RetryDelays 0.1.0

RetryDelays is a lightweight .NET library for generating retry delay strategies.
Simplifies retry logic by removing the need for manual delay calculations - just use the appropriate RetryDelay or RetryDelayOptions. Supports common retry patterns such as constant, linear, exponential, and time-series delays, including jitter and decorrelated jitter.

🚀 Quick start

Create a retry loop and let RetryDelay calculate delays between attempts.

var delay = ExponentialRetryDelay.Create(
    baseDelay: TimeSpan.FromMilliseconds(500),
    maxDelay: TimeSpan.FromSeconds(10),
    useJitter: true);

int attempt = 0;

while (attempt <= 10)
{
    try
    {
        var result = func();
        break;
    }
    catch (Exception)
    {
        attempt++;

        var nextDelay = delay.GetDelay(attempt);
        await Task.Delay(nextDelay, token);
    }
}

⭐ Key Features

• ⚙️ Supports multiple retry delay strategies - constant, linear, exponential, and time series
• 🎲 Built-in jitter support
• 📈 Decorrelated jitter implementation for exponential backoff, adapted from the Polly
• 🧩 Flexible configuration via RetryDelayOptions
• 🔄 Implicit conversion from RetryDelayOptions to RetryDelay
• 🧱 Ability to create custom delay strategies using Func<int, TimeSpan>

💡 Key Concepts

• RetryDelay: The base class responsible for calculating the exact TimeSpan to wait before the next attempt.
• RetryDelayOptions: Configuration objects (like LinearRetryDelayOptions or ConstantRetryDelayOptions) used to define the rules of your delay strategy.
• Attempts: Current retry attempt (an integer count starting from 0).

✨ Available delay types

All types respect the optional MaxDelay limit.

🎲 Jitter Strategies

Standard Jitter - Applies ±25% randomization to calculated delays to prevent synchronized retries across distributed systems.

Decorrelated Jitter - Available for exponential delays. Uses an advanced formula that produces smoother, more evenly distributed delays with fewer spikes than standard jitter. Based on techniques from Polly.

🛠️ Passing options where a RetryDelay is expected

Thanks to implicit conversion, you can write methods that accept a RetryDelay but pass configuration options directly:

// Method that accepts a RetryDelay
async Task<T> ExecuteWithRetry<T>(
    Func<Task<T>> action,
    RetryDelay retryDelay,
    CancellationToken cancellationToken = default)
{
    int attempt = 0;
    while (true)
    {
        try
        {
            return await action();
        }
        catch
        {
            attempt++;
            var delay = retryDelay.GetDelay(attempt);
            await Task.Delay(delay, cancellationToken);
        }
    }
}

// Call the method with an options object - it is implicitly converted to the corresponding RetryDelay
var options = new ExponentialRetryDelayOptions
{
    BaseDelay = TimeSpan.FromSeconds(1),
    ExponentialFactor = 2.0,
    UseJitter = true,
    MaxDelay = TimeSpan.FromSeconds(30)
};

var result = await ExecuteWithRetry(() => FetchDataAsync(), options);

The options are automatically converted to an ExponentialRetryDelay instance, so the method receives a fully functional delay calculator.

💡 Custom Delays

The RetryDelay base class is highly flexible. You can even create a delay strategy directly from a simple function:

Note: The RetryDelay base class can be instantiated from a Func<int, TimeSpan>, allowing for completely custom logic without sub-classing.

// Custom logic: delay is always (attempt * 2) seconds, but never more than 1 minute
RetryDelay customDelay = new RetryDelay(attempt => 
    TimeSpan.FromSeconds(Math.Min((attempt + 1) * 2, 60))
);

📌 When to use RetryDelays

RetryDelays focuses solely on delay calculation, allowing you to integrate it with any retry implementation or resilience framework.

🧪 Sample

The samples directory contains an ASP.NET Core Minimal API that demonstrates library usage with full configuration support.

Key highlights:

• IRetryConfiguration — interface that groups four named delay options (DelayOption1–DelayOption4, covering Constant, Exponential, Linear, and TimeSeries strategies)
• RetryConfiguration — concrete implementation bound from a single RetryConfiguration section in appsettings.json
• RetryDelayAnalysisService — service that accepts IRetryConfiguration and produces a per-attempt delay schedule and total wait time for all four strategies
• GET /retry-analysis?attempts=N — endpoint that runs the analysis and returns the full schedule as JSON