Changelog

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

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

[0.1.1] - 2026-03-16

Web Dashboard: Starlette-based real-time monitoring UI with dark theme
- Task list with state/queue/search filtering
- Task detail modal with result, error, traceback
- Cancel/revoke actions
- Auto-refresh (5s)
- CLI: flashq dashboard myapp:app --port 5555
Canvas Primitives: Celery-inspired task composition
- Signature: task reference with args/kwargs/options
- Chain: sequential execution, result piped to next
- Group: parallel execution
- Chord: group + callback when all complete
- .s() and .si() (immutable) shorthand on Task
Rate Limiting: Token bucket algorithm
- Per-task and global rate limits
- RateLimiter middleware
- Rate format: "100/m", "60/h", "10/s"
Dead Letter Queue: Permanent failure management
- DeadLetterQueue with add, list, replay, replay_all, purge
- _DLQMiddleware for automatic capture
Benchmark Suite: Throughput and latency measurements
- Enqueue throughput: ~19,000 tasks/s
- Concurrency scaling tests
- CPU-bound and I/O-bound workloads
20 production hardening edge case tests
Examples: basic, advanced (canvas + middleware + DLQ), FastAPI integration

Core Engine: FlashQ app class with @app.task() decorator
Task System: Type-safe .delay() and .apply() with PEP-612 ParamSpec
SQLite Backend: WAL journal mode, BEGIN IMMEDIATE for atomic dequeue, zero dependencies
PostgreSQL Backend: LISTEN/NOTIFY for real-time delivery, FOR UPDATE SKIP LOCKED
Redis Backend: Lists for queues, sorted sets for scheduling, Lua scripts for atomicity
Worker: Thread pool executor, async+sync task execution, graceful shutdown (SIGINT/SIGTERM)
Task Timeouts: Non-blocking ThreadPoolExecutor with shutdown(wait=False)
Middleware System: before_execute, after_execute, on_error, on_retry, on_dead
- Built-in: LoggingMiddleware, TimeoutMiddleware
Periodic Scheduler: every(hours=6) and cron("0 9 * * 1-5") syntax
Retry System: Exponential backoff, configurable max retries and delay
CLI: flashq worker, flashq info, flashq purge
Serializers: JSON (default) and Pickle
Type Safety: py.typed marker, full type annotations
CI/CD: GitHub Actions (Python 3.10–3.13, ubuntu/macos/windows) + PyPI publish on tag
168 core tests at ~95% coverage