* the type parameter is <E>, since otherwise the
         * MultiExceptionListIterator wouldn't be an Iterator<E>. The cast is
         * safe, even though javac can't tell.
         *
         * Sun bug 6665356 is an additional complication. Until OpenJDK 7, javac
         * doesn't recognize this kind of cast as unchecked cast. Neither does

   *
   * This has important consequences: it means that the RateLimiter doesn't remember the time of the
   * _last_ request, but it remembers the (expected) time of the _next_ request. This also enables
   * us to tell immediately (see tryAcquire(timeout)) whether a particular timeout is enough to get
   * us to the point of the next scheduling time, since we always maintain that. And what we mean by

            new AsyncCallable<String>() {
              @Override
              public ListenableFuture<String> call() throws Exception {
                // Make this executor terminate after this task so that the test can tell when
                // futureResult has received resultOfCombiner.
                executor.shutdown();
                callableBlocking.await();
                return resultOfCombiner;
              }

    GWT happens to use the prod .gwt.xml, so it looks for no supersource for
    tests, either. This causes it to fail to find AtomicLongMapTest.

    Our workaround is to tell GWT that util.concurrent and all other packages
    have prod supersource, even if they have none. GWT is happy to ignore us
    when we specify a nonexistent path.

    GWT happens to use the prod .gwt.xml, so it looks for no supersource for
    tests, either. This causes it to fail to find AtomicLongMapTest.

    Our workaround is to tell GWT that util.concurrent and all other packages
    have prod supersource, even if they have none. GWT is happy to ignore us
    when we specify a nonexistent path.

         * predictable what work might be done. (e.g., close a file and flush buffers to disk). To
         * protect ourselves from this, we park ourselves and tell our interrupter that we did so.
         */
        if (state == PARKED || compareAndSet(state, PARKED)) {
          // Interrupting Cow Says:
          //  ______
          // < Park >
          //  ------

             * corresponds to the least significant nonzero byte in lw ^ rw, since lw and rw are
             * little-endian. Long.numberOfTrailingZeros(diff) tells us the least significant
             * nonzero bit, and zeroing out the first three bits of L.nTZ gives us the shift to get
             * that least significant nonzero byte.
             */

  public void testCombinationsExhaustively_startingFromArbitrary() {
    Ordering<Object> arbitrary = Ordering.arbitrary();
    Object[] array = {1, "foo", new Object()};

    // There's no way to tell what the order should be except empirically
    Arrays.sort(array, arbitrary);
    testExhaustively(arbitrary, array);
  }

  /**

  public void testCombinationsExhaustively_startingFromArbitrary() {
    Ordering<Object> arbitrary = Ordering.arbitrary();
    Object[] array = {1, "foo", new Object()};

    // There's no way to tell what the order should be except empirically
    Arrays.sort(array, arbitrary);
    testExhaustively(arbitrary, array);
  }

  /**

    GWT happens to use the prod .gwt.xml, so it looks for no supersource for
    tests, either. This causes it to fail to find AtomicLongMapTest.

    Our workaround is to tell GWT that util.concurrent and all other packages
    have prod supersource, even if they have none. GWT is happy to ignore us
    when we specify a nonexistent path.