MikArt Europe

Adapters are intentionally thin. They translate the canonical schema and query model to a backend, expose capabilities, and provide native access for work that should stay backend-specific.

Modules

Module	Backend
`tava-postgres`	PostgreSQL and compatible JDBC endpoints
`tava-mysql`	MySQL and MariaDB
`tava-sqlite`	SQLite files and shared in-memory databases
`tava-h2`	H2, mainly useful for tests and local tools
`tava-sqlserver`	Microsoft SQL Server
`tava-oracle`	Oracle Database
`tava-mongodb`	MongoDB sync driver
`tava-dynamodb`	AWS SDK DynamoDB client

JDBC Adapters

try (Tava tava = Tava.open(Postgres.connect(
        "jdbc:postgresql://localhost:5432/app",
        "app",
        "secret"))) {
    // use Tava
}

try (Tava tava = Tava.open(Postgres.connect(
        "localhost",
        5432,
        "app",
        "app",
        "secret"))) {
    // same adapter, host-based helper
}

MongoDB

MongoConnection.java

import eu.mikart.tava.Tava;
import eu.mikart.tava.mongodb.Mongo;

try (Tava tava = Tava.open(Mongo.connect(
        "mongodb://localhost:27017",
        "app"))) {
    // Tava owns the MongoClient created from the connection string.
}

If your application already manages the client lifecycle, pass it in:

ExistingMongoClient.java

MongoClient client = MongoClients.create(connectionString);

try (Tava tava = Tava.open(Mongo.use(client, "app"))) {
    // Tava does not close the externally owned client.
}

DynamoDB

DynamoDbConnection.java

import eu.mikart.tava.Tava;
import eu.mikart.tava.dynamodb.DynamoDb;

try (Tava tava = Tava.open(DynamoDb.createDefault())) {
    // Uses DynamoDbClient.create().
}

DynamoDbClient client = DynamoDbClient.builder().build();
try (Tava tava = Tava.open(DynamoDb.use(client))) {
    // Tava does not own the externally supplied client.
}

Mark the partition key on the record component:

DynamoEntity.java

import eu.mikart.tava.dynamodb.DynamoPartitionKey;

record Device(
        @DynamoPartitionKey String tenantId,
        String id,
        String status
) {}

Native Type Defaults

Adapters choose sensible native defaults for logical types. A few examples:

Logical type	PostgreSQL	MySQL	SQLite	SQL Server	Oracle
`UUID`	`UUID`	adapter default	`TEXT`	`UNIQUEIDENTIFIER`	`VARCHAR2(36)`
`JSON`	`JSONB`	`JSON`	`TEXT`	`NVARCHAR(MAX)`	`CLOB`
`BINARY`	`BYTEA`	`LONGBLOB`	`BLOB`	`VARBINARY(MAX)`	`BLOB`
`INSTANT`	`TIMESTAMP WITH TIME ZONE`	adapter default	`TEXT`	`DATETIMEOFFSET`	`TIMESTAMP WITH TIME ZONE`

Use typed adapter annotations when the default is not the storage shape you want.

NativeTypeOverride.java

record AuditEvent(
        @PostgresType("JSONB")
        @OracleType("CLOB")
        Map<String, Object> payload
) {}

Managed services

Managed databases that expose the same protocol use the same adapter. For example, a managed PostgreSQL endpoint still uses tava-postgres.

Adapters

Modules

JDBC Adapters

MongoDB

DynamoDB

Native Type Defaults

On this page