bd sync: 2025-12-23 04:35:03

.beads/issues.jsonl

@@ -308,6 +308,7 @@
 {"id":"bd-n386","title":"Improve test coverage for internal/daemon (27.3% → 60%)","description":"The daemon package has only 27.3% test coverage. The daemon is critical for background operations and reliability.\n\nKey areas needing tests:\n- Daemon autostart logic\n- Socket handling\n- PID file management\n- Health checks\n\nCurrent coverage: 27.3%\nTarget coverage: 60%","status":"open","priority":2,"issue_type":"task","created_at":"2025-12-13T20:43:00.895238-08:00","updated_at":"2025-12-13T21:01:17.274438-08:00"}
 {"id":"bd-n3v","title":"Error committing to sync branch: failed to create worktree","description":"\u003e bd sync --no-daemon\n→ Exporting pending changes to JSONL...\n→ Committing changes to sync branch 'beads-sync'...\nError committing to sync branch: failed to create worktree: failed to create worktree parent directory: mkdir /var/home/matt/dev/beads/fix-ci/.git: not a directory","notes":"**Problem Diagnosed**: The `bd sync` command was failing with \"mkdir /var/home/matt/dev/beads/fix-ci/.git: not a directory\" because it was being executed from the wrong directory.\n\n**Root Cause**: The command was run from `/var/home/matt/dev/beads` (where the `fix-ci` worktree exists) instead of the main repository directory `/var/home/matt/dev/beads/main`. Since `fix-ci` is a git worktree with a `.git` file (not directory), the worktree creation logic failed when trying to create `\u003ccurrent_dir\u003e/.git/beads-worktrees/\u003cbranch\u003e`.\n\n**Solution Verified**: Execute `bd sync` from the main repository directory:\n```bash\ncd main \u0026\u0026 bd sync --dry-run\n```\n","status":"open","priority":2,"issue_type":"task","created_at":"2025-12-05T15:25:24.514998248-07:00","updated_at":"2025-12-05T15:42:32.910166956-07:00"}
 {"id":"bd-n4td","title":"Add warning when staleness check errors","description":"## Problem\n\nWhen ensureDatabaseFresh() calls CheckStaleness() and it errors (corrupted metadata, permission issues, etc.), we silently proceed with potentially stale data.\n\n**Location:** cmd/bd/staleness.go:27-32\n\n**Scenarios:**\n- Corrupted metadata table\n- Database locked by another process  \n- Permission issues reading JSONL file\n- Invalid last_import_time format in DB\n\n## Current Code\n\n```go\nisStale, err := autoimport.CheckStaleness(ctx, store, dbPath)\nif err \\!= nil {\n    // If we can't determine staleness, allow operation to proceed\n    // (better to show potentially stale data than block user)\n    return nil\n}\n```\n\n## Fix\n\n```go\nisStale, err := autoimport.CheckStaleness(ctx, store, dbPath)\nif err \\!= nil {\n    fmt.Fprintf(os.Stderr, \"Warning: Could not verify database freshness: %v\\n\", err)\n    fmt.Fprintf(os.Stderr, \"Proceeding anyway. Data may be stale.\\n\\n\")\n    return nil\n}\n```\n\n## Impact\nMedium - users should know when staleness check fails\n\n## Effort\nEasy - 5 minutes","status":"closed","priority":2,"issue_type":"bug","created_at":"2025-11-20T20:16:34.889997-05:00","updated_at":"2025-12-17T23:13:40.531031-08:00","closed_at":"2025-12-17T19:11:12.950618-08:00","dependencies":[{"issue_id":"bd-n4td","depends_on_id":"bd-2q6d","type":"blocks","created_at":"2025-11-20T20:18:20.154723-05:00","created_by":"stevey","metadata":"{}"}]}
+{"id":"bd-n6fm","title":"witness Handoff","description":"attached_molecule: bd-ndye\nattached_at: 2025-12-23T12:35:02Z","status":"pinned","priority":2,"issue_type":"task","created_at":"2025-12-23T04:35:02.675024-08:00","updated_at":"2025-12-23T04:35:02.99197-08:00"}
 {"id":"bd-n777","title":"Timer beads for scheduled agent callbacks","description":"## Problem\n\nAgents frequently need to wait for external events (CI completion, PR reviews, artifact builds) but have no good mechanism:\n- `sleep N` blocks and is unreliable (often times out at 8+ minutes)\n- Polling wastes context and is easy to forget\n- No way to survive session restarts\n\n## Proposal: Timer Beads\n\nA new bead type or field that represents a scheduled callback:\n\n### Creating timers\n```bash\nbd timer create --in 30s --callback \"Check CI run 12345\" --issue bd-xyz\nbd timer create --at \"2025-12-20T08:00:00\" --callback \"Morning standup\"\nbd timer create --in 5m --on-expire \"tmux send-keys -t dave 'bd show bd-xyz'\"\n```\n\n### Timer storage\n- Store in beads (survives restarts)\n- Fields: `expires_at`, `callback_description`, `on_expire_command`, `linked_issue`\n- Status: pending, fired, cancelled\n\n### Deacon integration\nThe Deacon daemon monitors timer beads:\n1. Wakes on next timer expiry\n2. Executes `on_expire` command (e.g., tmux send-keys to interrupt agent)\n3. Marks timer as fired\n4. Optionally updates linked issue\n\n### Use cases\n- CI monitoring: \"ping me when build completes\"\n- PR reviews: \"check back in 1 hour\"\n- Scheduled tasks: \"remind me at EOD to sync\"\n- Blocking waits: agent registers callback instead of sleeping\n\n## Acceptance criteria\n- [ ] Timer bead type or field design\n- [ ] `bd timer create/list/cancel` commands\n- [ ] Deacon timer monitoring loop\n- [ ] tmux integration for agent interrupts\n- [ ] Survives daemon restarts","status":"closed","priority":2,"issue_type":"feature","created_at":"2025-12-19T23:05:33.051861-08:00","updated_at":"2025-12-21T17:19:48.087482-08:00","closed_at":"2025-12-21T17:19:48.087482-08:00","close_reason":"Will not implement - Gas Town uses a different approach for timed events"}
 {"id":"bd-ncwo","title":"Ghost resurrection: remote status:closed wins during git merge","description":"During bd sync, the 3-way git merge sometimes keeps remote's status:closed instead of local's status:tombstone. This causes ghost issues to resurrect even when tombstones exist.\n\nRoot cause: Git 3-way merge doesn't understand tombstone semantics. If base had closed, local changed to tombstone, and remote has closed, git might keep remote's version.\n\nThe early tombstone check in importer.go only prevents CREATION when tombstones exist in DB. But if applyDeletionsFromMerge hard-deletes the tombstones before import runs (because they're not in the merged result), the check doesn't help.\n\nPotential fixes:\n1. Make tombstones 'win' in the beads merge driver (internal/merge/merge.go)\n2. Don't hard-delete tombstones in applyDeletionsFromMerge if they're in the DB\n3. Export tombstones to a separate file that's not subject to merge\n\nGhost issues: bd-cb64c226.*, bd-cbed9619.*","status":"tombstone","priority":1,"issue_type":"bug","created_at":"2025-12-16T22:01:03.56423-08:00","updated_at":"2025-12-17T16:11:17.070763-08:00","deleted_at":"2025-12-17T16:11:17.070763-08:00","deleted_by":"batch delete","delete_reason":"batch delete","original_type":"bug"}
 {"id":"bd-ndye","title":"mergeDependencies uses union instead of 3-way merge","description":"## Critical Bug\n\nThe `mergeDependencies` function in internal/merge/merge.go performs a UNION of left and right dependencies instead of a proper 3-way merge. This causes removed dependencies to be resurrected.\n\n### Root Cause\n\n```go\n// Current code (lines 795-816):\nfunc mergeDependencies(left, right []Dependency) []Dependency {\n    // Just unions left + right\n    // NEVER REMOVES anything\n    // Doesn't even look at base!\n}\n```\n\nAnd `mergeIssue` (line 579) doesn't pass `base`:\n```go\nresult.Dependencies = mergeDependencies(left.Dependencies, right.Dependencies)\n```\n\n### Impact\n\nIf:\n- Base has dependency D\n- Left removes D (intentional)\n- Right still has D (stale)\n\nCurrent: D is in result (resurrection!)\nCorrect: Left removed it, D should NOT be in result\n\nThis breaks Gas Town's workflow and data integrity. Closed means closed.\n\n### Fix\n\nChange `mergeDependencies` to take `base` and do proper 3-way merge:\n- If dep was in base and removed by left → exclude (left wins)\n- If dep was in base and removed by right → exclude (right wins)\n- If dep wasn't in base and added by either → include\n- If dep was in base and both still have it → include\n\nKey principle: **REMOVALS ARE AUTHORITATIVE**\n\n### Files to Change\n\n1. internal/merge/merge.go:\n   - `mergeDependencies(left, right)` → `mergeDependencies(base, left, right)`\n   - `mergeIssue` line 579: pass `base.Dependencies`\n\n### Related\n\nThis also explains why `ProtectLocalExportIDs` in importer is defined but never used - the protection was never actually implemented.","status":"closed","priority":0,"issue_type":"bug","created_at":"2025-12-18T23:15:54.475872-08:00","updated_at":"2025-12-18T23:21:10.709571-08:00","closed_at":"2025-12-18T23:21:10.709571-08:00"}