JDBC Store

ff4k-store-jdbc provides JdbcFeatureStore and JdbcPropertyStore for backend databases on top of any javax.sql.DataSource. PostgreSQL and MySQL dialects are built in; implement JdbcDialect for another database. Features and properties are stored as JSON documents in ff4k_features and ff4k_properties; feature updates use optimistic locking with retry.

Calls are suspend functions that run the blocking JDBC work on Dispatchers.IO.

Installation

dependencies {
    implementation("com.yonatankarp:ff4k-store-jdbc:<version>")
    implementation("org.postgresql:postgresql:<driver-version>")   // or com.mysql:mysql-connector-j
}

Usage

val dataSource: DataSource = HikariDataSource(config)   // any pooled DataSource
JdbcSchema.create(dataSource)                            // CREATE TABLE IF NOT EXISTS, once at startup

val ff4k = FF4k(
    features = JdbcFeatureStore(dataSource, JdbcDialect.Postgres).cached(),
    properties = JdbcPropertyStore(dataSource, JdbcDialect.Postgres).cached(),
)

Wrap the stores with cached() so flag checks do not hit the database on every call.

Dialect	Minimum version	Notes
JdbcDialect.Postgres	9.5	ON CONFLICT upsert
JdbcDialect.Mysql	8.0.19	row alias syntax in ON DUPLICATE KEY

Other databases

Only the upsert statement differs between databases:

object H2Dialect : JdbcDialect {
    override fun upsert(table: String) =
        "MERGE INTO $table (id, data, version) KEY (id) VALUES (?, ?, 1)"
}

Custom strategies and property types

Pass the same Json built on ff4kSerializersModule to both JdbcFeatureStore and JdbcPropertyStore, exactly as for the SQLite store.