Getting started

Installation

uv pip install redress
# or
pip install redress

Quick start (sync, canonical API)

from redress import Policy, Retry, default_classifier
from redress.strategies import decorrelated_jitter

policy = Policy(
    retry=Retry(
        classifier=default_classifier,
        strategy=decorrelated_jitter(max_s=5.0),
        max_attempts=5,
        deadline_s=30.0,
    ),
)

result = policy.call(lambda: do_work(), operation="sync_task")

Policy(retry=Retry(...)) is the primary API. It is the shape to use when you want a consistent model for retries, circuit breakers, shared hooks, and structured outcomes.

Quick start (async, canonical API)

import asyncio

from redress import AsyncPolicy, AsyncRetry, default_classifier
from redress.strategies import decorrelated_jitter

policy = AsyncPolicy(
    retry=AsyncRetry(
        classifier=default_classifier,
        strategy=decorrelated_jitter(max_s=5.0),
        max_attempts=5,
        deadline_s=30.0,
    ),
)

async def main() -> None:
    result = await policy.call(lambda: do_work_async(), operation="async_task")
    print(result)

asyncio.run(main())

Convenience shortcuts

from redress import RetryPolicy, default_classifier
from redress.strategies import decorrelated_jitter

policy = RetryPolicy(
    classifier=default_classifier,
    strategy=decorrelated_jitter(max_s=5.0),
)

result = policy.call(lambda: do_work(), operation="sync_task")

RetryPolicy and AsyncRetryPolicy are backward-compatible shortcuts for retry-only flows. They are sugar for Policy(retry=Retry(...)) and AsyncPolicy(retry=AsyncRetry(...)).

Decorator shortcut

from redress import retry

@retry
def fetch_user():
    ...

Use the decorator when you want the smallest amount of setup. Move to Policy(...) when you need circuit breakers, richer execution control, or shared configuration across multiple calls.