This Python SDK provides a framework for building distributed, scalable, and durable applications using the Temporal orchestration engine. It targets Python developers looking to implement complex, long-running business logic with resilience and fault tolerance.

How It Works

Workflows are defined as Python classes using async def and decorated with @workflow.defn. The SDK leverages a custom asyncio event loop to ensure deterministic execution, managing tasks, timers, and cancellations within a sandboxed environment. Activities, also defined with @activity.defn, can be synchronous or asynchronous and are executed by workers. The SDK handles serialization and deserialization of data through configurable data converters, with built-in support for JSON and Pydantic models.

Quick Start & Requirements

• Install: pip install temporalio
• Prerequisites: Python 3.9+, a running Temporal server (e.g., Temporalite).
• Setup: Minimal, typically involves running a worker and a client script.
• Docs: Application Development Guide, Python Code Samples, API Documentation

Highlighted Details

• Type-safe workflow and activity invocations with MyPy support.
• Flexible activity execution with support for async def, threaded, and multiprocess activities.
• Robust testing framework with automatic and manual time-skipping capabilities.
• Workflow sandbox for detecting non-deterministic code, with customization options.
• OpenTelemetry integration for metrics and tracing.

Maintenance & Community

The project is actively maintained by the Temporal team. Community support is available via Discord.

Licensing & Compatibility

MIT License. Compatible with commercial and closed-source applications.

Limitations & Caveats

The time-skipping test environment has known issues on ARM architectures. Nexus support is experimental and subject to change. Customizing the sandbox restrictions is an experimental API.