Skip to content

CHANGELOG.md

Latest commit

 

History

History
56 lines (39 loc) · 3.48 KB

CHANGELOG.md

File metadata and controls

56 lines (39 loc) · 3.48 KB
Raw

3.0.2

  • Add pgtyped-rescript diagnose for inspecting named SQL queries with describe, explain, and explain analyze, including support for parameter JSON, parameter files, processed SQL output, query listing, JSON output, and statement timeouts.
  • Reduce default CLI generation output to only recompiled files, errors, and concise unchanged-run summaries; pass --verbose to restore detailed per-file processing and skipped-file output.

3.0.1

  • Gracefully skip optional AST-based query analysis when pgsql-ast-parser cannot parse valid PostgreSQL syntax, so generation can continue with database-derived types.

3.0.0

  • BREAKING: Require ReScript >=12.0.0.
  • BREAKING: Remove the dependency on @rescript/core.
  • BREAKING: Remove the dependency on @rescript/tools; the CLI now uses the ReScript compiler toolchain from the installed rescript package.
  • Fix Pg.Pool.make(ConnectionString(...)) so it creates a pool with {connectionString} instead of passing a raw string to pg.Pool.
  • Fix Pg.Client.release binding to return unit, matching node-postgres.
  • Use chokidar for build-mode file matching so glob-style srcDir patterns are handled consistently with watch mode.
  • Make sure JSON.t is properly stringified so JSON arrays can be passed to params expecting JSON.t without them being confused for regular Postgres arrays.
  • Auto-escape all ReScript keywords in generated record field names.
  • Add automatic parsing of PostgreSQL check constraints to generate ReScript polyvariant types for enumeration-style constraints. Supports both column IN (value1, value2, ...) and column = ANY (ARRAY[value1, value2, ...]) patterns with string and integer values.
  • Add top-level literal inference for SELECT queries. When a query returns literal values with aliases (e.g. SELECT 'success' as status, 42 as code), PgTyped now automatically infers specific polyvariant types like [#"success"] and [#42] instead of generic string and int types. This provides better type safety and autocompletion. Also works with UNION queries where literals are consistent across all branches.
  • Add support for PostgreSQL JSON population functions (json_populate_record, json_populate_recordset, jsonb_populate_record, jsonb_populate_recordset, json_to_record, jsonb_to_recordset). This supports efficient bulk operations.
  • Use the queryConfig API in pgtyped-rescript-runtime.

2.6.0

  • Improve pg bindings.
  • Upgrade @rescript/core to 1.6.0.

2.5.0

  • Autoinsert trailing commas in embedded SQL blocks.
  • BREAKING CHANGE: Null.t is no longer emitted, all null values are autoconverted to option. This gives a much more idiomatic ReScript experience.
  • Emit actually runnable query in module comment for each query, instead of the original non-valid SQL query.
  • Relax requirement on providing query via @name comment.
  • Change expectOne to panic if not finding a single item.

2.4.0

  • Add mode for embedding %sql (with one, expectOne, many, and execute flavors) in ReScript directly.

2.3.1

  • Fix missing rescript.json in published package.

2.3.0

  • Fix type generation for arrays of types like JSON.t.
  • BREAKING: Up required ReScript version to >=11.1.0 and @rescript/core to >=1.3.0.
  • Proper bigint support.
  • Emit @gentype annotations for everything.
  • Add each query to its own ReScript module, and emit helpers many, one, expectOne and execute.

2.2.2

  • [CLI] bin is now named pgtyped-rescript so it won't clash with stock pgtyped.