return set;
  }

  /**
   * Creates a {@code CompactHashSet} instance, with a high enough "initial capacity" that it
   * <i>should</i> hold {@code expectedSize} elements without growth.
   *
   * @param expectedSize the number of elements you expect to add to the returned set
   * @return a new, empty {@code CompactHashSet} with enough capacity to hold {@code expectedSize}
   *     elements without resizing

TensorFlow was originally developed by researchers and engineers working within
the Machine Intelligence team at Google Brain to conduct research in machine
learning and neural networks. However, the framework is versatile enough to be
used in other areas as well.

TensorFlow provides stable [Python](https://www.tensorflow.org/api_docs/python)
and [C++](https://www.tensorflow.org/api_docs/cc) APIs, as well as a

        // Prepare test data
        ByteBuffer bb = ByteBuffer.wrap(testBuffer).order(ByteOrder.LITTLE_ENDIAN);

        // Header
        bb.putShort((short) 3); // version
        bb.putShort((short) 100); // size (large enough for all strings)
        bb.putShort((short) 1); // serverType
        bb.putShort((short) 0); // rflags (no name list)
        bb.putShort((short) 5); // proximity
        bb.putShort((short) 300); // ttl

    /**
     * Careful here: if threads are mutating the atomicLongArray while this method is executing, the
     * final long[] will be a "rolling snapshot" of the state of the bit array. This is usually good
     * enough, but should be kept in mind.
     */
    public static long[] toPlainArray(AtomicLongArray atomicLongArray) {
      long[] array = new long[atomicLongArray.length()];
      for (int i = 0; i < array.length; ++i) {

  }

  public void testGreatestOfIterable_simple() {
    /*
     * If greatestOf() promised to be implemented as reverse().leastOf(), this
     * test would be enough. It doesn't... but we'll cheat and act like it does
     * anyway. There's a comment there to remind us to fix this if we change it.
     */
    List<Integer> list = asList(3, 1, 3, 2, 4, 2, 4, 3);

  }

  public void testGreatestOfIterable_simple() {
    /*
     * If greatestOf() promised to be implemented as reverse().leastOf(), this
     * test would be enough. It doesn't... but we'll cheat and act like it does
     * anyway. There's a comment there to remind us to fix this if we change it.
     */
    List<Integer> list = asList(3, 1, 3, 2, 4, 2, 4, 3);

### Saved graphs and checkpoints

When loading untrusted serialized computation graphs (in form of a `GraphDef`,
`SavedModel`, or equivalent on-disk format), the set of computation primitives
available to TensorFlow is powerful enough that you should assume that the
TensorFlow process effectively executes arbitrary code.

The risk of loading untrusted checkpoints depends on the code or graph that you

      // if waiting works, this should get stuck
      Queues.drain(q, new ArrayList<>(), 1, MAX_VALUE, NANOSECONDS);
      fail();
    } catch (InterruptedException expected) {
      // we indeed waited; a slow thread had enough time to interrupt us
    }
  }

  // same as above; uninterruptible version
  @SuppressWarnings("ThreadPriorityCheck") // TODO: b/175898629 - Consider onSpinWait.

///

/// warning

These are very technical details about the standards **JSON Schema** and **OpenAPI**.

If the ideas above already work for you, that might be enough, and you probably don't need these details, feel free to skip them.

///

Before OpenAPI 3.1.0, OpenAPI used an older and modified version of **JSON Schema**.

    long nanosPerSecond = NANOSECONDS.convert(1, SECONDS);
    assertThat(nanosPerSecond * Long.MAX_VALUE).isLessThan(0L);

    // Check that we wait long enough anyway (presumably as long as MAX_VALUE nanos):
    TimedWaiterThread waiter = new TimedWaiterThread(future, Long.MAX_VALUE, SECONDS);
    waiter.start();
    waiter.awaitWaiting();

    future.set(1);