Fix #170: cycling/rowing workouts weren't being suggested

[abdulsaheel]

User description

Fixes #170.

Root cause

autoDetectWorkouts's motion-confirmation gate (motionConfirmMean = 0.15, mean L2 gravity-delta/s) is tuned for arm-swing activities like walking/running. It unconditionally drops any sustained-elevated-HR window where the wrist stays still — exactly what happens gripping a handlebar/oar during cycling/rowing — regardless of how strong the HR signal is. That's why a real 1h50m road cycling session (avg ~117W, HR 100-140, peak 157 bpm) produced no suggestion at all.

Fix

• Depends on autoDetectWorkouts: bypass motion gate on genuine HR onset (edge#170) analytics#32 (pinned above at its PR-branch head — repin to main once that merges): AutoWorkoutDetector now bypasses the motion gate on a genuine HR onset — mean bpm over the first 3 min of the candidate span must rise ≥25 bpm vs. the 3 min immediately before it (Whipp & Wasserman 1972 phase-II kinetics: real exertion shows a fast HR rise right at onset).
• A raw %HRR-level bypass was tried and rejected first: it reopens the exact false positive the motion gate exists to catch, since level alone can't distinguish "just started exercising" from "already elevated" (fever, anxiety, heat). The onset check does distinguish them — a slow drift shows no sharp rise and stays gated.
• kAlgoVersion 50 → 51 (this changes which suggestions autoDetectWorkouts emits).

Test plan

• dart test in analytics — 386 passed, no failures (see analytics#32 for the new onset-gate tests).
• flutter test --concurrency=1 — 1055 passed, no regressions.
• flutter analyze — clean.
• Verified the new pin resolves (pubspec.lock resolved-ref matches) with pubspec_overrides.yaml moved aside, per the repo's CI-reproduction convention.

---

PR Type

Bug fix

---

Description

• Fixes cycling/rowing workouts never being auto-detected (issue Auto-detect misses 1h50m road cycling session despite peak HR 157 bpm #170)

• Repins openstrap_analytics to PR-branch head of analytics#32 (HR-onset bypass)

• Bumps kAlgoVersion 50 → 51 (changes autoDetectWorkouts suggestions output)

---

Diagram Walkthrough

flowchart LR
  A["Motion gate\n(arm-swing tuned)"]
  B["Cycling/rowing window\n(wrist still on handlebar/oar)"]
  C["HR-onset check\n(≥25 bpm rise in first 3 min)"]
  D["Workout suggestion\nemitted"]
  E["Slow-drift elevation\n(fever/heat/anxiety)"]
  F["Suggestion suppressed\n(no onset detected)"]
  B -- "previously dropped" --> A
  B -- "now: onset check" --> C
  C -- "fast rise detected" --> D
  E -- "no sharp rise" --> F

Loading

File Walkthrough

	Relevant files
Bug fix	derivation_engine.dart Bump kAlgoVersion to 51 for workout detection fix                lib/compute/derivation_engine.dart Bumps kAlgoVersion from 50 to 51 Adds detailed changelog comment explaining the HR-onset bypass fix for cycling/rowing detection Documents that the analytics#32 pin is a PR-branch head pending merge to main +13/-1
Dependencies	pubspec.yaml Repin analytics to HR-onset bypass PR-branch head                pubspec.yaml Repins openstrap_analytics git ref from 5d421918... (analytics#31 tip) to 422c8b8b... (analytics#32 PR-branch head) Updates comment to note this is a PR-branch head, not main, and must be repinned to merge commit once analytics#32 lands Documents verification method for the new pin (onsetRiseBpm presence in auto_detect.dart) +10/-7

---

Summary by CodeRabbit

• Bug Fixes
  • Improved workout auto-detection for activities with limited lower-body movement.
  • Added a heart-rate rise threshold to distinguish genuine exercise from gradual, non-exercise elevations.
• Improvements
  • Updated analytics behavior to improve workout detection accuracy and verification.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 61d8deca-ee4b-410a-a74d-c21f72ae1973

📥 Commits

Reviewing files that changed from the base of the PR and between 4976e60 and 52b8e9b.

⛔ Files ignored due to path filters (1)

• pubspec.lock is excluded by !**/*.lock

📒 Files selected for processing (1)

• pubspec.yaml

---

📝 Walkthrough

Walkthrough

The PR increments the derivation algorithm version to 51 and pins openstrap_analytics to a PR branch that implements the HR-onset workout auto-detection behavior.

Changes

Workout detection analytics

Layer / File(s)	Summary
Align algorithm version and analytics revision lib/compute/derivation_engine.dart, pubspec.yaml	The algorithm version changes from 50 to 51. The analytics dependency changes to the PR branch that documents the three-minute, 25 BPM heart-rate rise condition for bypassing motion confirmation.

Estimated code review effort: 2 (Simple) | ~10 minutes

Suggested reviewers: dannymcc

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly identifies the fix for missing cycling and rowing workout suggestions, which matches the pull request objective.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

PR Reviewer Guide 🔍

(Review updated until commit 52b8e9b)

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 2 🔵🔵⚪⚪⚪

🧪 No relevant tests

🔒 No security concerns identified

⚡ Recommended focus areas for review PR-branch pin The analytics dependency is pinned to a PR-branch head (cbbe06a) rather than a merged main commit. AGENTS.md §3.5/§4.9 explicitly forbids this: "A bump citing a sibling change must be backed by the pin" and siblings must be "pinned to full commit SHAs, never branch refs." The comment itself acknowledges this ("PR-BRANCH HEAD, not main — repin once it lands"). If analytics#32 is rebased, force-pushed, or the branch is deleted after merge, flutter pub get will fail or silently resolve a different tree. More critically, kAlgoVersion 51 is now live in production builds pointing at an unmerged, mutable ref — if the analytics PR is amended before merge, the version bump and the actual behavior will diverge, exactly the scenario that kept v43's bug live for three releases. This should not ship until the analytics PR is merged and the pin is updated to the merge commit on main. # PR-BRANCH HEAD, not main — OpenStrap/analytics#32 (fix/issue-170-...) # is open, not yet merged. Repin to the merge commit on main once it # lands (same pattern v49 used briefly for protocol/analytics PR heads). # autoDetectWorkouts now bypasses the motion-confirmation gate on a # genuine HR-onset (fast rise vs. immediately-preceding baseline), # fixing low-limb-swing cardio (cycling/rowing) going undetected while # still rejecting slow-drifting elevations (fever/heat/anxiety) that # have no discernible onset. # Verified present at THIS sha: `git show <sha>:lib/src/onehz/workout/auto_detect.dart | grep onsetRiseBpm`. # Moved to the branch's new head (cbbe06a) after CodeRabbit caught a real # off-by-one in the onset window on analytics#32 (earlyMean was 181s vs # preMean's 180s) — fixed there, not worth a separate edge changelog line. ref: cbbe06addec1cb78b4ea2c75f64e8a281ac09294 Version bump without test kAlgoVersion is bumped from 50 to 51, which changes which workout suggestions autoDetectWorkouts emits. AGENTS.md §5 notes that behavior changes with no accompanying test in test/ are a real finding because no CI gate runs on PRs. The analytics-side tests live in the analytics repo and are not part of this repo's flutter test suite. There is no new test in test/ pinning the edge-side behavior change (e.g., that a cycling-shaped HR trace now produces a suggestion, or that a fever-shaped slow-drift trace still does not). Following the repo convention of naming tests for the bug they pin, a cycling_workout_detection_test.dart is missing. // v51: edge#170 — autoDetectWorkouts' motion-confirmation gate (tuned for // arm-swing activities) silently dropped every low-limb-swing cardio window // (cycling/rowing: the wrist stays still on a handlebar/oar) no matter how // strong the HR signal was. analytics#32 (PR-branch head, pinned above — // repin to main once merged) adds an HR-ONSET bypass: the gate is skipped // only when mean bpm over the candidate's first 3 min rises >=25 bpm versus // the 3 min immediately before it (Whipp & Wasserman 1972 phase-II kinetics), // which fires on genuine exercise starts but NOT on slow-drifting elevations // (fever/heat/anxiety) that have no discernible onset — so this changes which // suggestions autoDetectWorkouts emits without loosening the false-positive // gate it exists to protect. const int kAlgoVersion = 51;

[github-actions]

PR Code Suggestions ✨

Latest suggestions up to 52b8e9b
Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Pin sibling to merged main SHA, not branch head The comment explicitly states this is a PR-branch head ("not yet merged"), which violates the hard invariant §3.6 that siblings must be pinned to full commit SHAs on main, never branch refs or unmerged PR heads. Shipping an unmerged PR-branch SHA means the pin can be silently invalidated if the branch is force-pushed, and the changelog entry in kAlgoVersion v51 describes a fix that is not yet in main. This is the exact pattern that caused v43's bug to stay live for three releases. The PR should not be merged until analytics#32 lands on main and the pin is updated to that merge commit SHA. pubspec.yaml [64] -ref: cbbe06addec1cb78b4ea2c75f64e8a281ac09294 +# Repin to the merge commit SHA on analytics main once analytics#32 lands. +# Do NOT merge this edge PR until that SHA is available and verified: +# `git show <merge-sha>:lib/src/onehz/workout/auto_detect.dart | grep onsetRiseBpm` +ref: <analytics#32-merge-commit-sha-on-main> Suggestion importance[1-10]: 3 __ Why: The comment in the PR explicitly acknowledges this is a temporary PR-branch pin and instructs to repin once merged — this is intentional and documented. The suggestion's improved_code replaces the real SHA with a placeholder <analytics#32-merge-commit-sha-on-main>, which would break the build and is not a valid improvement over the current state.	Low

---

Previous suggestions

Suggestions up to commit 4976e60

Category	Suggestion	Impact
Possible issue	Pin sibling to merge commit, not branch head AGENTS.md §3 invariant 5 states "a bump citing a sibling change must be backed by the pin" and §4.9 documents that v43 shipped a changelog describing a fix whose pin never contained it. The kAlgoVersion bump to 51 cites analytics#32, but the pinned SHA 422c8b8b9631770b07bc571b6fbe904ef59e5052 is explicitly described as a PR-branch head that is "not yet merged" — if this SHA is force-pushed or the branch is rebased before the PR lands, the pin silently diverges from what was reviewed. This must be repinned to the actual merge commit SHA on main before this PR is merged, not after, to satisfy the invariant. pubspec.yaml [52-61] -# PR-BRANCH HEAD, not main — OpenStrap/analytics#32 (fix/issue-170-...) -# is open, not yet merged. Repin to the merge commit on main once it -# lands (same pattern v49 used briefly for protocol/analytics PR heads). -... -ref: 422c8b8b9631770b07bc571b6fbe904ef59e5052 +# Repin to the merge commit SHA on analytics main once OpenStrap/analytics#32 +# has landed. Do NOT merge this PR with a PR-branch head pinned here. +ref: <merge-commit-sha-of-analytics#32-on-main> Suggestion importance[1-10]: 5 __ Why: The suggestion correctly identifies that the ref is pinned to a PR-branch head rather than a merge commit, which could be problematic if the branch is force-pushed. However, the PR itself explicitly acknowledges this as a temporary state with instructions to repin, making this a known and documented trade-off rather than an oversight. The improved_code replaces the SHA with a placeholder, which is not actionable.	Low
	Version bump must coincide with verified merge-commit pin AGENTS.md §3 invariant 5 requires that a version bump citing a sibling change be backed by the actual pin, verified at the merge commit. The changelog here explicitly acknowledges the pin is a PR-branch head ("repin to main once merged"), meaning kAlgoVersion = 51 is being committed while the invariant is knowingly violated. Existing day_result rows at v50 will not recompute once the correct merge-commit SHA is repinned, because the version number was already bumped here — the repin will be a no-op for recomputation. The bump should only be committed together with the final merge-commit pin, not ahead of it. lib/compute/derivation_engine.dart [354-365] // v51: edge#170 — autoDetectWorkouts' motion-confirmation gate (tuned for // arm-swing activities) silently dropped every low-limb-swing cardio window // (cycling/rowing: the wrist stays still on a handlebar/oar) no matter how -// strong the HR signal was. analytics#32 (PR-branch head, pinned above — -// repin to main once merged) adds an HR-ONSET bypass: the gate is skipped -// only when mean bpm over the candidate's first 3 min rises >=25 bpm versus -// the 3 min immediately before it (Whipp & Wasserman 1972 phase-II kinetics), -// which fires on genuine exercise starts but NOT on slow-drifting elevations -// (fever/heat/anxiety) that have no discernible onset — so this changes which -// suggestions autoDetectWorkouts emits without loosening the false-positive -// gate it exists to protect. +// strong the HR signal was. analytics#32 adds an HR-ONSET bypass: the gate +// is skipped only when mean bpm over the candidate's first 3 min rises +// >=25 bpm versus the 3 min immediately before it (Whipp & Wasserman 1972 +// phase-II kinetics), which fires on genuine exercise starts but NOT on +// slow-drifting elevations (fever/heat/anxiety) that have no discernible +// onset — so this changes which suggestions autoDetectWorkouts emits without +// loosening the false-positive gate it exists to protect. +// Pinned at analytics merge commit: <merge-commit-sha-of-analytics#32-on-main> const int kAlgoVersion = 51; Suggestion importance[1-10]: 4 __ Why: The suggestion raises a valid concern that kAlgoVersion = 51 is bumped while the analytics pin is still a PR-branch head, but the improved_code only modifies comments and doesn't actually change the version number or resolve the core issue. The PR explicitly documents this as intentional with a repin planned, and the improved_code is nearly identical to existing_code with only minor comment rewording.	Low

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@pubspec.yaml`:
- Around line 60-61: Update the openstrap_analytics dependency pin in
pubspec.yaml to a full commit SHA containing the upstream three-minute
onset-window correction, replacing the current
422c8b8b9631770b07bc571b6fbe904ef59e5052 reference. Keep the dependency source
and surrounding configuration unchanged.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: 5a4e4c62-b553-477d-b6e3-e17203cc2ace

📥 Commits

Reviewing files that changed from the base of the PR and between 4cdb065 and 4976e60.

⛔ Files ignored due to path filters (1)

• pubspec.lock is excluded by !**/*.lock

📒 Files selected for processing (2)

• lib/compute/derivation_engine.dart
• pubspec.yaml

[github-actions]

Persistent review updated to latest commit 52b8e9b