feat(loop): periodic health sweep — self-heal stale drives + orphaned worktrees#11
Merged
Conversation
… worktrees A long-running loop accumulates two kinds of cruft that nothing reclaimed: features stuck in_progress after a drive died without finishing (holding a board slot), and feat-<id> worktrees left behind by a missed reap. A periodic sweep (health_sweep_interval_s, default 300s; 0 disables) now self-heals both: - in_progress features with no live drive (not in self._inflight_files) → reconciled the same way boot recovery does (adopt an open PR → in_review, else reset → ready). Boot recovery and the sweep now share _reconcile_orphan. - feat-<id> worktrees whose feature is gone or already done → reaped (in_review keeps its worktree for a CI-fail re-dispatch; a live drive's worktree is left alone). worktree.list_feature_worktrees enumerates the feat-<id> dirs under worktrees_root. Second of the P1 concurrency-correctness ports. Tests: +6 (110 total) — list_feature_worktrees (dirs only, absent dir); _sweep reconciles a driveless in_progress (skips a driven one) and reaps done/missing worktrees (keeps in_review); _maybe_sweep rate-limit + disabled. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
👀 Quinn is reviewing — verdict (PASS / WARN / FAIL) + findings to follow. |
mabry1985
added a commit
that referenced
this pull request
Jun 13, 2026
…w) (#12) By default a dependent waits for every blocks-dependency to MERGE (done) — `br ready` excludes a feature with any open blocker, so this falls out for free and stays the safe default (dep_gate: merge). The new opt-in dep_gate: review relaxes it: a NON-foundation blocker releases its dependents once it reaches in_review (build on code that's in review, not yet merged — more parallelism). Foundation features (foundation=True) ALWAYS gate dependents on merge. - store.ready_queue(relaxed=False): the relaxed branch adds dep-blocked ready features whose every still-open blocker is a non-foundation feature at in_review, resolved in Python from `br list` labels (board_state + foundation) + `br show` deps (`_open_blockers`), since `br ready` only knows "blocker closed". - create_feature(foundation=…) + the `foundation` label + _project exposure; the board_create_feature tool + the decompose skill gain the flag (the skill already marked milestones foundation in frontmatter — now it's wired to the gate). - loop: dep_gate config → relaxed_gate → ready_queue(relaxed=…). Verified against br 0.1.23: `br list --json` carries labels; `br show` deps carry status; `br ready` waits for blockers to close. Default behaviour is unchanged. Completes the P1 concurrency-correctness batch (hot-file guard #10, health sweep #11, this). Tests: +6 (116 total) — _project foundation, create_feature labels foundation, _open_blockers filtering, ready_queue(relaxed) releases only non-foundation in_review blockers, dep_gate config, _spawn_ready passes the gate through. Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
The second P1 concurrency-correctness port. A long-running loop accumulates two kinds of cruft nothing reclaimed: features stuck
in_progressafter a drive died without finishing (holding a board slot forever), andfeat-<id>worktrees left behind by a missed reap. A periodic sweep (health_sweep_interval_s, default 300s;0disables) self-heals both:in_progress(not inself._inflight_files→ no live drive) → reconciled the same way boot recovery does (adopt an open PR →in_review, else reset →ready). Boot recovery and the sweep now share_reconcile_orphan.feat-<id>dir whose feature is gone or alreadydone→ reaped. (in_reviewkeeps its worktree for a CI-fail re-dispatch; a live drive's worktree is left alone.)worktree.list_feature_worktreesenumerates thefeat-<id>dirs underworktrees_root.Tests
+6 (110 total), all green:
list_feature_worktrees(dirs only / absent dir);_sweepreconciles a drivelessin_progress(and skips a driven one) and reapsdone/missing worktrees (keepsin_review);_maybe_sweeprate-limit + disabled.ruffclean.P1 batch
🤖 Generated with Claude Code