feat(ext-cache-core): add the caching contract and a Caffeine backend

Split the way flash-ext-data and flash-ext-view are: cache-core defines
Cache, CacheManager, CacheSpec and CacheStats and talks to nothing;
cache-caffeine implements them in process.

    users = require(CacheManager.class).build("users", spec -> spec
            .maxSize(10_000).ttl(Duration.ofMinutes(10)));

    return users.get(id, repo::findById);

get(key, loader) is the only shape most code needs and the only one that is
hard to get right: the loader runs once per key across concurrent callers
rather than each racing its own. A null result stores nothing, because caching
absence is a decision rather than a default.

build(name, spec) is idempotent per name, so two handlers wanting one cache get
one cache without coordinating who creates it. Disagreeing about the spec
throws rather than resolving to whichever handler initialised first, which is a
bug that only surfaces under load.

recordStats() is opt-in — counting is two atomic increments per lookup, and a
cache nobody measures should not pay for numbers nobody reads. Unmeasured
caches return CacheStats.DISABLED rather than zeroes that look like a cold
cache.

Caffeine rather than a hand-rolled LRU: for genuinely low traffic
ConcurrentHashMap::computeIfAbsent is one line and needs no module at all, and
this exists for when that stops being true. W-TinyLFU admission, striped
counters and amortised eviction are not a weekend's work, and getting them
wrong yields a cache slower than no cache. The adapter is deliberately thin —
every method delegates, adding no wrapper, copy or locking of its own.

Caches are dropped through FlashContext.onClose, so values do not outlive the
app holding them. Invisible with one app per process; immediate under test.

flash-ext-cache-redis is designed but not built, and has docs only — no module,
no pom, no source. An empty module that builds an empty jar is dead weight in
the reactor. The docs record what changes once the cache can fail: get() must
decide whether to fall through to the loader, values need a codec,
invalidateAll needs a key prefix that becomes wire contract, and eviction stats
stop meaning anything. Those are decisions that want a real second replica to
check them against.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
Zakaria El Orche
2026-09-09 12:30:03 +00:00
co-authored by Claude Opus 5
parent 58bae41f7a
commit 24bb10175d
17 changed files with 753 additions and 1 deletions
@@ -0,0 +1,85 @@
# flash-ext-cache-caffeine
In-process caching backed by [Caffeine](https://github.com/ben-manes/caffeine). Implements
[`flash-ext-cache-core`](../../flash-ext-cache-core/docs/README.md).
## Dependency
```xml
<dependency>
<groupId>dev.relism</groupId>
<artifactId>flash-ext-cache-caffeine</artifactId>
<version>${flash.version}</version>
</dependency>
```
## Quick start
```java
FlashApp.create(8080)
.install(new CaffeineCacheExtension())
.scan("dev.example.api");
```
```java
@GET("/api/users/{id}")
public final class GetUser extends RequestHandler {
private Cache<String, User> users;
private UserRepository repo;
@Override protected void onInit() {
repo = require(UserRepository.class);
users = require(CacheManager.class).build("users", spec -> spec
.maxSize(10_000)
.ttl(Duration.ofMinutes(10)));
}
@Override public Object handle(Request req, Response res) {
return users.get(req.param("id"), repo::findById);
}
}
```
The extension takes no configuration. Each cache declares its own size and TTL where it is built.
## Why Caffeine and not a `LinkedHashMap`
An LRU on top of `LinkedHashMap` is about sixty lines, and for a cache that is genuinely
low-traffic it is the right answer — `ConcurrentHashMap::computeIfAbsent` is one line and has no
hit rate to get wrong.
This module exists for the case where that stops being true. Caffeine's W-TinyLFU admission,
striped frequency counters and amortised eviction are not a weekend's work to reproduce, and the
failure mode of getting them wrong is a cache that is *slower* than no cache — lock contention on
every lookup, or an eviction policy that throws away exactly the entries you were about to want.
## Lifecycle
Caches are released through `FlashContext.onClose`, so `app.stop()` drops every entry. That is
invisible in production with one app per process and matters immediately under test, where many
apps start and stop in one JVM.
## Statistics
```java
CacheStats stats = users.stats();
stats.hitRate(); // 0.0 until something is looked up
```
Requires `recordStats()` on the spec. Without it you get `CacheStats.DISABLED`, which is honest
about being unmeasured rather than reporting zeroes that look like a cold cache.
`manager.names()` lists every cache built so far, for an ops endpoint.
## What this is not
**HTTP caching.** If what you want is for the *client* to stop asking — `Cache-Control`, `ETag`,
`304 Not Modified` — that is a middleware, not an object cache, and it saves the whole request
rather than the lookup inside it. Reach for that first: it is cheaper, and the two solve different
problems.
**A shared cache.** Every replica has its own. Two instances will hold different values for the
same key, and an invalidation on one does not reach the other. When that becomes a problem the
answer is a networked backend — see the note on `flash-ext-cache-redis` — and the semantics change
with it: a cache that can fail is no longer transparent.
@@ -0,0 +1,39 @@
<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
<modelVersion>4.0.0</modelVersion>
<parent>
<groupId>dev.relism</groupId>
<artifactId>flash-extensions</artifactId>
<version>2.1.0-SNAPSHOT</version>
</parent>
<artifactId>flash-ext-cache-caffeine</artifactId>
<dependencies>
<dependency>
<groupId>dev.relism</groupId>
<artifactId>flash-ext-cache-core</artifactId>
</dependency>
<!--
Caffeine rather than a hand-rolled LRU: W-TinyLFU admission, striped counters and
amortised eviction are not a weekend's work to get right, and getting them wrong is a
cache that is slower than no cache.
-->
<dependency>
<groupId>com.github.ben-manes.caffeine</groupId>
<artifactId>caffeine</artifactId>
</dependency>
<dependency>
<groupId>org.junit.jupiter</groupId>
<artifactId>junit-jupiter</artifactId>
</dependency>
<dependency>
<groupId>dev.relism</groupId>
<artifactId>flash-testing</artifactId>
<scope>test</scope>
</dependency>
</dependencies>
</project>
@@ -0,0 +1,41 @@
package dev.relism.flash.ext.cache.caffeine;
import dev.relism.flash.ext.cache.Cache;
import dev.relism.flash.ext.cache.CacheStats;
import java.util.function.Function;
/**
* {@link Cache} over a Caffeine cache. A thin adapter by design: every method delegates directly,
* adding no wrapper object, no copy and no synchronisation of its own.
*/
final class CaffeineCache<K, V> implements Cache<K, V> {
private final com.github.benmanes.caffeine.cache.Cache<K, V> delegate;
private final boolean statsRecorded;
CaffeineCache(com.github.benmanes.caffeine.cache.Cache<K, V> delegate, boolean statsRecorded) {
this.delegate = delegate;
this.statsRecorded = statsRecorded;
}
@Override
public V get(K key, Function<? super K, ? extends V> loader) {
// Caffeine's own get(key, mappingFunction) already guarantees the loader runs once per key
// across concurrent callers; wrapping it in anything of ours would only add a race.
return delegate.get(key, loader);
}
@Override public V getIfPresent(K key) { return delegate.getIfPresent(key); }
@Override public void put(K key, V value) { delegate.put(key, value); }
@Override public void invalidate(K key) { delegate.invalidate(key); }
@Override public void invalidateAll() { delegate.invalidateAll(); }
@Override public long estimatedSize() { return delegate.estimatedSize(); }
@Override
public CacheStats stats() {
if (!statsRecorded) return CacheStats.DISABLED;
com.github.benmanes.caffeine.cache.stats.CacheStats snapshot = delegate.stats();
return new CacheStats(snapshot.hitCount(), snapshot.missCount(), snapshot.evictionCount());
}
}
@@ -0,0 +1,33 @@
package dev.relism.flash.ext.cache.caffeine;
import dev.relism.flash.ext.cache.CacheManager;
import dev.relism.flash.extension.FlashContext;
import dev.relism.flash.extension.FlashExtension;
import dev.relism.flash.extension.FlashRegistrar;
/**
* Installs an in-process {@link CacheManager} backed by Caffeine.
*
* <pre>{@code
* FlashApp.create(8080)
* .install(new CaffeineCacheExtension())
* .scan("dev.example.api");
* }</pre>
*
* <p>No configuration. Each cache declares its own size and TTL where it is built, because those
* are properties of what is being cached, not of the process caching it.
*
* <p>Caches are dropped through {@link FlashContext#onClose}, so a stopped app does not keep its
* values alive — which matters when many apps start and stop in one JVM, as they do under test.
*/
public final class CaffeineCacheExtension implements FlashExtension {
@Override
public void configure(FlashRegistrar<?> app, FlashContext ctx) {
ctx.supply(CacheManager.class, services -> {
CaffeineCacheManager manager = new CaffeineCacheManager();
services.onClose(manager::clear);
return manager;
});
}
}
@@ -0,0 +1,66 @@
package dev.relism.flash.ext.cache.caffeine;
import com.github.benmanes.caffeine.cache.Caffeine;
import dev.relism.flash.ext.cache.Cache;
import dev.relism.flash.ext.cache.CacheManager;
import dev.relism.flash.ext.cache.CacheSpec;
import java.util.Map;
import java.util.Set;
import java.util.concurrent.ConcurrentHashMap;
/** In-process {@link CacheManager} backed by Caffeine. */
final class CaffeineCacheManager implements CacheManager {
private final Map<String, Entry> caches = new ConcurrentHashMap<>();
@Override
@SuppressWarnings("unchecked")
public <K, V> Cache<K, V> build(String name, java.util.function.Consumer<CacheSpec> configure) {
CacheSpec spec = CacheSpec.of();
configure.accept(spec);
Entry entry = caches.computeIfAbsent(name, key -> new Entry(describe(spec), create(spec)));
// Two handlers sharing a cache is the point; two handlers disagreeing about its size or
// TTL is a bug that would otherwise resolve to whichever one ran first.
String requested = describe(spec);
if (!entry.signature.equals(requested))
throw new IllegalStateException("Cache '" + name + "' already exists as " + entry.signature
+ " but was requested as " + requested);
return (Cache<K, V>) entry.cache;
}
@Override
@SuppressWarnings("unchecked")
public <K, V> Cache<K, V> cache(String name) {
Entry entry = caches.get(name);
return entry == null ? null : (Cache<K, V>) entry.cache;
}
@Override
public Set<String> names() {
return Set.copyOf(caches.keySet());
}
/** Releases every entry so a stopped app does not keep its values alive. */
void clear() {
caches.values().forEach(entry -> entry.cache.invalidateAll());
caches.clear();
}
private static CaffeineCache<Object, Object> create(CacheSpec spec) {
Caffeine<Object, Object> builder = Caffeine.newBuilder();
if (spec.bounded()) builder.maximumSize(spec.maxSize());
if (spec.ttl() != null) builder.expireAfterWrite(spec.ttl());
if (spec.ttlAfterAccess() != null) builder.expireAfterAccess(spec.ttlAfterAccess());
if (spec.statsRecorded()) builder.recordStats();
return new CaffeineCache<>(builder.build(), spec.statsRecorded());
}
private static String describe(CacheSpec spec) {
return "maxSize=" + spec.maxSize() + " ttl=" + spec.ttl()
+ " ttlAfterAccess=" + spec.ttlAfterAccess() + " stats=" + spec.statsRecorded();
}
private record Entry(String signature, CaffeineCache<Object, Object> cache) {}
}
@@ -0,0 +1,160 @@
package dev.relism.flash.ext.cache.caffeine;
import dev.relism.flash.ext.cache.Cache;
import dev.relism.flash.ext.cache.CacheManager;
import dev.relism.flash.ext.cache.CacheStats;
import dev.relism.flash.extension.FlashApp;
import dev.relism.flash.extension.FlashConfiguration;
import dev.relism.flash.testing.FlashTest;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.extension.RegisterExtension;
import java.time.Duration;
import java.util.concurrent.CountDownLatch;
import java.util.concurrent.atomic.AtomicInteger;
import static org.junit.jupiter.api.Assertions.*;
class CaffeineCacheTest {
private static final AtomicInteger loads = new AtomicInteger();
@RegisterExtension
static FlashTest app = FlashTest.of(configured -> {
configured.install(new CaffeineCacheExtension());
configured.ctx().onReady(() -> {
Cache<String, String> users = configured.ctx().require(CacheManager.class)
.build("users", spec -> spec.maxSize(100).ttl(Duration.ofMinutes(5)).recordStats());
configured.get("/users/{id}", (req, res) ->
users.get(req.param("id"), id -> "loaded:" + id + ":" + loads.incrementAndGet()));
});
});
private static CacheManager manager() {
return app.app().ctx().require(CacheManager.class);
}
@Test
void aRepeatedRequestIsServedFromCache() {
String first = app.get("/users/alice").expectStatus(200).body();
String second = app.get("/users/alice").expectStatus(200).body();
assertEquals(first, second);
assertTrue(first.startsWith("loaded:alice:"));
}
@Test
void distinctKeysLoadSeparately() {
assertNotEquals(app.get("/users/bob").body(), app.get("/users/carol").body());
}
@Test
void statsCountHitsAndMisses() {
Cache<String, String> cache = manager().build("stats-probe", spec -> spec.maxSize(10).recordStats());
cache.get("k", key -> "v");
cache.get("k", key -> "v");
CacheStats stats = cache.stats();
assertEquals(1, stats.misses());
assertEquals(1, stats.hits());
assertEquals(0.5, stats.hitRate());
}
@Test
void statsAreDisabledUnlessAskedFor() {
Cache<String, String> cache = manager().build("no-stats", spec -> spec.maxSize(10));
cache.get("k", key -> "v");
assertEquals(CacheStats.DISABLED, cache.stats());
}
@Test
void theLoaderRunsOncePerKeyUnderConcurrency() throws Exception {
Cache<String, String> cache = manager().build("single-flight", spec -> spec.maxSize(10));
AtomicInteger invocations = new AtomicInteger();
int threads = 16;
CountDownLatch start = new CountDownLatch(1);
CountDownLatch done = new CountDownLatch(threads);
for (int i = 0; i < threads; i++) {
Thread.ofVirtual().start(() -> {
try {
start.await();
cache.get("hot", key -> {
invocations.incrementAndGet();
return "value";
});
} catch (InterruptedException e) {
Thread.currentThread().interrupt();
} finally {
done.countDown();
}
});
}
start.countDown();
assertTrue(done.await(5, java.util.concurrent.TimeUnit.SECONDS));
assertEquals(1, invocations.get(), "concurrent callers must share one load, not race");
}
@Test
void aNullLoaderResultStoresNothing() {
Cache<String, String> cache = manager().build("nulls", spec -> spec.maxSize(10));
assertNull(cache.get("missing", key -> null));
assertNull(cache.getIfPresent("missing"));
}
@Test
void invalidateDropsOneKeyAndInvalidateAllDropsEverything() {
Cache<String, String> cache = manager().build("invalidation", spec -> spec.maxSize(10));
cache.put("a", "1");
cache.put("b", "2");
cache.invalidate("a");
assertNull(cache.getIfPresent("a"));
assertEquals("2", cache.getIfPresent("b"));
cache.invalidateAll();
assertNull(cache.getIfPresent("b"));
}
@Test
void buildIsIdempotentPerName() {
Cache<String, String> first = manager().build("shared", spec -> spec.maxSize(10));
Cache<String, String> second = manager().build("shared", spec -> spec.maxSize(10));
assertSame(first, second, "two handlers asking for one cache must get one cache");
assertSame(first, manager().cache("shared"));
}
@Test
void disagreeingOnASharedCacheIsARejectedMistakeNotASilentWinner() {
manager().build("contested", spec -> spec.maxSize(10));
IllegalStateException conflict = assertThrows(IllegalStateException.class,
() -> manager().build("contested", spec -> spec.maxSize(999)));
assertTrue(conflict.getMessage().contains("contested"), conflict.getMessage());
}
@Test
void unknownNameReturnsNullRatherThanBuildingOne() {
assertNull(manager().cache("never-built"));
}
/** Why CaffeineCacheExtension registers onClose: values must not outlive the app holding them. */
@Test
void stoppingTheAppReleasesEveryCache() {
FlashApp standalone = FlashApp.create(FlashConfiguration.builder()
.port(0).host("127.0.0.1").shutdownDrainTimeoutMs(250).build())
.install(new CaffeineCacheExtension());
standalone.start();
CacheManager manager = standalone.ctx().require(CacheManager.class);
manager.build("scoped", spec -> spec.maxSize(10)).put("k", "v");
assertEquals(1, manager.names().size());
standalone.stop().join();
assertTrue(manager.names().isEmpty(), "caches must be released when the app stops");
}
}