// timed out. e.g. if we weren't event able to grab the lock within the timeout we would never
      // even check the guard. I don't think we care too much about this use case but it could lead
      // to a confusing error message.
      throw new TimeoutException("Timed out waiting for " + this + " to reach the RUNNING state.");
    }
  }

  @Override
  public final void awaitTerminated() {

 * <p><b>Warning:</b> The methods of {@code ForwardingMap} forward <i>indiscriminately</i> to the
 * methods of the delegate. For example, overriding {@link #put} alone <i>will not</i> change the
 * behavior of {@link #putAll}, which can lead to unexpected behavior. In this case, you should
 * override {@code putAll} as well, either providing your own implementation, or delegating to the
 * provided {@code standardPutAll} method.
 *

 * the methods of the delegate. For example, overriding {@link #add(Object, int)} alone <b>will
 * not</b> change the behavior of {@link #add(Object)}, which can lead to unexpected behavior. In
 * this case, you should override {@code add(Object)} as well, either providing your own
 * implementation, or delegating to the provided {@code standardAdd} method.
 *

      // timed out. e.g. if we weren't event able to grab the lock within the timeout we would never
      // even check the guard. I don't think we care too much about this use case but it could lead
      // to a confusing error message.
      throw new TimeoutException("Timed out waiting for " + this + " to reach the RUNNING state.");
    }
  }

  @Override
  public final void awaitTerminated() {

   * message must mention two values that were associated with the duplicate key in two different
   * calls to Builder.put (though we don't really care *which* two values if there were more than
   * two). These considerations lead us to have a field of type DuplicateKey in the Builder, which
   * will remember the first duplicate key we encountered. All later calls to buildOrThrow() can

              writeByteString(value)
              insertIntoDynamicTable(header)
            }
            name.startsWith(Header.PSEUDO_PREFIX) && TARGET_AUTHORITY != name -> {
              // Follow Chromes lead - only include the :authority pseudo header, but exclude all other
              // pseudo headers. Literal Header Field without Indexing - Indexed Name.
              writeInt(headerNameIndex, PREFIX_4_BITS, 0)

 * worry about being reentrant, unless also called from outside the EventBus.
 *
 * <h2>Dead Events</h2>
 *
 * <p>If an event is posted, but no registered subscribers can accept it, it is considered "dead."
 * To give the system a second chance to handle dead events, they are wrapped in an instance of
 * {@link DeadEvent} and reposted.
 *

   * 3. Tests on a mechanically generated dataset for chains starting with percentiles();
   * 4. Tests of illegal usages of the API.
   */

  /*
   * Covering every combination would lead to an explosion in the number of tests. So we cover only:
   * - median with compute taking a double-collection and with computeInPlace;
   * - quartiles with index and with indexes taking int-varargs, and with compute taking a

   * 3. Tests on a mechanically generated dataset for chains starting with percentiles();
   * 4. Tests of illegal usages of the API.
   */

  /*
   * Covering every combination would lead to an explosion in the number of tests. So we cover only:
   * - median with compute taking a double-collection and with computeInPlace;
   * - quartiles with index and with indexes taking int-varargs, and with compute taking a

   * higher value than you need can waste space and time, and a significantly lower value can lead
   * to thread contention. But overestimates and underestimates within an order of magnitude do not
   * usually have much noticeable impact. A value of one permits only one thread to modify the map
   * at a time, but since read operations can proceed concurrently, this still yields higher