the lobsters to the dance.  So they got thrown out to sea.  So
they had to fall a long way.  So they got their tails fast in
their mouths.  So they couldn't get them out again.  That's all.'

  `Thank you,' said Alice, `it's very interesting.  I never knew
so much about a whiting before.'

  `I can tell you more than that, if you like,' said the
Gryphon.  `Do you know why it's called a whiting?'

the lobsters to the dance.  So they got thrown out to sea.  So
they had to fall a long way.  So they got their tails fast in
their mouths.  So they couldn't get them out again.  That's all.'

  `Thank you,' said Alice, `it's very interesting.  I never knew
so much about a whiting before.'

  `I can tell you more than that, if you like,' said the
Gryphon.  `Do you know why it's called a whiting?'

   * are guarded by a lock and operations are applied in batches to avoid lock contention. The
   * penalty of applying the batches is spread across threads so that the amortized cost is slightly
   * higher than performing just the operation without enforcing the capacity constraint.
   *
   * This implementation uses a per-segment queue to record a memento of the additions, removals,

    derived.cancel(false);
    assertTrue(primary.isCancelled());
    assertFalse(primary.wasInterrupted());
  }

  // catching() uses a plain Function, so there's no testCatching_resultCancelledAfterFallback().

  // catching() uses a plain Function, so there's no testCatching_nullInsteadOfFuture().

  // Some tests of the exceptionType parameter:

  public void testCatching_Throwable() throws Exception {

 *
 * <p>A pipeline of {@code ClosingFuture}s is a tree of steps. Each step represents either an
 * asynchronously-computed intermediate value, or else an exception that indicates the failure or
 * cancellation of the operation so far. The only way to extract the value or exception from a step
 * is by declaring that step to be the last step of the pipeline. Nevertheless, we refer to the

    derived.cancel(false);
    assertTrue(primary.isCancelled());
    assertFalse(primary.wasInterrupted());
  }

  // catching() uses a plain Function, so there's no testCatching_resultCancelledAfterFallback().

  // catching() uses a plain Function, so there's no testCatching_nullInsteadOfFuture().

  // Some tests of the exceptionType parameter:

  public void testCatching_Throwable() throws Exception {

      assertThat(cache.getOrLoad("test")).isEqualTo("testLoad");

      ticker.advance(10, MILLISECONDS); // so that the next call will trigger refresh
      assertThat(cache.getOrLoad("test")).isEqualTo("testLoad");
      reloadStarted.await();
      ticker.advance(500, MILLISECONDS); // so that the entry expires during the reload
      threadAboutToBlockForRefresh.set(Thread.currentThread());

      assertThat(cache.getOrLoad("test")).isEqualTo("testLoad");

      ticker.advance(10, MILLISECONDS); // so that the next call will trigger refresh
      assertThat(cache.getOrLoad("test")).isEqualTo("testLoad");
      reloadStarted.await();
      ticker.advance(500, MILLISECONDS); // so that the entry expires during the reload
      threadAboutToBlockForRefresh.set(Thread.currentThread());

   * are guarded by a lock and operations are applied in batches to avoid lock contention. The
   * penalty of applying the batches is spread across threads so that the amortized cost is slightly
   * higher than performing just the operation without enforcing the capacity constraint.
   *
   * This implementation uses a per-segment queue to record a memento of the additions, removals,

         * That could change behavior if the input maps use different equivalence relations (and so
         * a key that appears once in `right` might appear multiple times in `left`). We don't
         * guarantee behavior in that case, anyway, and the current behavior is likely undesirable.
         * So that's either a reason to feel free to change it or a reason to not bother thinking
         * further about this.