The test guard was flagging false positives when SQLite shm/wal files
had their mtime updated from read-only operations (config loading, etc.)
without any actual content modification.
Now only reports when file size changes, which indicates actual content
modification rather than just file access.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add multiple layers of defense against misclassified wisps:
- Importer auto-detects -wisp- pattern and sets ephemeral flag
- GetReadyWork excludes -wisp- IDs via SQL LIKE clause
- Doctor check 26d detects misclassified wisps in JSONL
This addresses recurring issue where wisps with missing ephemeral
flag would pollute bd ready output after JSONL import.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- sync_git_test.go: getCurrentBranchOrHEAD now returns 1 value (not 2)
- daemon_basics_test.go: getPIDFileForSocket uses dbPath not socketPath
- daemon_autostart.go: mark unused socketPath param with underscore
These tests were broken by recent refactors that changed function signatures
but didn't update the corresponding test files.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
* feat: comprehensive NixOS support improvements
This commit adds full NixOS support and automates Nix package maintenance:
## Static Linux Binaries for NixOS
Changes .goreleaser.yml to create static Linux binaries:
- Set CGO_ENABLED=0 for Linux amd64 and arm64 builds
- Remove cross-compiler dependencies (aarch64-linux-gnu-gcc)
- Simplifies build process while fixing NixOS compatibility
Static binaries work on all Linux distributions including NixOS, Alpine,
and musl-based distros without dynamic linker dependencies.
## Automated default.nix Version Management
Adds default.nix to the version bump workflow:
- Updates default.nix version field in bump-version.sh (new step 9)
- Adds default.nix to version verification checks
- Prevents version drift (was 5 releases behind: 0.42.0 vs 0.47.0)
- Updates README.md to remove glibc 2.32+ requirement
## Automated vendorHash Management
Creates scripts/update-nix-vendorhash.sh to automate vendorHash updates:
- Automatically detects correct hash by triggering Nix build error
- Extracts hash from error message and updates default.nix
- Verifies update with clean build
- Eliminates error-prone manual copy-paste workflow
- Works with local Nix OR Docker (uses nixos/nix image automatically)
Integrates vendorHash check into bump-version.sh:
- Detects when go.mod or go.sum have changed
- Prompts to run vendorHash update script interactively
- Catches synchronization issues at release time
## Documentation
Updates AGENTS.md with Nix package maintenance guide:
- Documents when and how to update vendorHash
- Recommends automated script as primary method
- Provides manual and alternative methods as fallback
- Notes Docker fallback for maintainers without Nix
## Impact
- NixOS users can now install via standard methods
- Nix package version stays synchronized automatically
- vendorHash updates work without Nix installed (via Docker)
- vendorHash updates are caught during release workflow
- All Linux users benefit from more portable binaries
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
* fix: address PR review feedback for NixOS support
Fixes three issues identified in PR review:
1. Replace undefined log_* functions in bump-version.sh with existing
echo pattern (log_warning, log_info, log_success, log_error were
called but not defined)
2. Update default.nix version from 0.42.0 to 0.47.0 to fix version
drift with cmd/bd/version.go
3. Remove Nix Package Maintenance section from AGENTS.md per beads
architecture (use bd prime for dynamic context, keep AGENTS.md
minimal)
* fix: update versions to 0.47.1 after merge with main
- Update claude-plugin plugin.json to 0.47.1
- Update default.nix to 0.47.1
- Fixes version check failures after merging latest main branch
---------
Co-authored-by: Claude Sonnet 4.5 <noreply@anthropic.com>
When using short socket paths (workspaces with paths >103 chars),
filepath.Dir(socketPath) returns /tmp/beads-XXXXX/ instead of the
actual .beads/ directory. This caused TryDaemonLock to look in the
wrong location, always return false, clean up the startlock, and
recurse infinitely causing stack overflow.
Changed 4 occurrences in acquireStartLock, handleStaleLock,
handleExistingSocket, and getPIDFileForSocket to use
filepath.Dir(dbPath) which correctly points to .beads/.
Fixes: gt-qlt
Co-authored-by: furiosa <mayor@gastown.local>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
The function always returns nil for error since it is designed to
return HEAD on detached HEAD state. Simplify signature to just
return string.
Fixes golangci-lint unparam warning.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add bd doctor check that detects legacy gastown merge queue JSON files
in .beads/mq/. These files are local-only remnants from the old mrqueue
implementation and can safely be deleted since gt done already creates
merge-request wisps in beads.
- CheckStaleMQFiles() detects .beads/mq/*.json files
- FixStaleMQFiles() removes the entire mq directory
- Comprehensive tests for check and fix
This is the first step toward removing the mrqueue side-channel from
gastown. The follow-up convoy will update Refinery/Witness to use
beads exclusively.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add insertIssuesStrict function that uses plain INSERT instead of
INSERT OR IGNORE. Update bulkInsertIssues and transactional CreateIssues
to use the strict variant.
This fixes a race condition where INSERT OR IGNORE could silently skip
duplicate insertions, but the code would still attempt to record events
for those "inserted" issues, causing FOREIGN KEY constraint failures.
The strict INSERT will now fail explicitly if a duplicate is encountered,
which should never happen since checkForExistingIDs runs first within
the same transaction.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add bd doctor check that warns if .beads/last-touched is tracked by git.
This file is local runtime state that should never be committed, as it
causes spurious diffs in other clones.
- CheckLastTouchedNotTracked() detects if file is git-tracked
- FixLastTouchedTracking() untracks with git rm --cached
- Comprehensive tests for all scenarios
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The plugin.json was left at 0.47.0 while all other version files
were bumped to 0.47.1, causing CI failures on all PRs.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
chore(deps): bump fastmcp from 2.14.1 to 2.14.3 in /integrations/beads-mcp
Bug fixes in this release:
- Fix HTTP transport timeout defaulting to 5 seconds
- OAuth token storage TTL fix
- Redis key prefix for ACL isolation
- ContextVar propagation for ASGI-mounted servers
Add missing reconnectMu.RLock() protection to storage methods that were
vulnerable to the same race condition fixed in GH#607. The FreshnessChecker
can trigger reconnect() which closes s.db while queries are in flight,
causing "database is closed" errors during daemon export operations.
Protected methods:
- labels.go: GetLabelsForIssues (GetLabels intentionally unprotected - called from GetIssue which holds lock)
- comments.go: GetIssueComments, GetCommentsForIssues
- dependencies.go: GetDependencyCounts, GetDependencyRecords, GetAllDependencyRecords, GetDependencyTree, loadDependencyGraph
- config.go: SetConfig, GetConfig, GetAllConfig, DeleteConfig, SetMetadata, GetMetadata
- dirty.go: MarkIssueDirty, GetDirtyIssues, GetDirtyIssueHash, GetDirtyIssueCount
- events.go: GetEvents, GetStatistics, GetMoleculeProgress
- hash.go: All hash methods
- hash_ids.go: GetNextChildID, ensureChildCounterUpdated (getNextChildNumber unprotected - called internally)
Internal helpers called from already-locked contexts intentionally omit
RLock to avoid deadlock (Go's RWMutex doesn't support recursive locking).
Fixes: bd-vx7fp
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Executed-By: beads/crew/dave
Rig: beads
Role: crew
prek (https://prek.j178.dev) is a faster Rust-based alternative to
pre-commit that uses the same .pre-commit-config.yaml config files.
Changes:
- Add prek detection pattern in hookManagerPatterns (before pre-commit
to ensure correct detection since prek hooks may contain 'pre-commit')
- Handle prek in checkManagerBdIntegration using same config parser
- Update init_git_hooks.go to recognize prek-installed hooks
- Rename isPreCommit field to isPreCommitFramework for clarity
- Use regex pattern matching all pre-commit/prek signatures consistently
- Add test cases for prek run and prek hook-impl signatures
Co-authored-by: Ismar Iljazovic <ismar@gmail.com>
macOS allows file owners to write to their own read-only (0444) files,
so TestSetupGlobalGitIgnore_ReadOnly cannot exercise the "Unable to write"
code path on this platform.
Skip both test cases on darwin with an explanatory message.
The test was failing because it called 'git config --global core.excludesfile'
which returned the real user's gitignore path instead of using the test's
temp directory.
Fix: Set GIT_CONFIG_GLOBAL env var to an empty temp config file, ensuring
the test uses the temp directory's .config/git/ignore path as intended.
Also extracted the isolation logic into a reusable setupIsolatedGitConfig helper.
Co-authored-by: Ismar Iljazovic <ismar@gmail.com>
Fixes#1041
The pre-push hook was not passing arguments to chained hooks,
causing them to fail. This change:
1. Changes runPrePushHook() to accept args []string parameter
2. Passes args to runChainedHook() instead of nil
3. Updates call site to pass hookArgs
4. Renames local 'args' to 'statusArgs' to avoid variable shadowing
Co-authored-by: Ismar Iljazovic <ismar@gmail.com>
The export command was incorrectly listed in readOnlyCommands, causing the SQLite database to be opened in read-only mode. This prevented export from clearing dirty issues and updating the JSONL file hash.
Problem:
When a JSONL file hash mismatch was detected (e.g., after git operations
that modify the JSONL without updating export_hashes), the autoflush
system would trigger a full re-export. During this re-export, all issue
comments were silently dropped from the exported JSONL file.
Steps to reproduce:
1. Have a beads database with issues containing comments
2. Create a situation where JSONL hash doesn't match stored hash
(e.g., clone a repo, or manual JSONL edits)
3. Run any bd command that triggers autoflush (e.g., `bd create foo`)
4. Observe warning: "JSONL file hash mismatch detected"
5. Check .beads/issues.jsonl - all comments are now missing
Root cause:
Two different export code paths existed:
- exportToJSONLWithStore (daemon_sync.go) - correctly populated comments
- fetchAndMergeIssues (autoflush.go) - only fetched dependencies, NOT comments
When hash mismatch triggered a full re-export via the autoflush path,
fetchAndMergeIssues was called but it never populated issue.Comments,
resulting in all comments being lost.
Fix:
Add GetIssueComments call in fetchAndMergeIssues to populate comments
for each issue before export, matching the behavior of exportToJSONLWithStore.
Note: Labels were not affected because GetIssue() already populates them
internally. Comments are stored in a separate table and require explicit
fetching via GetIssueComments().
- Remove standalone beads-merge binary section (no longer maintained)
- Update Jujutsu config to use `bd merge` instead of `beads-merge`
- Fix config path to standard `~/.config/jj/config.toml`
- Add note that tool only works for .beads/issues.jsonl
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add getCurrentBranchOrHEAD() which returns "HEAD" when in detached HEAD
state instead of failing. This fixes bd sync --status for jj/jujutsu users.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
