* Note that when {@code containsKey(key)} is false, this returns an empty collection, not {@code
   * null}.
   *
   * <p>Changes to the returned collection will update the underlying multimap, and vice versa.
   */
  Collection<V> get(@ParametricNullness K key);

  /**
   * Returns a view collection of all <i>distinct</i> keys contained in this multimap. Note that the

   * natural ordering.
   *
   * <p><b>Note:</b> if mutability is not required, use {@link ImmutableSortedSet#copyOf(Iterable)}
   * instead.
   *
   * <p><b>Note:</b> If {@code elements} is a {@code SortedSet} with an explicit comparator, this
   * method has different behavior than {@link TreeSet#TreeSet(SortedSet)}, which returns a {@code
   * TreeSet} with that comparator.
   *

 * it being arbitrarily reordered, and you want to avoid that copy, you can use {@code
 * computeInPlace} instead of {@code compute}.
 *
 * <h3>Definition and notes on interpolation</h3>
 *
 * <p>The definition of the kth q-quantile of N values is as follows: define x = k * (N - 1) / q; if
 * x is an integer, the result is the value which would appear at index x in the sorted dataset

 *       so) before returning from the method. The {@code joinPool} method can be used to do this
 *       when using Executors.
 * </ol>
 *
 * <p><b>Other notes</b>
 *
 * <ul>
 *   <li>Usually, there is one testcase method per JSR166 method covering "normal" operation, and
 *       then as many exception-testing methods as there are exceptions the method can throw.

  //    Imagine:
  //    guard.lock();
  //    try { /* monitor locked and guard satisfied here */ }
  //    finally { guard.unlock(); }
  // Here are Justin's design notes about this:
  //
  // This idea has come up from time to time, and I think one of my
  // earlier versions of Monitor even did something like this. I ended
  // up strongly favoring the current interface.
  //

 * for comparisons. This class is not being deprecated, but we gently encourage you to migrate to
 * streams.
 *
 * <p><i>Performance notes:</i> Unless otherwise noted, all of the iterables produced in this class
 * are <i>lazy</i>, which means that their iterators only advance the backing iteration when
 * absolutely necessary.
 *
 * <p>See the Guava User Guide article on <a href=

 *   <li><a href="https://github.com/google/guava/wiki/GraphsExplained#synchronization">
 *       Synchronization policy</a>
 *   <li><a href="https://github.com/google/guava/wiki/GraphsExplained#notes-for-implementors">Notes
 *       for implementors</a>
 * </ul>
 *
 * @author James Sexton
 * @author Joshua O'Madadhain
 * @param <N> Node parameter type
 * @since 20.0
 */
@Beta

   *     // operate on tuple
   *   }
   * }
   * }</pre>
   *
   * <p>Note that if any input list is empty, the Cartesian product will also be empty. If no lists
   * at all are provided (an empty list), the resulting Cartesian product has one element, an empty
   * list (counter-intuitive, but mathematically consistent).
   *
   * <p><i>Performance notes:</i> while the cartesian product of lists of size {@code m, n, p} is a

 * as the traditional heap data structure used in {@link PriorityQueue}.
 *
 * <p>This class is not thread-safe, and does not accept null elements.
 *
 * <p><i>Performance notes:</i>
 *
 * <ul>
 *   <li>If you only access one end of the queue, and do use a maximum size, this class will perform
 *       significantly worse than a {@code PriorityQueue} with manual eviction above the maximum

        new ThreadFactoryBuilder()
            .setDaemon(true)
            .setThreadFactory(executor.getThreadFactory())
            .build());
  }

  // See newDirectExecutorService javadoc for behavioral notes.
  @J2ktIncompatible
  @GwtIncompatible // TODO
  private static final class DirectExecutorService extends AbstractListeningExecutorService {