uql-orm 0.7.5 → 0.7.9

package/CHANGELOG.md

@@ -4,6 +4,32 @@ All notable changes to this project will be documented in this file. Please add
 
 date format is [yyyy-mm-dd]
 
+## [0.7.8] - 2026-03-31
+
+### Breaking Changes
+- **Dialect barrel**: **`MongoDialect`** is no longer re-exported from **`uql-orm/dialect`** or the root **`uql-orm`** package (it previously pulled the Mongo dialect graph into SQL-only apps). Import it from **`uql-orm/mongo`**.
+- **Migrate and optional `mongodb` peer**: `createSchemaGenerator` is **SQL-only** (returns `undefined` for non-`AbstractSqlDialect`). For MongoDB use **`createSchemaGeneratorAsync`** from `uql-orm/migrate` (or the CLI re-export). **`MongoSchemaGenerator`** is exported from **`uql-orm/mongo`**, not from the `uql-orm/migrate` barrel, so importing migrate alone does not evaluate the Mongo generator graph. **`Migrator.findEntityForTable`** is now **async**. **`getSchemaGenerator`** (CLI) returns **`undefined`** for MongoDB; `uql-migrate` resolves the generator with **`createSchemaGeneratorAsync`**.
+
+## [0.7.7] - 2026-03-31
+
+### Breaking Changes (Internal API)
+- **Pools and config**: `QuerierPool` exposes `dialect` (renamed from `dialectInstance`). The engine id string is `dialect.dialectName` (replacing the old `dialect` property on the dialect instance). Removed top-level `dialect` from `Config` and `MigratorOptions`—migrations and `uql-migrate` use **`pool.dialect.dialectName`** only.
+- **Dialect API**: Removed `dialectConfig` / `DialectConfig` from the public dialect barrel. Configure behavior with `DialectOptions`, `DialectFeatures` on `AbstractDialect.features`, and **per-driver dialect classes** attached to each pool.
+
+### Improvements
+
+- **Per-driver dialect classes**: Pools use explicit dialect types (e.g. `PgDialect`, `NeonDialect`, `MySql2Dialect`, `BetterSqlite3Dialect`, `LibsqlDialect`, `D1SqliteDialect`, `MongodbNativeDialect`, `BunSqlPostgresDialect`). Migrator and schema generation read **`pool.dialect.dialectName`** instead of a separate config field.
+- **PostgreSQL (`pg` vs Bun SQL)**: `PgDialect` uses base Postgres defaults (native JS arrays for `ANY`/`ALL`, `$n::jsonb` for JSON). Bun SQL Postgres uses `BunSqlPostgresDialect` with `POSTGRES_WIRE_DRIVER_CAPABILITIES` from `uql-orm/postgres/wireCapabilities` plus `explicitJsonCast` where Bun’s bindings require `( $n::text )::jsonb` and array literal encoding.
+- **Migrations CLI**: Exported **`assertCliConfig`** (`uql-orm/migrate`) validates the default-exported config shape before commands run.
+- **Escaping**: Inlined **`escapeAnsiSqlLiteral`** for Postgres/SQLite literal escaping; dropped **`sqlstring-sqlite`** from those code paths. Prefer bound parameters for user input in production.
+- **Types and tests**: `Config.pool` is typed as `QuerierPool<Querier, AbstractDialect>`. **`createMockQuerierPool`** is available from `uql-orm/test`.
+
+## [0.7.6] - 2026-03-31
+### Improvements
+- **Bulletproof Postgres JSONB Serialization**: We've standardized how JSONB data travels to the database to ensure absolute reliability. By handling stringification at the ORM level and using explicit Postgres casts (`::text::jsonb`), we've eliminated subtle ambiguities between JSON arrays and native Postgres arrays. This ensures your data always arrives exactly as intended.
+- **Unified Driver Experience**: Whether you're running on `node-postgres` or the high-performance `bun:sql`, the behavior is now identical. This refactor definitively resolves "double-stringification" and "invalid syntax" edge cases, providing a truly seamless experience across all supported environments.
+- **Internal Architectural Polish**: Simplified the core normalization logic in the base SQL dialect, keeping the ORM engine lean while making the Postgres-specific implementation more declarative and easier to maintain.
+
 ## [0.7.5] - 2026-03-29
 ### New Features
 - **Type-Safe Aggregate Inference**: Enhanced `querier.aggregate()` to automatically infer return types from the `$group` definition. Aggregate results (e.g. `$sum`, `$avg`, `$max`) are now precisely typed without requiring manual casts or `any`.

package/README.md

@@ -284,6 +284,27 @@ export default {
 > **Notes:**
 > - Reuse one pool for both app queries and migrations to keep behavior (for example naming strategy) consistent.
 > - If your architecture spans backend + browser, `HttpQuerier` reduces custom API mapping and keeps query semantics aligned.
+> - **Upgrading:** use `pool.dialect` (not `pool.dialectInstance`), read `pool.dialect.dialectName`, and remove any top-level `dialect` from migrate config (see `CHANGELOG.md`).
+
+### Driver → pool → dialect class
+
+Migrations and the CLI read **`pool.dialect.dialectName`** only (there is no separate top-level `dialect` in config). Pick the pool (and thus the dialect class) for your driver.
+
+| Import path | Pool class | Dialect class (on the pool) | Notes |
+| :--- | :--- | :--- | :--- |
+| `uql-orm/postgres` | `PgQuerierPool` | `PgDialect` | Node `pg`; base Postgres capabilities (native arrays, `$n::jsonb`). |
+| `uql-orm/postgres` | — | `PostgresDialect` | Base Postgres AST; subclass for other Postgres wire drivers. |
+| `uql-orm/neon` | `NeonQuerierPool` | `NeonDialect` | `@neondatabase/serverless`; extends `PostgresDialect` with Neon-oriented defaults. |
+| `uql-orm/cockroachdb` | `CrdbQuerierPool` | `CockroachDialect` | `dialectName` is `cockroachdb`; extends `PostgresDialect`. |
+| `uql-orm/mysql` | `Mysql2QuerierPool` | `MySql2Dialect` | `mysql2` driver. |
+| `uql-orm/maria` | `MariadbQuerierPool` | `MariaDialect` | `mariadb` driver. |
+| `uql-orm/sqlite` | `Sqlite3QuerierPool` | `BetterSqlite3Dialect` | `better-sqlite3` (or Bun SQLite in that pool). |
+| `uql-orm/libsql` | `LibsqlQuerierPool` | `LibsqlDialect` | `@libsql/client`. |
+| `uql-orm/d1` | `D1QuerierPool` | `D1SqliteDialect` | Cloudflare D1. |
+| `uql-orm/mongo` | `MongodbQuerierPool` | `MongodbNativeDialect` | Official `mongodb` driver. |
+| `uql-orm/bunSql` | `BunSqlQuerierPool` | _inferred_ | Dialect class chosen from `SQL.Options` (Postgres / MySQL / MariaDB / SQLite / Cockroach). |
+
+For **wire-level** differences on PostgreSQL (e.g. Bun SQL vs `pg`), `BunSqlPostgresDialect` merges `POSTGRES_WIRE_DRIVER_CAPABILITIES` from `uql-orm/postgres/wireCapabilities` (array literals + JSON text cast); `PgDialect` does not.
 
 ## 4. Manipulate the Data