Access Cassandra From Scala – Cascal

I recently wrote about my experience using Hector (a Java-Cassandra client library) from Scala. Although I found it to be workable, I was looking for something even simpler and more in the spirit of Scala.

Chris Shorrock wrote a compact, pure Scala client library for Cassandra called Cascal. You can download a precompiled binary from GitHub but I chose to build the latest version myself. If you have Git and Maven already set up on your system you can just do:

$ git clone git://github.com/shorrockin/cascal.git
...
$ cd cascal
$ mvn package

After Maven is done building you’ll have a target directory. Add the main jar file (I got cascal-1.2-SNAPSHOT.jar) to the Build Path in Eclipse and make sure you have other Cassandra dependencies configured (thrift, commons-pool, slf4j, log4j, etc.) I then wrote a simple test object to perform the same functions as before and I think the resulting data access code is very clean and readable:

import com.shorrockin.cascal.utils.Conversions._
import com.shorrockin.cascal.session._

object CassandraCascalTester {
  def main(args : Array[String]) : Unit = {
    // Set up session pool
    val hosts  = Host("localhost", 9160, 250) :: Nil
    val params = new PoolParams(10, ExhaustionPolicy.Fail, 500L, 6, 2)
    val pool   = new SessionPool(hosts, params, Consistency.One)

    pool.borrow { session =>

      // Define column path
      val colPath = "Keyspace1" \ "Standard1" \ "key" \ "column-name"

      // insert
      session.insert(colPath \ "value")

      // read
      val col = session.get(colPath)
      System.out.println("Stored value: " + string(col.get.value))
    }
  }
}

As you can see, we don’t need to explicitly handle a try/catch block and Scala’s implicit type conversions take care of transforming strings to byte arrays and the directory-like column path syntax, all in a type-safe manner enforced by the compiler. For example, colPath will automatically become

Keyspace(value = Keyspace1) \ StandardColumnFamily(value = Standard1) \ StandardKey(value = key) \ Column(name = [B@6910fe28, value = null, time = 1272046381878)

and col will become

Some(Keyspace(value = Keyspace1) \ StandardColumnFamily(value = Standard1) \ StandardKey(value = key) \ Column(name = [B@30ec4a87, value = [B@6910fe28, time = 1272041137177))

Note that we do need to explicitly convert the retrieved value to a String because the basic get has no way of knowing what is stored in there. But Cascal has a solution for that too. Starting with version 1.1 Cascal includes a simple object mapping framework that can do the heavy lifting of populating objects from get and list results. I’ll explore this capability and its constraints in a future post.

Advertisement