v0.11.1

Bug Fixes

• Fix round-trip parsing for baseline tests
  • Strip quotes from parsed identifiers (trigger names, policy names, OWNED BY parts)
  • Normalize function return types to lowercase
  • Normalize view queries and function bodies whitespace
  • Normalize column defaults to lowercase cast types
  • Add support for parsing ALTER TABLE ADD CONSTRAINT for foreign keys
  • Fix function key format in introspection to use qualified names
  • Handle circular foreign key dependencies in topological sort

Internal

• Add normalize_sql_whitespace utility function
• Fix baseline tests using multi-statement SQL with sqlx::raw_sql

v0.11.0

What's New

PARTITION OF Support

• Full support for PostgreSQL table partitioning with PARTITION OF syntax
• Parse partition tables into schema.partitions with bound types:
  • FOR VALUES FROM ... TO ... (Range)
  • FOR VALUES IN (...) (List)
  • FOR VALUES WITH (MODULUS ..., REMAINDER ...) (Hash)
  • DEFAULT
• Generate CREATE TABLE ... PARTITION OF ... FOR VALUES ... SQL
• Proper dependency ordering (partitions created after parent tables)

sqlparser Upgrade

• Upgraded from sqlparser 0.52 to forked 0.60 with PARTITION OF parsing support
• Uses fmguerreiro/datafusion-sqlparser-rs partition-of-support branch

Note

This release uses a forked sqlparser and cannot be published to crates.io. Install from source:

cargo install --git https://github.com/fmguerreiro/pgmold

v0.10.0

What's New

Partitioned Tables Support (Phase 1)

• Parser: Parse PARTITION BY RANGE/LIST/HASH from SQL files
• Introspection: Extract partition info from pg_partitioned_table and pg_inherits catalogs
• SQL Generation: Generate PARTITION BY clause in CREATE TABLE statements
• Model: New types PartitionStrategy, PartitionKey, PartitionBound, Partition

Limitations

• PARTITION OF syntax (child partition creation) not yet supported in parser (sqlparser 0.52 limitation)
• Tracked in pgmold-32

Full Changelog

v0.9.0...v0.10.0

v0.9.0

• Add baseline command module for adopting existing databases
• Includes unsupported object detection (materialized views, domains, partitioned tables, etc.)
• Round-trip verification and zero-diff guarantees
• Text and JSON report formats

v0.8.0

• Add REFERENCING transition tables support for triggers (OLD TABLE, NEW TABLE)

v0.7.0

• Add trigger ENABLE/DISABLE state support (Origin, Disabled, Replica, Always)
• Parser handles ALTER TABLE ... ENABLE/DISABLE/ENABLE REPLICA/ENABLE ALWAYS TRIGGER
• Introspection reads tgenabled from pg_trigger catalog
• Diff emits AlterTriggerEnabled when only enabled state differs
• SQL generation for trigger enable/disable statements

v0.6.0

Breaking Changes

• Renamed Trigger.table_schema → target_schema
• Renamed Trigger.table → target_name
• Renamed MigrationOp::DropTrigger fields similarly

This fixes semantic accuracy: INSTEAD OF triggers target views (not tables), so "target" is more appropriate than "table".

Upgrading

Update any code that:

• Constructs Trigger structs directly
• Matches on MigrationOp::DropTrigger variants
• Accesses trigger.table or trigger.table_schema fields

v0.5.0

What's New

Schema Dump Command

New pgmold dump command to export database schemas to SQL DDL files.

# Dump schema to stdout
pgmold dump --database db:postgres://localhost/mydb

# Dump to file with specific schemas
pgmold dump --database db:postgres://localhost/mydb --target-schemas auth,api,public --output schema.sql

Features:

• Exports all schema objects (tables, indexes, foreign keys, enums, sequences, functions, views, triggers, policies)
• Dependency-ordered output (tables before foreign keys, etc.)
• Multi-schema support
• Optional header with metadata

Installation

cargo install pgmold

v0.4.0

Features

• SERIAL column support: Automatically expand SERIAL, BIGSERIAL, and SMALLSERIAL columns to their underlying integer type + sequence
• Sequence support: Full support for CREATE SEQUENCE including parsing, introspection, diff, and SQL generation
• Sequence lint rules: Added warnings for destructive sequence operations (DROP SEQUENCE, RESTART)

Improvements

• Stricter error handling: Removed silent fallbacks in favor of explicit errors for:
  • Unknown referential action codes from PostgreSQL
  • Unknown sequence data types
  • Missing function return types or bodies
  • Invalid UTF-8 paths
  • Missing tables during RLS introspection

Testing

• Added integration tests for sequence roundtrip and alterations
• Added parser tests for SERIAL variants with schema qualifiers

v0.3.1

• Fix function body comparison producing false diffs due to dollar-quote delimiters.