LRU cache implementation in Scala

LRUCache.scala

import akka.stm._
import scala.collection.immutable.ListMap

case class LRUCache[A, B](private val MAX_ENTRIES: Int)
{
  protected val cache = Ref(ListMap.empty[A, B]) 

  def getOrElse(key: A)(fn: => B): B = { 
    get(key).getOrElse {
      val result = fn
      put(key, result)
      result
    }
  }

  def get(key: A): Option[B] = atomic {
    cache.get.get(key)
  }

  def put(key: A, value: B) = atomic {
    cache.alter { current =>
      val altered = current + (key -> value)
      if(altered.size > MAX_ENTRIES) altered.takeRight(MAX_ENTRIES) else altered
    }   
  }

  def remove(key: A) = atomic {
    cache.alter { current => current - key }
  }
}

just a FIFO rather than a LRU?

What about:

import collection.JavaConversions._
def lruCache[A, B](maxEntries: Int): mutable.Map[A, B] = new java.util.LinkedHashMap[A, B]() {
  override def removeEldestEntry(eldest: java.util.Map.Entry[A,B]) = size > maxEntries
}

The code above is not LRU, in the sense that does not update the freshness on access.

It is possible to get LRU behaviour using a different constructor:

object LruMap {
  import collection.JavaConversions._
  def apply[A, B](maxEntries: Int): collection.mutable.Map[A, B] = new java.util.LinkedHashMap[A, B](16, 0.75F, true) {
    override def removeEldestEntry(eldest: java.util.Map.Entry[A,B]): Boolean = size > maxEntries
  }
}

What are you supposed to do now that akka has dropped stm?
(and twitter dropped util-collections which had an LRUCache implementation.)