* thus a motivated caller may be able to expose the reference to another thread that would then
   * call cancel() and be unable to cancel the delegate.
   * There are a number of ways to solve this, none of which are very pretty, and it is currently
   * believed to be a purely theoretical problem (since the other actions should supply sufficient
   * write-barriers).
   */

  @CheckForNull @LazyInit private ListenableFuture<V> delegateRef;

   * doing so would be to derive a separate test library by using remove_from_jar to strip out
   * ClassValueValidator.
   *
   * Fortunately, we get pretty good coverage "by accident": We run all these tests against the
   * *backport*, where ClassValueValidator is not present.
   */

    assertEquals(0, logHandler.getStoredLogRecords().size());
  }

  public void testPartiallyConstructedManager_transitionAfterAddListenerBeforeStateIsReady() {
    // The implementation of this test is pretty sensitive to the implementation :( but we want to
    // ensure that if weird things happen during construction then we get exceptions.
    final NoOpService service1 = new NoOpService();

   * doing so would be to derive a separate test library by using remove_from_jar to strip out
   * ClassValueValidator.
   *
   * Fortunately, we get pretty good coverage "by accident": We run all these tests against the
   * *backport*, where ClassValueValidator is not present.
   */

       * finalizable references could be enqueued subsequently (at which point the class loader
       * would be resurrected by virtue of us having a strong reference to it), we should pretty
       * much just shut down and make sure we don't keep it alive any longer than necessary.
       */
      return null;
    }
    try {
      return finalizableReferenceClass.getMethod("finalizeReferent");

    assertEquals(0, logHandler.getStoredLogRecords().size());
  }

  public void testPartiallyConstructedManager_transitionAfterAddListenerBeforeStateIsReady() {
    // The implementation of this test is pretty sensitive to the implementation :( but we want to
    // ensure that if weird things happen during construction then we get exceptions.
    final NoOpService service1 = new NoOpService();

   * 'affect' a bit(j) in the output if two inputs, identical but for bit(i), will differ at output
   * bit(j) about half the time
   *
   * <p>Funneling is pretty simple to detect. The key idea is to find example keys which
   * unequivocally demonstrate that funneling cannot be occurring. This is done bit-by-bit. For each

       * finalizable references could be enqueued subsequently (at which point the class loader
       * would be resurrected by virtue of us having a strong reference to it), we should pretty
       * much just shut down and make sure we don't keep it alive any longer than necessary.
       */
      return null;
    }
    try {
      return finalizableReferenceClass.getMethod("finalizeReferent");

   * 'affect' a bit(j) in the output if two inputs, identical but for bit(i), will differ at output
   * bit(j) about half the time
   *
   * <p>Funneling is pretty simple to detect. The key idea is to find example keys which
   * unequivocally demonstrate that funneling cannot be occurring. This is done bit-by-bit. For each