EDRI documentation

Event Driven Routing Infrastructure — a Python framework for building event-driven, parallel, and distributed applications.

EDRI is an application framework for Python that combines event-driven architecture, multiprocessing, and ASGI networking. It is designed for applications that need to route events between components running in separate processes, optionally distributed across machines.

When to use EDRI

• Event-driven applications with multiple processing stages

• Applications that distribute work across processes

• Systems requiring real-time communication between components (WebSocket, HTTP)

• Projects that need a structured event model with request/response patterns

• Distributed event routing across machines via TCP

When not to use EDRI

• Simple CRUD APIs — use FastAPI, Flask, or Django REST Framework

• Pub/Sub at scale — use Kafka, RabbitMQ, or Redis Streams

• Synchronous request-response only — a lightweight ASGI framework is sufficient

• Embedded or resource-constrained environments — EDRI's process model is designed for server-class deployments

Quick facts

Property	Detail
Language	Python 3.12+
Networking	ASGI (uvicorn, hypercorn, daphne)
Distribution	TCP Switch with pickle protocol
IPC	multiprocessing Queues, SharedMemoryPipe
Event model	Dataclass-based events with type-safe request/response
License	MIT

Getting Started

• Getting Started
• Architecture Overview

Core

• Event
• Response
• Directives
• Router
• Manager
• AsyncManagerBase
• Worker
• Worker types
• Worker lifecycle
• Communication model
• State and isolation
• SharedMemoryPipe in Workers
• Error handling
• Generic typing
• When NOT to use a Worker
• Minimal example
• Specialized managers
• SharedMemoryPipe

API

• API
• Broker
• API Event
• Downloading files
• Content Negotiation
• HeaderKit
• HTML API
• Static files
• WebSocket
• gRPC
• Validation & Injection
• Coercion
• Middleware
• Uploads
• JSON encoding

Distribution

• Connector
• Switch

Integrations

• Integrations
• SQLAlchemy (edri.integrations.sqlalchemy)
• Pydantic AI (edri.integrations.pydantic_ai)
• Writing your own integration

Operations

• Environment Variables
• Health checking
• Performance & scaling
• Deployment
• Testing
• Troubleshooting
• Watcher (Hot Reload)
• Utility

Project

• Changelog
• Versioned documentation
• EDRI
• Public API
• Contributing

Examples

• Examples
• REST CRUD with validation
• HTML form: GET the form, POST it, redirect on success
• Offloading CPU-bound work to a WorkerProcess
• SharedMemoryPipe: streaming a ZIP of a folder
• Streaming an upload straight into MinIO / S3
• Scatter–gather (fan-out / fan-in) with resolve_callback_worker
• Periodic refresh into the Key-Value Store
• Pydantic AI: a support agent with tools that dispatch to a different Manager
• Asynchronous File Processing with WebSocket Feedback (Ticket Pattern)
• Middleware: Token authentication end-to-end
• Middleware: deferred (async) authentication with replace_event + redirect_event
• Middleware: per-stage latency as a Server-Timing header

API Reference

• edri package