<version>${surefire.toolchain.version}</version>
            </jdkToolchain>
            <includes>
              <include>${test.include}</include>
            </includes>
            <!-- By having our own entries here, we also override the default exclusion filter, which excludes all nested classes. -->
            <excludes>
              <!-- https://github.com/google/guava/issues/2840 -->

              <version>${surefire.toolchain.version}</version>
            </jdkToolchain>
            <includes>
              <include>${test.include}</include>
            </includes>
            <!-- By having our own entries here, we also override the default exclusion filter, which excludes all nested classes. -->
            <excludes>
              <!-- https://github.com/google/guava/issues/2840 -->

   * <p>Note that, in cases where {@code lengthIfKnown} returns zero, it is <i>possible</i> that
   * chars are actually available for reading. This means that a source may return {@code true} from
   * {@code isEmpty()} despite having readable content.
   *
   * @throws IOException if an I/O error occurs
   * @since 15.0
   */
  public boolean isEmpty() throws IOException {
    Optional<Long> lengthIfKnown = lengthIfKnown();

   * <p>Note that, in cases where {@code lengthIfKnown} returns zero, it is <i>possible</i> that
   * chars are actually available for reading. This means that a source may return {@code true} from
   * {@code isEmpty()} despite having readable content.
   *
   * @throws IOException if an I/O error occurs
   * @since 15.0
   */
  public boolean isEmpty() throws IOException {
    Optional<Long> lengthIfKnown = lengthIfKnown();

          // flow-control window. This is necessary because we don't transmit window updates until
          // the application reads the data. If we throw this prior to updating the connection
          // flow-control window, we risk having it go to 0 preventing the server from sending data.
          throw errorExceptionToDeliver!!
        }

        return -1L // This source is exhausted.
      }
    }

   * objects per entry.  Instead we use an open-addressed hash table.  This design is basically
   * equivalent to RegularImmutableSet, save that instead of having a hash table containing the
   * elements directly and null for empty positions, we store indices of the keys in the hash table,
   * and ABSENT for empty positions.  We then look up the keys in alternatingKeysAndValues.
   *

 *     </ol>
 *     Unknown qualifiers are considered after known qualifiers,
 *     with lexical order (case-insensitive in the English locale).
 *     <code>ga</code> and <code>final</code> sort the same as not having a qualifier.
 *   </li>
 * <li>A hyphen usually precedes a qualifier, and is always less important than digits/number. For example
 *   {@code 1.0.RC2 < 1.0-RC3 < 1.0.1}; but prefer {@code 1.0.0-RC2} over {@code 1.0.0.RC2}, and more

Note that ExistingObjectReplication needs to be enabled in the config via `mc replicate [add|edit]` by passing `existing-objects` as one of the values to `--replicate` flag. Only those objects meeting replication rules and having existing object replication enabled will be re-synced.

### Multi destination replication

   * any of the provided futures fails or is canceled, its corresponding position will contain
   * {@code null} (which is indistinguishable from the future having a successful value of {@code
   * null}).
   *
   * <p>The list of results is in the same order as the input list.
   *
   * <p>This differs from {@link #allAsList(ListenableFuture[])} in that it's tolerant of failed

 * small portion of these would be in use.
 *
 * <p>Prior to this class, one might be tempted to use {@code Map<K, Lock>}, where {@code K}
 * represents the task. This maximizes concurrency by having each unique key mapped to a unique
 * lock, but also maximizes memory footprint. On the other extreme, one could use a single lock for
 * all tasks, which minimizes memory footprint but also minimizes concurrency. Instead of choosing