}

  @Override
  public boolean wasStreamClosed() {
    return byteSink.wasStreamClosed();
  }

  @Override
  public Writer openStream() throws IOException {
    // using TestByteSink's output stream to get option behavior, so flush to it on every write
    return new FilterWriter(new OutputStreamWriter(byteSink.openStream(), UTF_8)) {
      @Override
      public void write(int c) throws IOException {
        super.write(c);

   * Splitter.on(',').omitEmptyStrings().split(",a,,,b,c,,")} returns an iterable containing only
   * {@code ["a", "b", "c"]}.
   *
   * <p>If either {@code trimResults} option is also specified when creating a splitter, that
   * splitter always trims results first before checking for emptiness. So, for example, {@code

 *       <a href="https://developer.android.com/studio/build/shrink-code">R8 and Proguard</a>.
 *   <li>It doesn't offer a way to wait for multiple events before taking action. For example, it
 *       doesn't offer a way to wait for multiple producers to all report that they're "ready," nor
 *       does it offer a way to batch multiple events from a single producer together.

 * href="http://en.wikipedia.org/wiki/Null_Object_pattern">null object</a> for the type in question.
 *
 * <p>This class is not intended as a direct analogue of any existing "option" or "maybe" construct
 * from other programming environments, though it may bear some similarities.
 *
 * <p>An instance of this class is serializable if its reference is absent or is a serializable
 * object.
 *

   * TODO(kevinb): evaluate the trade-offs of going bimorphic to save these two fields from most
   * instances. Note that the instances that would get smaller are the right set to care about
   * optimizing, because the rest have the option of calling `trimmed`.
   */

  private final transient int start; // it happens that we only serialize instances where this is 0
  private final int end; // exclusive

  private ImmutableLongArray(long[] array) {

 * it is either open or unbounded. The construct {@code {x | statement}} is read "the set of all
 * <i>x</i> such that <i>statement</i>.")
 *
 * <blockquote>
 *
 * <table>
 * <caption>Range Types</caption>
 * <tr><th>Notation        <th>Definition               <th>Factory method
 * <tr><td>{@code (a..b)}  <td>{@code {x | a < x < b}}  <td>{@link Range#open open}

 *         }
 *         public void failure(Service service) {
 *           // Something failed, at this point we could log it, notify a load balancer, or take
 *           // some other action.  For now we will just exit.
 *           System.exit(1);
 *         }
 *       },
 *       MoreExecutors.directExecutor());
 *
 *     Runtime.getRuntime().addShutdownHook(new Thread() {
 *       public void run() {

   * backing set. The returned map does not support put operations.
   *
   * <p><b>Warning:</b> If the function rejects {@code null}, caution is required to make sure the
   * set does not contain {@code null}, because the view cannot stop {@code null} from being added
   * to the set.
   *

   *
   * <p><b>{@code Stream} equivalent:</b> {@code stream.filter(type::isInstance).map(type::cast)}.
   * This does perform a little more work than necessary, so another option is to insert an
   * unchecked cast at some later point:
   *
   * <pre>
   * {@code @SuppressWarnings("unchecked") // safe because of ::isInstance check
   * ImmutableList<NewType> result =

     * at table entries if it is 0.
     *
     * - All (synchronized) write operations should write to the "count" field after structurally
     * changing any bin. The operations must not take any action that could even momentarily cause a
     * concurrent read operation to see inconsistent data. This is made easier by the nature of the