Lightbend’s Activator server will be decommissioned by the end of 2017. This means the below examples are not available at the moment.
The easiest way to get started is with a working application in Activator. The following templates are created by the Slick team, with updated versions being made for new Slick releases:
- To learn the basics of Slick, start with the Hello Slick template. It contains an extended version of the tutorial and code from this page.
- The Slick Plain SQL Queries template shows you how to do SQL queries with Slick.
- The Slick Multi-DB Patterns template shows you how to write Slick applications that can use different database systems and how to use custom database functions in Slick queries.
- The Slick TestKit Example template shows you how to use Slick TestKit to test your own Slick profiles.
There are more Slick templates created by the community, as well as versions of our own templates for other Slick releases. You can find all Slick templates on the Lightbend web site.
To include Slick in an existing project use the library published on Maven Central. For sbt projects add the
following to your build definition -
For Maven projects add the following to your
<dependencies> (make sure to use the correct Scala
_2.12, to match your project’s Scala version):
Slick uses SLF4J for its own debug logging so you also need to add an SLF4J
implementation. Here we are using
slf4j-nop to disable logging. You have
to replace this with a real logging framework like Logback if you want to see
The Reactive Streams API is pulled in automatically as a transitive dependency.
If you want to use Slick’s connection pool support for HikariCP, you need to add
slick-hikaricp module as a dependency as shown above. It will automatically
provide a compatible version of HikariCP as a transitive dependency. Otherwise, you
might need to disable connection pooling or specify a third-party connection pool.
Note: The rest of this chapter is based on the Hello Slick template. The preferred way of reading this introduction is in Activator, where you can edit and run the code directly while reading the tutorial.
To use Slick you first need to import the API for the database you will be using, like:
Since we are using H2 as our database system, we need to import features
H2Profile. A profile’s
api object contains all commonly
needed imports from the profile and other parts of Slick such as
Slick’s API is fully asynchronous and runs database calls in a separate thread pool. For running
user code in composition of
Future values, we import the global
ExecutionContext. When using Slick as part of a larger application (e.g. with Play or
Akka) the framework may provide a better alternative to this default
In the body of the application we create a
Database object which specifies how to connect to a
database. In most cases you will want to configure database connections with Typesafe Config in
application.conf, which is also used by Play and Akka for their configuration:
For the purpose of this example we disable the connection pool (there is no point in using one for an embedded in-memory database) and request a keep-alive connection (which ensures that the database does not get dropped while we are using it). The database can be easily instantiated from the configuration like this:
Databaseobject usually manages a thread pool and a connection pool. You should always shut it down properly when it is no longer needed (unless the JVM process terminates anyway).
Before we can write Slick queries, we need to describe a database schema with
Table row classes
TableQuery values for our tables. You can either use the code generator
to automatically create them for your database schema or you can write them by hand:
All columns get a name (usually in camel case for Scala and upper case with underscores for SQL) and a
Scala type (from which the SQL type can be derived automatically). The table object also needs a Scala
name, SQL name and type. The type argument of the table must match the type of the special
In simple cases this is a tuple of all columns but more complex mappings are possible.
foreignKey definition in the
coffees table ensures that the
supID field can only contain values
for which a corresponding
id exists in the
suppliers table, thus creating an n to one relationship:
Coffees row points to exactly one
Suppliers row but any number of coffees can point to the same
supplier. This constraint is enforced at the database level.
The connection to the embedded H2 database engine provides us with an empty database. Before we can
execute queries, we need to create the database schema (consisting of the
and insert some test data:
schema method creates
DDL (data definition language) objects with the database-specific
code for creating and dropping tables and other database entities. Multiple
DDL values can be combined with
++ to allow all entities to be created and dropped in the correct order, even when they have circular
dependencies on each other.
Inserting the tuples of data is done with the
++= methods, similar to how you add data to mutable
++= methods return a
DBIOAction which can be executed on a database
at a later time to produce a result. There are several different combinators for combining multiple
DBIOActions into sequences, yielding another action. Here we use the simplest one,
can concatenate any number of actions, discarding the return values (i.e. the resulting
produces a result of type
Unit). We then execute the setup action asynchronously with
db.run, yielding a
Note: Database connections and transactions are managed automatically by Slick. By default connections are acquired and released on demand and used in auto-commit mode. In this mode we have to populate the
supplierstable first because the
coffeesdata can only refer to valid supplier IDs. We could also use an explicit transaction bracket encompassing all these statements (
db.run(setup.transactionally)). Then the order would not matter because the constraints are only enforced at the end when the transaction is committed.
The simplest kind of query iterates over all the data in a table:
This corresponds to a
SELECT * FROM COFFEES in SQL (except that the
* is the table’s
we defined earlier and not whatever the database sees as
*). The type of the values we get in the loop
is, unsurprisingly, the type parameter of
Let’s add a projection to this basic query. This is written in Scala with the
map method or a
The output will be the same: for each row of the table, all columns get converted to strings and concatenated
into one tab-separated string. The difference is that all of this now happens inside the database engine, and
only the resulting concatenated string is shipped to the client. Note that we avoid Scala’s
(which is already heavily overloaded) in favor of
++ (commonly used for sequence concatenation). Also,
there is no automatic conversion of other argument types to strings. This has to be done explicitly with the
type conversion method
This time we also use Reactive Streams to get a streaming result from the database and print the elements as they come in instead of materializing the whole result set upfront.
Joining and filtering tables is done the same way as when working with Scala collections:
Note the use of
==for comparing two values for equality and
!=for inequality. This is necessary because these operators are already defined (with unsuitable types and semantics) on the base type
Any, so they cannot be replaced by extension methods. The other comparison operators are the same as in standard Scala code:
The generator expression
suppliers if s.id === c.supID follows the relationship established by the foreign
Coffees.supplier. Instead of repeating the join condition here we can use the foreign key directly: