int dummy = 0;
    for (int i = 0; i < reps; i++) {
      StringBuilder sb = new StringBuilder();
      for (String comp : components) {
        if (sb.length() > 0) {
          sb.append(DELIMITER_STRING);
        }
        sb.append(comp);
      }
      dummy ^= sb.toString().length();
    }
    return dummy;
  }

  /**

    limiter.acquire();
    limiter.acquire();
    limiter.acquire();
    limiter.acquire();
    limiter.acquire();
    assertEvents(
        "R0.00", // First comes the saved-up burst, which defaults to a 1-second burst (2 requests).
        "R0.00", "R0.00", // Now comes the free request.
        "R0.50", // Now it's 0.5 seconds per request.
        "R0.50");

    limiter.setRate(Double.POSITIVE_INFINITY);
    limiter.acquire();

 *       testlib), but why bother?
 *   <li>Stripping code entirely might help us keep under the method limit someday. Even if it never
 *       comes to that, it may at least help with build and startup times.
 * </ul>
 */
@Retention(CLASS)
@Target({ANNOTATION_TYPE, CONSTRUCTOR, FIELD, METHOD, TYPE})
@GwtCompatible

   * ParametricNullness parametric nullness}. Sometimes, code may receive a null {@code T} but store
   * a "null sentinel" to take its place. When the time comes to convert it back to a {@code T} to
   * return to a caller, the code needs to a way to return {@code null} from a method that returns
   * "plain {@code T}." This API provides that.
   */

        return first.length() - second.length();
      } else {
        return first.compareTo(second);
      }
    }
  }

  /** Decreasing integer values. A {@code null} comes before any non-null value. */
  private static final Comparator<@Nullable Integer> DECREASING_INT_COMPARATOR =
      Ordering.<Integer>natural().reverse().<Integer>nullsFirst();

  private SetMultimap<String, Integer> create() {

    }
    if (that == aboveAll()) {
      return -1;
    }
    int result = Range.compareOrThrow(endpoint, that.endpoint);
    if (result != 0) {
      return result;
    }
    // same value. below comes before above
    return Booleans.compare(this instanceof AboveValue, that instanceof AboveValue);
  }

  C endpoint() {
    return endpoint;
  }

  @SuppressWarnings("unchecked") // catching CCE
  @Override

 * George: [Washington, Bush, Bush]
 * Grover: [Cleveland, Cleveland]        // Two, non-consecutive terms, rep'ing NJ!
 * ...
 * }</pre>
 *
 * <h3>Views</h3>
 *
 * <p>Much of the power of the multimap API comes from the <i>view collections</i> it provides.
 * These always reflect the latest state of the multimap itself. When they support modification, the

 * <p>In order to allow proper garbage collection of unused locks, the edges of the lock graph are
 * weak references.
 *
 * <p><strong>Performance</strong>
 *
 * <p>The extra bookkeeping done by cycle detecting locks comes at some cost to performance.
 * Benchmarks (as of December 2011) show that:
 *
 * <ul>
 *   <li>for an unnested {@code lock()} and {@code unlock()}, a cycle detecting lock takes 38ns as

        assertNull(beerPhase);
    }

    @Test
    void testFindLastInPhaseMisc() throws Exception {
        MavenExecutionPlan plan = LifecycleExecutionPlanCalculatorStub.getProjectAExecutionPlan();

        assertNull(plan.findLastInPhase("pacXkage"));
        // Beer comes straight after package in stub, much like real life.

   * QPS=5 (5 tokens per second), if we ensure that a request isn't granted earlier than 200ms after
   * the last one, then we achieve the intended rate. If a request comes and the last request was
   * granted only 100ms ago, then we wait for another 100ms. At this rate, serving 15 fresh permits
   * (i.e. for an acquire(15) request) naturally takes 3 seconds.
   *