this.height = 1 + Math.max(height(left), height(right));
    }

    private void recompute() {
      recomputeMultiset();
      recomputeHeight();
    }

    private AvlNode<E> rebalance() {
      switch (balanceFactor()) {
        case -2:
          // requireNonNull is safe because right must exist in order to get a negative factor.
          requireNonNull(right);
          if (right.balanceFactor() > 0) {

 * comparator utilities, see either {@code Comparator} itself (for Java 8+), or {@code
 * com.google.common.collect.Ordering} (otherwise).
 *
 * <h3>Relationship to {@code Ordering}</h3>
 *
 * <p>In light of the significant enhancements to {@code Comparator} in Java 8, the overwhelming
 * majority of usages of {@code Ordering} can be written using only built-in JDK APIs. This class is

     * all calling acquire() at these moments:
     *
     * T0 at 0 seconds
     * T1 at 1.05 seconds
     * T2 at 2 seconds
     * T3 at 3 seconds
     *
     * Due to the slight delay of T1, T2 would have to sleep till 2.05 seconds, and T3 would also
     * have to sleep till 3.05 seconds.
     */
    return create(permitsPerSecond, SleepingStopwatch.createFromSystemTimer());
  }

 * named analogously to their {@code BigInteger} counterparts.
 *
 * <p>The implementations of many methods in this class are based on material from Henry S. Warren,
 * Jr.'s <i>Hacker's Delight</i>, (Addison Wesley, 2002).
 *
 * <p>Similar functionality for {@code int} and for {@link BigInteger} can be found in {@link
 * IntMath} and {@link BigIntegerMath} respectively. For other common operations on {@code long}

 * randomly ordered data (the probability decreases faster than exponentially in N), but if you are
 * passing in unsanitized user data then a malicious user could force it. A light shuffle of the
 * data using an unpredictable seed should normally be enough to thwart this attack.
 *
 * <p>The time taken to compute multiple quantiles on the same dataset using {@link Scale#indexes

   * perform better than its longhand equivalent, but should not perform worse.
   *
   * @since 12.0
   */
  HashCode hashInt(int input);

  /**
   * Shortcut for {@code newHasher().putLong(input).hash()}; returns the hash code for the given
   * {@code long} value, interpreted in little-endian byte order. The implementation <i>might</i>

            "five", 5,
            "six", 6,
            "seven", 7,
            "eight", 8),
        "one",
        1,
        "two",
        2,
        "three",
        3,
        "four",
        4,
        "five",
        5,
        "six",
        6,
        "seven",
        7,
        "eight",
        8);
    assertMapEquals(
        ImmutableMap.of(
            "one", 1,

   *   <li>When the RateLimiter is not used, this goes right (up to maxPermits)
   *   <li>When the RateLimiter is used, this goes left (down to zero), since if we have
   *       storedPermits, we serve from those first
   *   <li>When _unused_, we go right at a constant rate! The rate at which we move to the right is
   *       chosen as maxPermits / warmupPeriod. This ensures that the time it takes to go from 0 to

            "five", 5,
            "six", 6,
            "seven", 7,
            "eight", 8),
        "one",
        1,
        "two",
        2,
        "three",
        3,
        "four",
        4,
        "five",
        5,
        "six",
        6,
        "seven",
        7,
        "eight",
        8);
    assertMapEquals(
        ImmutableBiMap.of(
            "one", 1,

     * 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
     * course stack overflows are bad news in general. For example, we may have overflowed in the