feat: Add bd compact --purge-tombstones for dependency-aware cleanup (hq-n19iv)

Unlike --prune which removes tombstones by age, --purge-tombstones removes
tombstones that have no open issues depending on them, regardless of age.
Also cleans stale deps from closed issues to tombstones.

Usage:
  bd compact --purge-tombstones --dry-run  # Preview what would be purged
  bd compact --purge-tombstones            # Actually purge

Note: Use --no-daemon to prevent daemon from re-exporting after cleanup.
This commit is contained in:
Steve Yegge
2025-12-28 23:46:05 -08:00
parent 7f1f906c28
commit 6c42b461a4
2 changed files with 291 additions and 26 deletions

View File

@@ -25,6 +25,7 @@ var (
compactApply bool compactApply bool
compactAuto bool compactAuto bool
compactPrune bool compactPrune bool
compactPurgeTombstones bool
compactSummary string compactSummary string
compactActor string compactActor string
compactLimit int compactLimit int
@@ -49,17 +50,26 @@ Tiers:
- Tier 1: Semantic compression (30 days closed, 70% reduction) - Tier 1: Semantic compression (30 days closed, 70% reduction)
- Tier 2: Ultra compression (90 days closed, 95% reduction) - Tier 2: Ultra compression (90 days closed, 95% reduction)
Tombstone Pruning: Tombstone Cleanup:
Tombstones are soft-delete markers that prevent resurrection of deleted issues. Tombstones are soft-delete markers that prevent resurrection of deleted issues.
The --prune mode removes expired tombstones (default 30 days) from issues.jsonl
to reduce file size and sync overhead. Use --older-than to customize the TTL. --prune: Remove tombstones by AGE (default 30 days). Safe but may keep
tombstones that could be deleted.
--purge-tombstones: Remove tombstones by DEPENDENCY ANALYSIS. More aggressive -
removes any tombstone that no open issues depend on, regardless of age.
Also cleans stale deps from closed issues to tombstones.
Examples: Examples:
# Prune tombstones only (recommended for reducing sync overhead) # Age-based pruning
bd compact --prune # Remove tombstones older than 30 days bd compact --prune # Remove tombstones older than 30 days
bd compact --prune --older-than 7 # Remove tombstones older than 7 days bd compact --prune --older-than 7 # Remove tombstones older than 7 days
bd compact --prune --dry-run # Preview what would be pruned bd compact --prune --dry-run # Preview what would be pruned
# Dependency-aware purging (more aggressive)
bd compact --purge-tombstones --dry-run # Preview what would be purged
bd compact --purge-tombstones # Remove tombstones with no open deps
# Agent-driven workflow (recommended) # Agent-driven workflow (recommended)
bd compact --analyze --json # Get candidates with full content bd compact --analyze --json # Get candidates with full content
bd compact --apply --id bd-42 --summary summary.txt bd compact --apply --id bd-42 --summary summary.txt
@@ -74,8 +84,8 @@ Examples:
bd compact --stats # Show statistics bd compact --stats # Show statistics
`, `,
Run: func(_ *cobra.Command, _ []string) { Run: func(_ *cobra.Command, _ []string) {
// Compact modifies data unless --stats or --analyze or --dry-run or --prune with --dry-run // Compact modifies data unless --stats or --analyze or --dry-run
if !compactStats && !compactAnalyze && !compactDryRun && !(compactPrune && compactDryRun) { if !compactStats && !compactAnalyze && !compactDryRun {
CheckReadonly("compact") CheckReadonly("compact")
} }
ctx := rootCtx ctx := rootCtx
@@ -95,12 +105,18 @@ Examples:
return return
} }
// Handle prune mode (standalone tombstone pruning) // Handle prune mode (standalone tombstone pruning by age)
if compactPrune { if compactPrune {
runCompactPrune() runCompactPrune()
return return
} }
// Handle purge-tombstones mode (dependency-aware tombstone cleanup)
if compactPurgeTombstones {
runCompactPurgeTombstones()
return
}
// Count active modes // Count active modes
activeModes := 0 activeModes := 0
if compactAnalyze { if compactAnalyze {
@@ -115,11 +131,11 @@ Examples:
// Check for exactly one mode // Check for exactly one mode
if activeModes == 0 { if activeModes == 0 {
fmt.Fprintf(os.Stderr, "Error: must specify one mode: --prune, --analyze, --apply, or --auto\n") fmt.Fprintf(os.Stderr, "Error: must specify one mode: --prune, --purge-tombstones, --analyze, --apply, or --auto\n")
os.Exit(1) os.Exit(1)
} }
if activeModes > 1 { if activeModes > 1 {
fmt.Fprintf(os.Stderr, "Error: cannot use multiple modes together (--prune, --analyze, --apply, --auto are mutually exclusive)\n") fmt.Fprintf(os.Stderr, "Error: cannot use multiple modes together (--prune, --purge-tombstones, --analyze, --apply, --auto are mutually exclusive)\n")
os.Exit(1) os.Exit(1)
} }
@@ -765,8 +781,9 @@ func init() {
compactCmd.Flags().BoolVar(&compactAnalyze, "analyze", false, "Analyze mode: export candidates for agent review") compactCmd.Flags().BoolVar(&compactAnalyze, "analyze", false, "Analyze mode: export candidates for agent review")
compactCmd.Flags().BoolVar(&compactApply, "apply", false, "Apply mode: accept agent-provided summary") compactCmd.Flags().BoolVar(&compactApply, "apply", false, "Apply mode: accept agent-provided summary")
compactCmd.Flags().BoolVar(&compactAuto, "auto", false, "Auto mode: AI-powered compaction (legacy)") compactCmd.Flags().BoolVar(&compactAuto, "auto", false, "Auto mode: AI-powered compaction (legacy)")
compactCmd.Flags().BoolVar(&compactPrune, "prune", false, "Prune mode: remove expired tombstones from issues.jsonl") compactCmd.Flags().BoolVar(&compactPrune, "prune", false, "Prune mode: remove expired tombstones from issues.jsonl (by age)")
compactCmd.Flags().IntVar(&compactOlderThan, "older-than", 0, "Prune tombstones older than N days (default: 30)") compactCmd.Flags().IntVar(&compactOlderThan, "older-than", 0, "Prune tombstones older than N days (default: 30)")
compactCmd.Flags().BoolVar(&compactPurgeTombstones, "purge-tombstones", false, "Purge mode: remove tombstones with no open deps (by dependency analysis)")
compactCmd.Flags().StringVar(&compactSummary, "summary", "", "Path to summary file (use '-' for stdin)") compactCmd.Flags().StringVar(&compactSummary, "summary", "", "Path to summary file (use '-' for stdin)")
compactCmd.Flags().StringVar(&compactActor, "actor", "agent", "Actor name for audit trail") compactCmd.Flags().StringVar(&compactActor, "actor", "agent", "Actor name for audit trail")
compactCmd.Flags().IntVar(&compactLimit, "limit", 0, "Limit number of candidates (0 = no limit)") compactCmd.Flags().IntVar(&compactLimit, "limit", 0, "Limit number of candidates (0 = no limit)")

View File

@@ -254,3 +254,251 @@ func runCompactPrune() {
} }
} }
} }
// PurgeTombstonesResult contains results of dependency-aware tombstone purging
type PurgeTombstonesResult struct {
TombstonesBefore int // Total tombstones before purge
TombstonesDeleted int // Tombstones deleted
TombstonesKept int // Tombstones kept (have open deps)
DepsRemoved int // Stale deps from closed issues to tombstones
OrphanDepsRemoved int // Orphaned deps cleaned up
DeletedIDs []string // IDs of deleted tombstones
KeptIDs []string // IDs of kept tombstones (for debugging)
}
// purgeTombstonesByDependency removes tombstones that have no open issues depending on them.
// This is more aggressive than age-based pruning because it removes tombstones regardless of age.
// Steps:
// 1. Find all tombstones
// 2. Build dependency graph to find which tombstones have open issues depending on them
// 3. Remove deps from closed issues to tombstones (stale historical deps)
// 4. Delete tombstones that have no remaining live open deps
// 5. Clean up any orphaned deps/labels
func purgeTombstonesByDependency(dryRun bool) (*PurgeTombstonesResult, error) {
beadsDir := filepath.Dir(dbPath)
issuesPath := filepath.Join(beadsDir, "issues.jsonl")
// Check if issues.jsonl exists
if _, err := os.Stat(issuesPath); os.IsNotExist(err) {
return &PurgeTombstonesResult{}, nil
}
// Read all issues
file, err := os.Open(issuesPath)
if err != nil {
return nil, fmt.Errorf("failed to open issues.jsonl: %w", err)
}
var allIssues []*types.Issue
issueMap := make(map[string]*types.Issue)
decoder := json.NewDecoder(file)
for {
var issue types.Issue
if err := decoder.Decode(&issue); err != nil {
if err.Error() == "EOF" {
break
}
continue
}
allIssues = append(allIssues, &issue)
issueMap[issue.ID] = &issue
}
if err := file.Close(); err != nil {
return nil, fmt.Errorf("failed to close issues file: %w", err)
}
// Identify tombstones and live issues
tombstones := make(map[string]*types.Issue)
liveOpen := make(map[string]bool) // Open, non-deleted issues
liveClosed := make(map[string]bool) // Closed, non-deleted issues
for _, issue := range allIssues {
if issue.DeletedAt != nil {
tombstones[issue.ID] = issue
} else if issue.Status == "open" {
liveOpen[issue.ID] = true
} else {
liveClosed[issue.ID] = true
}
}
result := &PurgeTombstonesResult{
TombstonesBefore: len(tombstones),
}
// Build reverse dependency map: tombstone_id -> list of issues that depend on it
depsToTombstone := make(map[string][]string)
for _, issue := range allIssues {
for _, dep := range issue.Dependencies {
if dep.DependsOnID != "" {
depsToTombstone[dep.DependsOnID] = append(depsToTombstone[dep.DependsOnID], issue.ID)
}
}
}
// Find tombstones safe to delete (no open issues depend on them)
safeToDelete := make(map[string]bool)
for tombstoneID := range tombstones {
hasOpenDep := false
for _, depID := range depsToTombstone[tombstoneID] {
if liveOpen[depID] {
hasOpenDep = true
break
}
}
if !hasOpenDep {
safeToDelete[tombstoneID] = true
}
}
// Calculate what we'll keep
for tombstoneID := range tombstones {
if safeToDelete[tombstoneID] {
result.DeletedIDs = append(result.DeletedIDs, tombstoneID)
} else {
result.KeptIDs = append(result.KeptIDs, tombstoneID)
}
}
result.TombstonesDeleted = len(result.DeletedIDs)
result.TombstonesKept = len(result.KeptIDs)
// Count stale deps (from closed issues to tombstones) that will be removed
for _, issue := range allIssues {
if liveClosed[issue.ID] {
for _, dep := range issue.Dependencies {
if tombstones[dep.DependsOnID] != nil {
result.DepsRemoved++
}
}
}
}
if dryRun {
return result, nil
}
// Actually modify: filter out deleted tombstones and clean deps
var kept []*types.Issue
for _, issue := range allIssues {
if safeToDelete[issue.ID] {
continue // Skip deleted tombstones
}
// Clean deps pointing to deleted tombstones
var cleanDeps []*types.Dependency
for _, dep := range issue.Dependencies {
if !safeToDelete[dep.DependsOnID] {
cleanDeps = append(cleanDeps, dep)
}
}
issue.Dependencies = cleanDeps
kept = append(kept, issue)
}
// Write back atomically
dir := filepath.Dir(issuesPath)
base := filepath.Base(issuesPath)
tempFile, err := os.CreateTemp(dir, base+".purge.*")
if err != nil {
return nil, fmt.Errorf("failed to create temp file: %w", err)
}
tempPath := tempFile.Name()
encoder := json.NewEncoder(tempFile)
for _, issue := range kept {
if err := encoder.Encode(issue); err != nil {
_ = tempFile.Close()
_ = os.Remove(tempPath)
return nil, fmt.Errorf("failed to write issue %s: %w", issue.ID, err)
}
}
if err := tempFile.Close(); err != nil {
_ = os.Remove(tempPath)
return nil, fmt.Errorf("failed to close temp file: %w", err)
}
if err := os.Rename(tempPath, issuesPath); err != nil {
_ = os.Remove(tempPath)
return nil, fmt.Errorf("failed to replace issues.jsonl: %w", err)
}
return result, nil
}
// runCompactPurgeTombstones handles the --purge-tombstones mode for dependency-aware cleanup.
// Unlike --prune which removes tombstones by age, this removes tombstones that have no
// open issues depending on them, regardless of age.
func runCompactPurgeTombstones() {
start := time.Now()
if compactDryRun {
result, err := purgeTombstonesByDependency(true)
if err != nil {
fmt.Fprintf(os.Stderr, "Error: failed to analyze tombstones: %v\n", err)
os.Exit(1)
}
if jsonOutput {
output := map[string]interface{}{
"dry_run": true,
"tombstones_before": result.TombstonesBefore,
"tombstones_to_delete": result.TombstonesDeleted,
"tombstones_to_keep": result.TombstonesKept,
"deps_to_remove": result.DepsRemoved,
"deleted_ids": result.DeletedIDs,
"kept_ids": result.KeptIDs,
}
outputJSON(output)
return
}
fmt.Printf("DRY RUN - Dependency-Aware Tombstone Purge\n\n")
fmt.Printf("Tombstones found: %d\n", result.TombstonesBefore)
fmt.Printf("Safe to delete: %d (no open issues depend on them)\n", result.TombstonesDeleted)
fmt.Printf("Must keep: %d (have open deps)\n", result.TombstonesKept)
fmt.Printf("Stale deps to clean: %d (from closed issues to tombstones)\n", result.DepsRemoved)
if len(result.KeptIDs) > 0 && len(result.KeptIDs) <= 10 {
fmt.Println("\nKept tombstones (have open deps):")
for _, id := range result.KeptIDs {
fmt.Printf(" - %s\n", id)
}
}
return
}
// Actually purge
result, err := purgeTombstonesByDependency(false)
if err != nil {
fmt.Fprintf(os.Stderr, "Error: failed to purge tombstones: %v\n", err)
os.Exit(1)
}
elapsed := time.Since(start)
if jsonOutput {
output := map[string]interface{}{
"success": true,
"tombstones_before": result.TombstonesBefore,
"tombstones_deleted": result.TombstonesDeleted,
"tombstones_kept": result.TombstonesKept,
"deps_removed": result.DepsRemoved,
"elapsed_ms": elapsed.Milliseconds(),
}
outputJSON(output)
return
}
if result.TombstonesDeleted == 0 {
fmt.Printf("No tombstones to purge (all %d have open deps)\n", result.TombstonesBefore)
return
}
fmt.Printf("✓ Purged %d tombstone(s)\n", result.TombstonesDeleted)
fmt.Printf(" Before: %d tombstones\n", result.TombstonesBefore)
fmt.Printf(" Deleted: %d (no open deps)\n", result.TombstonesDeleted)
fmt.Printf(" Kept: %d (have open deps)\n", result.TombstonesKept)
fmt.Printf(" Stale deps cleaned: %d\n", result.DepsRemoved)
fmt.Printf(" Time: %v\n", elapsed)
}