level; keep normal faults available with debug off and retain read-only settings changes in memory.
• Read static car footprints from the selected RSF installation with bounded persistent caching, source provenance and typed unknown geometry or pose correspondence.
• Run pages link directly to preselected telemetry with large stacked charts, bounded distance zoom and pan, shared chart/map cursor and a small route/driven map. Rust owns channel selection, typed missingness, native held gear, source units, acceleration signs and fixed percentile spike clipping; German and English explain unverified scales.
• Telemetry can overlay two recordings on one Rust-selected common distance grid, including partial attempts. A recording picker starts with the existing proposed reference; two labelled colours share chart zoom, pan and cursor. Typed coverage and stage refusals prevent extrapolation or incompatible overlays.
• The bundled German and English manual now has chapters, searchable topics and back/related-topic navigation. Contextual question marks show a short hover or keyboard-focus hint and open the matching topic in one movable, resizable in-page window. Escape closes it and restores focus; optional browser storage remembers its bounded position and size.
• Keep runs-library filters, sorting and search in the URL with Back/Forward and reload restoration. Copy the current link from the header without comparison consent; invalid filter syntax uses defaults with a localized status. Recording links can use opaque, root-independent recording references resolved locally with typed absence reasons. Moving a recording within telemetry changes its reference. Desktop links can be copied; link import is not supported.
• Unsigned desktop candidate build tooling for macOS DMG, Linux AppImage/deb and Windows NSIS, with source/tool manifests, checksums, license inventories and supplemental asset checks. Windows configuration embeds the WebView2 offline installer; real offline Windows acceptance remains required.
• Telemetry without imagery or route now draws a Rust-projected local recording line with distance marks and a shared chart cursor. Raw XY coordinates retain declared units or explicit unknown units, without claiming geographic placement; missing or invalid positions remain typed absences, alongside the stage-specific missing imagery reason.
• Generate the b4r changelog from validated per-bead fragments and check freshness plus explicit-base Git change coverage before landing.
• Read worked CLI examples generated from executed Rust Given/When/Then scenarios, with a live freshness check and a combined pre-landing documentation and changelog gate.
• Follow a real-recording CLI walkthrough with a reusable private request builder, exact input identities and observed analysis, saved-result and history outputs, including justified abstention.
• Browse from generated documentation back to BDD catalog entries, exact executable scenario functions, generators, Rust entry points and change fragments, with checked source targets and regenerated line anchors.
• Define BDD-first development and independent native Vega review with a target-owned project rubric, different-model reviewers, explicit evidence requirements and no routine review bypass.
• Add a private-input corpus audit and safe observations for ten distinct original recordings, with a fixed prospective development/held-out split, source-declared units and explicit capture/synchrony limitations. No product behavior or verified coaching claim changes.
• Inspect explicitly selected offline stage and car catalog claims in the headless workflow, preserving ambiguity and source identity across reopen without authenticating the recorded stage or car.
• Add a reusable offline reader for explicitly selected stage and car JSON catalog snapshots, retaining raw fields and source hashes, rejecting malformed or oversized inputs, and exposing duplicate, conflicting, missing and stale ID selections without authenticating a capture. Workflow integration is separate.
• Add the reusable b4r-analysis library for explicit inclusive native-window signed speed summaries and source raw brake falls/throttle transitions, preserving unit and provenance evidence, gaps, deterministic extrema/change endpoints and censored boundaries without inferring duration, physical pedal calibration or technique quality. Workflow and CLI integration are separate.
• Expose bounded optional recorded-signal question plans through the offline workflow and CLI, retaining typed source-derived reports, explicit unavailable outcomes and native-window review exercises across reopen. Public BDD covers the three source-signal questions and workbook-derived window-minimum baseline; numerical observations do not select a driving correction or establish benefit. Summaries distinguish raw baselines from relative extrema and withhold physical interpretations of negative recorded speed.
• Add the reusable relative_position question (Q-RAW-BASELINE): subtract the earliest observed-finite minimum of an explicit inclusive native window from source-scaled unverified driveline values. Compact reports retain the original baseline, relative extrema, raw adjacent changes, unknown units, partial coverage and censored boundar...

Rally Coach 0.1.0-dev.1807+g392ead6e883d — alpha

Rally Coach 0.1.0-dev.1807+g392ead6e883d — alpha

Build commit: 392ead6e883d9b518776d999bf962c4ab5b698a1

Publication commit: 392ead6e883d9b518776d999bf962c4ab5b698a1

Added

• Show a heading-based vehicle-width band and cursor car rectangle from installed body dimensions on telemetry maps and local recording plots, with remembered layer controls and explicit unknown position offset and missing geometry reasons.
• Local Rallys pages search and sort RaceStat rally name claims and show driven attempts in heuristic same-car passes with at most three days between runs. Repeated stages carry attempt numbers; source names are decoded for display. No official overall result or championship is inferred.
• Rally history offers pass sorting, car and repeated-pass filters and bounded continuation through every name. Passes show distinct-stage catalog kilometres and missing lengths, fastest finishes with at least one same-stage/car peer (ties share the mark), and links to available telemetry recordings.
• Stages includes installed and catalog routes without driving history, with a typed not-driven state, no invented times and working search, sorting and detail links. Local map descriptors and installed personal pacenotes provide declarative search filters using native stage identity resolution.
• Add an English/German static website with a six-step authored screenshot tour, platform download links, unsigned-alpha guidance and local-processing privacy information. The website does not change desktop behavior.
• Runs have a reloadable page linked from the latest recorded run and stage/car history, driver attempt labels, finish state, a single penalty summary, and recorded setup, conditions and splits. Debrief uses the shared exact recording-match verdict rather than an unconditional contradictory disclaimer; no finish tolerance was added.
• Show immediate accessible library loading with consistent page gutters, plain bilingual index phases and measured recording counts through one nonblocking web/native progress contract; retain the saved index and show a typed failure message when refresh fails.
• Telemetry now includes a grip scatter for the visible distance window and a measured gearing sheet for one attempt. Rust clips source acceleration spikes and calculates native powered RPM per km/h, sample counts and speed at observed maximum RPM. Typed absences and German/English notes retain unverified acceleration scale and recorded gear encoding; matching setup gearbox sections link to an attempt sheet.
• Keep bounded local diagnostics with release build identity and prepare a redacted feedback file from Help; open a prefilled GitHub issue only on request, with manual attachment and no app upload.
• Discover protected stage imagery from installed native stage files, send bounded read-only stage proofs without login, explain missing or unmatched versions, and retain cached maps offline while preserving Martin compatibility.
• Private release builds can provide a default imagery endpoint at compile time; saved settings override it, builds without it keep imagery unconfigured, and release manifests record only injection status and SHA256.
• After explicit distance-assumption consent, debrief offers up to three consistency review spots with measured spread, attempt counts and thin support for two measurements. Drivers can choose a spot, keep the existing topics or finish; unavailable evidence and the consistency-only scope remain explicit.
• Five reviewed built-in numeric mappings resolve only matching symbolic pacenote sections in the selected installation, with disclosed provenance. Conflicting installed sections remain explicit. Read-only note diagnostics linked from Settings and recce list installation and built-in resolutions and remaining undefined IDs with counts.
• Maps can display personal pacenotes from the selected RBR installation along a cached host route. Local Rust interpolates unchanged native distances with explicit unverified correspondence, preserves route/roadbook/notation hashes and opaque variant identity, and withholds unsupported positions with typed reasons. Drivers can choose an installed roadbook and reading system using the existing installed reader; a latest-modified file is only a current candidate, never a historical default. Pacenotes remain separate from host-derived corners and crests. No metre scale, fitted offset, authenticated direction, driving benefit or recce/coaching behavior change is claimed.
• Add a persisted bilingual Debug log switch, on by default for now and effective immediately, with private command/route/index/settings/panic diagnostics and map view/tile detail only at debug level; keep normal faults available with debug off and retain read-only settings changes in memory.
• Read static car footprints from the selected RSF installation with bounded persistent caching, source provenance and typed unknown geometry or pose correspondence.
• Run pages link directly to preselected telemetry with large stacked charts, bounded distance zoom and pan, shared chart/map cursor and a small route/driven map. Rust owns channel selection, typed missingness, native held gear, source units, acceleration signs and fixed percentile spike clipping; German and English explain unverified scales.
• Telemetry can overlay two recordings on one Rust-selected common distance grid, including partial attempts. A recording picker starts with the existing proposed reference; two labelled colours share chart zoom, pan and cursor. Typed coverage and stage refusals prevent extrapolation or incompatible overlays.
• The bundled German and English manual now has chapters, searchable topics and back/related-topic navigation. Contextual question marks show a short hover or keyboard-focus hint and open the matching topic in one movable, resizable in-page window. Escape closes it and restores focus; optional browser storage remembers its bounded position and size.
• Keep runs-library filters, sorting and search in the URL with Back/Forward and reload restoration. Copy the current link from the header without comparison consent; invalid filter syntax uses defaults with a localized status. Recording links can use opaque, root-independent recording references resolved locally with typed absence reasons. Moving a recording within telemetry changes its reference. Desktop links can be copied; link import is not supported.
• Unsigned desktop candidate build tooling for macOS DMG, Linux AppImage/deb and Windows NSIS, with source/tool manifests, checksums, license inventories and supplemental asset checks. Windows configuration embeds the WebView2 offline installer; real offline Windows acceptance remains required.
• Telemetry without imagery or route now draws a Rust-projected local recording line with distance marks and a shared chart cursor. Raw XY coordinates retain declared units or explicit unknown units, without claiming geographic placement; missing or invalid positions remain typed absences, alongside the stage-specific missing imagery reason.
• Generate the b4r changelog from validated per-bead fragments and check freshness plus explicit-base Git change coverage before landing.
• Read worked CLI examples generated from executed Rust Given/When/Then scenarios, with a live freshness check and a combined pre-landing documentation and changelog gate.
• Follow a real-recording CLI walkthrough with a reusable private request builder, exact input identities and observed analysis, saved-result and history outputs, including justified abstention.
• Browse from generated documentation back to BDD catalog entries, exact executable scenario functions, generators, Rust entry points and change fragments, with checked source targets and regenerated line anchors.
• Define BDD-first development and independent native Vega review with a target-owned project rubric, different-model reviewers, explicit evidence requirements and no routine review bypass.
• Add a private-input corpus audit and safe observations for ten distinct original recordings, with a fixed prospective development/held-out split, source-declared units and explicit capture/synchrony limitations. No product behavior or verified coaching claim changes.
• Inspect explicitly selected offline stage and car catalog claims in the headless workflow, preserving ambiguity and source identity across reopen without authenticating the recorded stage or car.
• Add a reusable offline reader for explicitly selected stage and car JSON catalog snapshots, retaining raw fields and source hashes, rejecting malformed or oversized inputs, and exposing duplicate, conflicting, missing and stale ID selections without authenticating a capture. Workflow integration is separate.
• Add the reusable b4r-analysis library for explicit inclusive native-window signed speed summaries and source raw brake falls/throttle transitions, preserving unit and provenance evidence, gaps, deterministic extrema/change endpoints and censored boundaries without inferring duration, physical pedal calibration or technique quality. Workflow and CLI integration are separate.
• Expose bounded optional recorded-signal question plans through the offline workflow and CLI, retaining typed source-derived reports, explicit unavailable outcomes and native-window review exercises across reopen. Public BDD covers the three source-signal questions and workbook-derived window-minimum baseline; numerical observations do not select a driving correction or establish benefit. Summaries distinguish raw baselines from relative extrema and withhold physical interpretations of negative recorded speed.
• Add the reusable relative_position question (Q-RAW-BASELINE): subtract the earliest observed-finite minimum of an explicit inclusive native window from source-scaled unverified driveline values. Compact reports retain the original baseline, relative extrema, raw adjacent changes, unknown units, partial coverage and censored boundar...

Rally Coach 0.1.0-dev.1779+g2f6268a0fe76 — alpha

Rally Coach 0.1.0-dev.1779+g2f6268a0fe76 — alpha

Build commit: 2f6268a0fe7623de797cc3e565e63472929bca5f

Publication commit: 2f6268a0fe7623de797cc3e565e63472929bca5f

Added

• Show a heading-based vehicle-width band and cursor car rectangle from installed body dimensions on telemetry maps and local recording plots, with remembered layer controls and explicit unknown position offset and missing geometry reasons.
• Local Rallys pages search and sort RaceStat rally name claims and show driven attempts in heuristic same-car passes with at most three days between runs. Repeated stages carry attempt numbers; source names are decoded for display. No official overall result or championship is inferred.
• Rally history offers pass sorting, car and repeated-pass filters and bounded continuation through every name. Passes show distinct-stage catalog kilometres and missing lengths, fastest finishes with at least one same-stage/car peer (ties share the mark), and links to available telemetry recordings.
• Stages includes installed and catalog routes without driving history, with a typed not-driven state, no invented times and working search, sorting and detail links. Local map descriptors and installed personal pacenotes provide declarative search filters using native stage identity resolution.
• Add an English/German static website with a six-step authored screenshot tour, platform download links, unsigned-alpha guidance and local-processing privacy information. The website does not change desktop behavior.
• Runs have a reloadable page linked from the latest recorded run and stage/car history, driver attempt labels, finish state, a single penalty summary, and recorded setup, conditions and splits. Debrief uses the shared exact recording-match verdict rather than an unconditional contradictory disclaimer; no finish tolerance was added.
• Telemetry now includes a grip scatter for the visible distance window and a measured gearing sheet for one attempt. Rust clips source acceleration spikes and calculates native powered RPM per km/h, sample counts and speed at observed maximum RPM. Typed absences and German/English notes retain unverified acceleration scale and recorded gear encoding; matching setup gearbox sections link to an attempt sheet.
• Discover protected stage imagery from installed native stage files, send bounded read-only stage proofs without login, explain missing or unmatched versions, and retain cached maps offline while preserving Martin compatibility.
• Private release builds can provide a default imagery endpoint at compile time; saved settings override it, builds without it keep imagery unconfigured, and release manifests record only injection status and SHA256.
• After explicit distance-assumption consent, debrief offers up to three consistency review spots with measured spread, attempt counts and thin support for two measurements. Drivers can choose a spot, keep the existing topics or finish; unavailable evidence and the consistency-only scope remain explicit.
• Five reviewed built-in numeric mappings resolve only matching symbolic pacenote sections in the selected installation, with disclosed provenance. Conflicting installed sections remain explicit. Read-only note diagnostics linked from Settings and recce list installation and built-in resolutions and remaining undefined IDs with counts.
• Maps can display personal pacenotes from the selected RBR installation along a cached host route. Local Rust interpolates unchanged native distances with explicit unverified correspondence, preserves route/roadbook/notation hashes and opaque variant identity, and withholds unsupported positions with typed reasons. Drivers can choose an installed roadbook and reading system using the existing installed reader; a latest-modified file is only a current candidate, never a historical default. Pacenotes remain separate from host-derived corners and crests. No metre scale, fitted offset, authenticated direction, driving benefit or recce/coaching behavior change is claimed.
• Read static car footprints from the selected RSF installation with bounded persistent caching, source provenance and typed unknown geometry or pose correspondence.
• Run pages link directly to preselected telemetry with large stacked charts, bounded distance zoom and pan, shared chart/map cursor and a small route/driven map. Rust owns channel selection, typed missingness, native held gear, source units, acceleration signs and fixed percentile spike clipping; German and English explain unverified scales.
• Telemetry can overlay two recordings on one Rust-selected common distance grid, including partial attempts. A recording picker starts with the existing proposed reference; two labelled colours share chart zoom, pan and cursor. Typed coverage and stage refusals prevent extrapolation or incompatible overlays.
• The bundled German and English manual now has chapters, searchable topics and back/related-topic navigation. Contextual question marks show a short hover or keyboard-focus hint and open the matching topic in one movable, resizable in-page window. Escape closes it and restores focus; optional browser storage remembers its bounded position and size.
• Keep runs-library filters, sorting and search in the URL with Back/Forward and reload restoration. Copy the current link from the header without comparison consent; invalid filter syntax uses defaults with a localized status. Recording links can use opaque, root-independent recording references resolved locally with typed absence reasons. Moving a recording within telemetry changes its reference. Desktop links can be copied; link import is not supported.
• Unsigned desktop candidate build tooling for macOS DMG, Linux AppImage/deb and Windows NSIS, with source/tool manifests, checksums, license inventories and supplemental asset checks. Windows configuration embeds the WebView2 offline installer; real offline Windows acceptance remains required.
• Telemetry without imagery or route now draws a Rust-projected local recording line with distance marks and a shared chart cursor. Raw XY coordinates retain declared units or explicit unknown units, without claiming geographic placement; missing or invalid positions remain typed absences, alongside the stage-specific missing imagery reason.
• Generate the b4r changelog from validated per-bead fragments and check freshness plus explicit-base Git change coverage before landing.
• Read worked CLI examples generated from executed Rust Given/When/Then scenarios, with a live freshness check and a combined pre-landing documentation and changelog gate.
• Follow a real-recording CLI walkthrough with a reusable private request builder, exact input identities and observed analysis, saved-result and history outputs, including justified abstention.
• Browse from generated documentation back to BDD catalog entries, exact executable scenario functions, generators, Rust entry points and change fragments, with checked source targets and regenerated line anchors.
• Define BDD-first development and independent native Vega review with a target-owned project rubric, different-model reviewers, explicit evidence requirements and no routine review bypass.
• Add a private-input corpus audit and safe observations for ten distinct original recordings, with a fixed prospective development/held-out split, source-declared units and explicit capture/synchrony limitations. No product behavior or verified coaching claim changes.
• Inspect explicitly selected offline stage and car catalog claims in the headless workflow, preserving ambiguity and source identity across reopen without authenticating the recorded stage or car.
• Add a reusable offline reader for explicitly selected stage and car JSON catalog snapshots, retaining raw fields and source hashes, rejecting malformed or oversized inputs, and exposing duplicate, conflicting, missing and stale ID selections without authenticating a capture. Workflow integration is separate.
• Add the reusable b4r-analysis library for explicit inclusive native-window signed speed summaries and source raw brake falls/throttle transitions, preserving unit and provenance evidence, gaps, deterministic extrema/change endpoints and censored boundaries without inferring duration, physical pedal calibration or technique quality. Workflow and CLI integration are separate.
• Expose bounded optional recorded-signal question plans through the offline workflow and CLI, retaining typed source-derived reports, explicit unavailable outcomes and native-window review exercises across reopen. Public BDD covers the three source-signal questions and workbook-derived window-minimum baseline; numerical observations do not select a driving correction or establish benefit. Summaries distinguish raw baselines from relative extrema and withhold physical interpretations of negative recorded speed.
• Add the reusable relative_position question (Q-RAW-BASELINE): subtract the earliest observed-finite minimum of an explicit inclusive native window from source-scaled unverified driveline values. Compact reports retain the original baseline, relative extrema, raw adjacent changes, unknown units, partial coverage and censored boundaries; all-invalid windows return an unavailable-baseline error. No meters, cumulative distance, timing or route alignment is inferred. CLI and real-data integration remain separate.
• Exercise explicit peak-to-first-observed-minimum raw brake windows on authored controls and two original recordings, distinguishing the selected after-peak question from a full-window fall that may precede its peak. Independent raw-decoded numeric oracles retain native indices, unknown units and censored event boundaries while excluding later raw braking episodes; no physical release or elapsed-time claim is added.
• Add an independent private-input numeric audit for signed recorded speed, raw brake/throttle changes and a window-minimum raw driveline baseline, retai...

Rally Coach 0.1.0-dev.1704+ge17a392b139e — alpha

Rally Coach 0.1.0-dev.1704+ge17a392b139e — alpha

Build commit: e17a392b139eb6fde868f61c19ecb48ec649f9ff

Publication commit: e17a392b139eb6fde868f61c19ecb48ec649f9ff

Added

• Show a heading-based vehicle-width band and cursor car rectangle from installed body dimensions on telemetry maps and local recording plots, with remembered layer controls and explicit unknown position offset and missing geometry reasons.
• Add an English/German static website with a six-step authored screenshot tour, platform download links, unsigned-alpha guidance and local-processing privacy information. The website does not change desktop behavior.
• Runs have a reloadable page linked from the latest recorded run and stage/car history, driver attempt labels, finish state, a single penalty summary, and recorded setup, conditions and splits. Debrief uses the shared exact recording-match verdict rather than an unconditional contradictory disclaimer; no finish tolerance was added.
• Telemetry now includes a grip scatter for the visible distance window and a measured gearing sheet for one attempt. Rust clips source acceleration spikes and calculates native powered RPM per km/h, sample counts and speed at observed maximum RPM. Typed absences and German/English notes retain unverified acceleration scale and recorded gear encoding; matching setup gearbox sections link to an attempt sheet.
• Discover protected stage imagery from installed native stage files, send bounded read-only stage proofs without login, explain missing or unmatched versions, and retain cached maps offline while preserving Martin compatibility.
• Private release builds can provide a default imagery endpoint at compile time; saved settings override it, builds without it keep imagery unconfigured, and release manifests record only injection status and SHA256.
• After explicit distance-assumption consent, debrief offers up to three consistency review spots with measured spread, attempt counts and thin support for two measurements. Drivers can choose a spot, keep the existing topics or finish; unavailable evidence and the consistency-only scope remain explicit.
• Five reviewed built-in numeric mappings resolve only matching symbolic pacenote sections in the selected installation, with disclosed provenance. Conflicting installed sections remain explicit. Read-only note diagnostics linked from Settings and recce list installation and built-in resolutions and remaining undefined IDs with counts.
• Maps can display personal pacenotes from the selected RBR installation along a cached host route. Local Rust interpolates unchanged native distances with explicit unverified correspondence, preserves route/roadbook/notation hashes and opaque variant identity, and withholds unsupported positions with typed reasons. Drivers can choose an installed roadbook and reading system using the existing installed reader; a latest-modified file is only a current candidate, never a historical default. Pacenotes remain separate from host-derived corners and crests. No metre scale, fitted offset, authenticated direction, driving benefit or recce/coaching behavior change is claimed.
• Read static car footprints from the selected RSF installation with bounded persistent caching, source provenance and typed unknown geometry or pose correspondence.
• Run pages link directly to preselected telemetry with large stacked charts, bounded distance zoom and pan, shared chart/map cursor and a small route/driven map. Rust owns channel selection, typed missingness, native held gear, source units, acceleration signs and fixed percentile spike clipping; German and English explain unverified scales.
• Telemetry can overlay two recordings on one Rust-selected common distance grid, including partial attempts. A recording picker starts with the existing proposed reference; two labelled colours share chart zoom, pan and cursor. Typed coverage and stage refusals prevent extrapolation or incompatible overlays.
• The bundled German and English manual now has chapters, searchable topics and back/related-topic navigation. Contextual question marks show a short hover or keyboard-focus hint and open the matching topic in one movable, resizable in-page window. Escape closes it and restores focus; optional browser storage remembers its bounded position and size.
• Keep runs-library filters, sorting and search in the URL with Back/Forward and reload restoration. Copy the current link from the header without comparison consent; invalid filter syntax uses defaults with a localized status. Recording links can use opaque, root-independent recording references resolved locally with typed absence reasons. Moving a recording within telemetry changes its reference. Desktop links can be copied; link import is not supported.
• Unsigned desktop candidate build tooling for macOS DMG, Linux AppImage/deb and Windows NSIS, with source/tool manifests, checksums, license inventories and supplemental asset checks. Windows configuration embeds the WebView2 offline installer; real offline Windows acceptance remains required.
• Telemetry without imagery or route now draws a Rust-projected local recording line with distance marks and a shared chart cursor. Raw XY coordinates retain declared units or explicit unknown units, without claiming geographic placement; missing or invalid positions remain typed absences, alongside the stage-specific missing imagery reason.
• Generate the b4r changelog from validated per-bead fragments and check freshness plus explicit-base Git change coverage before landing.
• Read worked CLI examples generated from executed Rust Given/When/Then scenarios, with a live freshness check and a combined pre-landing documentation and changelog gate.
• Follow a real-recording CLI walkthrough with a reusable private request builder, exact input identities and observed analysis, saved-result and history outputs, including justified abstention.
• Browse from generated documentation back to BDD catalog entries, exact executable scenario functions, generators, Rust entry points and change fragments, with checked source targets and regenerated line anchors.
• Define BDD-first development and independent native Vega review with a target-owned project rubric, different-model reviewers, explicit evidence requirements and no routine review bypass.
• Add a private-input corpus audit and safe observations for ten distinct original recordings, with a fixed prospective development/held-out split, source-declared units and explicit capture/synchrony limitations. No product behavior or verified coaching claim changes.
• Inspect explicitly selected offline stage and car catalog claims in the headless workflow, preserving ambiguity and source identity across reopen without authenticating the recorded stage or car.
• Add a reusable offline reader for explicitly selected stage and car JSON catalog snapshots, retaining raw fields and source hashes, rejecting malformed or oversized inputs, and exposing duplicate, conflicting, missing and stale ID selections without authenticating a capture. Workflow integration is separate.
• Add the reusable b4r-analysis library for explicit inclusive native-window signed speed summaries and source raw brake falls/throttle transitions, preserving unit and provenance evidence, gaps, deterministic extrema/change endpoints and censored boundaries without inferring duration, physical pedal calibration or technique quality. Workflow and CLI integration are separate.
• Expose bounded optional recorded-signal question plans through the offline workflow and CLI, retaining typed source-derived reports, explicit unavailable outcomes and native-window review exercises across reopen. Public BDD covers the three source-signal questions and workbook-derived window-minimum baseline; numerical observations do not select a driving correction or establish benefit. Summaries distinguish raw baselines from relative extrema and withhold physical interpretations of negative recorded speed.
• Add the reusable relative_position question (Q-RAW-BASELINE): subtract the earliest observed-finite minimum of an explicit inclusive native window from source-scaled unverified driveline values. Compact reports retain the original baseline, relative extrema, raw adjacent changes, unknown units, partial coverage and censored boundaries; all-invalid windows return an unavailable-baseline error. No meters, cumulative distance, timing or route alignment is inferred. CLI and real-data integration remain separate.
• Exercise explicit peak-to-first-observed-minimum raw brake windows on authored controls and two original recordings, distinguishing the selected after-peak question from a full-window fall that may precede its peak. Independent raw-decoded numeric oracles retain native indices, unknown units and censored event boundaries while excluding later raw braking episodes; no physical release or elapsed-time claim is added.
• Add an independent private-input numeric audit for signed recorded speed, raw brake/throttle changes and a window-minimum raw driveline baseline, retaining native boundaries and unknown physical semantics. Diagnostic evidence only; no product behavior, coaching benefit or pristine held-out validation claim.
• Add an actual-side diagnostic comparison of recorded-question library and CLI results against frozen independent original-file oracles. No product behavior or driving-benefit claim changes.
• Add Q-RACE-CLOCK-STEP for source-evidenced RaceTime with exact raw unit s. Reuse native adjacent analysis to retain separate largest positive counter steps and negative resets, raw endpoints, gaps, ties and censored bounds. Refuse unsupported units and quantities; no elapsed time, stage location, replay synchronization, penalty cause or exact recoverable gain is inferred. Workflow review requests manual replay or input-display confirmation.
• Add a private real-input regression evaluator over verified acquisition identities, reusi...