Mastering ReactiveUI.Extensions: A Comprehensive Guide to Async Reactive Programming in .NET

Author: Chris Pulman
Published: March 16, 2026

ReactiveUI.Extensions represents a significant evolution in reactive programming for .NET, providing a fully async-native observable framework that bridges the gap between traditional System.Reactive (Rx.NET) and modern async/await patterns. This library introduces IObservableAsync<T>, IObserverAsync<T>, and a comprehensive suite of operators that work seamlessly with ValueTask, CancellationToken, and IAsyncDisposable throughout the entire reactive pipeline.

This article provides an extensive, in-depth technical exploration of ReactiveUI.Extensions, covering every public function, advanced multi-threading scenarios, and practical patterns for building robust, scalable reactive applications.

---

Table of Contents

1. Introduction to ReactiveUI.Extensions
2. Understanding IObservableAsync vs IObservable
3. Understanding IObserverAsync vs IObserver
4. Core Async Interfaces and Types
5. Factory Methods and Observable Creation
6. Transformation Operators
7. Filtering Operators
8. Combining and Merging Operators
9. Error Handling and Resilience
10. Timing and Scheduling
11. Aggregation and Terminal Operators
12. Async Disposables and Resource Management
13. Subjects and Multicasting
14. Bridging Classic and Async Observables
15. Classic Reactive Extensions Operators
16. Advanced Multi-Threading Examples
17. Performance Considerations
18. Best Practices

---

1. Introduction to ReactiveUI.Extensions

Why ReactiveUI.Extensions?

Traditional System.Reactive (Rx.NET) was designed before the widespread adoption of async/await in C#. While Rx.NET provides excellent support for asynchronous data streams, it has several limitations in modern .NET development:

1. Synchronous Observer Callbacks: IObserver<T> methods (OnNext, OnError, OnCompleted) are synchronous, which can lead to thread pool starvation when performing async operations
2. No Built-in Cancellation: Traditional observables don't integrate naturally with CancellationToken
3. Synchronous Disposal: IDisposable doesn't support async cleanup operations
4. Task Integration Friction: Converting between Task, IAsyncEnumerable, and IObservable requires boilerplate

ReactiveUI.Extensions solves these problems by providing:

• End-to-End Async: Every notification (OnNextAsync, OnErrorResumeAsync, OnCompletedAsync) returns ValueTask
• Cancellation-First Design: Every operator accepts CancellationToken
• Async Disposal: Subscriptions return IAsyncDisposable for proper async resource cleanup
• Seamless Interop: Bidirectional bridging between IObservable and IObservableAsync
• Modern .NET Support: Requires .NET 8 or later, leveraging modern language features

Installation

dotnet add package ReactiveUI.Extensions

Supported Target Frameworks: .NET 4.6.2, .NET 4.7.2, .NET 4.8.1, .NET 8, .NET 9, .NET 10

---

2. Understanding IObservableAsync vs IObservable

IObservable (Traditional Rx.NET)

public interface IObservable<out T>
{
    IDisposable Subscribe(IObserver<T> observer);
}

Characteristics:

• Synchronous subscription
• Returns IDisposable for cleanup
• Observer callbacks are synchronous
• No built-in cancellation support
• Can block threads during async operations

IObservableAsync (ReactiveUI.Extensions)

public interface IObservableAsync<T>
{
    ValueTask<IAsyncDisposable> SubscribeAsync(
        IObserverAsync<T> observer, 
        CancellationToken cancellationToken);
}

Characteristics:

• Asynchronous subscription (ValueTask)
• Returns IAsyncDisposable for async cleanup
• Observer callbacks are asynchronous (ValueTask)
• Built-in CancellationToken support
• Non-blocking throughout the pipeline

Key Differences Table

Feature	IObservable	IObservableAsync
Subscription	Synchronous	Asynchronous (ValueTask)
Disposal	IDisposable	IAsyncDisposable
OnNext	void	ValueTask
OnError	void	ValueTask
OnCompleted	void	ValueTask
Cancellation	Manual	Built-in CancellationToken
Thread Blocking	Possible	Never
.NET Version	Any	.NET 8+

When to Use Each

Use IObservableAsync when:

• Building new async-first applications
• Working with I/O-bound operations (network, database, file system)
• Needing proper cancellation support
• Requiring async resource cleanup
• Targeting .NET 8 or later

Use IObservable when:

• Maintaining legacy code
• Working with CPU-bound operations
• Interoperating with existing Rx.NET libraries
• Targeting older .NET versions

---

3. Understanding IObserverAsync vs IObserver

IObserver (Traditional)

public interface IObserver<in T>
{
    void OnNext(T value);
    void OnError(Exception error);
    void OnCompleted();
}

Problems:

1. No Async Support: Cannot await async operations in callbacks
2. Exception Handling: Exceptions in callbacks can crash the application
3. Resource Cleanup: No async disposal mechanism
4. Backpressure: Difficult to implement async backpressure

IObserverAsync (ReactiveUI.Extensions)

public interface IObserverAsync<in T> : IAsyncDisposable
{
    ValueTask OnNextAsync(T value, CancellationToken cancellationToken);
    ValueTask OnErrorResumeAsync(Exception error, CancellationToken cancellationToken);
    ValueTask OnCompletedAsync(Result result);
}

Advantages:

1. Full Async Support: All callbacks are ValueTask-based
2. Error Recovery: OnErrorResumeAsync allows error recovery and continuation
3. Async Disposal: Implements IAsyncDisposable for proper cleanup
4. Cancellation: Every callback receives CancellationToken
5. Result Tracking: OnCompletedAsync receives Result indicating success/failure

Practical Example: Traditional vs Async Observer

// Traditional IObserver - Problem: Blocking async operations
public class TraditionalObserver : IObserver<string>
{
    public void OnNext(string value)
    {
        // BAD: Blocking async operation
        var result = SaveToDatabaseAsync(value).Result; // Thread pool starvation!
        Console.WriteLine($"Saved: {result}");
    }
    
    public void OnError(Exception error) => Console.WriteLine($"Error: {error}");
    public void OnCompleted() => Console.WriteLine("Completed");
}

// Async IObserverAsync - Solution: Non-blocking throughout
public class AsyncObserver : ObserverAsync<string>
{
    protected override async ValueTask OnNextAsyncCore(
        string value, 
        CancellationToken cancellationToken)
    {
        // GOOD: Non-blocking async operation
        var result = await SaveToDatabaseAsync(value, cancellationToken);
        Console.WriteLine($"Saved: {result}");
    }
    
    protected override async ValueTask OnErrorResumeAsyncCore(
        Exception error, 
        CancellationToken cancellationToken)
    {
        // Can attempt recovery
        await LogErrorAsync(error, cancellationToken);
        // Can choose to continue or stop
    }
    
    protected override async ValueTask OnCompletedAsyncCore(Result result)
    {
        await CleanupAsync();
        Console.WriteLine($"Completed: {result.IsSuccess}");
    }
    
    protected override ValueTask DisposeAsyncCore() => CleanupResourcesAsync();
}

Error Recovery with OnErrorResumeAsync

One of the most powerful features of IObserverAsync<T> is the ability to recover from errors:

public class ResilientObserver : ObserverAsync<int>
{
    private int _retryCount = 0;
    private const int MaxRetries = 3;
    
    protected override async ValueTask OnNextAsyncCore(
        int value, 
        CancellationToken cancellationToken)
    {
        try
        {
            await ProcessValueAsync(value, cancellationToken);
        }
        catch (Exception ex) when (ex is TimeoutException)
        {
            // Error will be sent to OnErrorResumeAsyncCore
            throw;
        }
    }
    
    protected override async ValueTask OnErrorResumeAsyncCore(
        Exception error, 
        CancellationToken cancellationToken)
    {
        if (_retryCount < MaxRetries)
        {
            _retryCount++;
            await Task.Delay(TimeSpan.FromSeconds(1), cancellationToken);
            // Continue processing - don't propagate error
            return;
        }
        
        // Max retries exceeded - log and continue
        await LogErrorAsync(error, cancellationToken);
    }
    
    protected override ValueTask OnCompletedAsyncCore(Result result)
    {
        Console.WriteLine($"Completed with {(result.IsSuccess ? "success" : "failure")}");
        return default;
    }
}

---

4. Core Async Interfaces and Types

IObservableAsync

public interface IObservableAsync<out T>
{
    ValueTask<IAsyncDisposable> SubscribeAsync(
        IObserverAsync<T> observer, 
        CancellationToken cancellationToken);
}

Purpose: Represents an asynchronous push-based notification provider.

Key Points:

• Subscription is asynchronous (ValueTask)
• Returns IAsyncDisposable for async cleanup
• Supports cancellation via CancellationToken
• Thread-safe for concurrent subscriptions

IObserverAsync

public interface IObserverAsync<in T> : IAsyncDisposable
{
    ValueTask OnNextAsync(T value, CancellationToken cancellationToken);
    ValueTask OnErrorResumeAsync(Exception error, CancellationToken cancellationToken);
    ValueTask OnCompletedAsync(Result result);
}

Purpose: Defines an asynchronous observer that receives notifications.

Key Points:

• All callbacks are asynchronous
• Implements IAsyncDisposable for cleanup
• OnErrorResumeAsync allows error recovery
• Result parameter indicates completion status

ObservableAsync (Abstract Base Class)

public abstract class ObservableAsync<T> : IObservableAsync<T>
{
    public async ValueTask<IAsyncDisposable> SubscribeAsync(
        IObserverAsync<T> observer, 
        CancellationToken cancellationToken)
    {
        var subscription = await SubscribeAsyncCore(observer, cancellationToken);
        return subscription;
    }
    
    protected abstract ValueTask<IAsyncDisposable> SubscribeAsyncCore(
        IObserverAsync<T> observer, 
        CancellationToken cancellationToken);
}

Purpose: Base class for creating custom async observables.

Example: Custom Async Observable

public class TickObservable : ObservableAsync<int>
{
    private readonly int _count;
    private readonly TimeSpan _interval;
    
    public TickObservable(int count, TimeSpan interval)
    {
        _count = count;
        _interval = interval;
    }
    
    protected override async ValueTask<IAsyncDisposable> SubscribeAsyncCore(
        IObserverAsync<int> observer, 
        CancellationToken cancellationToken)
    {
        var cts = CancellationTokenSource.CreateLinkedTokenSource(cancellationToken);
        
        _ = Task.Run(async () =>
        {
            try
            {
                for (int i = 0; i < _count && !cts.Token.IsCancellationRequested; i++)
                {
                    await observer.OnNextAsync(i, cts.Token);
                    await Task.Delay(_interval, cts.Token);
                }
                
                await observer.OnCompletedAsync(Result.Success);
            }
            catch (OperationCanceledException)
            {
                await observer.OnCompletedAsync(Result.Success);
            }
            catch (Exception ex)
            {
                await observer.OnErrorResumeAsync(ex, cts.Token);
            }
        }, cts.Token);
        
        return new CancellationDisposable(cts);
    }
}

// Usage
var ticks = new TickObservable(10, TimeSpan.FromMilliseconds(100));
await using var subscription = await ticks.SubscribeAsync(
    async (value, ct) => Console.WriteLine($"Tick: {value}"),
    CancellationToken.None);

ConnectableObservableAsync

public class ConnectableObservableAsync<T> : ObservableAsync<T>
{
    public ValueTask<IAsyncDisposable> ConnectAsync(CancellationToken cancellationToken);
}

Purpose: Extends ObservableAsync with deferred connection for multicasting.

Use Case: When you want to control when the source starts emitting (e.g., after multiple subscribers are ready).

Result

public readonly struct Result
{
    public bool IsSuccess { get; }
    public Exception? Exception { get; }
    
    public static Result Success { get; } = new(true, null);
    public static Result Failure(Exception exception) => new(false, exception);
}

Purpose: Carries completion status (success or failure) to OnCompletedAsync.

---

5. Factory Methods and Observable Creation

ObservableAsync.Create

Creates an observable from an async subscription delegate.

public static IObservableAsync<T> Create<T>(
    Func<IObserverAsync<T>, CancellationToken, ValueTask<IAsyncDisposable>> subscribeAsync)

Example:

var custom = ObservableAsync.Create<string>(async (observer, ct) =>
{
    await observer.OnNextAsync("Hello", ct);
    await observer.OnNextAsync("World", ct);
    await observer.OnCompletedAsync(Result.Success);
    return DisposableAsync.Empty;
});

await using var sub = await custom.SubscribeAsync(
    async (value, ct) => Console.WriteLine(value),
    CancellationToken.None);

ObservableAsync.CreateAsBackgroundJob

Runs the subscription logic as a background task.

public static IObservableAsync<T> CreateAsBackgroundJob<T>(
    Func<IObserverAsync<T>, CancellationToken, ValueTask> job,
    bool startSynchronously = false)

Example:

var background = ObservableAsync.CreateAsBackgroundJob<int>(async (observer, ct) =>
{
    for (int i = 0; i < 5; i++)
    {
        await observer.OnNextAsync(i, ct);
        await Task.Delay(100, ct);
    }
    await observer.OnCompletedAsync(Result.Success);
});

ObservableAsync.Return

Emits a single value and completes.

public static IObservableAsync<T> Return<T>(T value)

Example:

var single = ObservableAsync.Return(42);

ObservableAsync.Empty

Completes immediately without emitting.

public static IObservableAsync<T> Empty<T>()

ObservableAsync.Never

Never emits and never completes.

public static IObservableAsync<T> Never<T>()

ObservableAsync.Throw

Completes immediately with an error.

public static IObservableAsync<T> Throw<T>(Exception exception)

ObservableAsync.Range

Emits a sequence of integers.

public static IObservableAsync<long> Range(long start, long count)

Example:

var numbers = ObservableAsync.Range(1, 10); // 1, 2, 3, ..., 10

ObservableAsync.Interval

Emits incrementing values at regular intervals.

public static IObservableAsync<long> Interval(
    TimeSpan period,
    TimeProvider? timeProvider = null)

Example:

var timer = ObservableAsync.Interval(TimeSpan.FromSeconds(1));

ObservableAsync.Timer

Emits a value after a delay.

public static IObservableAsync<long> Timer(
    TimeSpan dueTime,
    TimeProvider? timeProvider = null)

Example:

var delayed = ObservableAsync.Timer(TimeSpan.FromSeconds(5));

ObservableAsync.Defer

Defers observable creation until subscription.

public static IObservableAsync<T> Defer<T>(
    Func<IObservableAsync<T>> factory)

Example:

var deferred = ObservableAsync.Defer(() => 
    ObservableAsync.Return(DateTime.Now.Second));

ObservableAsync.FromAsync

Wraps an async function as an observable.

public static IObservableAsync<T> FromAsync<T>(
    Func<CancellationToken, Task<T>> factory)

Example:

var fromTask = ObservableAsync.FromAsync(async ct =>
{
    await Task.Delay(100, ct);
    return 42;
});

Conversion Extensions

// Task to IObservableAsync
public static IObservableAsync<T> ToObservableAsync<T>(this Task<T> task)

// IAsyncEnumerable to IObservableAsync
public static IObservableAsync<T> ToObservableAsync<T>(
    this IAsyncEnumerable<T> asyncEnumerable)

// IEnumerable to IObservableAsync
public static IObservableAsync<T> ToObservableAsync<T>(
    this IEnumerable<T> enumerable)

// IObservable to IObservableAsync (Bridge)
public static IObservableAsync<T> ToObservableAsync<T>(
    this IObservable<T> observable)

// IObservableAsync to IObservable (Bridge)
public static IObservable<T> ToObservable<T>(
    this IObservableAsync<T> asyncObservable)

Example:

// From Task
var task = Task.FromResult(42);
var obs = task.ToObservableAsync();

// From IAsyncEnumerable
async IAsyncEnumerable<int> GenerateAsync()
{
    for (int i = 0; i < 5; i++)
    {
        await Task.Delay(50);
        yield return i;
    }
}

var fromAsyncEnum = GenerateAsync().ToObservableAsync();

// Bridge from classic IObservable
var classic = Observable.Interval(TimeSpan.FromMilliseconds(200)).Take(5);
var asyncVersion = classic.ToObservableAsync();

---

6. Transformation Operators

Select (Sync)

Projects each element using a synchronous function.

public static IObservableAsync<TResult> Select<TSource, TResult>(
    this IObservableAsync<TSource> source,
    Func<TSource, TResult> selector)

Example:

var source = ObservableAsync.Range(1, 5);
var doubled = source.Select(x => x * 2); // 2, 4, 6, 8, 10

Select (Async)

Projects each element using an async function.

public static IObservableAsync<TResult> Select<TSource, TResult>(
    this IObservableAsync<TSource> source,
    Func<TSource, CancellationToken, ValueTask<TResult>> asyncSelector)

Example:

var projected = source.Select(async (x, ct) =>
{
    await Task.Delay(10, ct);
    return x.ToString();
});

SelectMany

Flattens nested async observables.

public static IObservableAsync<TResult> SelectMany<TSource, TResult>(
    this IObservableAsync<TSource> source,
    Func<TSource, IObservableAsync<TResult>> selector)

Example:

var flattened = source.SelectMany(x => 
    ObservableAsync.Range(x * 10, 3));
// For x=1: 10, 11, 12
// For x=2: 20, 21, 22
// etc.

Scan (Sync)

Applies an accumulator over the sequence.

public static IObservableAsync<TAccumulate> Scan<TSource, TAccumulate>(
    this IObservableAsync<TSource> source,
    TAccumulate seed,
    Func<TAccumulate, TSource, TAccumulate> accumulator)

Example:

var runningTotal = source.Scan(0, (acc, x) => acc + x);
// 1, 3, 6, 10, 15

Scan (Async)

Async accumulator.

public static IObservableAsync<TAccumulate> Scan<TSource, TAccumulate>(
    this IObservableAsync<TSource> source,
    TAccumulate seed,
    Func<TAccumulate, TSource, CancellationToken, ValueTask<TAccumulate>> asyncAccumulator)

Cast

Casts each element to the specified type.

public static IObservableAsync<TResult> Cast<TSource, TResult>(
    this IObservableAsync<TSource> source)

OfType

Filters elements assignable to the specified type.

public static IObservableAsync<TResult> OfType<TSource, TResult>(
    this IObservableAsync<TSource> source)

GroupBy

Groups elements by key.

public static IObservableAsync<GroupedAsyncObservable<TKey, TSource>> GroupBy<TSource, TKey>(
    this IObservableAsync<TSource> source,
    Func<TSource, TKey> keySelector)

Example:

var grouped = source.GroupBy(x => x % 2 == 0 ? "even" : "odd");

---

7. Filtering Operators

Where (Sync)

Filters elements using a synchronous predicate.

public static IObservableAsync<T> Where<T>(
    this IObservableAsync<T> source,
    Func<T, bool> predicate)

Example:

var evens = source.Where(x => x % 2 == 0);

Where (Async)

Filters using an async predicate.

public static IObservableAsync<T> Where<T>(
    this IObservableAsync<T> source,
    Func<T, CancellationToken, ValueTask<bool>> asyncPredicate)

Example:

var asyncFiltered = source.Where(async (x, ct) =>
{
    await Task.Delay(1, ct);
    return x > 5;
});

Take

Takes the first N elements.

public static IObservableAsync<T> Take<T>(
    this IObservableAsync<T> source,
    int count)

Skip

Skips the first N elements.

public static IObservableAsync<T> Skip<T>(
    this IObservableAsync<T> source,
    int count)

TakeWhile

Takes elements while predicate holds.

public static IObservableAsync<T> TakeWhile<T>(
    this IObservableAsync<T> source,
    Func<T, bool> predicate)

SkipWhile

Skips elements while predicate holds.

public static IObservableAsync<T> SkipWhile<T>(
    this IObservableAsync<T> source,
    Func<T, bool> predicate)

TakeUntil (Multiple Overloads)

Takes elements until a signal.

// Until another observable
public static IObservableAsync<T> TakeUntil<T>(
    this IObservableAsync<T> source,
    IObservableAsync<Unit> other)

// Until a task
public static IObservableAsync<T> TakeUntil<T>(
    this IObservableAsync<T> source,
    Task other)

// Until cancellation token
public static IObservableAsync<T> TakeUntil<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

// Until predicate
public static IObservableAsync<T> TakeUntil<T>(
    this IObservableAsync<T> source,
    Func<T, bool> predicate)

Example:

using var cts = new CancellationTokenSource(TimeSpan.FromSeconds(5));
var bounded = ObservableAsync.Interval(TimeSpan.FromMilliseconds(100))
    .TakeUntil(cts.Token);

Distinct

Emits only unique elements.

public static IObservableAsync<T> Distinct<T>(
    this IObservableAsync<T> source)

DistinctBy

Emits only elements with unique keys.

public static IObservableAsync<T> DistinctBy<T, TKey>(
    this IObservableAsync<T> source,
    Func<T, TKey> keySelector)

DistinctUntilChanged

Suppresses consecutive duplicates.

public static IObservableAsync<T> DistinctUntilChanged<T>(
    this IObservableAsync<T> source)

DistinctUntilChangedBy

Suppresses consecutive elements with same key.

public static IObservableAsync<T> DistinctUntilChangedBy<T, TKey>(
    this IObservableAsync<T> source,
    Func<T, TKey> keySelector)

Example:

var items = new[] { 1, 2, 2, 3, 1, 3 }.ToObservableAsync();
var unique = items.Distinct(); // 1, 2, 3
var noConsecDups = items.DistinctUntilChanged(); // 1, 2, 3, 1, 3

---

8. Combining and Merging Operators

Merge

Merges multiple observables, interleaving values.

// Two sources
public static IObservableAsync<T> Merge<T>(
    this IObservableAsync<T> first,
    IObservableAsync<T> second)

// Multiple sources with concurrency limit
public static IObservableAsync<T> Merge<T>(
    this IObservableAsync<IObservableAsync<T>> sources,
    int maxConcurrency)

Example:

var a = ObservableAsync.Range(1, 3); // 1, 2, 3
var b = ObservableAsync.Range(10, 3); // 10, 11, 12
var merged = ObservableAsync.Merge(a, b); // Interleaved: 1, 10, 2, 11, 3, 12 (order may vary)

Concat

Concatenates observables sequentially.

public static IObservableAsync<T> Concat<T>(
    this IObservableAsync<T> first,
    IObservableAsync<T> second)

Example:

var sequential = ObservableAsync.Concat(a, b); // 1, 2, 3, 10, 11, 12

Switch

Switches to the most recent inner observable.

public static IObservableAsync<T> Switch<T>(
    this IObservableAsync<IObservableAsync<T>> sources)

Use Case: When you want to cancel previous operations when a new one arrives (e.g., search as you type).

CombineLatest

Combines latest values from multiple sources.

// Two sources
public static IObservableAsync<TResult> CombineLatest<T1, T2, TResult>(
    this IObservableAsync<T1> first,
    IObservableAsync<T2> second,
    Func<T1, T2, TResult> selector)

// Up to 8 sources
public static IObservableAsync<TResult> CombineLatest<T1, T2, T3, T4, T5, T6, T7, T8, TResult>(
    IObservableAsync<T1> first,
    IObservableAsync<T2> second,
    // ... up to 8 sources
    Func<T1, T2, T3, T4, T5, T6, T7, T8, TResult> selector)

Example:

var combined = a.CombineLatest(b, (x, y) => $"{x}+{y}");
// Emits whenever either source emits, with latest values from both

Zip

Pairs elements by index.

public static IObservableAsync<TResult> Zip<T1, T2, TResult>(
    this IObservableAsync<T1> first,
    IObservableAsync<T2> second,
    Func<T1, T2, TResult> selector)

Example:

var zipped = a.Zip(b, (x, y) => x + y); // 11, 13, 15

Prepend

Prepends a single value.

public static IObservableAsync<T> Prepend<T>(
    this IObservableAsync<T> source,
    T value)

StartWith

Prepends multiple values.

public static IObservableAsync<T> StartWith<T>(
    this IObservableAsync<T> source,
    params T[] values)

Example:

var withPrefix = a.Prepend(0); // 0, 1, 2, 3
var withMany = a.StartWith(-2, -1, 0); // -2, -1, 0, 1, 2, 3

---

9. Error Handling and Resilience

Catch

Catches errors and switches to fallback.

public static IObservableAsync<T> Catch<T>(
    this IObservableAsync<T> source,
    Func<Exception, IObservableAsync<T>> handler)

Example:

var flaky = ObservableAsync.Throw<int>(new InvalidOperationException("Oops"));
var safe = flaky.Catch(ex => ObservableAsync.Return(-1)); // emits -1

CatchAndIgnoreErrorResume

Suppresses error-resume notifications.

public static IObservableAsync<T> CatchAndIgnoreErrorResume<T>(
    this IObservableAsync<T> source)

Retry (Infinite)

Re-subscribes indefinitely on error.

public static IObservableAsync<T> Retry<T>(
    this IObservableAsync<T> source)

Retry (Count-Limited)

Re-subscribes up to N times.

public static IObservableAsync<T> Retry<T>(
    this IObservableAsync<T> source,
    int count)

Example:

var retried = flaky.Retry(3); // Retry up to 3 times

OnErrorResumeAsFailure

Converts error-resume to failure completion.

public static IObservableAsync<T> OnErrorResumeAsFailure<T>(
    this IObservableAsync<T> source)

---

10. Timing and Scheduling

Throttle

Suppresses rapid emissions (debounce).

public static IObservableAsync<T> Throttle<T>(
    this IObservableAsync<T> source,
    TimeSpan timeSpan,
    TimeProvider? timeProvider = null)

Example:

var source = ObservableAsync.Interval(TimeSpan.FromMilliseconds(50)).Take(10);
var throttled = source.Throttle(TimeSpan.FromMilliseconds(200));

Delay

Delays each emission.

public static IObservableAsync<T> Delay<T>(
    this IObservableAsync<T> source,
    TimeSpan timeSpan,
    TimeProvider? timeProvider = null)

Timeout

Raises TimeoutException if no value arrives.

// With exception
public static IObservableAsync<T> Timeout<T>(
    this IObservableAsync<T> source,
    TimeSpan timeSpan)

// With fallback
public static IObservableAsync<T> Timeout<T>(
    this IObservableAsync<T> source,
    TimeSpan timeSpan,
    IObservableAsync<T> fallback)

Example:

var guarded = source.Timeout(TimeSpan.FromSeconds(2));
var withFallback = source.Timeout(
    TimeSpan.FromSeconds(2),
    ObservableAsync.Return(999L));

ObserveOn

Shifts notifications to specified context.

public static IObservableAsync<T> ObserveOn<T>(
    this IObservableAsync<T> source,
    AsyncContext context)

public static IObservableAsync<T> ObserveOn<T>(
    this IObservableAsync<T> source,
    SynchronizationContext context)

public static IObservableAsync<T> ObserveOn<T>(
    this IObservableAsync<T> source,
    TaskScheduler scheduler)

public static IObservableAsync<T> ObserveOn<T>(
    this IObservableAsync<T> source,
    IScheduler scheduler)

Example:

var onContext = source.ObserveOn(
    new AsyncContext(SynchronizationContext.Current!));

Yield

Yields control between notifications.

public static IObservableAsync<T> Yield<T>(
    this IObservableAsync<T> source)

---

11. Aggregation and Terminal Operators

AggregateAsync

Applies accumulator and returns final result.

public static ValueTask<TAccumulate> AggregateAsync<TSource, TAccumulate>(
    this IObservableAsync<TSource> source,
    TAccumulate seed,
    Func<TAccumulate, TSource, CancellationToken, ValueTask<TAccumulate>> accumulator,
    CancellationToken cancellationToken)

Example:

var source = ObservableAsync.Range(1, 5);
int sum = await source.AggregateAsync(0, (a, x) => a + x, CancellationToken.None); // 15

CountAsync

Returns element count.

public static ValueTask<int> CountAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

LongCountAsync

Returns element count as long.

public static ValueTask<long> LongCountAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

AnyAsync

Returns true if any elements exist.

public static ValueTask<bool> AnyAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

public static ValueTask<bool> AnyAsync<T>(
    this IObservableAsync<T> source,
    Func<T, bool> predicate,
    CancellationToken cancellationToken)

AllAsync

Returns true if all elements match predicate.

public static ValueTask<bool> AllAsync<T>(
    this IObservableAsync<T> source,
    Func<T, bool> predicate,
    CancellationToken cancellationToken)

ContainsAsync

Returns true if sequence contains value.

public static ValueTask<bool> ContainsAsync<T>(
    this IObservableAsync<T> source,
    T value,
    CancellationToken cancellationToken)

FirstAsync

Returns first element or throws.

public static ValueTask<T> FirstAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

FirstOrDefaultAsync

Returns first element or default.

public static ValueTask<T> FirstOrDefaultAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

LastAsync / LastOrDefaultAsync

Returns last element.

public static ValueTask<T> LastAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

public static ValueTask<T> LastOrDefaultAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

SingleAsync / SingleOrDefaultAsync

Returns single element or throws.

public static ValueTask<T> SingleAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

public static ValueTask<T> SingleOrDefaultAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

ToListAsync

Collects all elements into List.

public static ValueTask<List<T>> ToListAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

Example:

List<int> all = await source.ToListAsync(CancellationToken.None); // [1, 2, 3, 4, 5]

ToDictionaryAsync

Collects into Dictionary.

public static ValueTask<Dictionary<TKey, T>> ToDictionaryAsync<TSource, TKey>(
    this IObservableAsync<TSource> source,
    Func<TSource, TKey> keySelector,
    CancellationToken cancellationToken)

ForEachAsync

Invokes async action for each element.

public static ValueTask ForEachAsync<T>(
    this IObservableAsync<T> source,
    Func<T, CancellationToken, ValueTask> action,
    CancellationToken cancellationToken)

Example:

await source.ForEachAsync(async (x, ct) =>
    Console.WriteLine($"Item: {x}"), CancellationToken.None);

WaitCompletionAsync

Awaits completion without capturing values.

public static ValueTask WaitCompletionAsync<T>(
    this IObservableAsync<T> source,
    CancellationToken cancellationToken)

---

12. Async Disposables and Resource Management

DisposableAsync

Provides async disposable utilities.

public static class DisposableAsync
{
    public static IAsyncDisposable Empty { get; }
    
    public static IAsyncDisposable Create(Func<ValueTask> disposeAsync);
}

CompositeDisposableAsync

Manages multiple async disposables.

public sealed class CompositeDisposableAsync : IAsyncDisposable
{
    public CompositeDisposableAsync();
    public CompositeDisposableAsync(int capacity);
    public CompositeDisposableAsync(params IAsyncDisposable[] disposables);
    public CompositeDisposableAsync(IEnumerable<IAsyncDisposable> disposables);
    
    public bool IsDisposed { get; }
    public int Count { get; }
    
    public ValueTask AddAsync(IAsyncDisposable item);
    public ValueTask<bool> RemoveAsync(IAsyncDisposable item);
    public ValueTask ClearAsync();
    public bool Contains(IAsyncDisposable item);
    public ValueTask DisposeAsync();
}

Example:

var composite = new CompositeDisposableAsync();
await composite.AddAsync(someResource1);
await composite.AddAsync(someResource2);

// All resources disposed together
await composite.DisposeAsync();

SerialDisposableAsync

Manages a single async disposable that can be replaced.

public class SerialDisposableAsync : IAsyncDisposable
{
    public ValueTask SetDisposableAsync(IAsyncDisposable? value);
    public ValueTask DisposeAsync();
}

Use Case: When you need to replace a resource and automatically dispose the old one.

Example:

var serial = new SerialDisposableAsync();

// Set initial resource
await serial.SetDisposableAsync(resource1);

// Replace - resource1 is automatically disposed
await serial.SetDisposableAsync(resource2);

// Dispose - resource2 is disposed
await serial.DisposeAsync();

SingleAssignmentDisposableAsync

Allows single assignment of async disposable.

public sealed class SingleAssignmentDisposableAsync : IAsyncDisposable
{
    public bool IsDisposed { get; }
    public IAsyncDisposable? GetDisposable();
    public ValueTask SetDisposableAsync(IAsyncDisposable? value);
    public ValueTask DisposeAsync();
}

Use Case: When a resource must be assigned exactly once.

---

13. Subjects and Multicasting

SubjectAsync

Async subject for multicasting.

public static class SubjectAsync
{
    public static ISubjectAsync<T> Create<T>();
    public static ISubjectAsync<T> CreateBehavior<T>(T initialValue);
    public static ISubjectAsync<T> CreateReplayLatest<T>();
}

ISubjectAsync

public interface ISubjectAsync<T> : IObservableAsync<T>, IObserverAsync<T>
{
    IObservableAsync<T> Values { get; }
}

ConcurrentSubjectAsync

Forwards notifications to observers concurrently.

public sealed class ConcurrentSubjectAsync<T> : BaseSubjectAsync<T>
{
    // Observers notified in parallel for high throughput
}

ConcurrentReplayLatestSubjectAsync

Replays latest value to new subscribers with concurrent notification.

ConcurrentStatelessSubjectAsync

Stateless subject with concurrent notification.

ConcurrentStatelessReplayLatestSubjectAsync

Stateless replay-latest subject with concurrent notification.

Multicasting Operators

// Publish with serial subject
public static ConnectableObservableAsync<T> Publish<T>(
    this IObservableAsync<T> source)

// Publish with stateless subject
public static ConnectableObservableAsync<T> StatelessPublish<T>(
    this IObservableAsync<T> source)

// Replay latest value
public static ConnectableObservableAsync<T> ReplayLatest<T>(
    this IObservableAsync<T> source)

// Auto-connect on first subscriber
public static IObservableAsync<T> RefCount<T>(
    this ConnectableObservableAsync<T> source)

// Custom subject factory
public static ConnectableObservableAsync<T> Multicast<T, TSubject>(
    this IObservableAsync<T> source,
    Func<TSubject> subjectFactory)
    where TSubject : ISubjectAsync<T>

Example:

var source = ObservableAsync.Interval(TimeSpan.FromMilliseconds(100)).Take(5);

// Publish + explicit connect
var published = source.Publish();
await using var sub1 = await published.SubscribeAsync(
    async (v, ct) => Console.WriteLine($"Sub1: {v}"), CancellationToken.None);
await using var sub2 = await published.SubscribeAsync(
    async (v, ct) => Console.WriteLine($"Sub2: {v}"), CancellationToken.None);
await using var connection = await published.ConnectAsync(CancellationToken.None);

// RefCount: auto-connect/disconnect
var shared = source.Publish().RefCount();

// ReplayLatest: new subscribers get most recent value
var replayed = source.ReplayLatest().RefCount();

---

14. Bridging Classic and Async Observables

ObservableBridgeExtensions

Provides bidirectional conversion.

// IObservable to IObservableAsync
public static IObservableAsync<T> ToObservableAsync<T>(
    this IObservable<T> observable)

// IObservableAsync to IObservable
public static IObservable<T> ToObservable<T>(
    this IObservableAsync<T> asyncObservable)

// IAsyncEnumerable to IObservableAsync
public static IObservableAsync<T> ToObservableAsync<T>(
    this IAsyncEnumerable<T> asyncEnumerable)

// IObservableAsync to IAsyncEnumerable
public static IAsyncEnumerable<T> ToAsyncEnumerable<T>(
    this IObservableAsync<T> asyncObservable)

Example:

// Bridge from classic IObservable
var classic = Observable.Interval(TimeSpan.FromMilliseconds(200)).Take(5);
var asyncVersion = classic.ToObservableAsync();

// Bridge back to classic
var backToClassic = asyncVersion.ToObservable();

// From IAsyncEnumerable
async IAsyncEnumerable<int> GenerateAsync()
{
    for (int i = 0; i < 5; i++)
    {
        await Task.Delay(50);
        yield return i;
    }
}

var fromAsyncEnum = GenerateAsync().ToObservableAsync();

---

15. Classic Reactive Extensions Operators

ReactiveUI.Extensions also provides valuable operators for traditional IObservable<T> that don't ship with System.Reactive.

Null & Signal Helpers

// Filter nulls
public static IObservable<T> WhereIsNotNull<T>(
    this IObservable<T?> source)
    where T : class

// Convert to Unit signal
public static IObservable<Unit> AsSignal<T>(
    this IObservable<T> source)

Timing & Scheduling

// Shared timer (one underlying timer per TimeSpan)
public static IObservable<long> SyncTimer(TimeSpan period)

// Schedule single value
public static IObservable<T> Schedule<T>(
    this T value, 
    TimeSpan dueTime, 
    IScheduler scheduler)

// Safe scheduling (handles null scheduler)
public static IDisposable ScheduleSafe(
    this IScheduler? scheduler, 
    Action action)

// ThrottleFirst: allow first item per window
public static IObservable<T> ThrottleFirst<T>(
    this IObservable<T> source, 
    TimeSpan window, 
    IScheduler? scheduler = null)

// ThrottleDistinct: throttle but only emit on change
public static IObservable<T> ThrottleDistinct<T>(
    this IObservable<T> source, 
    TimeSpan throttle, 
    IScheduler? scheduler = null)

// DebounceImmediate: emit first immediately then debounce
public static IObservable<T> DebounceImmediate<T>(
    this IObservable<T> source, 
    TimeSpan dueTime, 
    IScheduler? scheduler = null)

Inactivity / Liveness

// Heartbeat during quiet periods
public static IObservable<IHeartbeat<T>> Heartbeat<T>(
    this IObservable<T> source, 
    TimeSpan interval, 
    IScheduler? scheduler = null)

// Detect stale data
public static IObservable<IStale<T>> DetectStale<T>(
    this IObservable<T> source, 
    TimeSpan staleThreshold, 
    IScheduler? scheduler = null)

// Buffer until inactive
public static IObservable<IList<T>> BufferUntilInactive<T>(
    this IObservable<T> source, 
    TimeSpan inactivityPeriod, 
    IScheduler? scheduler = null)

Error Handling

// Ignore all errors
public static IObservable<T> CatchIgnore<T>(
    this IObservable<T> source)

// Return fallback on error
public static IObservable<T> CatchAndReturn<T>(
    this IObservable<T> source, 
    T fallback)

// Retry with error handler
public static IObservable<T> OnErrorRetry<T, TException>(
    this IObservable<T> source, 
    Action<TException> onError, 
    int retryCount = int.MaxValue, 
    TimeSpan delay = default, 
    IScheduler? delayScheduler = null)
    where TException : Exception

// Retry with exponential backoff
public static IObservable<T> RetryWithBackoff<T>(
    this IObservable<T> source, 
    int maxRetries, 
    TimeSpan initialDelay, 
    double backoffFactor = 2.0, 
    TimeSpan? maxDelay = null, 
    IScheduler? scheduler = null)

Combining & Aggregation

// All values are true
public static IObservable<bool> CombineLatestValuesAreAllTrue(
    this IEnumerable<IObservable<bool>> sources)

// All values are false
public static IObservable<bool> CombineLatestValuesAreAllFalse(
    this IEnumerable<IObservable<bool>> sources)

// Get max value
public static IObservable<T> GetMax<T>(
    this IObservable<T> source)

// Get min value
public static IObservable<T> GetMin<T>(
    this IObservable<T> source)

// Partition into two streams
public static (IObservable<T> True, IObservable<T> False) Partition<T>(
    this IObservable<T> source, 
    Func<T, bool> predicate)

Logical / Boolean

// Boolean negation
public static IObservable<bool> Not(
    this IObservable<bool> source)

// Filter true values
public static IObservable<bool> WhereTrue(
    this IObservable<bool> source)

// Filter false values
public static IObservable<bool> WhereFalse(
    this IObservable<bool> source)

Async / Task Integration

// Sequential async projection
public static IObservable<TResult> SelectAsyncSequential<TSource, TResult>(
    this IObservable<TSource> source, 
    Func<TSource, Task<TResult>> selector)

// Latest only (cancels previous)
public static IObservable<TResult> SelectLatestAsync<TSource, TResult>(
    this IObservable<TSource> source, 
    Func<TSource, Task<TResult>> selector)

// Limited parallelism
public static IObservable<TResult> SelectAsyncConcurrent<TSource, TResult>(
    this IObservable<TSource> source, 
    Func<TSource, Task<TResult>> selector, 
    int maxConcurrency)

// Async subscription
public static IDisposable SubscribeAsync<T>(
    this IObservable<T> source, 
    Func<T, Task> onNext)

// Synchronous gate
public static IDisposable SubscribeSynchronous<T>(
    this IObservable<T> source, 
    Func<T, Task> onNext)

// Convert to hot Task
public static Task<T> ToHotTask<T>(
    this IObservable<T> source)

Backpressure

// Conflate bursty updates
public static IObservable<T> Conflate<T>(
    this IObservable<T> source, 
    TimeSpan minimumPeriod, 
    IScheduler? scheduler = null)

Filtering / Conditional

// Filter strings by regex
public static IObservable<string> Filter(
    this IObservable<string> source, 
    string regexPattern)

// Take until predicate (inclusive)
public static IObservable<T> TakeUntil<T>(
    this IObservable<T> source, 
    Func<T, bool> predicate)

// Wait for first match
public static IObservable<T> WaitUntil<T>(
    this IObservable<T> source, 
    Func<T, bool> predicate)

// Sample latest on trigger
public static IObservable<T> SampleLatest<T>(
    this IObservable<T> source, 
    IObservable<Unit> trigger)

// Fallback if empty
public static IObservable<T> SwitchIfEmpty<T>(
    this IObservable<T> source, 
    IObservable<T> fallback)

// Drop if busy
public static IObservable<T> DropIfBusy<T>(
    this IObservable<T> source, 
    Func<T, Task> asyncAction)

Buffering & Transformation

// Buffer until delimiter
public static IObservable<IList<T>> BufferUntil<T>(
    this IObservable<T> source, 
    T startDelimiter, 
    T endDelimiter)

// Buffer until idle
public static IObservable<IList<T>> BufferUntilIdle<T>(
    this IObservable<T> source, 
    TimeSpan idlePeriod)

// Emit consecutive pairs
public static IObservable<(T Previous, T Current)> Pairwise<T>(
    this IObservable<T> source)

// Scan with initial value
public static IObservable<TAccumulate> ScanWithInitial<TSource, TAccumulate>(
    this IObservable<TSource> source, 
    TAccumulate initial, 
    Func<TAccumulate, TSource, TAccumulate> accumulator)

// Shuffle arrays in-place
public static IObservable<T[]> Shuffle<T>(
    this IObservable<T[]> source)

Subscription / Side Effects

// Action on subscribe
public static IObservable<T> DoOnSubscribe<T>(
    this IObservable<T> source, 
    Action action)

// Action on dispose
public static IObservable<T> DoOnDispose<T>(
    this IObservable<T> source, 
    Action disposeAction)

Utility & Miscellaneous

// ForEach with low allocations
public static IObservable<T> ForEach<T>(
    this IObservable<IEnumerable<T>> source)

// Using helper
public static IObservable<TResult> Using<TDisposable, TResult>(
    this TDisposable obj, 
    Func<TDisposable, TResult> function, 
    IScheduler? scheduler = null)
    where TDisposable : IDisposable

// While loop
public static IObservable<Unit> While(
    Func<bool> condition, 
    Action action, 
    IScheduler? scheduler = null)

// OnNext with params
public static void OnNext<T>(
    this IObserver<T> observer, 
    params T[] values)

// Read-only BehaviorSubject
public static (IObservable<T> Observable, IObserver<T> Observer) ToReadOnlyBehavior<T>(
    T initialValue)

// Property change observable
public static IObservable<TProperty> ToPropertyObservable<TSource, TProperty>(
    this TSource source, 
    Expression<Func<TSource, TProperty>> propertyExpression)
    where TSource : INotifyPropertyChanged

---

16. Advanced Multi-Threading Examples

Example 1: Parallel Data Processing Pipeline

using ReactiveUI.Extensions.Async;
using ReactiveUI.Extensions.Async.Subjects;

public class ParallelDataProcessor
{
    private readonly int _maxConcurrency;
    
    public ParallelDataProcessor(int maxConcurrency = 4)
    {
        _maxConcurrency = maxConcurrency;
    }
    
    public async Task ProcessDataStreamAsync(
        IAsyncEnumerable<string> input,
        Func<string, CancellationToken, Task<string>> processor,
        CancellationToken cancellationToken)
    {
        // Convert to async observable
        var observable = input.ToObservableAsync();
        
        // Process with limited concurrency
        var processed = observable
            .SelectMany(async (item, ct) =>
            {
                try
                {
                    return await processor(item, ct);
                }
                catch (Exception ex)
                {
                    Console.WriteLine($"Error processing {item}: {ex.Message}");
                    return null;
                }
            })
            .Where(x => x != null);
        
        // Subscribe and process results
        await processed.ForEachAsync(async (result, ct) =>
        {
            await SaveResultAsync(result!, ct);
        }, cancellationToken);
    }
    
    private async Task SaveResultAsync(string result, CancellationToken ct)
    {
        // Simulate async I/O
        await Task.Delay(10, ct);
        Console.WriteLine($"Saved: {result}");
    }
}

// Usage
var processor = new ParallelDataProcessor(maxConcurrency: 4);
var data = GenerateDataAsync(); // IAsyncEnumerable<string>

await processor.ProcessDataStreamAsync(
    data,
    async (item, ct) =>
    {
        await Task.Delay(100, ct); // Simulate processing
        return item.ToUpperInvariant();
    },
    CancellationToken.None);

Example 2: Real-Time Data Aggregation with Multiple Sources

public class RealTimeAggregator
{
    public async Task AggregateMultipleSourcesAsync(
        CancellationToken cancellationToken)
    {
        // Create multiple data sources
        var source1 = ObservableAsync.Interval(TimeSpan.FromMilliseconds(100))
            .Take(50)
            .Select(x => $"Source1-{x}");
        
        var source2 = ObservableAsync.Interval(TimeSpan.FromMilliseconds(150))
            .Take(50)
            .Select(x => $"Source2-{x}");
        
        var source3 = ObservableAsync.Interval(TimeSpan.FromMilliseconds(200))
            .Take(50)
            .Select(x => $"Source3-{x}");
        
        // Merge all sources with concurrency limit
        var merged = ObservableAsync.Merge(
            new[] { source1, source2, source3 },
            maxConcurrency: 3);
        
        // Buffer and process in batches
        var batches = merged
            .Buffer(TimeSpan.FromMilliseconds(500))
            .Where(batch => batch.Count > 0);
        
        // Process batches in parallel
        await batches.ForEachAsync(async (batch, ct) =>
        {
            Console.WriteLine($"Processing batch of {batch.Count} items");
            
            // Process batch items in parallel
            var tasks = batch.Select(async item =>
            {
                await Task.Delay(10, ct);
                return item.ToUpperInvariant();
            });
            
            var results = await Task.WhenAll(tasks);
            
            foreach (var result in results)
            {
                Console.WriteLine($"  {result}");
            }
        }, cancellationToken);
    }
}

Example 3: Cancellation-Cooperative Long-Running Operation

public class CancellableDataFetcher
{
    public async Task FetchDataWithCancellationAsync(
        string url,
        CancellationToken cancellationToken)
    {
        var fetchObservable = ObservableAsync.Create<string>(async (observer, ct) =>
        {
            var linkedCts = CancellationTokenSource.CreateLinkedTokenSource(ct, cancellationToken);
            
            try
            {
                for (int i = 0; i < 100 && !linkedCts.Token.IsCancellationRequested; i++)
                {
                    // Simulate async HTTP request
                    var data = await FetchPageAsync(url, i, linkedCts.Token);
                    await observer.OnNextAsync(data, linkedCts.Token);
                    
                    await Task.Delay(100, linkedCts.Token);
                }
                
                await observer.OnCompletedAsync(Result.Success);
            }
            catch (OperationCanceledException)
            {
                await observer.OnCompletedAsync(Result.Success);
            }
            catch (Exception ex)
            {
                await observer.OnErrorResumeAsync(ex, linkedCts.Token);
            }
            finally
            {
                linkedCts.Dispose();
            }
            
            return DisposableAsync.Empty;
        });
        
        // Subscribe with cancellation
        await using var subscription = await fetchObservable.SubscribeAsync(
            async (data, ct) =>
            {
                await ProcessDataAsync(data, ct);
            },
            cancellationToken);
    }
    
    private async Task<string> FetchPageAsync(string url, int page, CancellationToken ct)
    {
        await Task.Delay(50, ct); // Simulate network
        return $"Page {page} data";
    }
    
    private async Task ProcessDataAsync(string data, CancellationToken ct)
    {
        await Task.Delay(10, ct); // Simulate processing
        Console.WriteLine($"Processed: {data}");
    }
}

Example 4: Thread-Safe State Management with Async Observables

public class ThreadSafeStateManager<T>
{
    private readonly ConcurrentSubjectAsync<T> _subject = new();
    private T? _currentState;
    private readonly SemaphoreSlim _lock = new(1, 1);
    
    public IObservableAsync<T> State => _subject;
    
    public async ValueTask UpdateStateAsync(T newState, CancellationToken cancellationToken)
    {
        await _lock.WaitAsync(cancellationToken);
        try
        {
            _currentState = newState;
            await _subject.OnNextAsync(newState, cancellationToken);
        }
        finally
        {
            _lock.Release();
        }
    }
    
    public async ValueTask<T> GetStateAsync(CancellationToken cancellationToken)
    {
        await _lock.WaitAsync(cancellationToken);
        try
        {
            return _currentState!;
        }
        finally
        {
            _lock.Release();
        }
    }
    
    public async ValueTask DisposeAsync()
    {
        await _lock.WaitAsync();
        try
        {
            await _subject.OnCompletedAsync(Result.Success);
            await _subject.DisposeAsync();
        }
        finally
        {
            _lock.Release();
            _lock.Dispose();
        }
    }
}

// Usage
var stateManager = new ThreadSafeStateManager<int>();

// Subscribe from multiple threads
var tasks = Enumerable.Range(0, 5).Select(async i =>
{
    await using var sub = await stateManager.State.SubscribeAsync(
        async (state, ct) =>
        {
            Console.WriteLine($"Thread {i}: State = {state}");
        },
        CancellationToken.None);
    
    await Task.Delay(1000);
});

await Task.WhenAll(tasks);

// Update state
await stateManager.UpdateStateAsync(42, CancellationToken.None);

Example 5: Backpressure Handling with Conflation

public class BackpressureHandler
{
    public async Task HandleHighFrequencyDataAsync(
        CancellationToken cancellationToken)
    {
        // High-frequency source (1000 events/second)
        var highFrequency = ObservableAsync.Interval(TimeSpan.FromMilliseconds(1))
            .Take(10000);
        
        // Conflate to 100 events/second (keep latest)
        var conflated = highFrequency
            .Publish(shared => shared
                .Throttle(TimeSpan.FromMilliseconds(10))
                .Merge(shared.TakeLast(1)));
        
        // Process at manageable rate
        await conflated.ForEachAsync(async (value, ct) =>
        {
            await Task.Delay(5, ct); // Simulate processing
            Console.WriteLine($"Processed: {value}");
        }, cancellationToken);
    }
}

---

17. Performance Considerations

ValueTask vs Task

ReactiveUI.Extensions uses ValueTask throughout for better performance:

// ValueTask avoids heap allocation for synchronous completions
public ValueTask OnNextAsync(T value, CancellationToken cancellationToken)

Benefits:

• No heap allocation when operation completes synchronously
• Reduced GC pressure
• Better performance for hot paths

Allocation-Aware Operators

Many operators are designed to minimize allocations:

// ForEach with low allocations
public static IObservable<T> ForEach<T>(
    this IObservable<IEnumerable<T>> source)

Threading and Concurrency

Best Practices:

1. Use ConcurrentSubjectAsync for high throughput:

var subject = new ConcurrentSubjectAsync<string>();
// Observers notified in parallel

2. Limit concurrency with Merge:

var merged = sources.Merge(maxConcurrency: 4);

3. Use ObserveOn for context switching:

var onUiThread = source.ObserveOn(SynchronizationContext.Current!);

Memory Management

Dispose Properly:

// Always use await using for async disposables
await using var subscription = await observable.SubscribeAsync(...);

// Or use CompositeDisposableAsync for multiple resources
var composite = new CompositeDisposableAsync();
await composite.AddAsync(subscription1);
await composite.AddAsync(subscription2);
await composite.DisposeAsync();

---

18. Best Practices

1. Always Pass CancellationToken

// GOOD
await observable.ForEachAsync(async (item, ct) =>
{
    await ProcessAsync(item, ct);
}, cancellationToken);

// BAD - No cancellation support
await observable.ForEachAsync(async item =>
{
    await ProcessAsync(item, CancellationToken.None);
});

2. Use Async Disposables Properly

// GOOD
await using var subscription = await observable.SubscribeAsync(...);

// BAD - Synchronous disposal of async resources
using (await observable.SubscribeAsync(...)) { }

3. Handle Errors in OnErrorResumeAsync

public class ResilientObserver : ObserverAsync<T>
{
    protected override async ValueTask OnErrorResumeAsyncCore(
        Exception error, 
        CancellationToken cancellationToken)
    {
        // Log error
        await LogErrorAsync(error, cancellationToken);
        
        // Decide whether to continue or stop
        // Don't rethrow unless you want to terminate
    }
}

4. Use Appropriate Subjects

// For simple multicasting
var subject = SubjectAsync.Create<T>();

// For high-throughput scenarios
var subject = new ConcurrentSubjectAsync<T>();

// For replaying latest value
var subject = SubjectAsync.CreateReplayLatest<T>();

5. Limit Concurrency

// GOOD - Limited concurrency
var processed = source.SelectMany(
    async item => await ProcessAsync(item),
    maxConcurrency: 4);

// BAD - Unlimited concurrency
var processed = source.SelectMany(
    async item => await ProcessAsync(item));

6. Use ObserveOn for UI Updates

// Ensure UI updates happen on UI thread
var uiUpdates = source.ObserveOn(SynchronizationContext.Current!)
    .SubscribeAsync(async (value, ct) =>
    {
        UpdateUI(value);
    });

7. Bridge Carefully Between Classic and Async

// Convert classic to async when needed
var asyncObservable = classicObservable.ToObservableAsync();

// Convert back when interoperating with Rx.NET libraries
var classic = asyncObservable.ToObservable();

8. Test with TestScheduler

// Use TestScheduler for deterministic testing
var testScheduler = new TestScheduler();
var source = ObservableAsync.Interval(TimeSpan.FromSeconds(1), testScheduler);

// Advance time deterministically
testScheduler.AdvanceBy(TimeSpan.FromSeconds(5).Ticks);

---

Conclusion

ReactiveUI.Extensions represents a significant advancement in reactive programming for .NET, providing a fully async-native framework that addresses the limitations of traditional Rx.NET in modern async/await scenarios. By embracing ValueTask, CancellationToken, and IAsyncDisposable throughout the entire pipeline, it enables:

• True end-to-end async without thread blocking
• Proper cancellation support at every level
• Async resource cleanup for I/O-bound operations
• Seamless interop with both classic Rx.NET and modern async patterns
• High performance through allocation-aware design

The library's comprehensive operator set, combined with advanced features like concurrent subjects, async disposables, and bidirectional bridging, makes it an essential tool for building robust, scalable reactive applications in .NET 8 and beyond.

Whether you're building real-time data processing pipelines, handling high-frequency event streams, or simply need better async integration in your reactive code, ReactiveUI.Extensions provides the tools and patterns to succeed.

---

Additional Resources

• GitHub Repository: https://github.com/reactiveui/Extensions
• System.Reactive Documentation: https://github.com/dotnet/reactive
• ReactiveUI Documentation: https://www.reactiveui.net/docs/
• Introduction to Rx.NET eBook: Free 2nd Edition available

---

This article covers all public functions within the ReactiveUI.Extensions library as of version 2.2.x. For the most up-to-date information, please refer to the official GitHub repository.