rsschermer / entitytled   0.8.0

MIT License GitHub

Data access and persistence library build on top of Scala Slick.

Scala versions: 2.11 2.10

Entitytled

Entitytled is an ORM-like data access and persistence library build on top of Scala Slick. Entitytled adds "entity" and "relationship" concepts, with the aims of improving structure and consistency, and eliminating boilerplate.

Build Status

Table of contents

Artifact installation

The Entitytled artifact is available on Sonatype's central repository and is cross-build for Scala 2.10 and Scala 2.11.

To make SBT automatically use the correct build based on your project's Scala version, add the following to your build file:

libraryDependencies += "com.github.rsschermer" %% "entitytled-core" % "0.8.0"

To use the 2.10 build explicitly, add:

libraryDependencies += "com.github.rsschermer" % "entitytled-core_2.10" % "0.8.0"

To use the 2.11 build explicitly, add:

libraryDependencies += "com.github.rsschermer" % "entitytled-core_2.11" % "0.8.0"

Defining an Entity type

In Entitytled, entity types are case classes which extend the Entity base class. They need to be accompanied by an EntityTable definition, which describes the schema for the entity. They may additionally by accompanied by an EntityCompanion object, which provides a default set of methods for querying and persisting entities.

This example defines a Director entity, which models a movie director:

import entitytled.profile.H2Profile._
import entitytled.profile.H2Profile.driver.api._

case class Director(
    id: Option[Long],
    name: String,
    age: Int)
  extends Entity[Director, Long]

object Director extends EntityCompanion[Directors, Director, Long]

class Directors(tag: Tag) extends EntityTable[Director, Long](tag, "DIRECTORS") {
  def id = column[Long]("id", O.PrimaryKey, O.AutoInc)
  def name = column[String]("name")
  def age = column[Int]("age")

  def * = (id.?, name) <> ((Director.apply _).tupled, Director.unapply)
}

Lets break this example up and take a closer look at the individual pieces.

Importing a profile

Slick provides drivers for multiple databases. Entitytled can be used with any driver that implements scala.slick.driver.JdbcProfile. The Director example uses a predefined profile for the H2 database:

import entitytled.profile.H2Profile._
import entitytled.profile.H2Profile.driver.api._

The first line imports Entitytled's functionality for Slick's H2 driver and the second line imports Slick's own functionality for the H2 driver. (Note: Entitytled also supports dependency injection via the cake pattern. However, these examples will use hard-coded import statements instead, as the cake pattern adds a certain amount of complexity.)

Predefined profiles exist for several other databases:

  • entitytled.profile.PostgresProfile for use with PostgreSQL.
  • entitytled.profile.MySQLProfile for use with MySQL.
  • entitytled.profile.SQLiteProfile for use with SQLite.
  • entitytled.profile.DerbyProfile for use with Apache Derby.
  • entitytled.profile.HsqldbProfile for use with HyperSQL DataBase.

To use e.g. PostgreSQL instead of H2 use:

import entitytled.profile.PostgresProfile._
import entitytled.profile.PostgresProfile.driver.api._

You may also define your own profile for a driver that is not included in the standard distribution of Slick:

package models.meta

import entitytled.Entitytled
import my.driver.MyDriver

trait MyProfile extends Entitytled {
  val driver: MyDriver = MyDriver
}

object MyProfile extends MyProfile

And use it like this:

import models.meta.MyProfile._
import models.meta.MyProfile.driver.api._

Defining the Entity type

The second segment of interest in the Director example is the definition of the Director case class:

case class Director(
    id: Option[Long],
    name: String,
    age: Int)
  extends Entity[Director, Long]

The Director case class extends the Entity base class, which takes two type parameters: the Director type itself (a self bound), and the type of its ID:

extends Entity[Director, Long]

The id field then needs to be of type Option[IdType], which in this case means it needs to be of type Option[Long].

Defining the companion object

Next the Director example defines a Director companion object for the Director case class:

object Director extends EntityCompanion[Directors, Director, Long]

This companion object will be our entry point for querying and persisting directors. It extends EntityCompanion which takes the table type (Directors), the entity type (Director), and the entity's ID type (Long) as type parameters.

Defining the table type

The final part of the Director example defines the Directors table:

class Directors(tag: Tag) extends EntityTable[Director, Long](tag, "DIRECTORS") {
  def id = column[Long]("id", O.PrimaryKey, O.AutoInc)
  def name = column[String]("name")
  def age = column[Int]("age")

  def * = (id.?, name, age) <> ((Director.apply _).tupled, Director.unapply)
}

Table definition is essentially the same as it is in Slick and you can find more details in Slick's documentation on schemas. Instead of extending a regular Table, a table definition for an entity needs to extend the EntityTable base class. The EntityTable base class takes two type parameters, the entity type and the ID type, and requires that you at least define an id method that returns a value of type Column[IdType]:

def id = column[Long]("id", O.PrimaryKey, O.AutoInc)

Usually the id column is also the primary key of an entity table.

Safer IDs

To keep this example simple, the ID type was specified simply as Long. However, if we'd also have a Producer entity type, which also specified its ID type as a simple Long, we could be at risk of mixing up our director IDs with our producer IDs. Slick also supports using more precise types that wrap a primitive type. We could make use of this to make our id field a bit safer:

case class DirectorID(value: Long) extends MappedTo[Long]

We'd then have to update our Director entity definition as follows:

case class Director(
    id: Option[DirectorID],
    name: String,
    age: Int)
  extends Entity[Director, DirectorID]

object Director extends EntityCompanion[Directors, Director, DirectorID]

class Directors(tag: Tag) extends EntityTable[Director, DirectorID](tag, "DIRECTORS") {
  def id = column[DirectorID]("id", O.PrimaryKey, O.AutoInc)
  def name = column[String]("name")
  def age = column[Int]("age")

  def * = (id.?, name) <> ((Director.apply _).tupled, Director.unapply)
}

The Holywood example used for testing also uses this safer way of handling IDs.

Defining direct relationships

Direct relationships are relationships that involve at most two tables: one table for the owner entity type and one table for the target relation. The relationship is defined by a foreign key, either on the owner or the target.

Direct 'to one' relationships

Let's define a Movie entity type which has a 'to one' relationship with the Director entity type:

case class Movie(
    id: Option[Long],
    title: String,
    directorID: Long)(implicit includes: Includes[Movie])
  extends Entity[Movie, Long]
{
  val director = one(Movie.director)
}

object Movie extends EntityCompanion[Movies, Movie, Long] {
  val director = toOne[Directors, Director]
}

class Movies(tag: Tag) extends EntityTable[Movie, Long](tag, "MOVIES") {
  def id = column[Long]("id", O.PrimaryKey, O.AutoInc)
  def title = column[String]("title")
  def directorID = column[Long]("director_id")

  def * = (id.?, title, directorID) <> ((Movie.apply _).tupled, Movie.unapply)

  def director = foreignKey("MOVIES_DIRECTOR_FK", directorID, TableQuery[Directors])(_.id)
}

The Movie 'to one' Director relationship in this example is a direct relationship; the foreign key in this example is the directorID field on the Movie entity type. We've also specified a foreign key constraint in our table definition (see Slick's documentation for more details on table definition):

def director = foreignKey("MOVIES_DIRECTOR_FK", directorID, TableQuery[Directors])(_.id)

Apart from having a foreign key field, Movie differs in some other ways from the simpler Director example that did not have any relationships. The first is the addition of an implicit includes parameter to the case class constructor:

(implicit includes: Includes[Movie])

This is to support eager-loading (a topic discussed in more detail in the "Querying an Entity set" section of this guide). It does not have to be named includes, but it does need to be of type Includes[EntityType], which in this case means it has to be of type Includes[Movie].

The second difference is the addition of a director field in the case class body:

val director = one(Movie.director)

Let's skip that difference for a second and look at the third difference first, the addition of a director field on the companion object:

val director = toOne[Directors, Director]

This defines the relationship by calling the toOne function. The toOne function takes two type parameters: the target table type (Directors in this case) and the target type (Director in this case). You can also specify two optional parameters:

  • toQuery: a query identifying the set of all target relations of the target type that could possibly be related to a movie. In this case the target type is Director, so the toQuery must identify a set of directors.
  • joinCondition: the condition that joins a specific owner instance to the subset of relations identified by toQuery that belong to it. If you don't specify this parameter, Entitytled will attempt to infer a join condition at compile time, based on the foreign keys you defined on the owner table type and the target table type. For Entitytled to be able to do this, there must be exactly one candidate foreign key (a foreign from the owner table to the target table, or from the target table to the owner table). If there's more than one candidate, or if you did not define any candidate foreign keys, you will get a compiler error. You can always resolve this error by providing a join condition explicitly.

The example just uses the defaults for these parameters, which is essentially equivalent to:

val director = toOne[Directors, Director](
  toQuery       = TableQuery[Directors],
  joinCondition = (m: Movies, d: Directors) => m.directorID === d.id
)

The default toQuery allows any director to be a target for this relationship and the default join condition states that a director is related to a movie if the movie's directorID equals the director's id (which is exactly what the foreign key constraint we defined on the table describes).

Now, let's take another look at the director field on our Movie case class:

val director = one(Movie.director)

This calls the one function, which takes the relationship we defined in the companion object as an argument. If we have a specific movie instance, this field will represent the related director:

titanic.director // James Cameron

Navigating relationships is described in more detail in its own section.

Direct 'to many' relationships

Direct 'to many' relationships are very similar to 'to one' relationships. Here's a modified version of the simple Director example, except now with a 'to many' relationship to the Movie type:

case class Director(
    id: Option[Long],
    name: String,
    age: Int)(implicit includes: Includes[Director])
  extends Entity[Director, Long]
{
  val movies = many(Director.movies)
}

object Director extends EntityCompanion[Directors, Director, Long] {  
  val movies = toMany[Movies, Movie]
}

class Directors(tag: Tag) extends EntityTable[Director, Long](tag, "DIRECTORS") {
  def id = column[Long]("id", O.PrimaryKey, O.AutoInc)
  def name = column[String]("name")
  def age = column[Int]("age")

  def * = (id.?, name, age) <> ((Director.apply _).tupled, Director.unapply)
}

There's no foreign key field here, that's already covered by the foreign key on the Movie type. Just as with the Movie type, an implicit includes constructor argument was added to the case class to enable eager-loading.

We've added a movies field to case class body, except this time it calls the many function instead of the one function:

val movies = many(Director.movies)

many works exactly the same as one, except it is used for representing a related collection, instead of a single related instance.

The last change made, is the addition of the movies field on the companion object:

val movies = toMany[Movies, Movie]

This is basically the inverse of the director field on the Movie companion object described in the previous section. Together, this 'to many' relationship definition and the 'to one' relationship on the Movie companion object, form the full 'Director one-to-many Movie' relationship.

One thing better illustrated for a 'to many' relationship (but technically also possible for a 'to one' relationship) is that we can further constrain the set of possible related instances via the optional toQuery argument. We might for example define a relationship for only black-and-white movies:

val blackAndWhiteMovies = toMany[Movies, Movie](
  toQuery = TableQuery[Movies].filter(_.color === false)
)

Defining indirect relationships (many-to-many)

Indirect relationships rely on a 'join table'. Instead of one of the tables for the two related types defining a foreign key, a separate table is responsible for recording the relationships between these types. This is typically how 'many-to-many' relationships are defined.

As an example we'll set up a many-to-many relationship between the Movie type (which we already used in the direct relationships example) and a new Star type, which describes a movie star. This is what the join table looks like:

class MoviesStars(tag: Tag) extends Table[(Long, Long)](tag, "MOVIES_STARS") {
  def movieID = column[Long]("movie_id")
  def starID = column[Long]("star_id")

  def * = (movieID, starID)

  def pk = primaryKey("MOVIES_STARS_PK", (movieID, starID))
  def movie = foreignKey("MOVIES_STARS_MOVIE_FK", movieID, TableQuery[Movies])(_.id)
  def star = foreignKey("MOVIES_STARS_STAR_FK", starID, TableQuery[Stars])(_.id)
}

Note that this does not extend EntityTable, it just extends Slick's plain old Table. That's because the rows in this table do not represent entities, it exists only so we can describe our many-to-many relationship. There are 2 foreign keys, one for the star's ID and one for the movie's ID, and the (movieID, starID) pairing of these foreign keys makes up the table's primary key.

Now for the Movie type:

case class Movie(
    id: Option[Long],
    title: String)(implicit includes: Includes[Movie])
  extends Entity[Movie, Long]
{
  val stars = many(Movie.stars)
}

object Movie extends EntityCompanion[Movies, Movie, Long] {  
  val stars = toManyThrough[Stars, MoviesStars, Star]
}

class Movies(tag: Tag) extends EntityTable[Movie, Long](tag, "MOVIES") {
  def id = column[Long]("id", O.PrimaryKey, O.AutoInc)
  def title = column[String]("title")

  def * = (id.?, title) <> ((Movie.apply _).tupled, Movie.unapply)
}

It's very similar to the Movie type definition in the Direct relationships example. The Director relationship stuff (the foreign key and director fields on the case class and companion object) has been removed to keep it simple.

We've added a stars field to the case class body so we can navigate the relationship again:

val stars = many(Movie.stars)

This works exactly the same for indirect 'to many' relationships as it does for direct 'to many' relationships.

The only real difference comes with the stars field on the companion object:

val stars = toManyThrough[Stars, MoviesStars, Star]

Instead of calling toMany as we would do for a direct relationship, we call toManyThrough. toManyThrough takes 3 type parameters: the target table type (Stars), the join table type (MoviesStars) and the target type (Star).

Just as with direct relationships, toManyThrough may optionally be given a toQuery argument and a joinCondition argument. The above example is essentially equivalent to:

val stars = toManyThrough[Stars, MoviesStars, Star](
  toQuery       = TableQuery[MoviesStars] join TableQuery[Stars] on(_.starID === _.id),
  joinCondition = (m: Movies, t: (MovieStars, Stars)) => m.id === t._1.movieID
)

With indirect relationships, the toQuery argument's type is a bit more complicated. Instead of a query that represents a set of target instances, the toQuery for an indirect relationship represents a set of (JoinType, TargetType) pairs, in this case (MoviesStars, Stars). This is achieved by joining the join table onto the target table:

toQuery = TableQuery[MoviesStars] join TableQuery[Stars] on(_.starID === _.id)

If you don't specify the toQuery argument explicitly, Entitytled will attempt to infer a default toQuery at compile time by examining the foreign keys that were defined on the join table and the target table. For Entitytled to be able to do this, there must be exactly 1 candidate foreign key (a foreign key from the join table to the target table, or from the target table to the join table). If there's more than one candidate foreign key or if no candidate foreign keys were defined, a compiler error is raised. You can always resolve these errors by providing the toQuery explicitly.

The joinCondition argument is also slightly more complicated for indirect relationships. Instead of simply joining the owner type to the target type as with direct relationships, it now has to join the owner type to this joined (JoinType, TargetType) pair:

joinCondition = (m: Movies, t: (MovieStars, Stars)) => m.id === t._1.movieID

If you don't specify the joinCondition argument explicitly, Entitytled will attempt to infer a default joinCondition at compile time by examining the foreign keys that were defined on the owner table and the join table. For Entitytled to be able to do this, there must be exactly 1 candidate foreign key (a foreign key from the owner table to the join table, or from the join table to the owner table). Again, if there's more than one candidate foreign key or if no candidate foreign keys were defined, a compiler error is raised, which can always be resolved by providing the joinCondition explicitly.

To complete the example, here's the definition of the Star type:

case class Star(
    id: Option[Long],
    name: String)(implicit includes: Includes[Star])
  extends Entity[Star, Long]
{
  val movies = many(Star.movies)
}

object Star extends EntityCompanion[Stars, Star, Long] {
  val movies = toManyThrough[Movies, MoviesStars, Movie]
}

class Stars(tag: Tag) extends EntityTable[Star, Long](tag, "STARS") {
  def id = column[Long]("id", O.PrimaryKey, O.AutoInc)
  def name = column[String]("name")

  def * = (id.?, name) <> ((Star.apply _).tupled, Star.unapply)
}

There's nothing new going on here, it just defines the inverse relationship to complete the full many-to-many relationship.

Composing relationships

You may wish to define relationships that span across several tables. One option is to use an indirect relationship for this with a custom toQuery. If, for example, you want to define a relationship from stars to directors, via the movies they're both related to, you could add the following relationship definition to the Star companion object:

object Star extends EntityCompanion[Stars, Star, Long] {
  val movies = toManyThrough[Movies, MoviesStars, Movie]
  
  // Relationship with director using an indirect relationship with a custom toQuery
  val directors = toManyThrough[Directors, MoviesStars, Director](
    toQuery = TableQuery[MoviesStars].join(TableQuery[Movies]).on(_.movieID === _.id)
      .join(TableQuery[Directors]).on(_._2.directorID === _.id)
      .map(join => (join._1._1, join._2))
  )
}

We've joined together the MoviesStars, Movies and Directors tables using their foreign keys. This resulted in ((MoviesStars, Movies), Directors) pairs which we then mapped back to (MoviesStars, Directors) pairs, which conforms to the required query type for this relationship's toQuery.

This works, but since we've already defined both a movies relationship on the Star companion object, and a director relationship on the Movie companion object, we can achieve the same result in a more convenient way: relationship composition. Instead of defining a complicated toQuery as we did earlier, we could do the following:

object Star extends EntityCompanion[Stars, Star, Long] {
  val movies = toManyThrough[Movies, MoviesStars, Movie]
  
  // Composed relationship
  val directors = movies compose Movie.director
}

We've composed the movies relationship with the Movie.director relationship. Composing a 'to many' relationship with a 'to one' relationship will result in a new 'to many' relationship. Composing two 'to many' relationships will also result in a new 'to many' relationship. Composing two 'to one' relationships will result in a new 'to one' relationship.

Composed relationships behave just like normal relationships. We can add a directors field on the Star case class for navigating this relationship:

case class Star(
    id: Option[Long],
    name: String)(implicit includes: Includes[Star])
  extends Entity[Star, Long]
{
  val movies = many(Star.movies)
  val directors = many(Star.directors)
}

It's also possible to further compose composed relationships with other relationships (although at some point the queries for retrieving the composed relationship will become so monstrous, that you may want to consider creating a new join table to cache the relationship to improve performance).

Querying an entity set

The entry point for an entity query is the companion object for that entity. Although you may also use Slick's TableQuery directly, the companion object adds some additional functionality for working with entities.

As of Slick 3.0, all interactions with the database happen via a database I/O actions (DBIOAction). A DBIOAction represents one or more operations that are to be executed on a database, such as a read query, inserting a new record,
or deleting a record. You then use a specific database definition to run such
an action, which will produce a future holding the action's result:

db.run(someAction) // Future holding the action's result

If you're not familiar with Slick, I recommend you take some time to read the following chapters from Slick's documentation before you proceed:

Building read actions

The companion object defines 2 methods which can be used to start building a read action:

  • all returns an intermediate query for all entities of this type:

    Movie.all // Intermediate query for all movies
  • one takes an entity ID as its argument and returns an intermediate query for the specific entity with that ID:

    Movie.one(193) // Intermediate query for the movie with ID 193

Both produce intermediate query results which can be modified further. This works exactly the same as it does in Slick and you have access to all of Slick's result modifying operations (e.g. filter, sortBy, map, take, etc.). If, for example, you want to build a query for the top 10 movies with the highest rating, you could do this:

val topTenMoviesQuery = Movie.all.sortBy(_.rating.asc).take(10)

Just as in Slick, calling result on an query will build a database I/O action that can be run on the database:

val topTenMoviesAction = topTenMoviesQuery.result
val topTenMovies: Future[Seq[Movie]] = db.run(topTenMoviesAction)

Entitytled adds one special result modifying operation: include. include can be used to eager-load relationships. Eager-loading is a way to solve the n + 1 query problem:

db.run(Movie.all.result).onSuccess { _.foreach { m => 
    println(s"${m.name} was directed by:")
    println(m.director.map(_.name).getOrElse("Unknown"))
  }
}

This snippet runs a database I/O action which will produce a future holding all movies. If this future runs successfully, it will print out each movie's name and the name of the director who directed it. If there are 1000 movies, this will execute 1001 queries: 1 to retrieve all movies and then one for each movie to retrieve its director. This is obviously not desirable.

The solution to this problem is to eager-load the directors with include:

db.run(Movie.all.include(Movie.director).result).onSuccess { _.foreach { m =>
    println(s"${m.name} was directed by:")
    println(m.director.map(_.name).getOrElse("Unknown"))
  }
}

This modified version using include will execute only 2 queries: one to retrieve all the movies and one to retrieve all the directors related to these movies.

You may eager-load multiple relationships:

Movie.all.include(Movie.director, Movie.stars)

Nested eager-loading is also possible:

Director.all.include(Director.movies.include(Movie.stars))

You can eager-load an arbitrary number of relationships and nest eager-loads to arbitrary depth. The current implementation will execute one additional query for every eager-loaded relationship (both for sibling eager-loads and nested eager-loads).

include can be chained onto any query that produces an entity result:

val oldestTenMaleDirectorsWithMovies = 
  Director.all
    .filter(_.male === true)
    .sortBy(_.age.asc)
    .take(10)
    .include(Director.movies)
    .result

include cannot be chained onto queries that don't produce an entity result (e.g. Director.all.size, which will produce an integer result, not an entity result).

Actions that do eager-loading cannot be streamed.

Creating an insert action

An insert action can be created by calling the insert method on the companion object, which takes the new entity as an argument:

val insertAction = Star.insert(Star(None, "Marlon Brando"))

This action's result value will be the id of the newly inserted entity:

val id: Future[Long] = db.run(insertAction)

Creating an update action

An update action can be created by calling the update method on the companion object, which takes the updated entity as an argument:

val updateAction = Director.update(martinScorsese.copy(age = 72))

Only entity instances with an ID (the ID is not None) can be updated, otherwise an exception will be thrown.

Creating a delete action

A delete action can be created by calling the delete method on the companion object, which takes the ID of the entity to be deleted as an argument:

val deleteAction = Movie.delete(38)

Navigating relationship values

'To one' relationships (both direct and indirect) can be used as values of type Option[RelatedType]:

val director: Option[Director] = someMovie.director

'To many' relationships (both direct and indirect) can be used as values of type Seq[RelatedType]:

val stars: Seq[Star] = someMovie.stars

This behaviour is achieved through implicit conversions. In reality, eager-loaded relationship values are wrapped in OneFetched or ManyFetched for 'to one' and 'to many' relationships respectively; relationships that have not yet been loaded are represented by OneUnfetched or ManyUnfetched for 'to one' and 'to many' relationships respectively. These wrappers are collectively referred to as 'relationship value representations'.

Both fetched and unfetched relationship value representations expose the valueAction method, which returns a database I/O action which, when run, will result in a future holding the relationship's value:

db.run(someMovie.director.valueAction) // Future[Option[Director]]

However, for relationship values that were prefetched with include, this won't actually run a database query, but will simply return the prefetched value without making an extra round trip to the database.

The aforementioned implicit conversion will run this valueAction for you and then wait for the resulting future to resolve. For this to work, your database definition must be available as an implicit value in the current context:

import MyProfile.driver.api.Database

implicit val db: Database = ...

val movieDirector: Option[Director] = someMovie.director

The above example using the implicit conversion, is essentially equivalent to the following explicit example:

import scala.concurrent.Await
import MyProfile.driver.api.Database

val db: Database = ...

val movieDirector: Option[Director] = Await.result(db.run(someMovie.director.valueAction))

As in this explicit example, using the implicit conversion will block thread execution while it is waiting for the future result to resolve. When this is not desirable, it's best to run the valueAction explicitly and to handle the future result asynchronously:

db.run(someMovie.director.valueAction).onSuccess { director =>
  ...
}

Note however, that if you prefetched the relationship value with include, the valueAction will not actually result in a database round trip and the future will resolve immediately. In short: if you know upfront which relationship values you need, prefetch them with include and you can use the implicit conversions without worrying about blocking thread execution; if for some reason you can't prefetch the relationship values, consider running the valueAction explicitly and handling its future result asynchronously.

If you want to execute different code paths depending on whether or not the value was prefetched, you can pattern-match against the relationship value representation:

someMovie.director match {
  case OneFetched(relationship, value, ownerID) =>
    println(s"The director was eager-loaded: ${value.name}!")
  case _ =>
    println("The director hasn't been loaded yet...")
}

If you want to force the execution of a new query to retrieve a fresh value, regardless of whether it was already prefetched or not, first call asUnfetched:

// Will run a new database query, regardless of whether the director value was 
// already prefetched
val movieDirector: Option[Director] = someMovie.director.asUnfetched

Play Framework

Entitytled does not yet provide specific integration with Play Framework. However, using it together with play-slick is relatively straightforward. Add a play-slick dependency to your build for a matching version of Slick:

Entitytled version Slick version
<= 0.5.x 2.1.x
>= 0.6.0 < 0.8.0 3.0.0
>= 0.8.0 3.1.0

(The play-slick readme provides a table matching versions of the plugin to versions of Slick and versions of Play.)

Then define a profile as follows:

package models.meta

import entitytled.Entitytled
import play.api.db.slick.DatabaseConfigProvider

trait Profile extends Entitytled {
  val dbConfig = DatabaseConfigProvider.get[JdbcProfile](Play.current)
  
  val driver = dbConfig.driver
}

object Profile extends Profile

And use it like this:

import models.meta.Profile._
import models.meta.Profile.driver.api._

This will make Entitytled use the Slick driver you've configured in your Play config file.

Activator template

The simple activator sample gives an example of using Entitytled with Play Framework.