* this builder to sort entries by value.
   *
   * <p>Builder instances can be reused - it is safe to call {@link #buildOrThrow} multiple times to
   * build multiple bimaps in series. Each bimap is a superset of the bimaps created before it.
   *
   * @since 2.0
   */
  public static final class Builder<K, V> extends ImmutableMap.Builder<K, V> {
    /**

   *         .putAll("several", 1, 2, 3)
   *         .putAll("many", 1, 2, 3, 4, 5)
   *         .build();
   * }</pre>
   *
   * <p>Builder instances can be reused; it is safe to call {@link #build} multiple times to build
   * multiple multimaps in series. Each multimap contains the key-value mappings in the previously
   * created multimaps.
   *
   * @since 2.0
   */
  @DoNotMock
  public static class Builder<K, V> {

      // extensive testing
      try {
        return Float.parseFloat(string);
      } catch (NumberFormatException e) {
        // Float.parseFloat has changed specs several times, so fall through
        // gracefully
      }
    }
    return null;
  }

   * query than the original; your mileage may vary. Precomputation takes time and is likely to be
   * worthwhile only if the precomputed matcher is queried many thousands of times.
   *
   * <p>This method has no effect (returns {@code this}) when called in GWT: it's unclear whether a
   * precomputed matcher is faster, but it certainly consumes more memory, which doesn't seem like a

  private static void runUntimedInterruptsTest(int times)
      throws InterruptedException, ExecutionException, TimeoutException {
    SettableFuture<String> future = SettableFuture.create();
    FutureTask<Boolean> interruptReporter = untimedInterruptReporter(future, false);

    runNInterruptsTest(times, future, interruptReporter);
  }

  private static void runTimedInterruptsTest(int times)

    }

    @Override
    public int hashCode() {
      /*
       * The sum of the sums of the hash codes in each subset is just the sum of
       * each input element's hash code times the number of sets that element
       * appears in. Each element appears in exactly half of the 2^n sets, so:
       */
      return inputSet.keySet().hashCode() << (inputSet.size() - 1);
    }

    @Override

        "Should be able to peek() first element multiple times", "A", peekingIterator.peek());
    assertEquals(
        "next() should still return first element after peeking", "A", peekingIterator.next());

    assertEquals("Should be able to peek() at middle element", "B", peekingIterator.peek());
    assertEquals(
        "Should be able to peek() middle element multiple times", "B", peekingIterator.peek());
    assertEquals(

        Guava's main yardstick for evaluating proposed features can be summed up as [utility times
        ubiquity](https://github.com/google/guava/wiki/PhilosophyExplained#utility-times-ubiquity).


        #### Utility: compare with alternatives


        There is always *some* alternative to adding a new feature to Guava, even if it's just

   *
   * <p>This method should return promptly; prefer to do work on a different thread where it is
   * convenient. It is invoked exactly once on service startup, even when {@link #startAsync} is
   * called multiple times.
   */
  @ForOverride
  protected abstract void doStart();

  /**
   * This method should be used to initiate service shutdown. The invocation of this method should

   * original transformation and its inverse to a value gives you the original value give-or-take
   * numerical errors. Calling this method multiple times on the same instance will always return
   * the same instance. Calling this method on the result of calling this method on an instance will
   * always return that original instance.
   */