/** Performs an atomic compare-and-set of {@link AggregateFutureState#seenExceptionsField}. */
    abstract void compareAndSetSeenExceptions(
        AggregateFutureState<?> state, @Nullable Set<Throwable> expect, Set<Throwable> update);

    /** Performs an atomic decrement-and-get of {@link AggregateFutureState#remainingField}. */
    abstract int decrementAndGetRemainingCount(AggregateFutureState<?> state);

/**
 * Tests our AtomicHelper fallback strategy in AggregateFutureState.
 *
 * <p>On different platforms AggregateFutureState uses different strategies for its core
 * synchronization primitives. The strategies are all implemented as subtypes of AtomicHelper and
 * the strategy is selected in the static initializer of AggregateFutureState. This is convenient

     *
     * SynchronizedHelper also ends up getting used under the JVM during
     * AggregateFutureStateFallbackAtomicHelperTest, as discussed in a comment in
     * AggregateFutureState.
     *
     * Here, we check that we're able to force AbstractFutureState to select SynchronizedHelper, and
     * below, we actually run the AbstractFutureTest methods under that scenario.
     */

         * the case of the Samsung bug, and we do that by breaking AtomicReferenceFieldUpdater.
         * Breaking AtomicReferenceFieldUpdater not only forces AggregateFutureState to fall back to
         * another implementation but also forces AbstractFutureState to be able to do the
         * same—hence the try-catch here.
         *

   * since Android should normally succed in using UnsafeAtomicHelper and thus never even try the
   * problematic AtomicReferenceFieldUpdaterAtomicHelper code path. However, the same problem *does*
   * matter with AggregateFutureState, which does not have an Unsafe-based helper.
   *
   * This same problem is one of the reasons for us to likewise use package-private for the fields
   * in Waiter.
   */

  /**