Mostlylucid.Ephemeral

Fire... and Don't Quite Forget.

Bounded, observable, self-cleaning async execution with signal-based coordination.

dotnet add package mostlylucid.ephemeral

Quick Start

using Mostlylucid.Ephemeral;

// One-shot parallel processing
await items.EphemeralForEachAsync(
    async (item, ct) => await ProcessAsync(item, ct),
    new EphemeralOptions { MaxConcurrency = 8 });

// Long-lived coordinator
await using var coordinator = new EphemeralWorkCoordinator<WorkItem>(
    async (item, ct) => await ProcessAsync(item, ct),
    new EphemeralOptions { MaxConcurrency = 8 });

await coordinator.EnqueueAsync(new WorkItem("data"));

// See what's happening
var running = coordinator.GetRunning();
var failed = coordinator.GetFailed();
var pending = coordinator.PendingCount;

// Graceful shutdown
coordinator.Complete();
await coordinator.DrainAsync();

All Options

new EphemeralOptions
{
    // ═══════════════════════════════════════════════════════════════
    // CONCURRENCY
    // ═══════════════════════════════════════════════════════════════

    // Max parallel operations overall
    // Default: Environment.ProcessorCount
    MaxConcurrency = 8,

    // Allow runtime concurrency adjustment via SetMaxConcurrency()
    // Default: false (fastest hot-path)
    EnableDynamicConcurrency = false,

    // Max parallel operations per key (keyed coordinators only)
    // Default: 1 (sequential per key)
    MaxConcurrencyPerKey = 1,

    // ═══════════════════════════════════════════════════════════════
    // MEMORY / WINDOW
    // ═══════════════════════════════════════════════════════════════

    // Max operations retained in memory (LRU eviction)
    // Default: 200
    MaxTrackedOperations = 200,

    // Max age for tracked operations before cleanup
    // Default: 5 minutes
    MaxOperationLifetime = TimeSpan.FromMinutes(5),

    // ═══════════════════════════════════════════════════════════════
    // FAIR SCHEDULING (keyed coordinators only)
    // ═══════════════════════════════════════════════════════════════

    // Prevent hot keys from starving cold keys
    // Default: false (FIFO ordering)
    EnableFairScheduling = false,

    // Pending count before a key is deprioritized
    // Default: 10
    FairSchedulingThreshold = 10,

    // ═══════════════════════════════════════════════════════════════
    // SIGNALS
    // ═══════════════════════════════════════════════════════════════

    // Shared signal sink across coordinators
    // Default: null (isolated)
    Signals = new SignalSink(),

    // Sync callback on signal raise (keep fast!)
    // Default: null
    OnSignal = evt => Console.WriteLine($"Signal: {evt.Signal}"),

    // Async callback on signal raise (background queue, non-blocking)
    // Default: null
    OnSignalAsync = async (evt, ct) => await LogToService(evt, ct),

    // Sync callback on signal retract
    // Default: null
    OnSignalRetracted = evt => Console.WriteLine($"Retracted: {evt.Signal}"),

    // Async callback on signal retract
    // Default: null
    OnSignalRetractedAsync = async (evt, ct) => await NotifyService(evt, ct),

    // Max concurrent async signal handlers
    // Default: 4
    MaxConcurrentSignalHandlers = 4,

    // Max queued signals before dropping oldest
    // Default: 1000
    MaxQueuedSignals = 1000,

    // Self-documenting: signals this coordinator may emit
    // Default: null (no enforcement)
    Emits = new[] { "started", "completed", "error" },

    // Self-documenting: signals this coordinator listens for
    // Default: null (no enforcement)
    Listens = new[] { "backpressure", "shutdown" },

    // Constraints for signal propagation (cycles, depth, terminals)
    // Default: null
    SignalConstraints = new SignalConstraints { MaxDepth = 10 },

    // ═══════════════════════════════════════════════════════════════
    // SIGNAL-BASED CONTROL FLOW
    // ═══════════════════════════════════════════════════════════════

    // Skip new items when these signals are present
    // Use for circuit-breaker patterns
    // Default: null
    CancelOnSignals = new HashSet<string> { "circuit-open", "shutdown" },

    // Delay new items when these signals are present
    // Use for backpressure patterns
    // Default: null
    DeferOnSignals = new HashSet<string> { "backpressure", "rate-limited" },

    // How often to recheck when deferring
    // Default: 100ms
    DeferCheckInterval = TimeSpan.FromMilliseconds(100),

    // Max defer attempts before running anyway
    // Default: 50 (5 seconds at 100ms)
    MaxDeferAttempts = 50,

    // ═══════════════════════════════════════════════════════════════
    // SAMPLING
    // ═══════════════════════════════════════════════════════════════

    // Callback after each operation with window snapshot
    // Runs on caller's thread - keep it cheap!
    // Default: null
    OnSample = snapshot => metrics.Record(snapshot.Count)
}

Coordinators

EphemeralWorkCoordinator<T>

Long-lived work queue with bounded concurrency.

await using var coordinator = new EphemeralWorkCoordinator<Request>(
    async (req, ct) => await HandleAsync(req, ct),
    new EphemeralOptions { MaxConcurrency = 8 });

// Enqueue work
await coordinator.EnqueueAsync(request);
var id = await coordinator.EnqueueWithIdAsync(request);  // Get operation ID

// Query state
var running = coordinator.GetRunning();
var completed = coordinator.GetCompleted();
var failed = coordinator.GetFailed();
var snapshot = coordinator.GetSnapshot();
var byId = coordinator.GetById(id);

// Stats
int pending = coordinator.PendingCount;
int active = coordinator.ActiveCount;
int totalCompleted = coordinator.TotalCompleted;
int totalFailed = coordinator.TotalFailed;

// Dynamic concurrency (requires EnableDynamicConcurrency = true)
coordinator.SetMaxConcurrency(16);
int current = coordinator.CurrentMaxConcurrency;

// Signals
bool hasError = coordinator.HasSignal("error");
int errorCount = coordinator.CountSignals("error");
var errors = coordinator.GetSignalsByPattern("error.*");
var recent = coordinator.GetSignalsSince(DateTimeOffset.UtcNow.AddMinutes(-1));

// Shutdown
coordinator.Complete();          // Stop accepting new work
await coordinator.DrainAsync();  // Wait for in-flight to finish

EphemeralKeyedWorkCoordinator<T, TKey>

Per-key sequential processing with optional fair scheduling.

await using var coordinator = new EphemeralKeyedWorkCoordinator<Order, string>(
    keySelector: order => order.CustomerId,
    body: async (order, ct) => await ProcessOrder(order, ct),
    new EphemeralOptions
    {
        MaxConcurrency = 16,          // Total parallel
        MaxConcurrencyPerKey = 1,     // Sequential per customer
        EnableFairScheduling = true   // Prevent hot customer starvation
    });

await coordinator.EnqueueAsync(order);

EphemeralResultCoordinator<TInput, TResult>

Capture results from async operations.

await using var coordinator = new EphemeralResultCoordinator<Request, Response>(
    async (req, ct) => await FetchAsync(req, ct),
    new EphemeralOptions { MaxConcurrency = 4 });

var id = await coordinator.EnqueueAsync(request);
var snapshot = await coordinator.WaitForResult(id);

if (snapshot.HasResult)
    Console.WriteLine(snapshot.Result);
else if (snapshot.Exception != null)
    Console.WriteLine($"Failed: {snapshot.Exception.Message}");

PriorityWorkCoordinator<T>

Multiple priority lanes.

var coordinator = new PriorityWorkCoordinator<WorkItem>(
    body: async (item, ct) => await ProcessAsync(item, ct),
    new PriorityWorkCoordinatorOptions<WorkItem>(
        Lanes: new[]
        {
            new PriorityLane("critical", MaxDepth: 100),
            new PriorityLane("high"),
            new PriorityLane("normal"),
            new PriorityLane("low")
        }
    ));

await coordinator.EnqueueAsync(item, "critical");

Extension Method

One-shot parallel processing of collections.

await items.EphemeralForEachAsync(
    async (item, ct) => await ProcessAsync(item, ct),
    new EphemeralOptions
    {
        MaxConcurrency = 8,
        OnSignal = evt => Console.WriteLine(evt.Signal)
    });

Signals

Emitting Signals

// From within operation body (via ISignalEmitter passed to body)
await coordinator.EnqueueAsync(request);  // Body receives emitter

// From shared sink
var sink = new SignalSink();
sink.Raise("backpressure.downstream");
sink.Raise(new SignalEvent("error.timeout", operationId, key, DateTimeOffset.UtcNow));

// Retract signals
sink.Retract("backpressure.downstream");

Querying Signals

// Exact match
bool hasError = coordinator.HasSignal("error");
int count = coordinator.CountSignals("error");

// Pattern matching (glob-style: * and ?)
var errors = coordinator.GetSignalsByPattern("error.*");
var httpErrors = coordinator.GetSignalsByPattern("http.error.*");

// Time-based
var recent = coordinator.GetSignalsSince(DateTimeOffset.UtcNow.AddSeconds(-30));

// From shared sink
var snapshot = sink.Sense();
var filtered = sink.Sense(s => s.Signal.StartsWith("error"));

Shared Signal Sink

var sink = new SignalSink(maxCapacity: 1000, maxAge: TimeSpan.FromMinutes(5));

var coordinator1 = new EphemeralWorkCoordinator<A>(body, new EphemeralOptions { Signals = sink });
var coordinator2 = new EphemeralWorkCoordinator<B>(body, new EphemeralOptions { Signals = sink });

// Both coordinators see signals raised by either
sink.Raise("system.maintenance");

Molecules & Atom triggers

Use the mostlylucid.ephemeral.atoms.molecules package when you want to treat several atoms as one workflow. MoleculeBlueprintBuilder lets you describe each step, wire its signals into downstream steps, and build a blueprint that MoleculeRunner listens for (matching on a trigger signal and shared SignalSink). The runner raises MoleculeStarted, MoleculeCompleted, and MoleculeFailed, so you can observe what the chef (coordinator) dropped into the soup, and AtomTrigger gives you a lightweight watcher that starts additional atoms or molecules whenever a signal pattern fires.

Scheduled tasks

DurableTaskAtom + ScheduledTasksAtom turn cron/JSON schedules into durable work inside your coordinator window. Create ScheduledTaskDefinitions (cron expression, signal, optional key, payload, description, timeZone, format, runOnStartup, etc.), point the scheduler at SignalSink, and let it enqueue DurableTasks that emit the configured signals. Because the work runs inside the same coordinator, it inherits pinning, signal logging, and responsibility semantics along with your ad-hoc work.

Responsibility signals & echoes

Coordinators implement IOperationPinning and raise OperationFinalized when entries leave the window. ResponsibilitySignalManager provides PinUntilQueried so atoms can declare responsibility (pin) until a downstream ack signal arrives, preventing resources from disappearing while they await consumers. Subscribe to OperationFinalized to record “last words” (logs, diagnostics, signals) via LastWordsNoteAtom, or let OperationEchoMaker capture typed echoes for a configurable window so molecules, monitors, or auditors can still read the final signal wave.

Dependency Injection

// Register
services.AddEphemeralWorkCoordinator<WorkItem>(
    async (item, ct) => await ProcessAsync(item, ct),
    new EphemeralOptions { MaxConcurrency = 8 });

// Named coordinators
services.AddEphemeralWorkCoordinator<WorkItem>("priority",
    async (item, ct) => await ProcessPriorityAsync(item, ct),
    new EphemeralOptions { MaxConcurrency = 4 });

// Inject
public class MyService(IEphemeralCoordinatorFactory<WorkItem> factory)
{
    public async Task DoWork()
    {
        var coordinator = factory.CreateCoordinator();
        // or: factory.CreateCoordinator("priority")
        await coordinator.EnqueueAsync(new WorkItem());
    }
}

Related Packages

Package	Description
mostlylucid.ephemeral.atoms.fixedwork	Fixed worker pool with stats
mostlylucid.ephemeral.atoms.keyedsequential	Per-key sequential processing
mostlylucid.ephemeral.atoms.signalaware	Pause/cancel on signals
mostlylucid.ephemeral.atoms.batching	Time/size batching
mostlylucid.ephemeral.atoms.retry	Exponential backoff retry
mostlylucid.ephemeral.patterns.circuitbreaker	Signal-based circuit breaker
mostlylucid.ephemeral.patterns.backpressure	Signal-driven backpressure
mostlylucid.ephemeral.complete	All packages in one DLL

Target Frameworks

.NET 6.0, 7.0, 8.0, 9.0, 10.0

License

Unlicense (public domain)

Mostlylucid.Ephemeral

Fire... and Don't Quite Forget.

Bounded, observable, self-cleaning async execution with signal-based coordination.

dotnet add package mostlylucid.ephemeral

Quick Start

using Mostlylucid.Ephemeral;

// One-shot parallel processing
await items.EphemeralForEachAsync(
    async (item, ct) => await ProcessAsync(item, ct),
    new EphemeralOptions { MaxConcurrency = 8 });

// Long-lived coordinator
await using var coordinator = new EphemeralWorkCoordinator<WorkItem>(
    async (item, ct) => await ProcessAsync(item, ct),
    new EphemeralOptions { MaxConcurrency = 8 });

await coordinator.EnqueueAsync(new WorkItem("data"));

// See what's happening
var running = coordinator.GetRunning();
var failed = coordinator.GetFailed();
var pending = coordinator.PendingCount;

// Graceful shutdown
coordinator.Complete();
await coordinator.DrainAsync();

All Options

new EphemeralOptions
{
    // ═══════════════════════════════════════════════════════════════
    // CONCURRENCY
    // ═══════════════════════════════════════════════════════════════

    // Max parallel operations overall
    // Default: Environment.ProcessorCount
    MaxConcurrency = 8,

    // Allow runtime concurrency adjustment via SetMaxConcurrency()
    // Default: false (fastest hot-path)
    EnableDynamicConcurrency = false,

    // Max parallel operations per key (keyed coordinators only)
    // Default: 1 (sequential per key)
    MaxConcurrencyPerKey = 1,

    // ═══════════════════════════════════════════════════════════════
    // MEMORY / WINDOW
    // ═══════════════════════════════════════════════════════════════

    // Max operations retained in memory (LRU eviction)
    // Default: 200
    MaxTrackedOperations = 200,

    // Max age for tracked operations before cleanup
    // Default: 5 minutes
    MaxOperationLifetime = TimeSpan.FromMinutes(5),

    // ═══════════════════════════════════════════════════════════════
    // FAIR SCHEDULING (keyed coordinators only)
    // ═══════════════════════════════════════════════════════════════

    // Prevent hot keys from starving cold keys
    // Default: false (FIFO ordering)
    EnableFairScheduling = false,

    // Pending count before a key is deprioritized
    // Default: 10
    FairSchedulingThreshold = 10,

    // ═══════════════════════════════════════════════════════════════
    // SIGNALS
    // ═══════════════════════════════════════════════════════════════

    // Shared signal sink across coordinators
    // Default: null (isolated)
    Signals = new SignalSink(),

    // Sync callback on signal raise (keep fast!)
    // Default: null
    OnSignal = evt => Console.WriteLine($"Signal: {evt.Signal}"),

    // Async callback on signal raise (background queue, non-blocking)
    // Default: null
    OnSignalAsync = async (evt, ct) => await LogToService(evt, ct),

    // Sync callback on signal retract
    // Default: null
    OnSignalRetracted = evt => Console.WriteLine($"Retracted: {evt.Signal}"),

    // Async callback on signal retract
    // Default: null
    OnSignalRetractedAsync = async (evt, ct) => await NotifyService(evt, ct),

    // Max concurrent async signal handlers
    // Default: 4
    MaxConcurrentSignalHandlers = 4,

    // Max queued signals before dropping oldest
    // Default: 1000
    MaxQueuedSignals = 1000,

    // Self-documenting: signals this coordinator may emit
    // Default: null (no enforcement)
    Emits = new[] { "started", "completed", "error" },

    // Self-documenting: signals this coordinator listens for
    // Default: null (no enforcement)
    Listens = new[] { "backpressure", "shutdown" },

    // Constraints for signal propagation (cycles, depth, terminals)
    // Default: null
    SignalConstraints = new SignalConstraints { MaxDepth = 10 },

    // ═══════════════════════════════════════════════════════════════
    // SIGNAL-BASED CONTROL FLOW
    // ═══════════════════════════════════════════════════════════════

    // Skip new items when these signals are present
    // Use for circuit-breaker patterns
    // Default: null
    CancelOnSignals = new HashSet<string> { "circuit-open", "shutdown" },

    // Delay new items when these signals are present
    // Use for backpressure patterns
    // Default: null
    DeferOnSignals = new HashSet<string> { "backpressure", "rate-limited" },

    // How often to recheck when deferring
    // Default: 100ms
    DeferCheckInterval = TimeSpan.FromMilliseconds(100),

    // Max defer attempts before running anyway
    // Default: 50 (5 seconds at 100ms)
    MaxDeferAttempts = 50,

    // ═══════════════════════════════════════════════════════════════
    // SAMPLING
    // ═══════════════════════════════════════════════════════════════

    // Callback after each operation with window snapshot
    // Runs on caller's thread - keep it cheap!
    // Default: null
    OnSample = snapshot => metrics.Record(snapshot.Count)
}

Coordinators

EphemeralWorkCoordinator<T>

Long-lived work queue with bounded concurrency.

await using var coordinator = new EphemeralWorkCoordinator<Request>(
    async (req, ct) => await HandleAsync(req, ct),
    new EphemeralOptions { MaxConcurrency = 8 });

// Enqueue work
await coordinator.EnqueueAsync(request);
var id = await coordinator.EnqueueWithIdAsync(request);  // Get operation ID

// Query state
var running = coordinator.GetRunning();
var completed = coordinator.GetCompleted();
var failed = coordinator.GetFailed();
var snapshot = coordinator.GetSnapshot();
var byId = coordinator.GetById(id);

// Stats
int pending = coordinator.PendingCount;
int active = coordinator.ActiveCount;
int totalCompleted = coordinator.TotalCompleted;
int totalFailed = coordinator.TotalFailed;

// Dynamic concurrency (requires EnableDynamicConcurrency = true)
coordinator.SetMaxConcurrency(16);
int current = coordinator.CurrentMaxConcurrency;

// Signals
bool hasError = coordinator.HasSignal("error");
int errorCount = coordinator.CountSignals("error");
var errors = coordinator.GetSignalsByPattern("error.*");
var recent = coordinator.GetSignalsSince(DateTimeOffset.UtcNow.AddMinutes(-1));

// Shutdown
coordinator.Complete();          // Stop accepting new work
await coordinator.DrainAsync();  // Wait for in-flight to finish

EphemeralKeyedWorkCoordinator<T, TKey>

Per-key sequential processing with optional fair scheduling.

await using var coordinator = new EphemeralKeyedWorkCoordinator<Order, string>(
    keySelector: order => order.CustomerId,
    body: async (order, ct) => await ProcessOrder(order, ct),
    new EphemeralOptions
    {
        MaxConcurrency = 16,          // Total parallel
        MaxConcurrencyPerKey = 1,     // Sequential per customer
        EnableFairScheduling = true   // Prevent hot customer starvation
    });

await coordinator.EnqueueAsync(order);

EphemeralResultCoordinator<TInput, TResult>

Capture results from async operations.

await using var coordinator = new EphemeralResultCoordinator<Request, Response>(
    async (req, ct) => await FetchAsync(req, ct),
    new EphemeralOptions { MaxConcurrency = 4 });

var id = await coordinator.EnqueueAsync(request);
var snapshot = await coordinator.WaitForResult(id);

if (snapshot.HasResult)
    Console.WriteLine(snapshot.Result);
else if (snapshot.Exception != null)
    Console.WriteLine($"Failed: {snapshot.Exception.Message}");

PriorityWorkCoordinator<T>

Multiple priority lanes.

var coordinator = new PriorityWorkCoordinator<WorkItem>(
    body: async (item, ct) => await ProcessAsync(item, ct),
    new PriorityWorkCoordinatorOptions<WorkItem>(
        Lanes: new[]
        {
            new PriorityLane("critical", MaxDepth: 100),
            new PriorityLane("high"),
            new PriorityLane("normal"),
            new PriorityLane("low")
        }
    ));

await coordinator.EnqueueAsync(item, "critical");

Extension Method

One-shot parallel processing of collections.

await items.EphemeralForEachAsync(
    async (item, ct) => await ProcessAsync(item, ct),
    new EphemeralOptions
    {
        MaxConcurrency = 8,
        OnSignal = evt => Console.WriteLine(evt.Signal)
    });

Signals

Emitting Signals

// From within operation body (via ISignalEmitter passed to body)
await coordinator.EnqueueAsync(request);  // Body receives emitter

// From shared sink
var sink = new SignalSink();
sink.Raise("backpressure.downstream");
sink.Raise(new SignalEvent("error.timeout", operationId, key, DateTimeOffset.UtcNow));

// Retract signals
sink.Retract("backpressure.downstream");

Querying Signals

// Exact match
bool hasError = coordinator.HasSignal("error");
int count = coordinator.CountSignals("error");

// Pattern matching (glob-style: * and ?)
var errors = coordinator.GetSignalsByPattern("error.*");
var httpErrors = coordinator.GetSignalsByPattern("http.error.*");

// Time-based
var recent = coordinator.GetSignalsSince(DateTimeOffset.UtcNow.AddSeconds(-30));

// From shared sink
var snapshot = sink.Sense();
var filtered = sink.Sense(s => s.Signal.StartsWith("error"));

Shared Signal Sink

var sink = new SignalSink(maxCapacity: 1000, maxAge: TimeSpan.FromMinutes(5));

var coordinator1 = new EphemeralWorkCoordinator<A>(body, new EphemeralOptions { Signals = sink });
var coordinator2 = new EphemeralWorkCoordinator<B>(body, new EphemeralOptions { Signals = sink });

// Both coordinators see signals raised by either
sink.Raise("system.maintenance");

Molecules & Atom triggers

Scheduled tasks

Responsibility signals & echoes

Dependency Injection

// Register
services.AddEphemeralWorkCoordinator<WorkItem>(
    async (item, ct) => await ProcessAsync(item, ct),
    new EphemeralOptions { MaxConcurrency = 8 });

// Named coordinators
services.AddEphemeralWorkCoordinator<WorkItem>("priority",
    async (item, ct) => await ProcessPriorityAsync(item, ct),
    new EphemeralOptions { MaxConcurrency = 4 });

// Inject
public class MyService(IEphemeralCoordinatorFactory<WorkItem> factory)
{
    public async Task DoWork()
    {
        var coordinator = factory.CreateCoordinator();
        // or: factory.CreateCoordinator("priority")
        await coordinator.EnqueueAsync(new WorkItem());
    }
}

Related Packages

Package	Description
mostlylucid.ephemeral.atoms.fixedwork	Fixed worker pool with stats
mostlylucid.ephemeral.atoms.keyedsequential	Per-key sequential processing
mostlylucid.ephemeral.atoms.signalaware	Pause/cancel on signals
mostlylucid.ephemeral.atoms.batching	Time/size batching
mostlylucid.ephemeral.atoms.retry	Exponential backoff retry
mostlylucid.ephemeral.patterns.circuitbreaker	Signal-based circuit breaker
mostlylucid.ephemeral.patterns.backpressure	Signal-driven backpressure
mostlylucid.ephemeral.complete	All packages in one DLL

Target Frameworks

.NET 6.0, 7.0, 8.0, 9.0, 10.0

License

Unlicense (public domain)

mostlylucid/mostlylucid.ephemeralv2.3.2

Get Started

Readme

Mostlylucid.Ephemeral

Quick Start

All Options

Coordinators

EphemeralWorkCoordinator<T>

EphemeralKeyedWorkCoordinator<T, TKey>

EphemeralResultCoordinator<TInput, TResult>

PriorityWorkCoordinator<T>

Extension Method

Signals

Emitting Signals

Querying Signals

Shared Signal Sink

Molecules & Atom triggers

Scheduled tasks

Responsibility signals & echoes

Dependency Injection

Related Packages

Target Frameworks

License

mostlylucid/mostlylucid.ephemeralv2.3.2

Get Started

Readme

Mostlylucid.Ephemeral

Quick Start

All Options

Coordinators

EphemeralWorkCoordinator<T>

EphemeralKeyedWorkCoordinator<T, TKey>

EphemeralResultCoordinator<TInput, TResult>

PriorityWorkCoordinator<T>

Extension Method

Signals

Emitting Signals

Querying Signals

Shared Signal Sink

Molecules & Atom triggers

Scheduled tasks

Responsibility signals & echoes

Dependency Injection

Related Packages

Target Frameworks

License