Skip to content

fix(editor): suggest march backfills bridged-bar confidence once the far side confirms (c5 follow-up) - #272

Merged
byrongamatos merged 3 commits into
mainfrom
fix/editor-suggest-backfill
Jul 15, 2026
Merged

fix(editor): suggest march backfills bridged-bar confidence once the far side confirms (c5 follow-up)#272
byrongamatos merged 3 commits into
mainfrom
fix/editor-suggest-backfill

Conversation

@ChrisBeWithYou

@ChrisBeWithYou ChrisBeWithYou commented Jul 14, 2026

Copy link
Copy Markdown
Contributor

What

The explicitly-noted #231 scope gap ("c5's retroactive confidence backfill is a noted follow-up"): when the assisted fit marches across a sustained bar on bare prediction, it assigns the floor confidence — and left it there even after the next bar snapped cleanly, proving the marched path correct. The ghost over a held chord read as "guessing" when the evidence said "probably right".

  • Misses bridged between two corroborated hits (or a locked pin) are retroactively lifted to 75% of the weaker flank's intrinsic strength.
  • The subtle part: the closing hit's published confidence is already continuity-penalized for the very misses being backfilled — using it would double-count the gap (the first implementation did, and the fixture caught it as a no-op). The backfill therefore uses the hit's intrinsic strength (comb × consistency, continuity excluded).
  • Invariants kept: a backfilled bar never outranks a real corroborated hit; trailing misses that never re-corroborate keep the floor and are still dropped by the trailing-miss rule, bit-exactly; locked pins corroborate at full trust; _suggestFitPure's signature and public proposal shape are unchanged.

Tests

tests/suggest_backfill.test.mjs (3, fail on main — mid-run misses keep conf === missConf there): the bridged-bar lift with the never-outranks bound, the trailing-drop unchanged, and the locked-pin corroboration. Existing tempo_suggest pins all green unchanged. Full suite 147 green, lint 0 errors, routes.py untouched.

Notes

Purely a confidence-metadata change — proposal times, stops, and accept behaviour are untouched; the visible effect is honest ghost-alpha over bridged bars (and better zone-refine confidence readings downstream).

🤖 Generated with Claude Code

https://claude.ai/code/session_01EBQCHCNA81E9tHmSDHSe2Q

Summary by CodeRabbit

  • Improvements

    • Assisted tempo mapping now retroactively increases confidence for missed bars when later corroboration confirms the sequence.
    • Ghost markers over sustained chords appear more reliably while retaining confidence limits and miss-floor behavior.
    • Human-locked bars now provide full trust to preceding backfilled bars.
  • Tests

    • Added coverage for successful backfill, unresolved trailing misses, and human-locked confirmations.

The #231 scope-note follow-up: misses bridged BETWEEN two corroborated
hits are retroactively lifted to 75% of the weaker flank's INTRINSIC
strength (comb x consistency — the closing hit's published conf is
already continuity-penalized for the very gap being backfilled, so
using it would double-count). Trailing misses keep the floor and still
drop; locked pins corroborate at full trust; backfilled bars never
outrank a real hit.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EBQCHCNA81E9tHmSDHSe2Q
@coderabbitai

coderabbitai Bot commented Jul 14, 2026

Copy link
Copy Markdown

Review Change Stack

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: d5501436-b2dd-4507-b7e3-0d4f052cce85

📥 Commits

Reviewing files that changed from the base of the PR and between c85b7de and b6c7e25.

📒 Files selected for processing (3)
  • CHANGELOG.md
  • src/tempo-suggest.js
  • tests/suggest_backfill.test.mjs

📝 Walkthrough

Walkthrough

Assisted tempo suggestion now backfills confidence for missed bars when later corroboration or a locked bar confirms the march. New tests cover successful backfill, unreconfirmed misses, and full-trust locked-bar behavior.

Changes

Assisted tempo backfill

Layer / File(s) Summary
Backfill confidence engine
src/tempo-suggest.js
Tracks pending misses and retroactively raises their confidence using later corroboration or locked bars, while preserving capped intrinsic confidence and miss behavior when confirmation does not occur.
Backfill validation and changelog
tests/suggest_backfill.test.mjs, CHANGELOG.md
Adds deterministic coverage for backfilled marches, lost fits, and locked-bar trust, and documents the updated assisted tempo mapping behavior.

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

Possibly related PRs

🚥 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 describes the main change: retroactive backfilling of bridged-bar confidence in suggest.
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.
✨ Finishing Touches
📝 Generate docstrings
  • Create stacked PR
  • Commit on current branch
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Commit unit tests in branch fix/editor-suggest-backfill

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

ChrisBeWithYou and others added 2 commits July 14, 2026 16:31
@byrongamatos
byrongamatos merged commit 81bb3a2 into main Jul 15, 2026
3 checks passed
@byrongamatos
byrongamatos deleted the fix/editor-suggest-backfill branch July 15, 2026 20:24
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants