* benchmarking
   * on Open JDK 7, {@link #immutableSortedCopy} generally performs better (in both time and space)
   * than this method, and this method in turn generally performs better than copying the list and
   * calling {@link Collections#sort(List)}.
   */
  // TODO(kevinb): rerun benchmarks including new options
  public <E extends T> List<E> sortedCopy(Iterable<E> elements) {

                         * Apparently once a successful NTLMSSP login occurs, the
                         * server will return "Access denied" even if a logoff is
                         * sent. Unfortunately calling disconnect() doesn't always
                         * actually shutdown the connection before other threads
                         * have committed themselves (e.g. InterruptTest example).
                         */

  }

  /**
   * In reflection, {@code Foo<?>.getUpperBounds()[0]} is always {@code Object.class}, even when Foo
   * is defined as {@code Foo<T extends String>}. Thus directly calling {@code <?>.is(String.class)}
   * will return false. To mitigate, we canonicalize wildcards by enforcing the following
   * invariants:
   *
   * <ol>

 * ReentrantReadWriteLock} instances that detect potential deadlock by checking for cycles in lock
 * acquisition order.
 *
 * <p>Potential deadlocks detected when calling the {@code lock()}, {@code lockInterruptibly()}, or
 * {@code tryLock()} methods will result in the execution of the {@link Policy} specified when
 * creating the factory. The currently available policies are:
 *
 * <ul>

   *
   * <p>The resulting stream will only be as long as the shorter of the two input streams; if one
   * stream is longer, its extra elements will be ignored.
   *
   * <p>Note that if you are calling {@link Stream#forEach} on the resulting stream, you might want
   * to consider using {@link #forEachPair} instead of this method.
   *
   * <p><b>Performance note:</b> The resulting stream is not <a

        Segment<K, V> segment, K key, int hash, @Nullable ReferenceEntry<K, V> next);

    /**
     * Copies an entry, assigning it a new {@code next} entry.
     *
     * @param original the entry to copy. But avoid calling {@code getKey} on it: Instead, use the
     *     {@code key} parameter. That way, we prevent the key from being garbage collected in the

          implementation is trivial such as with getters and setters, or a
          method just delegates to another object.
       2. However, you should still add Javadoc if there are caveats around
          calling a method that are not immediately obvious from reading the
          method's implementation in isolation.
       3. You can omit Javadoc for simple classes, e.g. where they are a

- Fix a concurrent map writes error in kubelet ([#93773](https://github.com/kubernetes/kubernetes/pull/93773), [@knight42](https://github.com/knight42)) [SIG Node]
- Fix calling AttachDisk on a previously attached EBS volume ([#93567](https://github.com/kubernetes/kubernetes/pull/93567), [@gnufied](https://github.com/gnufied)) [SIG Cloud Provider, Storage and Testing]

    assertThat(consumingIterator.next()).isEqualTo("b");
    assertThat(list).isEmpty();

    assertFalse(consumingIterator.hasNext());
  }

  @GwtIncompatible // ?
  // TODO: Figure out why this is failing in GWT.
  public void testConsumingIterator_duelingIterators() {
    // Test data
    List<String> list = Lists.newArrayList("a", "b");

    // Test & Verify

      // a copy of the entries array. Likewise, localSize is the same as size except in that case.
      // It's possible to keep using this Builder after calling buildKeepingLast(), so we need to
      // ensure that its state is not corrupted by removing duplicates that should cause a later
      // buildOrThrow() to fail, or by changing the size.