ormed 0.4.0-dev+11

Strongly typed ORM core primitives (annotations, metadata, codecs, and generator) for the routed ecosystem.

Changelog

0.4.0-dev+11

• Added: Host-neutral atomic query-builder batches with native-driver execution, transactional fallback, ordered results, and per-statement metadata.
• Added: Repository batch staging plus generated identity results for atomic inserts and upserts.
• Fixed: updateBatch() now executes atomically and rejects rows missing their configured unique key instead of broadening the update predicate.
• Fixed: Failed native atomic batches are now recorded in query logs and connection events for every staged operation.
• Fixed: Native batch interceptors now receive each staged operation's query or mutation plan for policy enforcement.
• Fixed: Projected hard-delete batches retain the primary key required by query-delete compilers, and ID fallbacks reject auto-increment sentinels.

0.3.1

• Fixed: Codegen-free map inserts now keep inferred mutation columns separate from ad-hoc read and returning projections, preserving generated columns and later SELECT * queries.
• Fixed: Generated query helpers suppress the intentional unbounded-query analyzer warning using the namespaced plugin diagnostic.
• Fixed: Query safety analysis now handles generated sources without reporting the companion all helper.

0.3.0

• Added: OrmDatabase for direct, codegen-optional connections, raw SQL, schema plans, migrations, seeding, transactions, and lifecycle management.
• Added: Ordered QueryInterceptor middleware for query-builder queries, mutations, raw SQL, schema plans, migrations, streams, and transactions.
• Added: Built-in OrmOpenTelemetryInterceptor backed by dartastic_opentelemetry, with SQL text excluded by default.
• Added: Reactive Query.watch() / watchRows() snapshots with table-aware invalidation and an opt-in driver change-feed contract.
• Added: Reactive changes are held across managed and manually controlled transactions until commit, with direct invalidation helpers for integrations.
• Added: Optional DriverLifecycle support lets external executor adapters open their backend during DataSource initialization.
• Breaking: Requires Dart 3.12 or newer.
• Breaking: Migrated the analyzer integration to Analyzer 14, including the new argument and formal-parameter AST APIs.
• Updated: Refreshed build, source generation, testing, and analyzer tooling dependencies to their latest Dart 3.12-compatible releases.
• Updated: Internal driver and test-package constraints now support the full 0.x release range.

0.2.0

• Added: OrmedEnvironment with .env loading, variable interpolation support (${VAR} / ${VAR:-default}), and typed helpers for required values.
• Added: MigrationEntry.named(...) helper for easier manual migration registration.
• Improved: DataSource.fromConfig now falls back to an empty ModelRegistry when not explicitly provided.
• Changed: Default ORM registry generation output path moved to lib/src/database/orm_registry.g.dart.
• Fixed: Registry generator now emits package imports when needed so generated files continue to compile from nested output paths.
• Docs: Updated quickstart and migration helper guidance to use make:migration and document makemigrations as model-diff generation plus registry sync.
• Docs: Clarified migrations:sync as the registry-only command and aligned command examples with the current ormed_cli executable usage.

0.1.0

• Release: Promote Ormed to a stable 0.1.0 release.
• Improved: Align analyzer tooling with analyzer 10 (analysis_server_plugin/analyzer_plugin/analyzer_testing).

0.1.0-dev+10

• Added: Query cache invalidation policy flushOnWrite to clear cached results after mutations, plus warning hooks for risky query features like fallback row identifiers.
• Added: disableAutoHydration to opt out of automatic hydration when building queries.
• Improved: raw_sql_interpolation analyzer lint is now configurable (or disable-able) via analysis_options.yaml.
• Fixed: Non-nullable enum defaults now parse CAST() expressions correctly, and generated tracked models no longer duplicate relation getters.
• Fixed: SQL preview now surfaces unsupported query features instead of masking errors.

0.1.0-dev+9

• Added: test helpers now generated via ormed init include dual SQLite datasources, primaryTestConfig/analyticsTestConfig, and connection helpers so user tests can rely on shared migrations without managing the multi-datasource wiring manually.
• Added: bootstrap test harness writes and runs a helper-driven test while the CLI template now offers --only=tests and a reusable helper to prove connection isolation in generated apps.
• Improved: setUpOrmed runs migrations for the provided DataSource by default, adds deterministic parallel groupIds, and optionally provisions a Postgres schema per config key so shared manager reuse stays conflict-free.
• Improved: TestDatabaseManager accepts an optional baseSchema, ensures it is created/selected before migrations, and tracks it for cleanup.
• Switched: internal hashing helpers now use hashlib instead of crypto.
• Refactor: Decoupled core package from driver-specific dependencies (ormed_sqlite, ormed_mysql, ormed_postgres).
• Added: MockDriver and InMemoryQueryExecutor for isolated grammar and primitive validation without driver dependencies.
• Improved: Schema migration runner now ensures boarder compatibility by avoiding implicit transactions for DDL (standardized across all drivers).
• Fixed: PostgreSQL JSON encoding now recursively handles AdHocRow, ModelAttributes, and DTOs within nested Maps and Iterables.

0.1.0-dev+8

• Added: Structured OrmException/DriverException for richer runtime and CLI messaging.
• Added: dart run ormed now prints guidance for running the CLI package.
• Added: External model factory definitions with named states and registry-based overrides.
• Fixed: Generated tracked model constructors now mirror user constructor shapes.

0.1.0-dev+7

• Added: Transaction outcome hooks/events that distinguish commit vs rollback (with savepoint scope details).
• Added: Transaction boundary logging in query logs (BEGIN/COMMIT/ROLLBACK and savepoints).
• Added: Analyzer plugin is now included in ormed (no separate ormed_analyzer package).
• Added: Analyzer diagnostics for select/order/group/having validation, nested relation checks, relation callback field mismatch, missing pivot fields, type-aware predicates, query safety (limit/order + unbounded get/all), raw SQL alias checks, DTO validation, and soft-delete/timestamp misuse; supports split-chain tracking.
• Fixed: copyWith now casts non-nullable Object fields to avoid analyzer errors.

0.1.0-dev+6

• Added: Typed predicate field accessors for where callbacks (e.g. q.title.eq(...)) and typed relation helpers (withPosts, whereHasPosts).
• Added: Generated copyWith extensions for user-defined models with sentinel-based null handling, including constructor-backed relation fields.
• Added: Generated toMap/fromMap helpers for user and tracked models using column-name maps.
• Added: Driver extension registry and query builder hooks (selectExtension, whereExtension, orderByExtension, groupByExtension, havingExtension, join constraints).
• Added: Full-text query compilation for whereFullText across Postgres, MySQL/MariaDB, and SQLite, including optional index targeting.
• Docs: Added driver extension tutorial and examples.
• Added: DataSource logging can now use an injected contextual logger and exposes it for inspection.

0.1.0-dev+5

• Fixed: Treat driver SQL NULL wrapper values as null when applying timestamp defaults.
• Improved: Generated codecs now omit virtual attributes (timestamps/soft deletes) unless explicitly set.
• Added: Core tests for SQL NULL handling helpers.
• Fixed: JSON casts now accept raw strings and mixed JSON values for Object? fields.
• Fixed: JSON cast decoding now handles byte payloads from MySQL/MariaDB drivers.
• Fixed: Map/DTO insert & upsert inputs now apply field codecs (casts) consistently.
• Changed: Automatic timestamps now use plain UTC time without monotonic state.

0.1.0-dev+4

• Added: Relationship touching with touches, touch(), and touchOwners() for keeping related updated_at values in sync.
• Added: withoutTouchingOn / withoutTimestampsOn scopes to suppress automatic touching or timestamp updates.
• Added: @OrmModel(timestamps: false) to disable automatic timestamp updates while still using timestamp fields.
• Improved: Timestamp casting accepts numeric epoch inputs (seconds) in addition to DateTime/Carbon types.
• Added: Through relations support with hasOneThrough and hasManyThrough relation types.
• Added: Polymorphic relations with morphTo, morphToMany, and morphedByMany support.
• Added: Attribute accessors and mutators for get/set interception.
• Added: Appended/computed attributes for serialization.
• Added: Extended attribute casting system with custom cast handlers.
• Added: Pivot enhancements including withPivot field selection and pivot timestamps.
• Added: Raw query helpers orderByRaw and groupByRaw.
• Fixed: updatedAt now respects explicitly provided values during updates (dirty-check aware).
• Added: Object input support for fill, forceFill, and fillIfAbsent methods.
• Refactored: Decoupled database drivers from the core package. Drivers now register themselves via DriverAdapterRegistry.
• Added: DataSource.fromConfig as the standard, driver-agnostic entry point for ORM initialization.
• Added: syncWithoutDetaching, syncWithPivotValues, toggle, and updateExistingPivot methods for ManyToMany relationships.
• Added: Support for eager loading nested relations (e.g., ds.query<User>().with('posts.comments').get()).
• Added: copyWith method generation for DTOs and partial models.
• Added: suppressEvents option to Query for optimized bulk operations.
• Added: loadOrmConfig() and findOrmConfigFile() convenience helpers for configuration management.
• Fixed: Eager loading for models with custom primary keys (Issue #12).
• Fixed: Improved relation inference for complex naming conventions and ambiguous foreign keys (Issue #7).
• Fixed: Missing relation accessors in generated models for certain edge cases (Issue #6).
• Fixed: Resolved bugs #4, #11, and #13 related to query builder edge cases.
• Improved: Code generator now emits dartdoc comments for all generated models and members.
• Improved: Reached 160/160 pub score with expanded documentation and examples.

0.1.0-dev+3

• Fixed: Timestamp getters (createdAt, updatedAt, deletedAt) now return immutable Carbon instances to prevent accidental mutation of model state when chaining date methods like subDay().
• Refactored Timestamps and SoftDeletes mixins to use CarbonInterface for getters and Object? for setters.
• Exported carbonized package directly from ormed.dart.
• Improved DateTimeCodec to handle Carbon instances during decoding.
• Fixed mass assignment to handle both field and column names for excluded attributes.
• Optimized Postgres test performance using transactional isolation.
• Fixed SQLite test cleanup issue where test_g* files were left behind.
• Expanded testing documentation with isolation strategies and concurrency guides.
• Updated doc comments to follow Effective Dart guidelines.

0.1.0-dev+2

• Synchronized release with ormed_cli rebranding.

0.1.0-dev+1

• Added automatic snake_case column name inference for model fields.
• Fixed DatabaseSeeder.seed<T>() to use repositories for correct data encoding.
• Fixed bootstrapOrm() to ensure model registration in existing registries.
• Improved documentation for naming conventions and CLI-first workflow.

0.1.0-dev

• Initial release.