Skip to content

CHANGELOG.md

Latest commit

 

History

History
114 lines (67 loc) · 3.87 KB

CHANGELOG.md

File metadata and controls

114 lines (67 loc) · 3.87 KB
Raw

Changelog

All notable changes to this project will be documented in this file.

[0.1.3] - 2026-06-06

Added

  • PostgreSQL dialect — full DDL support via -d postgresql (CLI), dialect.set("postgresql") (Gradle plugin), or PostgreSqlDialect (programmatic)
    • Double-quoted identifiers
    • BIGSERIAL / SERIAL / SMALLSERIAL for auto-increment columns
    • Native SEQUENCE support (@GeneratedValue(strategy = SEQUENCE))
    • Type mappings: BOOLEAN, uuid, BYTEA, DOUBLE PRECISION, REAL, NUMERIC(p,s), TIMESTAMP WITH TIME ZONE, TEXT (unknown-type fallback)
    • Primary key constraint drop uses {table}_pkey naming convention
  • FK dependency-based table ordering (DependencyResolver) — CREATE TABLE and DROP TABLE statements are now automatically ordered using a topological sort over foreign key relationships; no manual ordering required regardless of entity declaration order
    • Supported topologies: linear chains, diamond shapes, join tables, multiple independent trees
    • noConstraint = true relationships are excluded from the dependency graph
    • Circular FK detection: logs a warning to stderr and falls back to the original order safely
  • Renamed table handling — tables detected as renamed are now generated as DROP TABLE + CREATE TABLE instead of RENAME TABLE, consistent with the intended production behaviour

Fixed

  • toLowerCase() in DependencyResolver now uses Locale.ROOT to prevent incorrect FK matching under locale-sensitive JVM environments (e.g. Turkish locale I → ı)
  • BFS queue initialisation in DependencyResolver now uses LinkedHashMap insertion order for deterministic output when no FK dependency exists between independent tables

[0.1.2] - 2026-03-09

Added

  • naming.strategy can now be configured via jinx.yaml (previously only supported through Gradle DSL or -A compiler options)

Changed

  • All annotation processor option parsing (maxLength, strategy) is now consolidated in JpaSqlGeneratorProcessor.init(). ProcessingContext no longer reads from jinx.yaml directly.

[0.1.1] - 2026-03-08

Fixed

  • Fixed SQL generation order bug on column rename: indexes referencing the renamed column were generated before the column existed
  • Resolved ModifyContributor dependency ordering issue in MySQL migrations by splitting modify operations into separate Drop and Add contributors

Changed

  • Index, constraint, and foreign key modify operations now follow explicit Drop → Column change → Add ordering

[0.1.0] - 2025-03-07

Fixed

  • Fixed duplicate FK generation in JOINED inheritance hierarchies
  • Fixed FK mismatch in multi-level JOINED inheritance
  • Fixed AttributeDescriptorFactory incorrectly recursing into @Entity superclasses
  • Fixed TEXT type being generated for AttributeConverter output types — now correctly resolves the converter's DB-side type (Y in AttributeConverter<X, Y>)
  • Fixed deferred PK validation timing for JOINED hierarchies and @ElementCollection

[0.0.22] - 2025-12-03

Fixed

  • Fixed constraint and index position bug in DDL generation
  • Fixed duplicate column generation for boolean fields

Changed

  • Deprecated IndexModifyContributor, ConstraintModifyContributor, RelationshipModifyContributor

[0.0.17]

Added

  • NamingStrategy SPI — SNAKE_CASE and NO_OP strategies included

[0.0.14]

Changed

  • Deferred FK processing now uses dynamic retry count with progress detection to handle complex circular references

[0.0.13]

Added

  • FK post-create phase separation for circular reference support

Fixed

  • OneToOne missing UNIQUE constraint on inverse side
  • ToOne FK missing when processed before referenced entity

[0.0.9]

Fixed

  • MySQL ENUM literal generation for EnumType.STRING
  • Primitive and enum field SQL type mapping errors