* we've chosen to instead use a signature that benefits existing users -- and is still safe.
   *
   * For convertAll, I haven't looked as closely at *how* much existing users benefit, so we should
   * keep an eye out for problems that new users encounter. Note also that convertAll could support

      // We're looking for the smallest charsPerChunk such that bitsPerChar * charsPerChunk is a
      // multiple of 8. A multiple of 8 has 3 low zero bits, so we just need to figure out how many
      // extra zero bits we need to add to the end of bitsPerChar to get 3 in total.
      // The logic here would be wrong for bitsPerChar > 8, but since we require distinct ASCII
      // characters that can't happen.

 *
 * <p>This multimap allows duplicate key-value pairs. After adding a new key-value pair equal to an
 * existing key-value pair, the {@code ArrayListMultimap} will contain entries for both the new
 * value and the old value.
 *
 * <p>Keys and values may be null. All optional multimap methods are supported, and all returned
 * views are modifiable.
 *

  public void testCompare() {
    // max value
    assertThat(UnsignedLongs.compare(0, 0xffffffffffffffffL)).isLessThan(0);
    assertThat(UnsignedLongs.compare(0xffffffffffffffffL, 0)).isGreaterThan(0);

    // both with high bit set
    assertThat(UnsignedLongs.compare(0xff1a618b7f65ea12L, 0xffffffffffffffffL)).isLessThan(0);
    assertThat(UnsignedLongs.compare(0xffffffffffffffffL, 0xff1a618b7f65ea12L)).isGreaterThan(0);

    return new AndPredicate<T>(defensiveCopy(components));
  }

  /**
   * Returns a predicate that evaluates to {@code true} if both of its components evaluate to {@code
   * true}. The components are evaluated in order, and evaluation will be "short-circuited" as soon
   * as a false predicate is found.
   */

     *
     * - If the stack overflowed after set()/setValue() stored its result, then a call to
     * setException() will be a no-op because the Future is already done.
     *
     * Both scenarios are bad: The output Future might never complete, or, if it does complete, it
     * might not run some of its listeners. The likely result is that the app will hang. (And of

    for (int k = 0; k < 50; k++) {
      // For each listener we add a task that submits it to the executor directly for the blocking
      // get use case and another task that adds it as a listener to the future to exercise both
      // racing addListener calls and addListener calls completing after the future completes.
      final Runnable listener =
          k % 2 == 0 ? collectResultsRunnable : collectResultsTimedGetRunnable;

  /**
   * Returns {@code true} if {@code object} is an {@code Optional} instance, and either the
   * contained references are {@linkplain Object#equals equal} to each other or both are absent.
   * Note that {@code Optional} instances of differing parameterized types can be equal.
   *
   * <p><b>Comparison to {@code java.util.Optional}:</b> no differences.
   */
  @Override

     * contrast to the "V?" type that we're using. As a result, Kotlin sees a conflict between the
     * nullness annotations in ImmutableMap and those in its own Map type. In response, it considers
     * the parameter and return type both to be platform types. As a result, Kotlin permits calls
     * that can lead to NullPointerException. That's unfortunate. But hopefully most Kotlin callers
     * use `get(key) ?: defaultValue` instead of this method, anyway.

   * @throws IllegalArgumentException if {@code occurrences} is negative
   */
  /*
   * TODO(cpovirk): remove and removeExactly currently accept null inputs only
   * if occurrences == 0. This satisfies both NullPointerTester and
   * CollectionRemoveTester.testRemove_nullAllowed, but it's not clear that it's
   * a good policy, especially because, in order for the test to pass, the