Changelog

Added recovery assessment and migration inspection CLI commands

Added MigrationCommandResponse dataclass for framework-neutral CLI output

Added StorageFingerprint serialization methods (to_mapping, from_mapping)

Added CLI error types for migration operations

Added test suite for migration recovery scenarios

Improved migration phase transitions to allow recovery from failed state

Improved item state transitions for direct pending-to-complete and stage-verified-to-activation paths

Changed CLI migrate module to add recovery assessment and inspection commands

Fixed pytest failures in migration module

Fixed Windows test failures in migration CLI tests

Added DestinationPrecondition dataclass for typed before-state fingerprints

Added MigrationPhase and MigrationItemState enums for schema-3 state machine

Added Schema3MigrationJournal, Schema3ItemJournalState, Schema3ConfigSwitchState dataclasses

Added write_schema3_journal and _parse_schema3_journal for schema-3 TOML persistence

Added StorageMigrationHooks for lifecycle callbacks (quiescence, staged/activated validation, finalize)

Added MigrationLock for advisory project migration locking

Added _check_same_filesystem and _validate_same_filesystem for cross-filesystem detection

Added _durable_rename for fsync-backed atomic renames

Added render_ledger_manifest and render_ledger_local_config for pure TOML rendering

Added strict plan-structure validation (_validate_plan_structure) for migration_id, policies, strategies, kinds

Added stage/backup collision detection in _check_path_overlaps

Added source fingerprint population during planning

Changed file fingerprints (sha256-file-v1) to be path-independent (content hash only, no filename)

Changed fingerprint failures to fail closed: state=’invalid’ instead of ‘owned’ with missing fingerprint

Changed config inspection to check parent binding even when file is absent

Changed noop-if-exact to compare against expected_target_fingerprint (not before-state)

Changed replace-owned to use expected_before.fingerprint for before-state validation

Changed executor to call validate_storage_migration_plan before first journal write

Fixed fail-open validation where fingerprint errors allowed owned+replaceable state

Fixed missing config file bypassing foreign parent detection

Fixed unknown destination policies/strategies falling through to default action

BREAKING: Added a schema-3 storage model with derived paths, binding markers, TOML ownership, and migration

Documented per-ledger sibling storage paths and shared project UUID under one root

Classified incomplete migration recovery as manual intervention (STORAGE_MIGRATION_MANUAL_INTERVENTION_REQUIRED)

Disabled explicit mode=move execution (raises STORAGE_MIGRATION_MOVE_DISABLED)

Fixed codecov test configuration

Documented Ledgercore 0.5.1 migration containment strategy

Migrated releaseledger storage to project-local ledger directory

Added project badges

Fixed documentation typo

Fixed pytest configuration

Added tests to reproduce migration journal validation defects

Fixed completed recovery to return truthful source_removed (False for schema-2 copy, None for schema-1)

Fixed schema-1 journal inspection to use None for unknown bindings instead of fabricating storage=project

Fixed migration progress metadata in schema-2 journals (items_completed, mode, verify, project_root)

Fixed exact migration journal binding identity round-trip (source and destination bindings persisted in schema-2 journals)

Added shared ledger configuration helpers for locating ledger.toml files and selecting project or tool tables

Changed core storage and validation behavior for atomic writes, front matter, IDs, and source-tree versions

Documented Sphinx builds with MyST Markdown and myst-parser dependencies

JSON Lines object storage with deterministic writing, comment support, and recoverable line-level error handling

SHA-256 content fingerprinting with front matter decomposition for document and metadata hashing

Path text normalization for Unicode escapes, punctuation, and filesystem-safe slug generation

FrontMatterRenderOptions with key ordering, remaining key sort control, scalar styling, and template placeholder modes

arc42 architecture documentation covering system context, building blocks, runtime, deployment, and quality requirements

Version sourcing switched to hatch-vcs for automatic version derivation from git tags

Version smoke test and wheel distribution validation script

Pre-commit formatting and test assertion fixes

write_text uses write_bytes on Windows to avoid encoding issues with special characters

Atomic text write and create primitives with fsync guarantees and temp-file safety

YAML front matter reader/writer with body mode control and source file iteration

Prefixed numeric ID formatting, parsing, and next-ID generation with configurable width and separator

UTF-8 text helpers including newline normalization, content hashing, and text merging

Validated JSON object/array loading and deterministic JSON writing with sorted keys

Safe relative POSIX path validation and config-relative path resolution

LedgerIdFormat with configurable segment support for compound IDs like prefix_segment_number

YAML store module with validated object/array loading and atomic writing

Cross-ledger resource reference system with canonical, file-safe, and local ref formats

atomic_create_text uses O_CREAT|O_EXCL for race-safe file creation instead of check-then-create

API reference documentation for all public modules

Usage examples for front matter, resource references, and storage operations

CI workflows for testing, coverage, and PyPI publishing