Mostlylucid.Ephemeral.Atoms.SlidingCache 2.3.2

Mostlylucid.Ephemeral.Atoms.SlidingCache

Caches work results with sliding expiration - accessing a result resets its TTL. Results that haven't been accessed expire and are recomputed on next request.

dotnet add package mostlylucid.ephemeral.atoms.slidingcache

Quick Start

using Mostlylucid.Ephemeral.Atoms.SlidingCache;

await using var cache = new SlidingCacheAtom<string, UserProfile>(
    async (userId, ct) => await LoadUserProfileAsync(userId, ct),
    slidingExpiration: TimeSpan.FromMinutes(5));

// First call: computes and caches
var profile = await cache.GetOrComputeAsync("user-123");

// Second call within 5 minutes: returns cached, resets TTL
var cached = await cache.GetOrComputeAsync("user-123");

// After 5 minutes of no access: entry expires, recomputes

All Options

new SlidingCacheAtom<TKey, TResult>(
    // Required: async factory to compute values
    factory: async (key, ct) => await ComputeAsync(key, ct),

    // Time without access before entry expires
    // Default: 5 minutes
    slidingExpiration: TimeSpan.FromMinutes(5),

    // Maximum time entry can live regardless of access
    // Default: 1 hour
    absoluteExpiration: TimeSpan.FromHours(1),

    // Maximum cache entries
    // Default: 1000
    maxSize: 1000,

    // Max concurrent factory calls
    // Default: Environment.ProcessorCount
    maxConcurrency: 8,

    // Signal sampling rate (1 = all, 10 = 1 in 10)
    // Default: 1
    sampleRate: 1,

    // Shared signal sink
    // Default: null (creates internal)
    signals: sharedSink
)

API Reference

// Get or compute value (resets sliding expiration on hit)
Task<TResult> GetOrComputeAsync(TKey key, CancellationToken ct = default);

// Try get without triggering computation (still resets TTL on hit)
bool TryGet(TKey key, out TResult? value);

// Invalidate specific entry
void Invalidate(TKey key);

// Clear all entries
void Clear();

// Get statistics
CacheStats GetStats(); // (TotalEntries, ValidEntries, ExpiredEntries, HotEntries, MaxSize)

// Get signals
IReadOnlyList<SignalEvent> GetSignals();
IReadOnlyList<SignalEvent> GetSignals(string pattern);

// Dispose
ValueTask DisposeAsync();

How It Works

Sliding vs Absolute Expiration

Entry created at T=0, slidingExpiration=5min, absoluteExpiration=1hr

T=0:   [Created] ─────────────────────────────────────────> Absolute deadline: T=60min
       LastAccess=T=0

T=3min: [Access] ─> LastAccess=T=3min ─> Sliding deadline: T=8min

T=7min: [Access] ─> LastAccess=T=7min ─> Sliding deadline: T=12min

T=15min: [No access since T=7min] ─> Entry EXPIRED (sliding)

T=59min: [Access after recompute] ─> New entry, LastAccess=T=59min

T=61min: Entry EXPIRED (absolute deadline from T=59min creation)

Eviction Strategy

When cache exceeds maxSize:

1. First pass: Remove all expired entries
2. Second pass: Remove coldest entries (lowest access count, then oldest access time)

Signals Emitted

Example: API Response Caching

await using var cache = new SlidingCacheAtom<string, ApiResponse>(
    async (endpoint, ct) =>
    {
        var response = await httpClient.GetAsync(endpoint, ct);
        return await response.Content.ReadFromJsonAsync<ApiResponse>(ct);
    },
    slidingExpiration: TimeSpan.FromMinutes(2),
    absoluteExpiration: TimeSpan.FromMinutes(30),
    maxConcurrency: 4);

// Multiple concurrent requests for same endpoint are deduplicated
var tasks = Enumerable.Range(0, 10)
    .Select(_ => cache.GetOrComputeAsync("/api/users"));

var results = await Task.WhenAll(tasks);
// Only 1 HTTP call made, all 10 tasks get same result

Example: Database Query Caching

await using var cache = new SlidingCacheAtom<int, Order>(
    async (orderId, ct) => await db.Orders.FindAsync(orderId, ct),
    slidingExpiration: TimeSpan.FromMinutes(10),
    maxSize: 5000,
    sampleRate: 10);  // Sample 1 in 10 for high-volume

// Hot orders stay cached, cold orders expire
var order = await cache.GetOrComputeAsync(orderId);

// Monitor cache health
var stats = cache.GetStats();
Console.WriteLine($"Hit rate estimate: {stats.HotEntries}/{stats.TotalEntries} hot");

// Check for errors
var errors = cache.GetSignals("cache.error:*");
if (errors.Any())
    logger.LogWarning("Cache errors: {Count}", errors.Count);

Example: With Shared Signal Sink

var sink = new SignalSink();

await using var userCache = new SlidingCacheAtom<string, User>(
    LoadUserAsync,
    signals: sink);

await using var orderCache = new SlidingCacheAtom<int, Order>(
    LoadOrderAsync,
    signals: sink);

// Monitor all cache activity
var allMisses = sink.Sense(s => s.Signal.StartsWith("cache.miss"));
var allErrors = sink.Sense(s => s.Signal.StartsWith("cache.error"));

Related Packages

Cache Strategy Comparison

Use the right cache for the job:

Tip: Default SQLite helper uses EphemeralLruCache for hot-key bias; reach for SlidingCacheAtom when you need async factories with sliding expiration and dedupe.

License

Unlicense (public domain)