Skip to content

Changelog

1.0.0-b4.5

Update now: 

pip install --upgrade chacc-api

Changed

  • Module naming convention – Module directories now use underscores instead of hyphens (e.g., chacc_file_manager instead of chacc-file-manager) to align with Python naming standards. This enables consistent use of the standard import system throughout the module loading pipeline. Module metadata name fields must also use underscores. Migration required: Rename your module directories and update module_meta.json files.

  • Code formatting tool – Switched from Black to Ruff formatter. All code is now formatted using ruff format with the same 100-character line length. You can run ruff format . to auto-format and ruff format --check . to verify formatting.

  • Documentation workflow – Documentation Docker images are now built and pushed automatically when a release is published. Manual documentation builds can be triggered by including "build docs" in a commit message on the develop or main branch, or via the workflow_dispatch workflow in GitHub Actions.

  • Changelog location – The changelog has been moved to chacc-docs/docs/changelog.md. The root CHANGELOG.md file has been removed. See the changelog on chacc.dev for the complete history.

  • ChaCC Theme Applied in Swagger UI and ReDoc - Added custom ChaCC theme styling to Swagger UI and ReDoc interfaces for consistent branding.

Added

  • Manual Docker workflows – Added separate workflows for manual Docker builds:
  • .github/workflows/docker-manual.yml – Build and push main Docker image on demand
  • .github/workflows/docker-docs-manual.yml – Build and push documentation Docker image on demand

Fixed

  • Module loading duplicate class registration – Fixed "Multiple classes found for path" SQLAlchemy errors when loading modules by refactoring the discovery and setup phases to use a consistent import mechanism. Both phases now use the same module objects from sys.modules, eliminating duplicate SQLAlchemy declarative registry entries.

  • AutoIncrement Indexed ID – Fixed auto-increment behavior for indexed primary keys from ChaCCBaseModel.

AutoIncrement Indexed ID is a breaking change. Ensure to backup your database before upgrading.

1.0.0-b4.2

Beta4.2 fixes a critical startup crash, resolves the audit-schema chicken-and-egg problem, and cleans up model discovery. We have brought you better landing page. If you've had issues with module loading or SQLite table detection, this release is for you.

Added

  • Smarter model discovery – Modules now load their dependencies in the correct order automatically, even when imports are nested.
  • Better migration safety – New dependency resolver and operation executor handle PostgreSQL enum conflicts and ensure migrations run correctly across SQLite and PostgreSQL.
  • Comprehensive test coverage – Tests now cover migration ordering, safe operation filtering, schema reflection, and model discovery.
  • HTML welcome page – Adds a dark-mode HTML welcome page with links to Swagger UI, ReDoc, and chacc.dev documentation, replacing the raw JSON root endpoint.

Fixed

  • Startup crash – Fixed "Multiple classes found for some modules" crash that occurred when plugin discovery was disabled in development mode.
  • Double model registration – Removed the old _model_registry and now rely exclusively on SQLAlchemy's declarative metadata. No more duplicate table errors.
  • SQLite table detection – Fixed a bug where table_exists() returned incorrect results for existing tables.
  • Audit schema chicken-and-egg – Split schema initialization into two passes (before and after entry points) so audit fields are applied correctly even when the enable_audit_fields service is registered after startup.
  • Module load failure handling – Failed modules are now properly disabled, preventing repeated crash loops on restart.
  • Logging crashes – Fixed crashes when logging routes that had None paths or methods.

Changed

  • Module loading sequence – Both dev and prod modes now follow the same sequence: discover models → initialize database → run migrations → load entry points → apply deferred schema changes → optional follow-up migration.
  • Database initialization – Now uses ChaCCBaseModel subclass enumeration instead of the removed _model_registry.

Removed

  • Redundant startup schema pass – Removed a duplicate migration run that caused unnecessary migrations in some environments.
  • _model_registry – The old model registry has been removed in favor of SQLAlchemy's declarative metadata.

Notes

  • Versioning: The package metadata uses 1.0.0-b4.post2 for PyPI; the user-facing version is 1.0.0-b4.2. Both refer to the same release.
  • Port default: All references now consistently use port 8085 (previously some docs mentioned 8080).
  • .env behavior: The .env.sample file is copied from the package to your working directory if missing. This is intentional and unchanged from previous versions.

1.0.0-b4.1

Added

  • PostgreSQL enum migration support through alembic-postgresql-enum, including handling for create_enum, sync_enum_values, and drop_enum operations.
  • Migration runner support for Alembic PostgreSQL enum operation objects generated by alembic-postgresql-enum.
  • Automatic enum type creation before adding enum-backed columns or tables in PostgreSQL migrations to prevent UndefinedObject failures.
  • uuid7-based default UUID generation for ChaCCBaseModel, using Python 3.12+ uuid.uuid7 with uuid-utils fallback on older Python versions.
  • uuid-utils runtime dependency for environments without native uuid7 support.
  • SQLITE_DATABASE_NAME and SQLITE_DATABASE_PATH environment variables for custom SQLite database names and storage locations.
  • CHACC_VERBOSE and CHACC_DEBUG environment controls for runtime log level selection.
  • CLI chacc run server -v/--verbose support that now propagates verbose logging into the server subprocess.
  • Docker health checks and exposed ports updated to the current default server port.
  • Production Docker Compose now uses the published Docker Hub image directly instead of local image build comments.

Fixed

  • PostgreSQL migration failures when Alembic detects enum changes by importing and enabling alembic-postgresql-enum hooks.
  • PostgreSQL UndefinedObject crashes when adding enum-backed columns or tables by creating enum types before table/column operations.
  • Migration operation ordering to apply enum creation before tables, columns, constraints, indexes, and enum synchronization.
  • Migration descriptions for enum create, sync, and drop operations.
  • CLI verbose flag propagation so chacc run server -v affects the child server process.
  • Docker and Docker Compose default port mismatch by replacing stale 8080 references with 8085.
  • Alembic log noise by forcing the alembic logger to WARNING.
  • Default logging behavior so normal server runs are quieter while CHACC_VERBOSE and CHACC_DEBUG still enable detailed logs.
  • Repeated logger imports by centralizing default log level detection in get_default_log_level().
  • SQLite database path handling so SQLITE_DATABASE_PATH and SQLITE_DATABASE_NAME configuration takes effect.

Removed

  • Removed redundant LogLevels.INFO logger setup across server, database, migration, module loader, Redis, health, and environment validation modules.
  • Removed stale Docker 8080 exposed port and health check references.
  • Removed local-build instructions from the production Docker Compose file.

Changed

  • README file has been updated to focus on ChaCC brief intro and link the entire guidance to chacc.dev

1.0.0-b4

Added

  • Breaking: GUID TypeDecorator for cross-database UUID support (PostgreSQL UUID type / SQLite TEXT storage)
  • SQLite batch operation support for constraints, indexes, and columns via batch_alter_table()
  • Automatic .env file creation from .env.sample on application startup if .env does not exist
  • .env.sample included in package distribution for ready-to-use configuration reference
  • Migration version counter suffix to prevent duplicate migration versions
  • New database engine detection to distinguish PostgreSQL vs other databases

Fixed

  • Breaking: Migration tracker table rollback_available column conversion from BOOLEAN to INTEGER for PostgreSQL, and compatibility for SQLITE
  • Migration runner now uses run_in_executor() for synchronous DB operations to prevent blocking async event loop
  • Made tracker/backup lazy properties to defer DB initialization
  • Catch ProgrammingError/OperationalError for already-existing resources to make migrations idempotent
  • GUID TypeDecorator now returns UUID instances directly without unnecessary conversion
  • Tracker table now filtered from migration detection to prevent accidental drops
  • Resolved requirements.txt path issue for installed package layout

Removed

  • Removed redundant database type conversion in GUID TypeDecorator

Caution: Users migrating to 1.0.0-b4 will need to drop their existing databases and start afresh. The rollback_available column type changed from BOOLEAN to INTEGER, and GUID column changes require a clean schema migration that cannot be automatically applied to existing data.