Skip to content

Changelog

Changelog

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

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.15.3] - 2026-04-11

Changed

  • Queue dequeue polling now keeps polling state in the caller lifecycle, with builder-style until(...).poll(&store) semantics and bounded wait coverage
  • pgqrs distribution is now library-only; the CLI binary, CLI docs, and CLI-only test paths were removed
  • Worker lifecycle internals were simplified by flattening admin/producer/consumer worker ownership and sharing worker-table transition logic across SQL backends
  • S3 test harnesses now require explicit cache directories so tests avoid accidental cache sharing and clean up their temporary state more reliably

Fixed

  • Polling lifecycle transitions now preserve ready, timeout, heartbeat, and interrupt behavior more consistently in one-shot and looped dequeue paths

[0.15.2] - 2026-04-07

Changed

  • Worker identity now uses explicit worker names across Rust, Python, migrations, and examples instead of host/port fields
  • S3 sync-state handling now distinguishes remote-only, local-only, and concurrent changes so durable reads choose snapshot vs sync more predictably
  • S3 process isolation support now scopes cache usage through config and adds process-isolated helper/test coverage for distinct local caches

Fixed

  • Stale dirty S3 durable reads now surface a conflict instead of silently serving out-of-date local state
  • Backend test defaults and S3 cache-id reuse were corrected so backend suites keep passing after the worker identity migration

[0.15.1] - 2026-04-02

Added

  • Python S3 store handles via as_s3(store) with backend-specific sync() and snapshot() support for durable S3 queues
  • S3 benchmark harness support with LocalStack/Toxiproxy bring-up, an S3 smoke path, and dashboard/docs coverage for backend-level results

Changed

  • Python binding enums now expose workflow and status values as typed Python enums instead of raw strings
  • SQLite/Turso SQL dialect handling now routes queue, workflow, and run queries through the shared dialect layer for more consistent backend behavior

Fixed

  • Python S3 casting now fails fast for non-S3 stores instead of exposing unsupported durability operations
  • Shared benchmark executor builds remain compatible with non-S3 backends while S3 benchmark support is enabled

[0.15.0] - 2026-03-23

Added

  • S3-backed queue storage via s3://bucket/key.sqlite DSNs for queue state backed by object storage
  • Explicit S3 durability lifecycle with bootstrap(), snapshot(), sync(), sync-state reporting, and Local vs Durable write modes
  • Benchmark documentation with curated queue benchmark baselines and scenario writeups for backend behavior

Changed

  • Python configuration parity by exposing validation and enqueue rate-limit settings on py-pgqrs.Config
  • Dequeue builder behavior aligned across Rust and Python with a simpler poller model and consistent worker stop handling
  • Admin behavior across backends unified so PostgreSQL, SQLite, Turso, and S3 share the same store-backed admin paths
  • SQLite/Turso locking internals refactored to serialize table access more reliably under contention

Fixed

  • Single-message dequeue handlers now validate batch_size == 1 instead of relying on hidden coercion
  • Polling loops now exit cleanly for interrupted and suspended workers instead of diverging by backend or wrapper path

[0.14.0] - 2026-02-21

Added

  • Trigger/Worker workflow architecture with explicit workflow definitions and run execution
  • Durable workflow lifecycle states including QUEUED, PAUSED, SUCCESS, and ERROR
  • Retry and pause semantics using message visibility for backoff and external-event pauses
  • Result retrieval APIs for non-blocking status checks and blocking waits
  • Python workflow bindings aligned with the new workflow API patterns

Changed

  • BREAKING: Replaced legacy workflow creation APIs with workflow("name").trigger(...).execute(...)
  • BREAKING: Replaced global step acquisition with ctx.step("id", ...) on workflow handlers
  • BREAKING: Updated workflow schema to use definitions, runs, and step state tables
  • Workflow documentation overhaul to match the v0.14 API and retry model
  • Worker registration now uses fluent consumer().handler().create() pattern

Fixed

  • Turso retry safety to prevent duplicate inserts during visibility-based retries
  • Runtime initialization safety to avoid panic in Python bindings

[0.13.0] - 2025-10-03

Added

  • Turso backend support for distributed SQLite deployments
  • Turso backend documentation covering configuration and usage

Changed

  • Test infrastructure refactor to support Turso integration

[0.12.0] - 2025-12-XX

Changed

  • BREAKING: Migrated from a two-table model (pgqrs_messages + pgqrs_archive) to a single-table model. All messages now reside in the pgqrs_messages table.
  • BREAKING: Removed ArchivedMessage and ArchiveTable types. Use QueueMessage and MessageTable::list_archived_by_queue instead.
  • Updated dequeued_at semantics to represent the first lease time only (never reset).
  • Optimized indexes for the "hot set" (archived_at IS NULL) across all backends.

[0.11.0] - 2025-12-19

Added (Python Bindings)

  • Advanced Configuration: Fine-tune specific settings like lock_time, batch_size, pre_fetch count, and heartbeat_interval via the Config object.
  • Archive Access: Read-only access to archived messages via get_archived_message and list_archived_messages.
  • Worker API: Python API for registering and managing workers.
  • Delayed Messages: Support for enqueue_delayed to schedule messages for future delivery.
  • Batch Operations: Efficient batch enqueue and delete operations.
  • Message Ownership: Enforced ownership checks when deleting or updating messages.
  • Visibility Extension: Ability to extend the visibility timeout of a message using extend_visibility.

Documentation

  • Comprehensive Guide: Added a complete User Guide with "Getting Started", "Concepts", and API references for both Rust and Python.
  • Catppuccin Theme: Updated documentation theme to Catppuccin with multiple color schemes.

Chores

  • Build System: Improved build system and dependency management.
  • Release Workflow: Added manual release workflow for coordinated Rust and Python releases.

[0.4.0] - 2025-12-12

Added

  • Workspace Support: Restructured the repository into a Cargo workspace to support multiple crates (core library and python bindings).
  • Python Bindings: Initial structure for py-pgqrs crate.
  • Zombie Reclamation: Admin command to recover tasks from dead workers.
  • Worker Health: New health check command for workers.
  • System Stats: Admin command to view global system statistics.
  • Queue Metrics: CLI command to view metrics for specific queues.

Changed

  • Renaming: Removed Pgqrs prefix from internal struct names for cleaner API.

[0.3.0] - 2024-11-13

Added

  • Producer/Consumer architectural separation for cleaner role-based interfaces
  • Unified Table trait with consistent CRUD operations across all database tables
  • Count methods (count() and count_by_fk()) for efficient metrics collection
  • Full Table trait implementation for pgqrs_archive with CRUD operations
  • Enhanced schema design with proper foreign key relationships
  • Comprehensive documentation updates and examples

Changed

  • BREAKING: Replaced monolithic Queue struct with dedicated Producer and Consumer structs
  • Refactored database architecture to four-table design (queues, workers, messages, archive)
  • Updated workers table to use queue_id instead of queue_name for better normalization
  • Enhanced error handling and validation across all interfaces

Improved

  • Type-safe Producer for message creation with validation and rate limiting
  • Optimized Consumer for message processing with batch operations
  • Updated README.md with detailed architecture diagrams and usage examples
  • Enhanced CLAUDE.md with current code structure and development patterns

Fixed

  • All existing tests updated and passing with new architecture
  • Integration tests for Producer/Consumer workflows
  • Documentation consistency across all modules

[0.2.0] - Previous Release

Added

  • Initial PostgreSQL-backed job queue implementation
  • Basic queue operations and message handling
  • CLI interface for queue management
  • Schema installation and verification
  • Worker registration and tracking

[0.1.0] - Initial Release

Added

  • Basic project structure
  • Core queue functionality
  • PostgreSQL integration