Cached

Wraps a cached implicit T.

Looking for an implicit Cached[T] first triggers a look for an implicit T, caches the resulting tree, and returns it immediately and in subsequent look ups for an implicit Cached[T]. Thus, subsequent look ups do not trigger looking for an implicit T, only returning the instance kept in cache.

Beware that if the contexts in which two subsequent look ups are different, so that looking for a T in each of them doesn't return the same result, this change would be ignored by caching. Looking for a Cached[T] in the first context would put the implicit T of this context in cache, and then looking for a Cached[T] in the second context would return the former instance from the first context. E.g.

trait TC[T] {
  def msg: String
}

object First {
  implicit val tc: TC[Int] = new TC[Int] {
    val msg = "first"
  }

  def print() = println(implicitly[TC[Int]].msg)
  def printCached() = println(Cached.implicitly[TC[Int]].msg)
}

object Second {
  implicit val tc: TC[Int] = new TC[Int] {
    val msg = "second"
  }

  def print() = println(implicitly[TC[Int]].msg)
  def printCached() = println(Cached.implicitly[TC[Int]].msg)
}

First.print()
Second.print()
First.printCached()
Second.printCached()

would print "first" then "second" (non cached TC[Int] instances), then "first" twice (first instance, returned the second time too through the cache).