Skip to content

backlog(B-0534): file backlog-ready-notifier stale-local-checkout bug (P2) #3551

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
AceHack merged 1 commit into from
May 15, 2026
Merged

backlog(B-0534): file backlog-ready-notifier stale-local-checkout bug (P2) #3551

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions docs/BACKLOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -576,6 +576,7 @@ are closed (status: closed in frontmatter)._
- [ ] **[B-0498](backlog/P2/B-0498-substrate-evolution-algebra-rule-promotion-after-cooling-period-2026-05-14.md)** Substrate-evolution algebra (6 invariants + HKT − Rodney's Razor) rule-promotion after cooling period
- [ ] **[B-0524](backlog/P2/B-0524-manifesto-v2-verbatim-grok-fetch-2026-05-14.md)** Manifesto V2 verbatim Grok fetch — convert shadow-lock to full lock
- [ ] **[B-0529](backlog/P2/B-0529-tick-shard-schema-validator-vs-practice-drift-2026-05-15.md)** Tick-shard schema validator vs current practice drift
- [ ] **[B-0534](backlog/P2/B-0534-backlog-ready-notifier-reads-stale-local-checkout-2026-05-15.md)** backlog-ready-notifier reads stale local working-tree files → publishes work-assignment envelopes for closed rows

## P3 — convenience / deferred

Expand Down Expand Up @@ -659,5 +660,6 @@ are closed (status: closed in frontmatter)._
- [ ] **[B-0528](backlog/P3/B-0528-shadow-launchd-installer-unit-tests-2026-05-15.md)** Unit tests for tools/shadow/launchd/install-launchagent.ts
- [ ] **[B-0530](backlog/P3/B-0530-cron-sentinel-mutex-prevent-otto-cli-self-contention-2026-05-15.md)** Cron-sentinel mutex — prevent multi-Otto-CLI self-contention on .git/objects/pack
- [ ] **[B-0532](backlog/P3/B-0532-backlog-graph-consistency-lint-parent-child-status-mismatch-2026-05-15.md)** Backlog-graph consistency lint — flag parent `status: closed` while declared child is `status: open`
- [ ] **[B-0533](backlog/P3/B-0533-section33-migration-dead-xref-sweep-and-lint-2026-05-15.md)** §33 migration dead-xref sweep + static lint — live-nav pointers to docs/research/ paths that have been migrated to memory/persona/<name>/conversations/

<!-- END AUTO-GENERATED -->
65 changes: 65 additions & 0 deletions ...cklog/P2/B-0534-backlog-ready-notifier-reads-stale-local-checkout-2026-05-15.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,65 @@
---
id: B-0534
priority: P2
status: open
title: "backlog-ready-notifier reads stale local working-tree files → publishes work-assignment envelopes for closed rows"
tier: factory-infrastructure
effort: S
created: 2026-05-15
last_updated: 2026-05-15
depends_on: []
composes_with: [B-0441, B-0442, B-0532]
tags: [background-service, bus, drift-detection, multi-otto, mechanization]
type: bug
---

# backlog-ready-notifier reads stale local working-tree files

## Origin

Surfaced 2026-05-15T18:18Z. Running `bun tools/bg/backlog-ready-notifier.ts --once` from the primary worktree returned ready-candidate IDs including `B-0442` and `B-0503`, both of which were closed earlier in the same session via [PR #3518](https://github.com/Lucent-Financial-Group/Zeta/pull/3518) (`627e797` on main).

Root cause: the notifier reads YAML frontmatter from local working-tree files. The primary worktree was on peer Otto-CLI's branch `feat/persona-vera-migrate-conversations-otto-cli-2026-05-15`, which doesn't contain my B-0442/B-0503 closure commits. The local files therefore showed `status: open`, even though `origin/main` correctly shows `status: closed`.

The notifier filtered by `status === "open"` correctly. The bug is the source of truth: **local working tree under multi-Otto branch contention is not a reliable view of what's actually open on main**.

## Impact

- Peer agents subscribing to the bus topic `work-assignment` could pick up already-done work
- Wasted compute on already-merged rows
- Confusion when the picked row has no open work to do

Three already-merged rows were published as work-assignments at the observed tick (envelopes `292f1c57`, `20f991c8`, `aa5060ff`).

## Acceptance criteria

- [ ] Modify `tools/bg/backlog-ready-notifier.ts` to read backlog row state from `origin/main` (via `git show origin/main:<path>` or equivalent) instead of the local working-tree file. Fetch `origin/main` at the start of each poll iteration.
- [ ] Preserve existing behavior for local-only invocations (e.g., `--backlog-dir <path>` for testing) — only the default-no-flag path should read from `origin/main`.
- [ ] Add a CLI flag `--source local|origin` defaulting to `origin` for the prod path; `local` for testing / pre-merge previews.
- [ ] Tests cover:
- `--source origin` reads from `origin/main` and skips rows whose merged-state status is `closed`
- `--source local` reads from working tree (existing behavior)
- Both modes return the same shape (`PollResult`)
- [ ] The output's `note` field surfaces which source was used (`"read from origin/main"` vs `"read from local working tree"`)

## Composes with

- [B-0441](../P1/B-0441-backlog-ready-notifier-2026-05-13.md) (the notifier this row fixes)
Comment thread
AceHack marked this conversation as resolved.
- [B-0442](../P1/B-0442-missed-substrate-cascade-detector-background-service-2026-05-13.md) (closed by PR #3518; surfaced this bug)
- [B-0532](../P3/B-0532-backlog-graph-consistency-lint-parent-child-status-mismatch-2026-05-15.md) (related: backlog-state consistency under multi-Otto contention)
- [`.claude/rules/refresh-before-decide.md`](../../../.claude/rules/refresh-before-decide.md) (the discipline this row mechanizes for the notifier)
- [`.claude/rules/otto-channels-reference-card.md`](../../../.claude/rules/otto-channels-reference-card.md) (ID-allocation section: "the local working tree may be on a stale HEAD ... gives misleading 'what's free' answers" — same failure mode, different scope)

## Why P2

Real bug with concrete failure mode (false-positive work-assignments for closed rows). Not P1 because: the failure mode is detectable + recoverable on the receiver side (agent picks up B-0442, sees the row is `closed` in their own checkout if fresh, declines the work); the bus-envelope TTL is 2h so stale assignments expire on their own; no production-critical path depends on the notifier today (it's advisory only). Bumping to P1 if peer agents start acting on the false-positive envelopes without re-verifying.

## Pre-start checklist

- [ ] Prior-art search: `tools/bg/missed-substrate-detector.ts` uses `gh pr view --json` + `git log <headRefOid>..origin/<branch>` to compare branch vs merged state — similar pattern; possible composition
- [ ] Verify `git fetch origin` in pre-poll doesn't hit the wedge documented in `otto-channels-reference-card.md` (or if it does, fall back to `FETCH_HEAD`)
- [ ] Confirm `git show origin/main:<path>` returns stable output even under concurrent fetch by peer Otto-CLI sessions

## Substrate-honest caveat

The bug is real but the impact is currently low (advisory-only, TTL-expired envelopes). Filing as P2 captures the design without committing to immediate implementation. Future-Otto reading this row sees the failure-mode trace + the empirical evidence (envelope IDs) without needing to re-derive it.
Loading