Pluggable read-through/write-through caching layer for database operations with interchangeable backends, volatile query caching auto-invalidated on writes, optional write-behind, and per-entry TTL support.
A pluggable read-through/write-through cache layer for Kotlin database operations. Pick a cache backend, pick a database adapter, wrap your queries — no boilerplate.
Documentation · KDoc · Releases
Pick one cache backend and one database adapter. kachecontroller-core is pulled in transitively.
Replace LATEST_VERSION with the version from the badge above.
Kotlin Gradle
dependencies {
// cache backend — pick one
implementation("com.funyinkash:kachecontroller-cache-redis:LATEST_VERSION")
// implementation("com.funyinkash:kachecontroller-cache-memory:LATEST_VERSION")
// implementation("com.funyinkash:kachecontroller-cache-sqlite:LATEST_VERSION")
// database adapter — pick one
implementation("com.funyinkash:kachecontroller-mongo:LATEST_VERSION")
// implementation("com.funyinkash:kachecontroller-exposed:LATEST_VERSION")
}
Groovy Gradle
dependencies {
implementation 'com.funyinkash:kachecontroller-cache-redis:LATEST_VERSION'
implementation 'com.funyinkash:kachecontroller-mongo:LATEST_VERSION'
}
Maven
Also add the kotlinx-serialization plugin — all models must be @Serializable:
plugins {
kotlin("plugin.serialization") version "1.9.0"
}
@Serializable
data class User(
@SerialName("_id")
override val id: String = ObjectId().toHexString(),
val firstName: String,
val lastName: String,
) : Model
val cache = RedisCacheClient("redis://localhost:6379")
val controller = MongoKacheController(cache = cache)
See the example module for a complete runnable demo with MongoDB and Redis.
Every collection gets two Redis hashes:
"myApp:users" ← primary hash — one field per document (id → json)
"myApp:users:volatile" ← volatile hash — cached query results (key → json array)
Full breakdown: How It Works
Implement CacheClient to bring your own backend.
Full documentation including installation, API reference, examples, and internal architecture:
funyin.github.io/KacheController
KDoc (generated API reference):
funyin.github.io/KacheController/kdoc/latest/
| Module | Artifact | Purpose |
|---|
kachecontroller-core | kachecontroller-core | Model, CacheClient interface, KacheController base |
kachecontroller-cache-redis | kachecontroller-cache-redis | Redis backend via Lettuce |
kachecontroller-cache-memory | kachecontroller-cache-memory | In-memory backend (no external deps) |
kachecontroller-cache-sqlite | kachecontroller-cache-sqlite | SQLite backend via sqlite-jdbc |
kachecontroller-mongo | kachecontroller-mongo | MongoKacheController — MongoDB adapter |
kachecontroller-exposed | kachecontroller-exposed | ExposedKacheController — any Exposed-compatible DB |
<dependency>
<groupId>com.funyinkash</groupId>
<artifactId>kachecontroller-cache-redis-jvm</artifactId>
<version>LATEST_VERSION</version>
</dependency>
<dependency>
<groupId>com.funyinkash</groupId>
<artifactId>kachecontroller-mongo-jvm</artifactId>
<version>LATEST_VERSION</version>
</dependency>
val db = mongoClient.getDatabase("myApp")
val users = db.getCollection<User>("users")
// write-through
controller.set(users, User.serializer()) {
findOneAndUpdate(filter, update, options)
}
// read-through
val user = controller.get(userId, users, User.serializer()) {
find(Filters.eq("_id", userId)).firstOrNull()
}
// bulk write-through
controller.setAll(users, User.serializer()) {
if (insertMany(newUsers).wasAcknowledged()) newUsers else emptyList()
}
// bulk read-through
val allUsers = controller.getAll(users, User.serializer()) {
find().toList()
}
// filtered query — result cached as volatile, auto-invalidated on writes
val admins = controller.getAll(users, User.serializer(), cacheKey = "users:role:admin") {
find(Filters.eq("role", "admin")).toList()
}
// delete and evict
controller.remove(userId, users) {
deleteOne(Filters.eq("_id", userId)).wasAcknowledged()
}
| Backend | Class | Persistence | TTL support |
|---|
| Redis | RedisCacheClient | Shared, survives restarts | Full (per-field via HEXPIRE, Redis 7.4+) |
| In-memory | InMemoryCacheClient | Process-local | None |
| SQLite | SQLiteCacheClient | File-based | None |
"<dbName>:<collectionName>" (MongoDB) or "<schema>:<table>" (Exposed).getAll keys are volatile — stored in the volatile hash, auto-invalidated on writes.__kache_empty__ sentinel so repeated getAll calls don't hit the DB.removeAll uses DEL (atomic single command) to drop the entire collection hash.setAsync/setAllAsync) — updates cache immediately, queues the DB write. Not for transactional data.cacheEnabled — a () -> Boolean evaluated on every call; wire to a feature flag for runtime toggling.Surfaced from shared tags and platforms — no rankings paid for.